ABSTRACT
Transcription factor E4TF1 is the human homolog of GABP and has been renamed hGABP (human GABP). hGABP is composed of two types of subunits; hGABP[beta]1/E4TF1-53 and the ets-related protein hGABP[alpha]/E4TF1-60. Both bind together to form an ([alpha])2([beta]1)2 heterotetrameric complex on DNA and activate transcription at specific promoters in vitro. Tetramer formation depends on two regions of hGABP[beta]1; the N-terminal region containing the Notch/ankyrin-type repeats is necessary for binding to hGABP[alpha] and the C-terminal region is necessary for homodimerization. In this report, we constructed various deletion mutants of hGABP[beta]1 in order to delimit the functional regions required for nuclear localization and transcription activity. We found that hGABP[beta]1 localization in the nucleus is dependent on a region located between amino acids 243 and 330 and that the presence of hGABP[beta]1 influences the efficiency of hGABP[alpha] transport into the nucleus. Next, we demonstrated that the hGABP complex composed of [alpha] and [beta]1 subunits activates transcription from the adenovirus early 4 promoter in vivo. This transcription activation needs the C-terminal region of hGABP[beta]1 and is consistent with results obtained with the in vitro assay. Furthermore, site-directed mutagenesis analysis of the C-terminal region reveals that the [alpha]-helix structure and the leucine residues are important for formation of a heterotetrameric complex with hGABP[alpha]in vitro and for transcription activation in vivo. These results suggest that hGABP[beta]1 stimulates transcription as part of a heterotetrameric complex with hGABP[alpha]in vivo.
In eukaryotes, many gene-specific transcription factors regulate transcription initiation by RNA polymerase II. These transcription factors function cooperatively by forming large complexes with one another at promoter sequences. Within these complexes protein-protein interactions between DNA binding transcription factors and non-DNA binding transcription factors have been shown to be responsible in most instances for transcriptional regulation of gene expression. Small regions termed activation domains were found to interact with certain general transcription factors and/or co-activators and to be both necessary and sufficient for transcription activation.
Transcription factor E4TF1 was originally purified to homogeneity from HeLa cells on the basis of its ability to bind to and stimulate transcription from the adenovirus early 4 (E4) promoter (1 ,2 ). Further characterization of E4TF1 revealed the presence of two distinct subunits, an ets-related DNA binding protein, E4TF1-60, and a non-DNA binding factor, E4TF1-53. The N-terminal region of the latter contains four tandemly repeated motifs homologous to Notch/ankyrin. Both subunits interacted with one another to form the E4TF1 heterotetrameric complex (E4TF1-60)2(E4TF1-53)2 on the specific DNA sequence 5'-CGGAAGTG-3'. This was shown to result in efficient activation of transcription in vitro (2 -5 ). Another E4TF1 subunit, E4TF1-47, is identical to E4TF1-53 in the N-terminal region but contains a distinct C-terminal 15 amino acid sequence. It can complex with E4TF1-60 but the complex does not stimulate transcription in vitro as efficiently as E4TF1 complexes made up of E4TF1-53 (6 ). Sequence data of cDNA clones from a HeLa cDNA library showed that E4TF1-60, E4TF1-53 and E4TF1-47 are highly homologous to GABP[alpha], GABP[beta]1-1 and GABP[beta]1-2 respectively. For this reason, E4TF1 is thought to be the human homolog of rat GA binding protein (GABP), which was purified as a factor that binds to a cis-regulatory DNA sequence important for herpes simplex virus type 1 (HSV-1) immediate early (IE) gene activation (8 -10 ,13 ). E4TF1 and its subunits have thus been renamed according to the human GABP nomenclature (hGABP). hGABP[alpha], hGABP[beta]1 and hGABP[gamma]1 correspond to E4TF1-60, E4TF1-53 and E4TF1-47, as shown in Figure 1 A. The genes for the hGABP subunits, hGABP[alpha] and hGABP[beta]1, were mapped to human chromosomes 21.q21.2-q21.3 and 7.q11.21 respectively, while the genes for the mouse GABP subunits, GABP[alpha] and [beta], were mapped to mouse chromosomes 6 and 2 respectively (11 -13 ). Recently, the transcription factors EF-1A (14 ), NRF-2 (15 ), XrpFI (16 ), RBF-1 (17 ) and [beta] factor (18 ) have been found to be immunologically related to GABP. This is especially the case for the NRF-2 subunits [alpha], [beta]2 and [gamma]2, which are identical to hGABP[alpha], hGABP[beta]1 and hGABP[gamma]1 respectively at the level of cDNA. The NRF-2 [beta]1 and [gamma]1 subunits are variants of [beta]2 and [gamma]2 possessing an additional 12 amino acid insertion (19 ). GABP has been shown to be involved in the expression of certain cellular genes, for instance the male-specific steroid 16[alpha]-hydroxylase gene (20 ) and the leukocyte-specific adhesion molecule CD18 ([beta]2 leukocyte integrin; 21 ).
pET53 (6 ) was completely digested with SacI and partially digested with PstI to generate an 879 bp fragment that carries a partial cDNA of hGABP[beta]1. In order to generate plasmids pET[beta]1QK339GT and pET[beta]1GL341GT expressing [beta]1QK339GT and [beta]1GL341GT respectively (Fig. 4 ), the DNA fragments encoding the N-terminal flanking region of the mutants 339-N and 341-N were synthesized and annealed. The DNA fragments encoding the C-terminal flanking regions of the mutants were amplified using two primers, [beta]BamHI and 339-C or 341-C respectively. The PCR products were digested with KpnI and BamHI and purified using agarose gel electrophoresis. The two fragments above and the 879 bp fragment were inserted into the SacI and BamHI sites of pET53. To generate plasmids pET[beta]1KL369GT and pET[beta]1EA371GT expressing [beta]1KL369GT and [beta]1EA371GT, the DNA fragments encoding the C-terminal flanking regions of mutants 369-C and 371-C were synthesized and annealed. The DNA fragments encoding the N-terminal flanking regions of the mutants were amplified using two primers, [beta]sppstI and 369-N or 371-N respectively. The PCR products were digested with PstI and KpnI and purified by agarose gel electrophoresis. The two fragments above and the 879 bp fragment were inserted intothe SacI and BamHI sites of pET53. The other plasmids that express hGABP[beta] site-directed mutants in Escherichia coli were constructed as follows. The DNA fragment encoding the N-terminal flanking region of the mutant was amplified using two primers, Number-N and [beta]sppstI. The DNA fragment encoding the C-terminal flanking region was amplified by PCR using two primers, Number-C and [beta]sppstI. The PCR products were digested with KpnI and BamHI or PstI and purified by agarose gel electrophoresis. The two PCR fragments and the 879 bp fragmentwere inserted into the SacI and BamHI sites of pET53. The construction of hGABP[beta]1 deletion mutant expression vectors in E.coli was as described previously (7 ).
All plasmids that expressed hGABP[beta]1 mutants in Drosophila melanogaster Schneider line 2 (SL2) cells (22 ) were constructed as follows. pET[beta] mutants were digested with BamHI and BglII. The DNA fragments containing the region that codes for the hGABP[beta] mutants were subcloned into the BamHI site of A5C[Delta]P, which contains the Drosophila actin 5C promoter (23 ).
All plasmids that expressed mutant hGABP[beta]1 in COS-1 cells were constructed as follows. pET[beta] mutant plasmids were all digested with BamHI and BglII. The DNA fragments containing the coding regions of the corresponding mutants were subcloned into the BglII site of the mammalian expression vector pCAGGS (24 ).
The pET[beta]1I243/NLS/330 plasmid was constructed as follows. The C-terminal flanking region was amplified using two primers, [beta]BamHI and NLS-243, which contains the nuclear localization signal (NLS) sequence of the SV40 large T antigen. The PCR product was digested with BamHI and PstI and purified by agarose gel electrophoresis. The DNA fragment coding for a part of hGABP[beta] was prepared by digesting pET53 with SacI and PstI and purifying the DNA fragment by agarose gel electrophoresis. Then, the PCR product and the fragment produced by SacI and PstI digestion were subcloned into the SacI and BamHI sites of pET53. Next, to construct the expression vector for SL2 and COS-1 cells, pET[beta]1I243/NLS/330 was digested with BamHI and BglII and the fragment was subcloned into the BamHI site of A5C[Delta]P and the BglII site of the pCAGGS vector.
All synthesized oligonucleotides to construct the plasmids are shown in Table 2 .
Table 1
All DNAs of hGABP[beta] mutants were sequenced using a 373A-18 sequencer with a fluorescence detection system (Applied Biosystems).
Table 2 DNA fragment and PCR primer of sequenceTransfected COS-1 cells were placed onto a micro cover glass. The cells were washed twice with sterile PBS(-) and fixed in 3.7% formalin/PBS(-) for 10 min. The cells were washed with PBS(-) and 0.1% Triton X-100/PBS(-) was added for 10 min to permeabilize the cells. After blocking with 1% skimmed milk/PBS(-) for 10 min, the first antibody (anti-hGABP monoclonal antibody) was added for 1 h. The monoclonal antibodies were 3A4G7G3H11 and 4F3HF12E12, directed against hGABP[beta], and 5B8A12D7C12, directed against hGABP[alpha]. The cells were washed with 0.1% NP-40/PBS(-) three times for 5 min with agitation, followed by treatment with 5 [mu]g/ml TRITC-conjugated anti-mouse secondary antibody (Chemicon) for 20 min in a dark box. They were subsequently washed with 0.1% NP-40/PBS(-) three times for 5 min with agitation. Cells were stained with 0.1 mg/ml diaminophenolindole (DAPI) for 5 min and washed twice with PBS(-) and mounted in 90% glycerol, 10 mg/ml p-phenylenediamine, 50 mM Na2CO3-NaHCO3, pH 8.0. Samples were examined and photographed using a Carl Zeiss microscope equipped for fluorescence photomicroscopy and a Fuji NEOPAN 400. All procedures were carried out at room temperature.
SL2 cells were maintained in tissue culture flasks containing Schneider's Drosophila Medium (Gibco BRL) supplemented with 10% fetal bovine serum (FBS), 50 mg/ml streptomycin sulfate and 50 U/ml penicillin G. At 2-3 h before transfection, the cells were replated onto 60 mm polystyrene dishes at a density of 3 * 106 cells/5 ml medium/dish. Transfections were carried out by the calcium phosphate method (25 ). The cells received variable amounts of hGABP expression vectors, the luciferase reporter vector, the [beta]-galactosidase vector and A5C[Delta]P DNA, so that the total concentration of DNA was 12 [mu]g/dish. After addition of DNA, cells were incubated at 27oC and left undisturbed until the time of harvest 40 h later.
COS-1 cells were maintained in Dulbecco's modified Eagle's medium (Gibco BRL) supplemented with 10% FBS. Transfections were carried out by electroporation. A total of 6 [mu]g DNA was transfected into a volume of 250 ml containing ~1 * 106 cells using a Bio-Rad Gene Pulser set at 220 V and 960 [mu]F. The width of the cuvette was 0.4 cm. Transfected cells were plated onto a micro cover glass in a 60 mm tissue culture dish containing 5 ml medium and the dishes were incubated at 37oC with 5% CO2 until harvesting 40 h later.
Cell extracts were prepared as follows. Transfected SL2 cells grown in 60 mm dishes were washed three times with PBS(-) and then lysed by the addition of 400 ml cell lysis buffer (Toyo-ink PGK-L-500). Cell lysates were collected in 1.5 ml tubes and centrifuged at 12 000 g for 5 min at 4oC. The supernatant of the cell lysate was diluted 1:10 in cell lysis buffer containing 1 mg/ml bovine serum albumin fraction V. For the luciferase assay, 20 ml diluted cell lysate were mixed with 100 ml luminescence reagent (Toyo-ink) and luciferase activity was measured in a Lumat LB 9501 luminometer (Berthod). For the [beta]-galactosidase assay, 20 ml diluted cell lysate were mixed with 500 ml buffer Z (10 mM KCl, 1 mM MgSO4, 50 mM 2-mercaptoethanol, 0.1 M NaHPO4-Na2HPO4, pH 7.5) and 5 ml chlorophenyl red [beta]-D- galactopyranoside (Boehringer Mannheim) and then incubated at 30oC for ~20 min. The absorbency of each sample was measured at 574 nm. The luciferase activity in each assay was standardized against the corresponding [beta]-galactosidase activity.
The cell lysates were loaded onto 10% SDS-PAGE gels. After electrophoresis proteins were transferred to immobilon transfer membranes (Millipore) and the hGABP protein was detected using a blotting detection kit for mouse antibodies (Amersham) and the monoclonal antibody anti-hGABP antibody.
Proteins were expressed in E.coli BL21(DE3). Purification and renaturation of hGABP subunits and hGABP[beta]1 mutant polypeptides were performed as previously described (6 ). The concentration of each renatured protein was determined by silver staining of SDS-PAGE gels.
Gel shift assays were performed as previously described (2 ) except that gel electrophoresis was carried out at 4oC. The DNA probe for this assay contained a single hGABP recognition site and was prepared as previously described (7 ). About 1 ng DNA probe was used for the binding reactions.
We isolated two cDNA clones coding for the two newly described subunits of hGABP (19 ). They contain an additional 12 amino acid insertion into hGABP[beta]1 and hGABP[gamma]1. These subunits have therefore been termed hGABP[beta]2 and hGABP[gamma]2, as indicated in Figure 1 A.
Before examining the transcription activation domain of the hGABP complex, it was essential first of all to identify the region of the hGABP complex required for transport into the nucleus. We constructed expression vectors of hGABP[alpha], [beta]1, [gamma]1 and the various hGABP[beta]1 mutant constructs (Fig. 1 ) and transfected them into COS-1 cells. Their location in the cell was detected by an immunofluorescent assay using specific monoclonal antibodies as described in Materials and Methods. hGABP[alpha] was found to be localized in the entire cell when expressed alone (Fig. 2 a and s). However, hGABP[alpha] became localized in the nucleus upon co-transfection with the expression vectors for hGABP[beta]1 or hGABP[gamma]1 (Fig. 2 b and c). On the other hand, full-length hGABP[beta]1 was observed only in the nucleus, whether expressed alone or co-expressed with hGABP[alpha] (Fig. 2 j and k).
To examine whether the hGABP complex can stimulate transcription in vivo, we carried out transient transfection assays with D.melanogaster SL2 cells (22 ) as described in Material and Methods. We chose these cells because they are highly responsive to exogenous transcription factors, in contrast to mammalian cells (26 ,27 ). The luciferase reporter plasmid used in these experiments contains an intact (-324 to +39) adenovirus E4 promoter (Fig. 3 A). As shown in Figure 3 B, no activation was observed when each subunit was present alone (lanes 2-4), consistent with our previous results in vitro (2 ). However, strong activation of transcription from the E4 promoter was observed when hGABP[alpha] was co-expressed with increasing amounts of hGABP[beta]1 or [beta]2 (Fig. 3 B, lanes 7-12). In this assay, hGABP[beta]1 was more efficient in activation of transcription compared with hGABP[beta]2. This was not due to differences in the amount of expressed protein, as demonstrated by a quantitative Western blot assay (data not shown). On the other hand, transcription activation was not observed when hGABP[alpha] was co-expressed with hGABP[gamma]1 or [gamma]2. These results indicate that the hGABP complexes composed of hGABP[alpha] and hGABP[beta]1 or hGABP[beta]2 are transcriptional activators and that transcription activation requires the non-DNA binding hGABP[beta]1 or [beta]2 subunits, but not the hGABP[gamma]1 or [gamma]2 subunits.
To analyze further the region of hGABP[beta]1 necessary for transcription activation from the E4 promoter in vivo, a series of expression vectors containing deletions in hGABP[beta]1 (Fig. 1 B) were examined by transient transfection assay. The level of expression was determined by Western blot assay and transcriptional activity was measured by luciferase assay. As shown in Figure 2 C, none of the deletion mutants could stimulate transcription in the presence of hGABP[alpha], even though these truncated molecules were expressed at levels comparable with that of full-length hGABP[beta]1 (data not shown). Previously, a gel shift assay showed that the N-terminal deletion mutant [beta]1N133 could not bind hGABP[alpha] and that the C-terminal deletion mutants [beta]1C248 and [beta]1C332 could bind hGABP[alpha] but could not form heterotetrameric complexes functional in transcription activation in vitro. Note that the internal deletion mutant [beta]1I153/267 could not stimulate transcription in this assay, although it showed a slight capacity to stimulate transcription activation in vitro (7 ).
Co-expression of hGABP[alpha] with [beta]1I243/NLS/330, which has a NLS derived from the SV40 large T antigen inserted into the deletion region of [beta]1I243/330 (Fig. 1 B), resulted in a ~10-fold increase in luciferase activity compared with the expression vector containing no hGABP[beta] or [beta]1I243/330 construct cDNA (Fig. 4 , lanes 1 and 4). This NLS-fused mutant was observed only in the nucleus by immunofluorescence assay (Fig. 2 r). These results suggest that the C-terminal sequence (amino acids 330-353) of hGABP[beta]1 is required for hGABP-induced transcription activation in vivo and that transcription activation is relatively independent of amino acids 243-330, which are necessary for nuclear localization of the hGABP complex.
The results above indicate that the C-terminal region of hGABP[beta]1 (amino acids 330-353) is important for transcription activation in vivo, which is consistent with previous in vitro data (7 ). This leads to the conclusion that this region is important for both heterotetrameric complex formation and transcription activation by hGABP. To further dissect the functional domains in this region, we carried out site-directed mutagenesis. In vitro analysis revealed that this region has homodimerization activity (7 ,13 ). Besides, it is free of [alpha]-helix destabilizing residues and has a hydrophobic phase composed of leucine and alanine residues when displayed on an idealized [alpha]-helix projection. Therefore, it may function as a leucine-zipper structure for homodimerization (28 ,29 ). We systematically constructed a series of vectors expressing hGABP[beta]1 mutants having two sequential amino acids substituted by the [alpha]-helix destabilizing residues glycine and threonine within this region (Fig. 4 ). These substitution mutants were expressed in E.coli and purified as described in Materials and Methods. Their capacity to form heterotetramers with hGABP[alpha] was examined by gel shift assay, as shown in Figure 5 . Although [beta]1QK339GT (lanes 1-3) and [beta]1EA371GT (lanes 49-51) could form a heterotetramer with hGABP[alpha] as efficiently as wild-type hGABP[beta]1 (lanes 52 and 53), other mutants and especially those that contained a mutations within the hydrophobic phase of the hypothetical [alpha]-helix structure, such as [beta]1AN345GT (lanes 10-12), [beta]1KY351GT (lanes 19-21) and [beta]1KE359GT (lanes 31-33), could not form a heterotetrameric complex with hGABP[alpha] efficiently. Three leucine-defective mutants, [beta]1GL341GT (lanes 4-6), [beta]1QL355GT (lanes 26-27) and [beta]1KL369GT (lanes 46-48), all failed to form the heterotetrameric complex, but two alanine-defective mutants, [beta]1AQ349GT (lanes 16-18) and [beta]1AE363GT (lanes 37-39), could form the complex, albeit less efficiently than wild-type hGABP[beta]1. These results suggest that amino acids 341-369 are important for formation of the hGABP heterotetrameric complex. Also, it seems possible that this region forms an [alpha]-helix structure and that the three leucine residues are part of a leucine zipper structure necessary for homodimerization.
The studies reported here further extend our understanding of the relationship between structure and function in hGABP[beta]1 (summarized in Fig. 7 ). The identification of two functional regions in hGABP[beta]1 was previously indicated by in vitro analysis (7 ). The N-terminal region containing four tandem repeats with homology to Notch/ankyrin was shown to be required for binding to hGABP[alpha]. The C-terminal region containing the leucine zipper-like motif was found to be responsible for transcription activation and homodimerization. Here, we identify a new functional region necessary for nuclear localization of hGABP[beta]1 and show that this region is also necessary for efficient nuclear localization of hGABP[alpha]. Also, we have extended our study of the C-terminal region of hGABP[beta]1 by identifying key amino acids necessary for both stimulation of transcription and homodimerization. This further underlines the coincidental nature of these two activities within the C-terminal region of hGABP[beta]1.
The contribution of M.Ikeda and K.Tamai to the production of monoclonal antibodies against hGABP[alpha] and [beta] is gratefully acknowledged. We thank Dr J.-i.Inoue for the gift of SL2 cells and SL2 expression vectors. We also thank Dr S.-i.Hisanaga for technical advice on DAPI staining. We are grateful to Drs T.Wada and T.Imai for helpful discussions and M.Usui for reviewing the manuscript. This work was supported by Research Fellowships from the Japan Society for the Promotion of Science for Young Scientists.
*To whom correspondence should be addressed. Tel: +81 45 924 5797; Fax: +81 45 923 0380; Email: hhanda@bio.titech.ac.jp
REFERENCES
Return