The transcription factor GATA-1 regulates the promoter activity of the platelet glycoprotein IIb gene.

Glycoprotein IIb (GPIIb) is an early and specific marker of the megakaryocytic lineage. We have previously shown that a fragment extending 643 base pairs upstream the transcription start site of the human GPIIb promoter was able to control the tissue-specific expression of the CAT gene in transfection experiments. Four potential GATA-binding sites, located at positions -463, -376, -243, and -54 are present within this fragment. Gel shift analysis revealed that nuclear extracts from the erythroleukemic cell line HEL contain a DNA-binding protein that recognizes these GATA sites. Using an antiserum raised to an hydrophilic region of the transcription factor GATA-1, the HEL GATA-binding protein was found to be GATA-1. Point mutations of the different GATA sites indicated that they did not equally contribute to GPIIb promoter activity. The -463 GATA motif located in an enhancer region is essential for full transcription activity and was found to be dominant upon the other GATA motifs. When this site is mutated, the -54 GATA site appears to be essential for the remaining CAT activity. These results indicate that the transcription factor GATA-1 plays an important role in the regulation of the transcription of the megakaryocyte specific GPIIb gene.

family of DNA-binding proteins including GATA-2 and GATA-3, which are related by two highly conserved zinc finger DNA recognition domains (4-6). These GATA factors have a different cellular distribution although coexpression in several cell types has also been observed. GATA-1 has been found in erythroid (7), mast (8), and megakaryocytic (8,9) cells, suggesting that this factor may be a regulator common to these lineages. To verify this hypothesis, we have used a megakaryocyte-specific marker gene to analyze the role of GATA-1 in the transcriptional activity of a megakaryocytic promoter. We have previously cloned the human GPIIb gene (lo), an early and specific marker of megakaryocytopoiesis (11). We have also shown that a -643 to +33 fragment of the GPIIb promoter was able to drive the cell-specific expression of a reporter gene in megakaryocytic cell lines (12) and contained an enhancer active in both erythroid and megakaryocytic cell lines (13). In the present study, we have dissected the transcriptional activity of four GATA sites present in the promoter region of this gene. Using an anti-GATA-1 antiserum, we demonstrate that all the GATA sites bind GATA-1 in vitro but do not equally contribute to the promoter activity.

MATERIALS AND METHODS
DNA Probes-The synthetic oligonucleotides used in mobility shift assays were prepared with an Applied Biosystem DNA synthesizer. The first strand was 5'-end-labeled and annealed with an excess of the non-labeled second strand at a one to four ratio to optimize annealing. The following oligonucleotides were prepared 5'-CTGATGGGCCTTATCTCTTTACCCACCT-3' from the erythroid promoter of human porphobilinogen deaminase gene (PBGD) was used as a GATA-1 standard binding site (14) and 5'-GGCCTGGCTTATCTCCGGCTGC-3' from the promoter of preproendothelin-1 gene (PPET-1) used as a GATA-2 standard binding site (15); the HNF-1-binding site from the fibrinogen 0 promoter was 5'-GATCTCAAACTGTCAAATATTAACTAAAGGGAG-3' (16); the CCAAT-binding site from GPIIb promoter was 5"TAGGCA-GAACCAATAGGACATGGTA-3' (12); the wild type and the mutated GATA-binding sites of human GPIIb promoter were, respectively, 5'-CCAGTTTGATAAGAAAAGAC-3' and 5"CCAGTTT GACGAGAAAAGAC-3' for positions -61 to -42 (-54 GATA probe), Cells and Nuclear Extracts-The nuclear extracts were prepared from HEL (erythroleukemia) and Hela (epithelial carcinoma) cells as described elsewhere ( E ) , according to the method of Dignam et al. .

1606
The human endothelial cells (HUVEC)' were collected from umbilical veins as described by Jaffe (18) and were grown on purified fibronectin-coated plates in medium 199 supplemented with 2% endothelial cells growth substance (ULTROSER IBF) and 20% fetal calf serum. A rapid nuclear protein preparation was performed by the method of Schreiber et al.(19).
Gel Retardation Assays-The gel retardation assays were performed by a combination of the procedures of Halligan and Desiderio (20) and of Singh et al. (21).
For the binding reaction, 0.2-0.5 ng of radiolabeled DNA fragment (20,000 counts/min) were mixed with 5 pg of nuclear extracts in a final volume of 30 p1 containing: 10 mM Tris-HC1, pH 7.5, 25 mM KC1, 1 mM EDTA, 1 mM dithiothreitol, 15% glycerol, and 2 pg of poly(d1)-(dC) used as non-specific competitor. Specific competitors and antisera were added to each reaction as described in individual experiments. Samples were incubated for 20 min at room temperature and analyzed on 6% polyacrylamide gels in 0.5 X TBE buffer (TBE, 0.089 M Tris-HCI, 0.089 M boric acid, 0.002 M EDTA).
Plasmid Construction and DNA Transfection-The basic CAT plasmid used was pBLCAT3 (22). The mutated GPIIb fragments were obtained by using the "oligonucleotide-directed in uitro mutagenesis system, version 2" (Amersham Corp.). A fragment of GPIIb promoter extending from -777 to +33 was subcloned at restricted sites SstI and BamHI of M13mp18 vector. Oligonucleotides used for mutagenesis are described above. Wild type and mutated promoter fragments were inserted between XbaI and BamHI sites of pBLCAT3 vector. The IIb414 construct containing a promoter fragment extending from -414 to +33 was obtained as already described (13). The pRSV-luciferase plasmid was used as internal standard expressing firefly luciferase under the control of the Rous sarcoma virus promoter (23). HEL cells were transfected by electroporation as already described (13).
Luciferase and CATAssays-Cells were harvested 48 h after transfection. CAT assays were performed after normalization in function of luciferase activity as previously described (13).
Antiserum Preparation-Antiserum against GATA-1 (aG1) was prepared by immunizing rabbits (25) with the synthetic peptide TREDSPPQAVEDLDGKGSTS corresponding to a hydrophilic region of GATA-1 and chemically cross-linked to bovine serum albumin (26). The rabbits were given subcutaneous booster injections at 2week intervals and weekly test bleeds were begun 3 weeks after the initial injection. The amount of peptide used for immunization was 50 pg by injection. Production of polyclonal antiserum was controlled by ELISA.
ELZSA-Polyvinyl microtiter wells were coated with 100 p1 of a lo-' M GATA-1 peptide solution or with the same amount of an irrelevant peptide solution. After an overnight incubation at 37 "C, the plates were postcoated with 0.1 M sodium carbonate, pH 9.6, containing 1% ovalbumin, washed with phosphate-buffered saline, 0.05% Tween 20, and incubated for 1.5 h with the aG1 antiserum at dilutions varying from 1/200 to 1/6400. Extensive washing was performed with phosphate-buffered saline, 0.05% Tween 20. 100 pl of a 1/500 dilution of goat anti-rabbit IgG conjugated with alkaline phosphatase (Dakopatts, Denmark) was then added. Incubation was performed for 1 h, and after washing the binding of aG1 antiserum was detected using p-nitrophenylphosphate as a substrate for alkaline phosphatase and measuring absorbance a t 410 nm after 20 min of enzyme reaction.

R E S U L T S
Production and Specificity of Anti-GATA-1 Antibody-To identify factors that interact with the GATA protein putative binding sites of the GPIIb promoter, a specific anti-GATA-1 polyclonal antibody was produced (aG1). The antibody was raised to a potential antigenic site present on GATA-1 having no homology with any sequence of other GATA proteins. The following experiments were designed to establish the specificity of this antiserum.
In ELISA, aG1 reacted with the GATA-1 peptide but did not react with an irrelevant peptide sequence (Fig. 1). The The abbreviations used are: HUVEC, human endothelial cells; bp, base pair(s); ELISA, enzyme-linked immunosorbant assay; CAT, chloramphenicol acetyltransferase. Characterization of the aG1 antiserum raised to a GATA-1 immunogenic peptide. Three different end-labeled DNA fragments were incubated with nuclear extracts from HEL or HUVE cells. HEL cells nuclear extracts were incubated with a PBGD promoter fragment (-83 to -56) containing a GATA-1-binding site (lanes 1-5) or with a GPIIb promoter fragment (-243 to -219) containing a CCAAT motif (lanes 11-13). HUVE cell nuclear extracts were incubated with a PPET-1 promoter fragment (-144 to -123) containing a GATA-2-binding site (lanes 6 -10). The binding reactions were performed without competitor and without antiserum, lanes I, 6, and ZZ; with a 100-fold excess of the unlabeled competitors: PPET-1 promoter fragment homologous to the probe, lane 7; heterologous promoter fragment carrying the HNF-1-binding site, lane 8. Prior to polyacrylamide gel electrophoresis, some of the mixtures were incubated for 15 min on ice with 1 pl of undiluted preimmune antiserum, lanes 2, 9, and 12; with 1 pl of undiluted aC1 antiserum, lanes 3 4 8 , and I I. The protein-DNA complex specific of the PBGD GATA probe is indicated by the closed arrow I , and the protein-DNA complex specific of the PPET-1 GATA probe is indicated by the closed arrow 2. The change in mobility of the complex detected with the PBGD probe is indicated with an open arrow. This supershifted complex was inhibited when GATA-1 immunogenic peptide was added (1 pl of a M solution), lane 4; i t was still detected in presence of the same amount of an irrelevant peptide, lane 5.
antiserum was then tested in band shift assays using a PBGD GATA-1 probe. After incubation with HEL nuclear extracts, a major DNA-protein complex was detected (Fig. 2, lane I ).
The formation of this complex was not affected by the addition of preimmune antiserum to the mixture (Fig. 2, lane 2). In the presence of aG1, the intensity of the retarded band was considerably reduced, and a supershifted complex was formed -4 6 3 -3 1 6 -2 4 3 ( Fig. 2, lane 3 ) . The formation of this supershifted band was inhibited by the addition of GATA-1 immunogenic peptide (Fig. 2, lane 4 ) , but not by the addition of an irrelevant peptide (Fig. 2, lane 5 ) .

FIG
The same experiment was performed with a GPIIb CCAAT probe. This probe was able to form two complexes with HEL nuclear proteins (Fig. 2, lane 11 ). These interactions were neither affected by the addition of the preimmune antiserum (Fig. 2,lane 12) nor by the addition of aG1 (Fig. 2, lane 13). These results showed that the aG1 antiserum recognizes specifically a GATA binding factor present in HEL cells.
To further characterize the specificity of aG1, nuclear extracts from HUVEC which produce GATA-2 but do not express GATA-1 (27) were used in gel shift assays. In these experiments, a GATA-2-binding site of the PPET-1 promoter was used as a specific probe (28). The PPET-1 GATA-2 probe formed a complex with the HUVEC extracts (Fig. 2, lune 6). This complex had a molecular weight slightly higher than the complex formed with the PBGD GATA-1 probe incubated with HEL nuclear extracts. The difference in gel mobility was consistent with the difference in size between GATA-1 and GATA-2 containing 409 and 480 amino acids, respectively (4,5). The DNA-protein complex observed with HUVEC nuclear extracts was specific of the PPET-1 probe: it was competed by an excess of the unlabeled homologous oligonucleotide (Fig. 2, lane 7), and it was not affected by an heterologous competitor containing the HNF-1-binding site (Fig. 2,lane 8). No change in the mobility of the gel shift complex was observed, either with the preimmune antiserum (Fig. 2,lane 9) or with aG1 (Fig. 2, lane 10).
From these results we concluded that aG1 recognizes the GATA-1 molecule present in HEL cells.
Identification of GATA-I-binding Sites on the GPIIb Promoter Sequence-Analysis of the nucleotide sequence of a 777-bp DNA fragment extending upstream the transcription start site of GPIIb gene indicates the presence of four putative binding sites for a GATA protein (Fig. 3). Two of these sites centered a t positions -54 and -243 exhibited the strict consensus sequence WGATAR and are in the direct orientation. A third site, a t position -376 contains the motif GATTAG which diverges slightly from WGATAR, but has been described as a possible GATA-binding site (2). The fourth site contains the CGATAA motif a t position -463 in the reverse orientation, with a C instead of a W. Using DNase-I footprint analysis, we have shown in previous studies that the -54 and -463 sites interact with a factor present in nuclear protein extracts of erythroid and megakaryocytic cell lines (12). Using similar DNase-I protection assays, we have shown that the domains corresponding to sites -376 and -243 are also protected by nuclear extracts from HEL cells (data not shown). Thus, all the sites are potential binding sequences for nuclear proteins.

21609
T o explore more precisely the DNA-protein interactions, a series of mobility shift assays using oligonucleotides corresponding to each of the GPIIb GATA sequence was performed. Results of these experiments are summarized in Fig.  4. In each case, a similar DNA-protein complex was formed when nuclear extracts of HEL cells were used (Fig. 4, panels A-D, lanes 1 ), but this complex was not detected with extracts from HeLa cells. The formation of this complex was inhibited in the presence of an excess of the unlabeled homologous oligonucleotide (lanes 2), and in the presence of the unlabeled PBGD GATA-1 oligonucleotide as well (lanes 3 ) . In contrast, the formation of this complex was not affected by unlabeled mutated GPIIb GATA oligonucleotides (lanes 4) or by the unlabeled heterologous oligonucleotide (lanes 5 ) , used as competitors. Several additional bands were detected with the -54 and -243 GATA probes (Fig. 4, panels A and B ) . With the -54 probe, the major additional band was also detected with HeLa extracts and could not correspond to a HEL-specific GATA-binding protein. With the -243 probe, all the additional bands were also present in HeLa nuclear extracts, and the formation of these complexes were not inhibited by competition with the PBGD GATA oligonucleotide, demonstrating that they could not correspond to a GATA protein.
From these results we conclude that the different putative GATA sites of the GPIIb promoter may interact with a DNA binding factor, present in HEL but not in HeLa cells, and that these interactions are specifically inhibited by the GATA motif. We suggest that this factor could be GATA-1.
T o verify this hypothesis, we used aG1 in mobility shift experiments (Fig. 5). Nuclear extracts from HEL cells were incubated with the different GPIIb GATA probes (Fig. 5,  panels A-D, lanes 1 ). An excess of the HNF-1 oligonucleotide used as heterologous competitor was added to reduce the formation of nonspecific complexes observed with the -54 and the -243 GATA probes. Formation of the specific DNAprotein complexes was not affected by the preimmune antiserum (lanes 2). In contrast when aG1 was used, formation of this complex was either inhibited or markedly reduced with the appearance of a weak supershifted band (lanes 3).
Thus, these results strongly suggest that GATA-1 is the transcription factor that binds to the GPIIb promoter GATA sites.
Functional Analysis of the Different GATA Sites on the GPZZb Promoter Activity-We have checked by mobility shift assays that the different probes corresponding to the GPIIb GATA sites failed to interact with nuclear proteins when mutated at the GA positions (not shown). Thus, each site was mutated within the promoter context, and the functional consequences of these mutations were examined in CAT assays after transfection in HEL cells. In these experiments, the wild type IIb777 construct corresponded to a transcriptional activity of 100% (Fig. 6).
Mutation at the -463 GATA site in the IIb777 construct reduced the activity down to 39 k 9%, indicating that this site is critical for full promoter activity.
Mutation of the -376 GATA site resulted in a weak reduction of the relative CAT activity, from 100% to 85 f 11%, suggesting that the binding of GATA-1 to this site has a minor contribution to the promoter activity in uitro.
Mutation of the -243 site resulted in an increased activity from 100% to 143 f 46%, suggesting a negative effect of this site on the promoter activity. However, the standard deviation observed on an average of four independent CAT assays was significantly high as compared to the dispersity of the values obtained with the other constructs. Mutation of GATA-54 within the IIb777 construct resulted in a CAT activity of 108 f 16%, indicating a null contribution of this GATA site to the promoter activity. A plasmid construct, IIb414, which did not contain the enhancer region was tested for its transcriptional activity. This 414-bp region of the promoter was able to direct the expression of the CAT enzyme a t a lower but significant level (32% compared to 100% with the enhancer containing fragment). When this IIb414 construct was mutated a t position -54, a drop of the CAT activity from 32 to 4% was observed. These results suggested that the -54 GATA site is active only when the -463 GATA site was not functional. In support to this conclusion was the observation that mutations at -54 and -463 reduced the CAT activity down to 15 & 3%, compared to 39% obtained with the single mutation a t -463. Effect of point mutations in GATA motifs on GPIIb promoter transcriptional activity. The plasmid constructs contained either 777 bp (IIb777 construct) or 414 bp (IIb414 construct) upstream from the transcription start site (+1) and 33 bp downstream. The GATA and the point mutated sites are represented by hatched and crossed boxes, respectively. Each plasmid was transfected into HEL cells and the CAT activities were measured 48 h after transfection. In each assay, pRSVL plasmid was cotransfected and the CAT assays were normalized according to the luciferase activity. The CAT values obtained with the different plasmids were expressed relatively to the non mutated IIb777 construct which was taken as the 100% value. The promoterless plasmid pBLCAT3 was used to measure the background level. The values of relative CAT activity are averages _t standard deviations of four individual transfection experiments.
Finally, mutation of the four GATA sites induced an important decrease of the CAT activity 100% to 26 f 4%. This drop of activity was less marked than that observed with the construct where only GATA -463 and -54 were mutated (26% compared to 15%). This can be explained by the negative effect of the -243 GATA site, and the very weak contribution of the GATA -376 site.

DISCUSSION
GATA-1 has been detected in several early myeloid cell lines but this expression is turned off in the presence of myeloid growth factors (29). This suggests that GATA-1 is probably expressed at an early stage of hematopoiesis in different pluripotent precursors and that this expression is only maintained in selected lineages. The role of GATA-1 in the development of erythroid cells has been extensively investigated. Implication of this factor in the regulation of erythroid genes has been demonstrated at the level of promoters (7), enhancers (30), or erythroid dominant region, DCR (31)(32)(33). Knock out experiments have established the crucial role of GATA-1 in the erythroid differentiation process (34).
Although the role of GATA-1 in the regulation of megakaryocytic genes is less understood, recent observations indicate that it may also be implicated in the control of the differentiation of this lineage. GATA-1 was found in megakaryocytes, and it was shown that overexpression of this factor in the myeloid cell line 416B induces megakaryocytic markers (35). This suggests that GATA-1 can act as an important regulator of megakaryocytopoiesis, and its implication in the transcriptional regulation of megakaryocytic genes has been proposed (9,13,36,37). To verify this hypothesis we have studied the interaction of GATA-1 with the promoter region of the megakaryocyte-specific GPIIb gene.
To clearly identify GATA-1 as the implicated factor, a polyclonal anti-GATA-1 antibody was produced, using a synthetic peptide containing 20 amino acid residues of a GATA-1-specific hydrophilic domain, with no homology with other members of the GATA family. The specificity of this antibody was analyzed using ELISA and mobility shift experiments and was based on the following observations: 1) on ELISA, the antibody reacted with the GATA-1 peptide but did not interact with an irrelevant peptide; 2) the antibody was able to form a super-shifted DNA-protein complex in mobility shift assays using a probe containing the GATA-1-binding site of the erythroid-specific PBGD promoter and HEL nuclear proteins; 3) the antibody failed to react with a DNAprotein complex formed between a specific GATA-2 probe of the PPET-1 gene and nuclear proteins of HUVEC which express GATA-2 but do not produce GATA-1. Since GATA-1 and GATA-2 but not GATA-3 are present in HEL cells (4,5,38), we concluded from these results that the antibody was specific to GATA-1. Although this antibody was able to form a supershifted complex with a WGATAR probe, the intensity of the shifted band was considerably reduced, indicating that the antibody was also able to inhibit the formation of the complex. The promoter domain of the GPIIb gene contains four potential GATA-binding sites centered at -54, -243, -376, and -463. Mobility shift assays have shown that each GATA site interacted with a GATA protein present in HEL nuclear extracts. This protein was identified as GATA-1 using the specific anti-GATA-1 antibody in supershift experiments.
The relative contribution of each GATA-1-binding site to the transcriptional activity of the promoter was established by the mutation of each site either alone or in combination, and the effect of these mutations was examined in CAT assays.
We have already shown that the -463 GATA site is contained within a specific erythromegakaryocytic enhancer (13). Consistent with this observation we found that this GATA site is critical. Its mutation induces a drop of 60% in the promoter activity. The present study shows that one of the critical factors that interacts with this enhancer is GATA-1 and establishes the role of this transcription factor in the activity of a megakaryocyte-specific promoter.
Mutation of the -376 site produced a drop of about 15% in the CAT activity, suggesting that the binding of GATA-1 to this site does not directly contribute to the promoter function. The sequence of this site contains a GATTAG motif which diverges from the consensus WGATAR motif and may not be functionally active within the GPIIb promoter context.
In previous CAT assays using a 77-bp sequence upstream from the transcription start site of the GPIIb promoter, we have shown that the GATA site at position -54 was functional (9). Mutation of this site, however, within the context of the whole promoter, did not affect the transcriptional activity. This suggests that the -54 GATA site has a minor contribution to the overall activity. Nevertheless, the mutation of this site in the promoter deleted from its enhancer region reduced considerably the residual activity. These results indicate that the site at -54 may be functional and responsible for the residual activity of about 30%, providing the enhancer is deleted. One possible interpretation of this observation could be that the site at position -463 has a dominant effect on the -54 site. Observation that a double mutation at position -463 and -54 resulted in a 85% inhibition of the promoter activity supports this conclusion.
Recent studies on the p-globin and the PF4 promoters suggest that GATA sites located at the expected position for a TATA box could be implicated in the control of the cell specificity (36,39). Timoty and Emerson (39) propose that interactions between a GATA site located at -30 and a distal enhancer containing another GATA site controls the erythroid specificity of the p globin promoter. Such a situation is encountered on GPIIb promoter, with the proximal -54 GATA site and the erythromegakaryocytic enhancer that have been previously identified. This enhancer contains the -463 GATA site that is essential for the enhancer function. We have checked if the megakaryocytic specificity of GPIIb promoter depends on a cooperativity between the -54 and the -463 GATA sites by changing the -54 GATA sequence into a TATA sequence. When the construction bearing this mutation was transfected in HeLa or K562 cells, the expression remained close to the background level and was not significantly different than that observed with the wild type promoter.' This suggests that GATA-1 is implicated in the regulation of transcription level of GPIIb gene but not in the control of the tissue specificity.
Interestingly, both -463 and -54 GATA sites are flanked by a region corresponding to the consensus binding sites of transcription factors of the Ets family (for review see Ref. 40). These Ets-binding sites react with nuclear factors as shown in DNase-I footprinting experiments (12,37). We have shown that the -563 Ets site was important for the enhancer function and for the promoter activity (13). The same observations were made by Lemarchandel et a1 (37) for the proximal GATA/Ets association. A promoter fragment containing only the GATA-and Ets-binding sites was able to confer activity to an heterologous promoter in megakaryocytic cells. Several Ets factors like Ets-1, Fli-1, and Pu-l/Spi-l are responsible for the erythroleukemic transformation (41)(42)(43)(44), indicating that members of the Ets family could be implicated in the hematopoietic differentiation and development. The association between GATA and Ets cis-acting elements has been described for the GATA-1 promoter (45). These two GATA/ Ets regions may be conserved through evolution since they * F. Martin, M.-H. Prandini, D. Thevenon, G. Marguerie, G. Uzan, unpublished results. are present on the murine GPIIb pr~moter,~ and the proximal GATA/Ets sequence is present on the rat GPIIb promoter (46). Transfection experiments suggested a negative effect of the GATA-1-binding site located a t -243 on the promoter activity. Since a high standard deviation of the CAT values has been observed, another interpretation is that this site is not functional. Further investigation is still needed to understand the precise contribution of this site to the transcriptional activity. Furthermore, the promoter mutated on the four GATA sites was more active than the promoter mutated on GATA -463 and GATA -54. Taken together these observations confirm the hypothesis of a negative effect of the -243 GATA site on the promoter activity.
In summary, the results presented in this paper demonstrate that GATA-1 is a major regulator of GPIIb gene expression in uitro. Not all GATA sites are functionally equivalent, and they may differentially modulate the promoter activity during megakaryocytopoiesis.