Ets family members share a conserved DNA-binding ETS domain, and serve a variety of roles in development, differentiation and oncogenesis. Besides DNA binding, the ETS domain also participates in protein–protein interactions with other structurally unrelated transcription factors. Although this mechanism appears to confer tissue- or development stage-specific functions on individual Ets proteins, the biological significance of many of these interactions remains to be evaluated, because their molecular basis has been elusive. We previously demonstrated a direct interaction between the ETS domain of the widely expressed GABPα (GA-binding protein α) and the granulocyte inducer C/EBPα (CCAAT/enhancer-binding protein α), and suggested its involvement in co-operative transcriptional activation of myeloid-specific genes, such as human FCAR encoding FcαR [Fc receptor for IgA (CD89)]. By deletion analysis, we identified helix α3 and the β3/β4 region as the C/EBPα-interacting region. Domain-swapping of individual sub-domains with those of other Ets proteins allowed us to highlight β-strand 3 and the subsequent loop, which when exchanged by those of Elf-1 (E74-like factor 1) reduced the ability to recruit C/EBPα. Further analysis identified a four-amino acid swap mutation of this region (I387L/C388A/K393Q/F395L) that reduces both physical interaction and co-operative transcriptional activation with C/EBPα without affecting its transactivation capacity by itself. Moreover, re-ChIP (re-chromatin immunoprecipitation) analysis demonstrated that GABPα recruits C/EBPα to the FCAR promoter, depending on these residues. The identified amino acid residues could confer the specificity of the action on the Ets proteins in diverse biological processes through mediating the recruitment of its partner factor.

INTRODUCTION

Ets proteins are a family of related transcription factors that are involved in various cellular functions, including development, cellular differentiation, apoptosis and carcinogenesis [1]. There are approx. 30 distinct mammalian Ets proteins, which share an evolutionarily conserved 85-amino acid DNA-binding domain (ETS domain) that preferentially binds to purine-rich sequences containing a GGAA/T core [1,2]. Although most Ets proteins are oncogenic, there is growing evidence that Ets proteins are also essential in the haematopoietic system [1,3]. For example, PU.1 is required for the development of multiple haematopoietic lineages, especially myeloid and B-cell development [4,5]. Fli-1 (Friend leukaemia integration 1) appears to be essential for development of megakaryocytes, because Fli-1-knockout mice exhibit severe haemorrhage in the brain on the basis of a lack of platelets [6,7]. Elf-1 (E74-like factor 1) probably has a role in T-cell proliferation and differentiation, because it has been shown to regulate T-cell expression of immunological genes such as those encoding IL-2 (interleukin-2) [8], IL-2 receptor α chain [9] and CD4 [10].

Among Ets proteins, GABP (GA-binding protein) is unique because it is the only known multimeric member, consisting of GABPα and GABPβ [11,12]. GABPα includes an ETS DNA-binding domain, and GABPβ contains ankyrin repeats and the transcriptional activation domain. GABP is expressed in a wide variety of tissues, and regulates numerous targets including so-called housekeeping genes. Basic cellular activities, including mitochondrial function, protein synthesis and cell cycle events, are dependent on GABP transcriptional regulation [13]. Furthermore, despite its wide expression patterns, many lineage-restricted genes have also been identified as GABP targets [13]. In the haematopoietic system, GABP regulates transcription of several myeloid genes, such as the genes encoding FcαR [Fc receptor for IgA (CD89)] [14], CD18 [15] and neutrophil elastase [16]. In addition, the involvement of GABP in mechanisms regulating T-cell and B-cell expression of the Il7ra (IL-7 receptor α chain) gene has previously been demonstrated in relation to development of these lineages [1719]. Maturation stage-specific regulation of megakaryopoiesis has also been reported to involve GABP in addition to Fli-1 [20].

Molecular mechanisms accounting for how the widely expressed GABP controls such lineage-specific gene expression remain unknown. We previously showed in this regard that GABP co-operates with C/EBPα (CCAAT/enhancer-binding protein α) to activate the myeloid-specific promoter of FCAR encoding FcαR [14]. C/EBPα is a bZIP (basic leucine zipper) family member [21], which is implicated as the granulocyte inducer in the haematopoietic system [22,23]. Moreover, we demonstrated that GABPα but not GABPβ physically interacts with C/EBPα through the ETS domain [14]. This strongly suggests that GABP is employed in a more specific manner by directly interacting with C/EBPα, which enables GABP function to be incorporated into a myeloid-specific transcription complex. Supporting this notion, a GABPα mutant lacking the C/EBPα-interacting ETS domain could not synergistically activate the FCAR promoter in combination with C/EBPα [14]. However, because the ETS domain is also essential for DNA binding, it remains unclear whether the observed loss of functional synergy resulted from a defect in the protein–protein interaction with C/EBPα or in DNA-binding ability.

To evaluate the functional significance of the GABPα–C/EBPα interaction, we sought to identify key residues of the GABPα ETS domain that are critical for its interaction with C/EBPα. For this purpose, we performed domain-swap experiments, in which individual sub-domains in the GABPα ETS domain were replaced by the homologous amino acid sequences of other Ets proteins. This approach allowed us to highlight β-strand 3 and the subsequent loop, which when exchanged by those of Elf-1 reduced the ability to recruit C/EBPα. We identified a four-amino acid swap mutation of this region that reduces both physical interaction and co-operative transcriptional activation with C/EBPα without affecting its transactivation capacity. These results not only demonstrate that protein–protein interaction between the ETS domain of GABPα and C/EBPα is necessary for proper synergistic activation of the FCAR promoter by GABP and C/EBPα, but will also provide a molecular basis to evaluate the biological significance of protein–protein interactions between Ets proteins and their given partners.

MATERIALS AND METHODS

Plasmid constructions

The human C/EBPα expression vector pD3C/EBPαKoz contains the CEBPA gene from hCMV-C/EBPα [24] with the perfect Kozak sequence (CCACCATGG; the translation start site is underlined) [25] in pcDNA3.1(+) (Invitrogen) [14]. To construct pD3C/EBPα273–358, the fragment encoding the C/EBPα C-terminal bZIP domain (amino acids 273–358) was amplified by PCR from pD3C/EBPαKoz using a 5′-primer containing the perfect Kozak sequence, and subcloned into the NheI/BamHI site of pcDNA3.1(+).

pGST-GAα318-399, in which GST (glutathione transferase) is fused to the GABPα ETS domain, has been described previously [14]. To create the pGST-Elf-ETS, pGST-PU-ETS and pGST-Fli-ETS plasmids, in which GST is fused to the Elf-1, PU.1 and Fli-1 ETS domains, the fragments spanning these domains were amplified by PCR from pcDNAf-Elf-1 (the human Elf-1 expression vector) [8], pRc/CMV-PU.1 (the human PU.1 expression vector) [26] and pC3Fli-1S [the human Fli-1 expression vector that was created by subcloning of the FLI1 cDNA amplified from U937 cells into pCR3.1 (Invitrogen)] respectively. These fragments were subcloned into the SalI/NotI site of pGEX-4T3 (GE Healthcare) in-frame. The GABPα ETS domain deletion mutants were obtained by PCR amplification, and subcloned into the SalI/NotI site of pGEX-4T3 in-frame to create the GST-fused deletion mutants. To construct GST-fused chimaeric or amino acid substitution Ets domain mutants, DNA fragments of chimaeras or point mutations were generated by two-step PCR mutagenesis and subcloned into the SalI/NotI site of pGEX-4T3 in-frame. To construct GST–C/EBPα273–358, the fragment coding the C/EBPα C-terminal bZIP domain (amino acids 273–358) was amplified from pD3C/EBPαKoz, and subcloned into the SalI/NotI site of pGEX-4T1 (GE Healthcare) in-frame.

The human GABPα and GABPβ1 expression vectors pC3E4TF1-60S and pC3E4TF1-53S respectively, and their corresponding empty vector pCR3.1E have been described previously [14]. To construct expression vectors of full-length GABPα containing chimaeric and/or amino acid substitution ETS domain mutants, DNA fragments spanning the GABPα C-terminal region of chimaeras or point mutations were generated by two-step PCR mutagenesis using the GST-fused ETS domain constructs containing the corresponding mutations and pC3E4TF1-60S as the templates for the first PCR. The GABPα C-terminal region in pC3E4TF1-60S was then replaced by the resulting fragments using the XhoI site located at 5′ outside of the β-strand 3′-coding region in the ETS domain and the XhoI or ApaI site in multicloning sites downstream of the GABPα termination codon. To exactly fit the 3′-end of the WT (wild-type) construct to these mutants, the corresponding WT fragment was amplified from pC3E4TF1-60S, and the GABPα C-terminal region in pC3E4TF1-60S was replaced by this fragment using the identical sites (the XhoI or ApaI site) to create pC3GAα.

The mutated −259 bp FCAR promoter–luciferase construct pGLmCE12-259 has been described previously [14,27]. The pC3HA-GABPα and its mutant plasmids, in which HA (haemagglutinin) with the perfect Kozak sequence (CCACCATGG) is fused to their N-terminus, were generated by the two-step PCR method.

Recombinant protein preparation

GST-fused proteins were expressed in Escherichia coli JM109 or BL21. E. coli was suspended in NETN-100 buffer (50 mM Tris/HCl, pH 8.0, 1 mM EDTA, 100 mM NaCl and 0.5% Nonidet P40), lysed by sonication and centrifuged at 20000 g at 4 °C for 10 min. The supernatants were stored at −80 °C until they were used for protein interaction assays (for Figures 1C and 2C, and Supplementary Figure S1) or subjected to further purification using glutathione–Sepharose 4B beads (GE Healthcare). The purified proteins were dialysed against NETN-100 buffer supplemented with 1 mM PMSF, and stored at −80 °C until they were used for protein interaction assays (for Figures 3B, 3D and 4B) or EMSAs (electrophoretic mobility-shift assays) (for Figure 1D).

The ETS domain of GABPα physically interacts with the bZIP domain of C/EBPα to form the GABPα/GABPβ–C/EBPα bZIP–DNA ternary complex

Figure 1
The ETS domain of GABPα physically interacts with the bZIP domain of C/EBPα to form the GABPα/GABPβ–C/EBPα bZIP–DNA ternary complex

(A) Functional domains of GABPα and its ETS domain (amino acids 318–399) fused to GST. The schematic diagram depicts the structure/function domains of GABPα described previously [30]. The ETS domain is represented with its three α-helices (boxes) and four β-strands (arrows). Numbers indicate the amino acid sequence number. (B) Structure of C/EBPα and its deletion mutant. Numbers indicate the amino acid sequence number. (C) Interaction of C/EBPα and its bZIP domain (amino acids 273–358) with ETS domains. GST-fused ETS domains of GABPα (lanes 3 and 4), Elf-1 (lanes 5 and 6), PU.1 (lanes 7 and 8) and Fli-1 (lanes 9 and 10) or GST alone (lanes 11 and 12) were tested for association with in vitro translated 35S-labelled C/EBPα (lanes 3, 5, 7, 9 and 11) or C/EBPα273–358 (lanes 4, 6, 8, 10 and 12). (D) Simultaneous DNA binding of GABPα/GABPβ heterodimer and C/EBPα bZIP on the FCAR promoter sequence. GABPα and GABPβ proteins prepared by in vitro translation were incubated with or without purified GST-fused C/EBPα273–358 in the absence (lanes 1–4 and 8–10) or presence of the indicated antibodies (lanes 5 to 7). Lanes 1–8 contained the labelled oligonucleotide spanning the FCAR promoter sequence nt −99/−60 (W), which contains both binding sites for C/EBPs (CE3) and GABP [14]. Lanes 9 and 10 contained the labelled oligonucleotide of the mutated FCAR promoter sequence with mutated GABP-binding site (mE) and mutated C/EBP-binding site (mC) respectively. Bands corresponding to the GABPα/GABPβ–DNA complex, the C/EBPα273–358–DNA complex, and the ternary complex GABPα/GABPβ–C/EBPα273–358–DNA are indicated. ss, supershift; ori, origin of electrophoresis. Complexes that did not enter into the gel appear to be antibody-containing complexes (lanes 5 and 6).

Figure 1
The ETS domain of GABPα physically interacts with the bZIP domain of C/EBPα to form the GABPα/GABPβ–C/EBPα bZIP–DNA ternary complex

(A) Functional domains of GABPα and its ETS domain (amino acids 318–399) fused to GST. The schematic diagram depicts the structure/function domains of GABPα described previously [30]. The ETS domain is represented with its three α-helices (boxes) and four β-strands (arrows). Numbers indicate the amino acid sequence number. (B) Structure of C/EBPα and its deletion mutant. Numbers indicate the amino acid sequence number. (C) Interaction of C/EBPα and its bZIP domain (amino acids 273–358) with ETS domains. GST-fused ETS domains of GABPα (lanes 3 and 4), Elf-1 (lanes 5 and 6), PU.1 (lanes 7 and 8) and Fli-1 (lanes 9 and 10) or GST alone (lanes 11 and 12) were tested for association with in vitro translated 35S-labelled C/EBPα (lanes 3, 5, 7, 9 and 11) or C/EBPα273–358 (lanes 4, 6, 8, 10 and 12). (D) Simultaneous DNA binding of GABPα/GABPβ heterodimer and C/EBPα bZIP on the FCAR promoter sequence. GABPα and GABPβ proteins prepared by in vitro translation were incubated with or without purified GST-fused C/EBPα273–358 in the absence (lanes 1–4 and 8–10) or presence of the indicated antibodies (lanes 5 to 7). Lanes 1–8 contained the labelled oligonucleotide spanning the FCAR promoter sequence nt −99/−60 (W), which contains both binding sites for C/EBPs (CE3) and GABP [14]. Lanes 9 and 10 contained the labelled oligonucleotide of the mutated FCAR promoter sequence with mutated GABP-binding site (mE) and mutated C/EBP-binding site (mC) respectively. Bands corresponding to the GABPα/GABPβ–DNA complex, the C/EBPα273–358–DNA complex, and the ternary complex GABPα/GABPβ–C/EBPα273–358–DNA are indicated. ss, supershift; ori, origin of electrophoresis. Complexes that did not enter into the gel appear to be antibody-containing complexes (lanes 5 and 6).

Identification of GABPα ETS domain regions essential for interaction with C/EBPα

Figure 2
Identification of GABPα ETS domain regions essential for interaction with C/EBPα

(A) Amino acid sequences of the ETS domain of GABPα and other members of the family. Numbers above the sequences indicate GABPα amino acid sequence number, and numbers on the left indicate numbers of the initial residue given for each of the family members. Secondary structure elements are indicated by boxes (α-helices) and arrows (β-strands). Strictly conserved residues that make DNA contacts are indicated by open triangles [30]. The NCBI (National Center for Biotechnology Information) accession numbers for human GAPBα, Glf-1, PU.1 and Fli-1 proteins are BAA02575.1, AAH30507.1, CAA36281.1 and AAH10115.1 respectively. (B) Structure of the GST-fused GABPα ETS domain deletion mutants. Amino acid numbers of encoded proteins are indicated for each construct. (C) GST pull-down assay showing the interaction between C/EBPα and GST-fused GABPα ETS domain using the mutant proteins shown in (B). The upper panel shows [35S]C/EBPα bound to each GST-fusion protein beads analysed by SDS/PAGE and autoradiography. The numbers below the panel indicate the relative intensity quantified by Image J software. Lower panel, the input GST-fused proteins were analysed by SDS/PAGE and Coomassie Blue staining. Molecular mass markers are shown in lane 1; molecular masses are given in kDa.

Figure 2
Identification of GABPα ETS domain regions essential for interaction with C/EBPα

(A) Amino acid sequences of the ETS domain of GABPα and other members of the family. Numbers above the sequences indicate GABPα amino acid sequence number, and numbers on the left indicate numbers of the initial residue given for each of the family members. Secondary structure elements are indicated by boxes (α-helices) and arrows (β-strands). Strictly conserved residues that make DNA contacts are indicated by open triangles [30]. The NCBI (National Center for Biotechnology Information) accession numbers for human GAPBα, Glf-1, PU.1 and Fli-1 proteins are BAA02575.1, AAH30507.1, CAA36281.1 and AAH10115.1 respectively. (B) Structure of the GST-fused GABPα ETS domain deletion mutants. Amino acid numbers of encoded proteins are indicated for each construct. (C) GST pull-down assay showing the interaction between C/EBPα and GST-fused GABPα ETS domain using the mutant proteins shown in (B). The upper panel shows [35S]C/EBPα bound to each GST-fusion protein beads analysed by SDS/PAGE and autoradiography. The numbers below the panel indicate the relative intensity quantified by Image J software. Lower panel, the input GST-fused proteins were analysed by SDS/PAGE and Coomassie Blue staining. Molecular mass markers are shown in lane 1; molecular masses are given in kDa.

Identification of GABPα ETS domain regions critical for interaction with C/EBPα using chimaeric ETS domains

Figure 3
Identification of GABPα ETS domain regions critical for interaction with C/EBPα using chimaeric ETS domains

(A and C) Structure of the GST-fused chimaeric ETS domains. (B and D) GST pull-down assay showing the interaction between C/EBPα and GST-fused GABPα ETS domain using the chimaeric proteins shown in (A) and (C) respectively. The upper panels show [35S]C/EBPα bound to each GST-fusion protein beads analysed by SDS/PAGE and autoradiography. The numbers below the panel indicate the relative intensity quantified by Image J software. Lower panels, the input GST-fused proteins were analysed by SDS/PAGE and Coomassie Blue staining. Molecular mass markers are shown in lane 1; molecular masses are given in kDa.

Figure 3
Identification of GABPα ETS domain regions critical for interaction with C/EBPα using chimaeric ETS domains

(A and C) Structure of the GST-fused chimaeric ETS domains. (B and D) GST pull-down assay showing the interaction between C/EBPα and GST-fused GABPα ETS domain using the chimaeric proteins shown in (A) and (C) respectively. The upper panels show [35S]C/EBPα bound to each GST-fusion protein beads analysed by SDS/PAGE and autoradiography. The numbers below the panel indicate the relative intensity quantified by Image J software. Lower panels, the input GST-fused proteins were analysed by SDS/PAGE and Coomassie Blue staining. Molecular mass markers are shown in lane 1; molecular masses are given in kDa.

Identification of amino acids in GABPα ETS domain regions critical for interaction with C/EBPα using the four-amino acid swap ETS domain mutants

Figure 4
Identification of amino acids in GABPα ETS domain regions critical for interaction with C/EBPα using the four-amino acid swap ETS domain mutants

(A) Structure of the GST-fused four-amino acid swap ETS domain mutants. (B) GST pull-down assay showing the interaction between C/EBPα and GST-fused GABPα ETS domain using the mutant proteins shown in (A). The upper panel shows [35S]C/EBPα bound to each GST-fusion protein beads analysed by SDS/PAGE and autoradiography. The numbers below the panel indicate the relative intensity quantified by Image J software. Lower panel, the input GST-fused proteins were analysed by SDS/PAGE and Coomassie Blue staining. Molecular mass markers are shown in lane 1; molecular masses are given in kDa.

Figure 4
Identification of amino acids in GABPα ETS domain regions critical for interaction with C/EBPα using the four-amino acid swap ETS domain mutants

(A) Structure of the GST-fused four-amino acid swap ETS domain mutants. (B) GST pull-down assay showing the interaction between C/EBPα and GST-fused GABPα ETS domain using the mutant proteins shown in (A). The upper panel shows [35S]C/EBPα bound to each GST-fusion protein beads analysed by SDS/PAGE and autoradiography. The numbers below the panel indicate the relative intensity quantified by Image J software. Lower panel, the input GST-fused proteins were analysed by SDS/PAGE and Coomassie Blue staining. Molecular mass markers are shown in lane 1; molecular masses are given in kDa.

Protein interaction assays

[35S]C/EBPα or [35S]C/EBPα273–358 protein was synthesized from pD3C/EBPαKoz or pD3C/EBPα273–358 respectively using the TnT®-coupled reticulocyte lysate system (Promega). GSTfused proteins in bacterial extracts (10 μg) or purified GST-fused proteins (5 μg) were immobilized on 5 μl of glutathione–Sepharose 4B beads. The relative amount of proteins bound to the beads was determined by SDS/PAGE using Coomassie Blue with BSA as a standard. For interaction assays, the beads bound by GST-fused proteins were incubated with 35S-labelled C/EBPα or its mutant in NETN-50 buffer (50 mM Tris/HCl, pH 8.0, 1 mM EDTA, 50 mM NaCl and 0.5% Nonidet P40) containing 100 μg/ml ethidium bromide [28] at room temperature (22 °C) for 1 h. The beads were then washed with binding buffer four times in the presence of ethidium bromide, and subsequently twice in its absence. Bound proteins were eluted by boiling in SDS sample buffer, resolved by SDS/PAGE, visualized by autoradiography, and quantified using Image J 1.39u software (http://rsb.info.nih.gov/ij/; National Institutes of Health, Bethesda, MA, U.S.A.).

EMSAs

In vitro translated GABPα and GABPβ proteins were prepared with the supercoiled pC3GAα and pC3E4TF1-53S (1 μg) respectively, as described above. EMSAs were performed as described previously [27]. Briefly, EMSA probes were 5′-end-labelled with T4 polynucleotide kinase and [γ-32P]ATP (PerkinElmer). Probes used were as follows: FCAR WT nt −99/−60 spanning both FCAR GABP-binding site and C/EBP-binding site (CE3) (indicated in bold) [5′-TCCTCATACTTCCTGCGGAGCTTATTGTCGTAAGAATATC-3′ (W)]; FCAR GABP site mutation [5′-TCCTCATACTTAGTGCGGAGCTTATTGTCGTAAGAATATC-3′ (mE)] (underlining indicates mutated sequences); FCAR CE3 site mutation [5′-TCCTCATACTTCCTGCGGAGCTTATTGTCCACCAAATATC-3′ (mC)]; FCAR GABP site nt −98/−79 (5′-CCTCATACTTCCTGCGGAGC-3′). The radiolabelled probe (25 fmoles) was incubated on ice with in vitro translated proteins (1 μl each) and/or purified GST-fusion protein (approx. 50 ng) and poly(dI-dC)·(dI-dC) (0.5 μg) as a non-specific competitor. Protein–DNA complexes were separated on 4% non-denaturing polyacrylamide gels in 0.25× TBE buffer (1×TBE is 90 mM Tris-base, 90 mM boric acid and 2 mM EDTA) at room temperature. Antibodies used for supershift EMSAs were rabbit antiserum against human GABPα (H-180; Santa Cruz Biotechnology) and rabbit antiserum against the C-terminal region of human C/EBPβ (Δ180; Santa Cruz Biotechnology), which cross-reacts with C/EBPα.

For Figure 5 and Supplementary Figure S2, in vitro translated GABPα mutants were prepared in the presence of L-[35S]methionine as described above, and analysed using SDS/PAGE. Labelled proteins were quantified by Image J software, and used for EMSA after being diluted to normalize protein concentrations. Autoradiography was performed using an extra film to quench radioactivity arising from the 35S-labelled proteins.

DNA-binding ability of the four-amino acid swap mutant proteins

Figure 5
DNA-binding ability of the four-amino acid swap mutant proteins

(A) Structure of the four-amino acid swap mutants. (B) DNA binding of GABPα mutants to the FCAR GABP-binding site. A labelled oligonucleotide spanning the FCAR promoter sequence nt −98/−79, which contains the GABP-binding site, was incubated with in vitro translated GABPβ and GABPα mutants shown in (A). The numbers below each lane number indicate the relative intensity quantified by Image J software.

Figure 5
DNA-binding ability of the four-amino acid swap mutant proteins

(A) Structure of the four-amino acid swap mutants. (B) DNA binding of GABPα mutants to the FCAR GABP-binding site. A labelled oligonucleotide spanning the FCAR promoter sequence nt −98/−79, which contains the GABP-binding site, was incubated with in vitro translated GABPβ and GABPα mutants shown in (A). The numbers below each lane number indicate the relative intensity quantified by Image J software.

Transfection

Plasmids were purified using a Maxiprep kit (Qiagen). HeLa human cervical carcinoma cells were transfected using Superfect transfection reagent (Qiagen) as described previously [27]. Briefly, cells plated on 24-well plates were transfected with the FCAR promoter–luciferase construct pGLmCE12-259 (0.7 μg) in the presence or absence of expression plasmids for GABPβ (pC3E4TF1-53S) and GABPα (pC3GAα) or its mutants (0.1 μg each) along with the C/EBPα expression plasmid pD3C/EBPαKoz or its empty vector pcDNA3.1(+) (1.5 ng). Each sample was co-transfected with pRL-CMV plasmid encoding Renilla luciferase (Promega) (2.5 ng) to normalize for transfection efficiency, and also with the additional empty vector pCR3.1E to bring the amount of transfected DNA in each sample to 1 μg. All transfections were performed in duplicate. Cells were harvested 24 h after transfection, and assayed for firefly and Renilla luciferases in a luminometer (Centro LB960; Berthold) using a dual-luciferase reporter assay system (Promega). Data were corrected for transfection efficiency and presented as the fold stimulation by the transcription factor(s) compared with the promoter activity seen without the expression vector(s).

ChIP (chromatin immunoprecipitation) and re-ChIP assays

HeLa cells plated on 10-cm-diameter dishes were transfected with 2.5 μg each of pGLmCE12-259, GABPβ (pC3E4TF1-53S) and HA-tagged GABPα (pC3HA-GABPα) or its mutant (pC3HA-GAα-β3-QQL) along with the C/EBPα expression plasmid pD3C/EBPαKoz as described above. The first ChIP was performed using rabbit polyclonal antiserum against HA (Santa Cruz Biotechnology) or rabbit IgG as described previously [14], except that ChIP complexes were eluted by incubation at 37 °C for 30 min in elution buffer (100 mM NaHCO3 and 1% SDS) supplemented with 10 mM dithiothreitol. After centrifugation, the supernatant was diluted 50 times with ChIP dilution buffer [14] and subjected to another round of immunoprecipitation using rabbit polyclonal antiserum against HA or human C/EBPα (Santa Cruz Biotechnology), or rabbit IgG. ChIP primers for the FCAR promoter–reporter plasmid were: 5′primer from the FCAR promoter −94/−71 (5′-ATACTTCCTGCGGAGCTTATTGTC-3′); and 3′ primer specific for the reporter construct pGL3-dBH (5′-CTTTACCAACAGTACCGGAATGC-3′) [29].

RESULTS

Protein domains required for GABPα/GABPβ–C/EBPα–DNA ternary complex formation

We previously demonstrated that GABPα protein sequence amino acids 318–399 corresponding to the ETS domain physically interacts with C/EBPα [14]. GABPα can also recruit C/EBPβ [14] and C/EBPδ (results not shown), suggesting that their interaction is mediated by the highly conserved bZIP domain of each C/EBP factor. To better characterize protein sequences important for C/EBPα recruitment, we prepared a recombinant protein GST–GAα318–399 [the GABPα ETS domain (amino acids 318–399) fused to GST] (Figure 1A) and C/EBPα and C/EBPα273–358 proteins [the full-length protein and the bZIP domain (amino acids 273–358) translated in vitro using rabbit reticulocyte lysates respectively] (Figure 1B). GST pull-down assay showed that 35S-labelled full-length C/EBPα and C/EBPα273–358 bind to GST–GAα318–399 (Figure 1C, lanes 3 and 4) but not to GST alone (Figure 1C, lanes 11 and 12), confirming that the GABPα ETS domain recruits the bZIP domain of C/EBPα.

The ETS domain of GABPα contains three α-helices (α1–α3) and four-stranded β-sheets (β1–β4) arranged in the order α1-β1-β2-α2-α3-β3-β4 folding into a ‘winged-helix-turn-helix’ configuration (Figures 1A and 2A) [30]. The structure is essentially identical in topology with structures of other Ets proteins, including Fli-1 [31], Ets-1 [32,33] and PU.1 [34]. This suggests that other Ets proteins also recruit C/EBPα. Indeed, the PU.1 ETS domain has been reported to physically interact with C/EBPα [35]. In addition, we observed that the Elf-1 and Fli-1 ETS domains also interact with the C/EBPα bZIP domain (Figure 1C, lanes 5 and 6, and 9 and 10), suggesting that physical interactions between Ets and C/EBP family members represent an evolutionarily conserved mechanism for transcriptional regulation in a wide variety of cell types.

We next performed EMSA using the FCAR promoter region (nt −99/−60 relative to the translation start ATG), which contains both binding sites for C/EBPs (previously named as CE3) and GABP [14]. This probe was incubated with in vitro translated GABPα and GABPβ proteins and/or purified GST recombinant protein containing the C/EBPα bZIP domain (C/EBPα273–358). In the presence of both GABPα/GABPβ and C/EBPα bZIP, we observed a complex migrating more slowly than the complex formed with GABPα/GABPβ alone (Figure 1D, compare lane 4 with lane 2). This complex formation was prevented by mutation of either binding site for GABP or C/EBP (Figure 1D, lane 9 or 10 respectively). In addition, the addition of anti-GABPα antibody abolished formation of the slower migrating complex and the complex forming with GABPα/GABPβ (Figure 1D, lane 5). Conversely, the slower migrating complex and the complex forming with C/EBPα273–358 were prevented by an antibody against amino acids 199–345 of C/EBPβ (Figure 1D, lane 6), which cross-reacts with the C/EBPα bZIP domain. Taken together, these results demonstrate that the GABPα/GABPβ heterodimer and the C/EBPα bZIP domain form a ternary complex with the FCAR promoter region.

The region spanning helix α3 and the β3/β4 region in the GABPα ETS domain is required for C/EBPα recruitment

To identify amino acid sequences allowing the recruitment of C/EBPα, we performed a pull-down assay using a set of differentially truncated GABPα ETS domain mutants that were fused to GST (Figure 2B). Even when both helices α1 and α2, and β-strands 1 and 2, were deleted, the ETS domain retained the ability to interact with 35S-labelled C/EBPα (Figure 2C, lanes 2–5). On the other hand, the recruitment of C/EBPα was considerably decreased by deletion of either helix α3 or the β3/β4 region (Figure 2C, lanes 6 and 7), and completely lost by removing both (Figure 2C, lane 8). These results indicate that each of helix α3 and the β3/β4 region is able to interact with C/EBPα by itself with very low affinity, and contributes to the GABPα–C/EBPα interaction by co-operative action.

Interaction of C/EBPα with chimaeric molecules of the GABPα ETS domain

Considering the evolutionarily conserved structure of the ETS domain, it is tempting to speculate that other Ets factors also use the region spanning both helix α3 and the β3/β4 region to interact with C/EBPα. However, for the interaction of PU.1 with C/EBPα, it has been reported that the PU.1 β3/β4 region itself is necessary and sufficient to interact with C/EBPα [35], indicating that the C/EBPα-interacting surfaces on Ets proteins are not always identical with those on GABPα. We therefore designed ‘domain-swap’ mutants to focus on amino acid residues that participate in the C/EBPα recruitment.

In the first instance, we divided the C/EBPα-interacting region into three regions: (i) helix α3; (ii) the loop joining helix α3 and β-strand 3; and (iii) the β3/β4 region. Each of them was replaced with the homologous amino acid sequences of other Ets proteins, including Elf-1, PU.1 and Fli-1 (Figure 3A). The GST-fused chimaeric ETS domain in which GABPα helix α3 was exchanged for that of Elf-1, PU.1 or Fli-1 showed an affinity for C/EBPα similar to the WT GABPα ETS domain (Figure 3B, compare lanes 3–5 with lane 2). On the other hand, replacement of the GABPα loop joining helix α3 and β-strand 3 by that of Elf-1, PU.1 or Fli-1 increased the ability to interact with C/EBPα (Figure 3B, lanes 6–8). Also, the chimaeric ETS domain in which the GABPα β3/β4 region was replaced by that of PU.1 or Fli-1 interacted with C/EBPα with a higher affinity (Figure 3B, lanes 10 and 11). However, the replacement of this region by that of Elf-1 (GST–GAαETS-GGE) resulted in an approx. 3–4-fold reduction in the ability to recruit C/EBPα (Figure 3B, compare lane 9 with lane 2). These results suggest that the Elf-1 ETS domain interacts with C/EBPα through interacting surfaces different from those on the GABPα, PU.1 and Fli-1 ETS domains, and highlights the β3/β4 region as an important structural determinant for the GABPα–C/EBPα interaction.

Because no crystallographic data for the structure of Elf-1 are as yet available, we could not predict the structural basis explaining why exchanging by the Elf-1 β3/β4 region but not by other Ets β3/β4 regions reduced the affinity for C/EBPα. Within the β3/β4 region, there are six amino acid differences between GABPα and Elf-1 (Figure 2A). However, single-amino acid swapping of these residues (I387L, C388A, Q391E, K393Q, F395L or K398Q) resulted in little if any effects on the GABPα–C/EBPα interaction (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/430/bj4300129add.htm), suggesting an importance of overall structure formed by the β3/β4 region rather than the specific amino acid residues. To more accurately define this C/EBPα-interacting region, we further divided it into three sub-domains based on structural features: (i) β-strand 3; (ii) the loop connecting β-strands 3 and 4; and (iii) β-strand 4. Regarding β-strand 4, there is only one amino acid difference between GABPα and Elf-1 (Figure 2A), whose swapping K398Q had an only marginal effect (see Supplementary Figure S1B, lane 8). Therefore we replaced the two sub-domains, (i) β-strand 3 and (ii) the subsequent loop, with those of Elf-1 (Figure 3C). GST pull-down assay showed that replacement of either sub-domain by that of Elf-1 (GST–GAαETS-β3 or GST–GAαETS-L) resulted in an only slight decrease in affinity for C/EBPα (Figure 3D, compare lane 4 or 5 with lane 2). However, the chimaeric ETS domain GST–GAαETS-β3-L in which both were swapped had a 6-fold lower affinity (Figure 3D, lane 6), indicating that these sub-domains contribute to the C/EBPα recruitment complementarily.

Within β-strand 3 and the subsequent loop, there are five residues that are not present in Elf-1: Ile387 and Cys388 in β-strand 3; and Gln391, Lys393 and Phe395 in the loop (Figures 2A and 4A). We therefore generated four-amino acid swap mutants by swapping each of the two residues in β-strand 3 (I387L or C388A) in the context of GST–GAαETS-L or two of the three within the loop (Q391E/K393Q, Q391E/F395L or K393Q/F395L) in the context of GST–GAαETS-β3 (Figure 4A). As shown in Figure 4(B), all of the mutants were observed to be deleterious for the interaction with C/EBPα, but the effects were weaker than those seen with GST–GAαETS-β3-L. This suggests that these residues all contribute to the interaction with C/EBPα. It should be noted, however, that GST–GAαETS-β3-QQL showed an approx. 3.6-fold decrease in the affinity for C/EBPα (Figure 4B, lane 8; see below).

DNA-binding capacity of chimaeric GABPα molecules

The initial goal of the present study was to distinguish between ETS domain functions involved in transcriptional activation by itself and transcriptional synergy with C/EBPα. We observed a considerable decrease in the ability to interact with C/EBPα when both β-strand 3 and the subsequent loop in the β3/β4 domain were replaced with those of Elf-1 (Figure 3D). However, such replacement in the context of full-length GABPα resulted in a complete loss of DNA-binding capacity, probably due to the replacement of the loop (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/430/bj4300129add.htm).

Within the loop connecting β-strands 3 and 4, there are three amino acid differences between GABPα and Elf-1 (Figures 2A and 5A). Therefore we exchanged two of the three for those of Elf-1 (Q391E/K393Q, Q391E/F395L or K393Q/F395L) in the context of the chimaera where β-strand 3 is replaced by that of Elf-1 (Figure 5A, GAα-β3-EQF, GAα-β3-EKL or GAα-β3-QQL respectively). Although the mutation Q391E/K393Q or Q391E/F395L resulted in a complete loss of DNA-binding capacity (Figure 5B, lanes 3 and 4), GAα-β3-QQL containing the mutation K393Q/F395L retained DNA-binding ability with an approx. 10-fold lower DNA binding in comparison with the WT GABPα protein (Figure 5B, lane 5). Despite this defect, however, GAα-β3-QQL appeared to entirely retain the transcriptional activity (see below).

Transcriptional activity of GABPα mutants

The ‘domain-swap’ experiments demonstrated critical amino acid residues in the β-strand 3 and the subsequent loop of the ETS domain for interaction with C/EBPα. To determine the effects of their mutation on transcriptional activity, transfection assays were performed for the three GABPα mutants, GAα-β3-EQF, GAα-β3-EKL and GAα-β3-QQL (Figure 5A). Expression vectors for these mutants and GABPβ were co-transfected into HeLa cells with a reporter gene containing the FCAR promoter. In these experiments, we used pGL-mCE12-259, where two upstream C/EBP-binding sites (CE1 and CE2) of the three sites were mutated to investigate the co-operative effect only on CE3-mediated activation (Figure 6A) [14,27]. WT GABPα activated the FCAR promoter 4.2-fold over background levels (Figure 6B). As expected, GAα-β3-EQF and GAα-β3-EKL, which contain DNA binding-deficient Q391E/K393Q and Q391E/F395L respectively, were inactive for transactivation. On the other hand, the K393Q/F395L mutant GAα-β3-QQL activated the FCAR promoter (3.8-fold) at the same levels as the WT protein, indicating that this mutant substantially binds to the promoter in vivo. These results cannot be a consequence of different protein stability because similar levels of the mutant and WT proteins were expressed in transfected cells (see Supplementary Figure S3 at http://www.BiochemJ.org/bj/430/bj4300129add.htm).

Functional co-operation between C/EBPα and various GABPα mutants

Figure 6
Functional co-operation between C/EBPα and various GABPα mutants

(A) A schematic representation for the structure of the mutated FCAR promoter–luciferase (Luc) construct pGL-mCE12-259 [14,27]. × indicates C/EBPα-binding site mutation. (B and C) GABPβ and different GABPα mutants shown in Figure 5(A) were co-transfected into HeLa cells with pGL-mCE12–259 in the absence (B) or presence (C) of C/EBPα. Results are expressed as fold activation relative to basal promoter activity. Values are means±S.E.M. (n=7). Statistical analysis (paired t test) was performed to show the significant difference between GAα-β3-QQL and WT GABPα (*P<0.02). ns, not significant.

Figure 6
Functional co-operation between C/EBPα and various GABPα mutants

(A) A schematic representation for the structure of the mutated FCAR promoter–luciferase (Luc) construct pGL-mCE12-259 [14,27]. × indicates C/EBPα-binding site mutation. (B and C) GABPβ and different GABPα mutants shown in Figure 5(A) were co-transfected into HeLa cells with pGL-mCE12–259 in the absence (B) or presence (C) of C/EBPα. Results are expressed as fold activation relative to basal promoter activity. Values are means±S.E.M. (n=7). Statistical analysis (paired t test) was performed to show the significant difference between GAα-β3-QQL and WT GABPα (*P<0.02). ns, not significant.

We next examined functional interaction with C/EBPα (Figure 6C). Whereas C/EBPα alone activated the FCAR promoter 1.4-fold, it synergistically activated 54-fold in the presence of GABP. On the other hand, the GABPα mutants GAα-β3-EQF or GAα-β3-EKL co-operated only 16-fold or 8-fold respectively. Interestingly, the GAα-β3-QQL mutant, which showed no decrease in transactivation capacity, displayed a significant decrease in functional synergism with C/EBPα (30-fold compared with 54-fold; P<0.02). In order to more accurately quantify these synergistic effects, we calculated a true fold synergy as reported previously [36,37]. The fold synergy was defined as fold activation in the presence of a combination of GABPα/GABPβ and C/EBPα divided by the sum of the fold activation by each factor individually. In co-transfection of WT GABP and C/EBPα, the level of activation was calculated to be 9.1±2.4-fold greater than the sum of the individual responses. Although no transactivation capacity was observed with GAα-β3-EQF and GAα-β3-EKL, these retained lower levels of synergistic activation in combination with C/EBPα with fold synergy values of 6.0±2.1 and 3.2±0.9 respectively. On the other hand, GAα-β3-QQL had a 1.7-fold lower level of synergistic activation compared with WT GABPα (5.5±1.4-fold compared with 9.1±2.4-fold synergy; P<0.02). These results are consistent with our observation that this mutant showed approx. 3.6-fold lower affinities for C/EBPα than the WT protein in the GST pull-down assay (Figure 4B, lane 8).

It is possible that these functional changes are due to an alteration of the overall structure of the mutants that affects their ability to interact with the transactivation subunit GABPβ, although the domain responsible for binding to GABPβ lies downstream of the ETS domain. However, we observed by GST pull-down assay and co-immunoprecipitation experiments that these mutants are still able to interact with GABPβ at the same levels as the WT protein, indicating that the GABPα–GABPβ interaction is independent of the mutated residues (see Supplementary Figure S4 at http://www.BiochemJ.org/bj/430/bj4300129add.htm). Thus physical interaction between the ETS domain of GABPα and C/EBPα is necessary for proper synergistic activation of the FCAR promoter by GABP and C/EBPα.

In vivo recruitment of C/EBPα by GABPα

To investigate the in vivo interaction between GABPα and C/EBPα, and the importance of their interaction for promoter recruitment of C/EBPα, we co-transfected HeLa cells with the FCAR promoter-reporter construct pGL-mCE12-259, and expression plasmids encoding C/EBPα, GABPβ, and the WT GABPα or mutant GAα-β3-QQL, and performed re-ChIP analysis. As shown in Figure 7, when HA-tagged GABPα and its mutant HA–GAα-β3-QQL were first immunoprecipitated, subsequent re-ChIP analysis showed that the transfected FCAR promoter is equally co-immunoprecipitated with HA-tagged GABPα and its mutant HA–GAα-β3-QQL by anti-HA antibody. This indicates that in contrast with the data obtained in EMSA, the WT GABPα and mutant GAα-β3-QQL can equally bind to the FCAR promoter in vivo, which is consistent with the transactivation data in Figure 6(B) that WT and GAα-β3-QQL equally activates the FCAR promoter. Furthermore, the re-ChIP analysis showed that C/EBPα was associated with immunoprecipitated GABPα–promoter complex. More importantly, we observed a decreased level of C/EBPα associated with the GAα-β3-QQL mutant on the transfected FCAR promoter in the re-ChIP analysis. These results demonstrate that GABPα recruits C/EBPα to the promoter, and this recruitment requires the amino acid residues that mediate the GABPα–C/EBPα direct interaction.

In vivo recruitment of C/EBPα by GABPα

Figure 7
In vivo recruitment of C/EBPα by GABPα

HA-tagged GABPα or mutant GAα-β3-QQL were co-transfected into HeLa cells with the FCAR promoter–luciferase construct pGL3mCE12–259 in the presence of C/EBPα. At 24 h following transfection, cells were cross-linked and re-ChIP analysis was performed using primer sequencing specific for the transfected FCAR promoter–luciferase construct. Assays were performed using anti-HA or control IgG as the first immunoprecipitating (1st IP) antibody. Secondary antibodies used in the re-ChIP analysis are shown. The numbers below the panel indicate the relative intensity quantified by Image J software.

Figure 7
In vivo recruitment of C/EBPα by GABPα

HA-tagged GABPα or mutant GAα-β3-QQL were co-transfected into HeLa cells with the FCAR promoter–luciferase construct pGL3mCE12–259 in the presence of C/EBPα. At 24 h following transfection, cells were cross-linked and re-ChIP analysis was performed using primer sequencing specific for the transfected FCAR promoter–luciferase construct. Assays were performed using anti-HA or control IgG as the first immunoprecipitating (1st IP) antibody. Secondary antibodies used in the re-ChIP analysis are shown. The numbers below the panel indicate the relative intensity quantified by Image J software.

DISCUSSION

Many transcription factors very often use their DNA-binding domains for interacting with other transcription factors. Regarding Ets proteins, the ETS domains engage physical interactions with various DNA-binding motifs such as bZIP (c-Jun [37,38] and C/EBPs [14,35,39,40]), zinc finger {Sp1 (specificity protein 1) [41,42] and GATA proteins [4345]}, runt homology (AML1) [36] and Rel homology domains [NF (nuclear factor)-κB/NF-AT] [9,46]. However, to evaluate the functional significance of these direct interactions, one must distinguish between functions of the ETS domain involved in transcriptional activation by themselves and transcriptional synergy with their partners. By mutational study, we demonstrated that helix α3 and the subsequent β3/β4 region in the GABPα ETS domain co-operatively contribute to the recruitment of the C/EBPα bZIP domain (Figure 2C). Subsequent domain swapping allowed us to highlight β-strand 3 and the subsequent loop, which when exchanged by those of Elf-1 reduced the ability to recruit C/EBPα (Figure 3D). Of five residues in this region that differ between GABPα and Elf-1, swapping of four residues I387L/C388A/K393Q/F395L appeared to substantially preserve DNA-binding capacity enough to activate the FCAR promoter at the level similar to that seen with the WT protein (Figures 5B and 6B). The more important aspect is that this mutant GAα-β3-QQL displayed significantly decreased functional synergy with C/EBPα (Figure 6C), and this defect was demonstrated by the re-ChIP analysis to be due to decreased recruitment of C/EBPα to the promoter by this mutant (Figure 7). These results indicate that Ile387, Cys388, Lys393 and Phe395 of the GABPα ETS domain are required for both the physical and functional interactions with C/EBPα, and that there are likely to be exacting tertiary structural requirements in β-strand 3 and the subsequent loop for GABPα to functionally interact with C/EBPα that are different from those required for transcriptional activation by itself.

GABPα functions through recruiting the transactivation subunit GABPβ. This raises a question whether the GABPα–C/EBPα and GABPα–GABPβ interactions are mutually permissible on the FCAR promoter. The crystal structure of GABPα/GABPβ ETS domain-ankyrin repeat heterodimer bound to DNA has revealed that helix α3 is the major DNA recognition component whose amino acid side chains contact the core GGA motif at the centre of the GABPα recognition sequence [30]. On the other hand, the β3/β4 region contacts phosphates 5′ of the GGA motif. In the FCAR promoter, the core GGA motif for GABPα lies on the antisense strand, and C/EBPα binds to its binding motif CE3 on the sense strand located to the 5′ side of the GGA motif for GABP. Interestingly, the GABPβ subunit interacts with GABPα on DNA being to the 3′ side of the GGA motif, where GABPβ lies in the opposite side to the β3/β4 region across helix α3 [30]. Thus if the C/EBPα bZIP domain contacts the β-strand 3 and the subsequent loop on GABPα, it can be incorporated into the GABP–DNA complex without preventing the GABPα–GABPβ interaction. Consistent with this, we could indeed detect the GABPα/GABPβ–C/EBPα bZIP–DNA ternary complex in EMSA (Figure 1D).

Despite their complete defect in DNA binding and transcriptional activation capacities (probably due to having the substitution Q391E) (Figures 5B and 6B), the four-amino acid swap mutants GAα-β3-EQF and GAα-β3-EKL retained functional synergistic effects with C/EBPα to some extent, albeit their ability was severely decreased (Figure 6C). This suggests that these mutants could functionally interact with C/EBPα without binding to promoter DNA. However, we previously showed that mutation of the GABP-binding site in the FCAR promoter abolished co-operative transcriptional activation by GABP and C/EBPα [14], indicating that it is not the case. These mutants may interact with DNA in vivo to some extent, possibly through being stabilized by ‘higher order’ interaction with other factors such as C/EBPα and the cofactor p300, which has been reported to interact with GABPα [4750].

Regarding the interactions between the ETS and bZIP domains, the amino acid residue Tyr371 of the Ets protein Erg, which locates in helix α3 of the ETS domain, has been demonstrated to mediate c-Jun recruitment: its mutation to valine decreases transcriptional synergy as well as the ability to physically interact with the Jun/Fos heterodimer without affecting its transcriptional activation capacity [37]. Although this tyrosine residue (Tyr380 in GABPα) is conserved among Ets family members except for PU.1 (Figure 2A), the GABPα mutation Y380V completely abolished DNA-binding capacity (T. Shimokawa and C. Ra, unpublished work), indicating that the transcriptional conformations of the complex determined on the Erg–Jun/Fos–DNA complex could not be extended to all the Ets–bZIP transcription complexes.

Interestingly, chemical-shift perturbation studies have previously shown that the direct interaction of the PU.1 ETS domain with the C-terminal zinc finger motif of GATA1 is mediated by not only amino acid residues in the β3/β4 region but also residues in the N-terminal end of helix α3 and in helices α1 and α2 [51], despite that the previous pull-down data indicated that the PU.1 β3/β4 region was necessary and sufficient for interacting with GATA1 [43]. However, similar to our data for the GABPα–C/EBPα interaction (see Supplementary Figure S1), none of the single point mutations in these residues abrogated the PU.1–GATA1 interaction [51]. These observations suggest that several sub-domains in the ETS domain possess potential interaction elements with varying affinities for each interacting partner. Among them, the sub-domain that possesses the highest affinity for a given interaction partner may be used as a main interacting surface, which appears to vary from member to member even among Ets family members. In respect of the binding surfaces of GABPα and PU.1 for C/EBPα, whereas the β3/β4 region of PU.1 is necessary and sufficient to recruit C/EBPα [35], helix α3 in addition to the β3/β4 region of GABPα is required to efficiently interact with C/EBPα (Figure 2C). Interestingly, the functional consequences of the C/EBPα recruitment by GABPα and PU.1 clearly differ. GABPα activates transcription in co-operation with C/EBPα to probably promote granulocyte gene expression [14]. On the other hand, C/EBPα recruited by PU.1 inhibits the PU.1 function through competing with its cofactor c-Jun for the β3/β4 region to possibly repress monocyte gene expression and differentiation [35]. Such functional diversity might determine ETS domain structural components interacting with its partner. In this context, domain swapping among Ets proteins based on such diversity is useful for identifying ETS sub-domains that determine specific protein–protein interactions with a given partner.

Ets family members are intimately involved in important cellular functions, including development, cellular differentiation, apoptosis and carcinogenesis. However, because many cell types simultaneously express several Ets proteins and these proteins bind similar DNA sequences, how individual Ets proteins achieve specificity of function has long been elusive. One of the mechanisms that confer tissue- or development stage-specific functions on individual Ets proteins appears to be protein–protein interactions with other key transcription factors. However, the biological significance of many of these interactions remains to be evaluated, because their molecular basis has been elusive. In the present study, we succeeded in identifying sub-domains and amino acid residues in the Ets domain that are critical for physical and functional interaction with the key myeloid Ets partner factor C/EBPα, but do not perturb the capacity of GABPα to function alone. Although whether our findings could be extended to other Ets proteins remains to be assessed, the identified protein–protein interface may provide a unique target for intervention to distinguish the individual functional properties of Ets proteins, which provides an approach for correlating the specific physical interaction with biological function in development, cellular differentiation, apoptosis and carcinogenesis. With regard to GABPα, targeted disruption of Gabpa (encoding mouse GABPα) resulted in a peri-implantation embryonic lethal defect [52], possibly due to roles of its target genes in essential cellular processes, including mitochondrial function, protein synthesis and cell-cycle events [13]. Thus direct evaluation of the role of GABP needs effort to manipulate only specific functions of GABP without affecting functions other than those to be examined. The present study will provide a framework for understanding the molecular mechanism of the physical interaction between GABP, C/EBPα and DNA, which could potentially lead to the generation of GABPα variants specifically designed to retain GABP function in the absence of properties requiring interaction with C/EBPα. Such variants could allow us to determine GABP biological function specifically mediated by the interaction with C/EBPα.

Abbreviations

     
  • bZIP

    basic leucine zipper

  •  
  • C/EBP

    CCAAT/enhancer-binding protein

  •  
  • ChIP

    chromatin immunoprecipitation

  •  
  • Elf-1

    E74-like factor 1

  •  
  • EMSA

    electrophoretic mobility-shift assay

  •  
  • FcαR

    Fc receptor for IgA

  •  
  • Fli-1

    Friend leukaemia integration 1

  •  
  • GABP

    GA-binding protein

  •  
  • GST

    glutathione transferase

  •  
  • HA

    haemagglutinin

  •  
  • IL

    interleukin

  •  
  • WT

    wild-type

AUTHOR CONTRIBUTION

Toshibumi Shimokawa designed and performed the study and experiments, analysed the data and wrote the paper. Satoshi Nunomura and Yukinori Enomoto performed experiments and analysed the data. Chisei Ra designed the study, analysed the data and wrote the paper.

We thank Dr Gretchen J. Darlington (Bayer College of Medicine, Houston, TX, U.S.A.) for the C/EBPα expression vector, Dr Jeffrey M. Leiden (Clarus Ventures, Cambridge, MA, U.S.A.) for human Elf-1 cDNA, Dr Michael J. Klemsz (Indiana University School of Medicine, Indianapolis, IN, U.S.A.) and Dr Shoichi Suzuki (Saga Medical School, Saga, Japan) for human PU.1 cDNA.

FUNDING

This work was supported in part by a grant for the ‘Strategic Research Base Development’ Program for Private Universities subsidized by the Ministry of Education, Culture, Sports, Science and Technology, Japan [grant number S0801033], a grant for Scientific Research from the Japanese Society for the Promotion of Science [grant number 15390163], and a grant from Nihon University School of Medicine.

References

References
1
Oikawa
T.
Yamada
T.
Molecular biology of the Ets family of transcription factors
Gene
2003
, vol. 
303
 (pg. 
11
-
34
)
2
Janknecht
R.
Nordheim
A.
Gene regulation by Ets proteins
Biochim. Biophys. Acta
1993
, vol. 
1155
 (pg. 
346
-
356
)
3
Bartel
F. O.
Higuchi
T.
Spyropoulos
D. D.
Mouse models in the study of the Ets family of transcription factors
Oncogene
2000
, vol. 
19
 (pg. 
6443
-
6454
)
4
Scott
E. W.
Simon
M. C.
Anastasi
J.
Singh
H.
Requirement of transcription factor PU.1 in the development of multiple hematopoietic lineages
Science
1994
, vol. 
265
 (pg. 
1573
-
1577
)
5
McKercher
S. R.
Torbett
B. E.
Anderson
K. L.
Henkel
G. W.
Vestal
D. J.
Baribault
H.
Klemsz
M.
Feeney
A. J.
Wu
G. E.
Paige
C. J.
Maki
R. A.
Targeted disruption of the PU.1 gene results in multiple hematopoietic abnormalities
EMBO J.
1996
, vol. 
15
 (pg. 
5647
-
5658
)
6
Hart
A.
Melet
F.
Grossfeld
P.
Chien
K.
Jones
C.
Tunnacliffe
A.
Favier
R.
Bernstein
A.
Fli-1 is required for murine vascular and megakaryocytic development and is hemizygously deleted in patients with thrombocytopenia
Immunity
2000
, vol. 
13
 (pg. 
167
-
177
)
7
Spyropoulos
D. D.
Pharr
P. N.
Lavenburg
K. R.
Jackers
P.
Papas
T. S.
Ogawa
M.
Watson
D. K.
Hemorrhage, impaired hematopoiesis, and lethality in mouse embryos carrying a targeted disruption of the Fli1 transcription factor
Mol. Cell. Biol.
2000
, vol. 
20
 (pg. 
5643
-
5652
)
8
Thompson
C. B.
Wang
C. Y.
Ho
I. C.
Bohjanen
P. R.
Petryniak
B.
June
C. H.
Miesfeldt
S.
Zhang
L.
Nabel
G. J.
Karpinski
B.
, et al. 
cis-acting sequences required for inducible interleukin-2 enhancer function bind a novel Ets-related protein, Elf-1
Mol. Cell. Biol.
1992
, vol. 
12
 (pg. 
1043
-
1053
)
9
John
S.
Reeves
R. B.
Lin
J. X.
Child
R.
Leiden
J. M.
Thompson
C. B.
Leonard
W. J.
Regulation of cell-type-specific interleukin-2 receptor alpha-chain gene expression: potential role of physical interactions between Elf-1, HMG-I(Y), and NF-kappa B family proteins
Mol. Cell. Biol.
1995
, vol. 
15
 (pg. 
1786
-
1796
)
10
Wurster
A. L.
Siu
G.
Leiden
J. M.
Hedrick
S. M.
Elf-1 binds to a critical element in a second CD4 enhancer
Mol. Cell. Biol.
1994
, vol. 
14
 (pg. 
6452
-
6463
)
11
LaMarco
K.
Thompson
C. C.
Byers
B. P.
Walton
E. M.
McKnight
S. L.
Identification of Ets- and notch-related subunits in GA binding protein
Science
1991
, vol. 
253
 (pg. 
789
-
792
)
12
Watanabe
H.
Wada
T.
Handa
H.
Transcription factor E4TF1 contains two subunits with different functions
EMBO J.
1990
, vol. 
9
 (pg. 
841
-
847
)
13
Rosmarin
A. G.
Resendes
K. K.
Yang
Z.
McMillan
J. N.
Fleming
S. L.
GA-binding protein transcription factor: a review of GABP as an integrator of intracellular signaling and protein-protein interactions
Blood Cells Mol. Dis.
2004
, vol. 
32
 (pg. 
143
-
154
)
14
Shimokawa
T.
Ra
C.
C/EBPalpha functionally and physically interacts with GABP to activate the human myeloid IgA Fc receptor (Fc alphaR, CD89) gene promoter
Blood
2005
, vol. 
106
 (pg. 
2534
-
2542
)
15
Rosmarin
A. G.
Caprio
D. G.
Kirsch
D. G.
Handa
H.
Simkevich
C. P.
GABP and PU.1 compete for binding, yet cooperate to increase CD18 (beta 2 leukocyte integrin) transcription
J. Biol. Chem.
1995
, vol. 
270
 (pg. 
23627
-
23633
)
16
Nuchprayoon
I.
Simkevich
C. P.
Luo
M.
Friedman
A. D.
Rosmarin
A. G.
GABP cooperates with c-Myb and C/EBP to activate the neutrophil elastase promoter
Blood
1997
, vol. 
89
 (pg. 
4546
-
4554
)
17
Xue
H. H.
Bollenbacher
J.
Rovella
V.
Tripuraneni
R.
Du
Y. B.
Liu
C. Y.
Williams
A.
McCoy
J. P.
Leonard
W. J.
GA binding protein regulates interleukin 7 receptor alpha-chain gene expression in T cells
Nat. Immunol.
2004
, vol. 
5
 (pg. 
1036
-
1044
)
18
Xue
H. H.
Bollenbacher-Reilley
J.
Wu
Z.
Spolski
R.
Jing
X.
Zhang
Y. C.
McCoy
J. P.
Leonard
W. J.
The transcription factor GABP is a critical regulator of B lymphocyte development
Immunity
2007
, vol. 
26
 (pg. 
421
-
431
)
19
DeKoter
R. P.
Schweitzer
B. L.
Kamath
M. B.
Jones
D.
Tagoh
H.
Bonifer
C.
Hildeman
D. A.
Huang
K. J.
Regulation of the interleukin-7 receptor alpha promoter by the Ets transcription factors PU.1 and GA-binding protein in developing B cells
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
14194
-
14204
)
20
Pang
L.
Xue
H. H.
Szalai
G.
Wang
X.
Wang
Y.
Watson
D. K.
Leonard
W. J.
Blobel
G. A.
Poncz
M.
Maturation stage-specific regulation of megakaryopoiesis by pointed-domain Ets proteins
Blood
2006
, vol. 
108
 (pg. 
2198
-
2206
)
21
Lekstrom-Himes
J.
Xanthopoulos
K. G.
Biological role of the CCAAT/enhancer-binding protein family of transcription factors
J. Biol. Chem.
1998
, vol. 
273
 (pg. 
28545
-
28548
)
22
Zhang
D. E.
Zhang
P.
Wang
N. D.
Hetherington
C. J.
Darlington
G. J.
Tenen
D. G.
Absence of granulocyte colony-stimulating factor signaling and neutrophil development in CCAAT enhancer binding protein alpha-deficient mice
Proc. Natl. Acad. Sci. U.S.A.
1997
, vol. 
94
 (pg. 
569
-
574
)
23
Radomska
H. S.
Huettner
C. S.
Zhang
P.
Cheng
T.
Scadden
D. T.
Tenen
D. G.
CCAAT/enhancer binding protein alpha is a regulatory switch sufficient for induction of granulocytic development from bipotential myeloid progenitors
Mol. Cell. Biol.
1998
, vol. 
18
 (pg. 
4301
-
4314
)
24
Timchenko
N.
Wilson
D. R.
Taylor
L. R.
Abdelsayed
S.
Wilde
M.
Sawadogo
M.
Darlington
G. J.
Autoregulation of the human C/EBP alpha gene by stimulation of upstream stimulatory factor binding
Mol. Cell. Biol.
1995
, vol. 
15
 (pg. 
1192
-
1202
)
25
Kozak
M.
The scanning model for translation: an update
J. Cell Biol.
1989
, vol. 
108
 (pg. 
229
-
241
)
26
Suzuki
S.
Kumatori
A.
Haagen
I. A.
Fujii
Y.
Sadat
M. A.
Jun
H. L.
Tsuji
Y.
Roos
D.
Nakamura
M.
PU.1 as an essential activator for the expression of gp91(phox) gene in human peripheral neutrophils, monocytes, and B lymphocytes
Proc. Natl. Acad. Sci. U.S.A.
1998
, vol. 
95
 (pg. 
6085
-
6090
)
27
Shimokawa
T.
Ra
C.
C/EBP alpha and Ets protein family members regulate the human myeloid IgA Fc receptor (Fc alpha R, CD89) promoter
J. Immunol.
2003
, vol. 
170
 (pg. 
2564
-
2572
)
28
Lai
J. S.
Herr
W.
Ethidium bromide provides a simple tool for identifying genuine DNA-independent protein associations
Proc. Natl. Acad. Sci. U.S.A.
1992
, vol. 
89
 (pg. 
6958
-
6962
)
29
Shimokawa
T.
Tsuge
T.
Okumura
K.
Ra
C.
Identification and characterization of the promoter for the gene encoding the human myeloid IgA Fc receptor (FcalphaR, CD89)
Immunogenetics
2000
, vol. 
51
 (pg. 
945
-
954
)
30
Batchelor
A. H.
Piper
D. E.
de la Brousse
F. C.
McKnight
S. L.
Wolberger
C.
The structure of GABPalpha/beta: an ETS domain-ankyrin repeat heterodimer bound to DNA
Science
1998
, vol. 
279
 (pg. 
1037
-
1041
)
31
Liang
H.
Mao
X.
Olejniczak
E. T.
Nettesheim
D. G.
Yu
L.
Meadows
R. P.
Thompson
C. B.
Fesik
S. W.
Solution structure of the ets domain of Fli-1 when bound to DNA
Nat. Struct. Biol.
1994
, vol. 
1
 (pg. 
871
-
875
)
32
Donaldson
L. W.
Petersen
J. M.
Graves
B. J.
McIntosh
L. P.
Solution structure of the ETS domain from murine Ets-1: a winged helix-turn-helix DNA binding motif
EMBO J.
1996
, vol. 
15
 (pg. 
125
-
134
)
33
Werner
M. H.
Clore
M.
Fisher
C. L.
Fisher
R. J.
Trinh
L.
Shiloach
J.
Gronenborn
A. M.
The solution structure of the human ETS1-DNA complex reveals a novel mode of binding and true side chain intercalation
Cell
1995
, vol. 
83
 (pg. 
761
-
771
)
34
Kodandapani
R.
Pio
F.
Ni
C. Z.
Piccialli
G.
Klemsz
M.
McKercher
S.
Maki
R. A.
Ely
K. R.
A new pattern for helix-turn-helix recognition revealed by the PU.1 ETS-domain-DNA complex
Nature
1996
, vol. 
380
 (pg. 
456
-
460
)
35
Reddy
V. A.
Iwama
A.
Iotzova
G.
Schulz
M.
Elsasser
A.
Vangala
R. K.
Tenen
D. G.
Hiddemann
W.
Behre
G.
Granulocyte inducer C/EBPalpha inactivates the myeloid master regulator PU.1: possible role in lineage commitment decisions
Blood
2002
, vol. 
100
 (pg. 
483
-
490
)
36
Petrovick
M. S.
Hiebert
S. W.
Friedman
A. D.
Hetherington
C. J.
Tenen
D. G.
Zhang
D. E.
Multiple functional domains of AML1: PU.1 and C/EBPalpha synergize with different regions of AML1
Mol. Cell. Biol.
1998
, vol. 
18
 (pg. 
3915
-
3925
)
37
Verger
A.
Buisine
E.
Carrere
S.
Wintjens
R.
Flourens
A.
Coll
J.
Stehelin
D.
Duterque-Coquillaud
M.
Identification of amino acid residues in the ETS transcription factor Erg that mediate Erg-Jun/Fos-DNA ternary complex formation
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
17181
-
17189
)
38
Bassuk
A. G.
Leiden
J. M.
A direct physical association between ETS and AP-1 transcription factors in normal human T cells
Immunity
1995
, vol. 
3
 (pg. 
223
-
237
)
39
Nagulapalli
S.
Pongubala
J. M.
Atchison
M. L.
Multiple proteins physically interact with PU.1. Transcriptional synergy with NF-IL6 beta (C/EBP delta, CRP3)
J. Immunol.
1995
, vol. 
155
 (pg. 
4330
-
4338
)
40
Yang
Z.
Wara-Aswapati
N.
Chen
C.
Tsukada
J.
Auron
P. E.
NF-IL6 (C/EBPbeta) vigorously activates il1b gene expression via a Spi-1 (PU.1) protein-protein tether
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
21272
-
21277
)
41
Galvagni
F.
Capo
S.
Oliviero
S.
Sp1 and Sp3 physically interact and co-operate with GABP for the activation of the utrophin promoter
J. Mol. Biol.
2001
, vol. 
306
 (pg. 
985
-
996
)
42
Takahashi
K.
Hayashi
N.
Shimokawa
T.
Umehara
N.
Kaminogawa
S.
Ra
C.
Cooperative regulation of Fc receptor gamma-chain gene expression by multiple transcription factors, including Sp1, GABP, and Elf-1
J. Biol. Chem.
2008
, vol. 
283
 (pg. 
15134
-
15141
)
43
Zhang
P.
Behre
G.
Pan
J.
Iwama
A.
Wara-Aswapati
N.
Radomska
H. S.
Auron
P. E.
Tenen
D. G.
Sun
Z.
Negative cross-talk between hematopoietic regulators: GATA proteins repress PU.1
Proc. Natl. Acad. Sci. U.S.A.
1999
, vol. 
96
 (pg. 
8705
-
8710
)
44
Rekhtman
N.
Radparvar
F.
Evans
T.
Skoultchi
A. I.
Direct interaction of hematopoietic transcription factors PU.1 and GATA-1: functional antagonism in erythroid cells
Genes Dev.
1999
, vol. 
13
 (pg. 
1398
-
1411
)
45
Nerlov
C.
Querfurth
E.
Kulessa
H.
Graf
T.
GATA-1 interacts with the myeloid PU.1 transcription factor and represses PU.1-dependent transcription
Blood
2000
, vol. 
95
 (pg. 
2543
-
2551
)
46
Bassuk
A. G.
Anandappa
R. T.
Leiden
J. M.
Physical interactions between Ets and NF-kappaB/NFAT proteins play an important role in their cooperative activation of the human immunodeficiency virus enhancer in T cells
J. Virol.
1997
, vol. 
71
 (pg. 
3563
-
3573
)
47
Bannert
N.
Avots
A.
Baier
M.
Serfling
E.
Kurth
R.
GA-binding protein factors, in concert with the coactivator CREB binding protein/p300, control the induction of the interleukin 16 promoter in T lymphocytes
Proc. Natl. Acad. Sci. U.S.A.
1999
, vol. 
96
 (pg. 
1541
-
1546
)
48
Bush
T. S.
St Coeur
M.
Resendes
K. K.
Rosmarin
A. G.
GA-binding protein (GABP) and Sp1 are required, along with retinoid receptors, to mediate retinoic acid responsiveness of CD18 (beta 2 leukocyte integrin): a novel mechanism of transcriptional regulation in myeloid cells
Blood
2003
, vol. 
101
 (pg. 
311
-
317
)
49
Resendes
K. K.
Rosmarin
A. G.
GA-binding protein and p300 are essential components of a retinoic acid-induced enhanceosome in myeloid cells
Mol. Cell. Biol.
2006
, vol. 
26
 (pg. 
3060
-
3070
)
50
Kang
H. S.
Nelson
M. L.
Mackereth
C. D.
Scharpf
M.
Graves
B. J.
McIntosh
L. P.
Identification and structural characterization of a CBP/p300binding domain from the ETS family transcription factor GABP alpha
J. Mol. Biol.
2008
, vol. 
377
 (pg. 
636
-
646
)
51
Liew
C. W.
Rand
K. D.
Simpson
R. J.
Yung
W. W.
Mansfield
R. E.
Crossley
M.
Proetorius-Ibba
M.
Nerlov
C.
Poulsen
F. M.
Mackay
J. P.
Molecular analysis of the interaction between the hematopoietic master transcription factors GATA-1 and PU.1
J. Biol. Chem.
2006
, vol. 
281
 (pg. 
28296
-
28306
)
52
Ristevski
S.
O'Leary
D. A.
Thornell
A. P.
Owen
M. J.
Kola
I.
Hertzog
P. J.
The ETS transcription factor GABPalpha is essential for early embryogenesis
Mol. Cell. Biol.
2004
, vol. 
24
 (pg. 
5844
-
5849
)

Supplementary data