Characterization of a Specific Erythromegakaryocytic Enhancer within the Glycoprotein IIb Promoter*

The gene coding for glycoprotein IIb (GPIIb), the a subunit of platelet integrin GPIIbflIIa is an early and specific marker of the megakaryocytic lineage. Thus, studies on the regulation of this gene may provide helpful information on the mechanisms controlling cell specificity and differentiation in this lineage. The promoter region of this gene was isolated and analyzed to understand its tissue-specific transcriptional activity. A region GATAl, results the promoter of a megakaryocytic gene DNA-protein autoradiog-raphy, protein-bound free oligonucleotide out electroe-lution, cleaved piperidine at for electrophoresed on acrylamide, autoradiographed.

The commitment of a totipotent hematopoietic stem cell and its terminal development into the different hematopoietic lineages are certainly controlled by switching on and off a number of genes and by controlling the extent of transcription of these genes. Modulation of gene expression in response to intra-or extracellular cues can be influenced by different combinations of nuclear transcription factors. Despite the bulk of information concerning the effect of growth factors and cytokines on hematopoiesis (l), the role of transcription factors in the developmental program of hematopoietic cells is less documented than in the hepatocyte or the muscle cell system. One of the few transcription factors that have been identified and most extensively characterized is GATAl (2-6). This factor is implicated in the transcription of erythroid genes and was shown to play a key role in the development of the erythroid lineage (7).
Identification of lineage restricted factors can be achieved * The costs of publication of this article were defrayed in part by the payment of page charges, This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
through their binding capacity to cis-acting elements of promoter regions of cell-specific genes. To characterize factors that may be implicated in the establishment of the megakaryocyte phenotype, we have used the gene encoding the platelet glycoprotein IIb (GPIIb).' This glycoprotein is the a subunit of the platelet adhesion receptor GPIIb-IIIa and belongs to the family of RGD-sensitive integrins (8-12). While GPIIIa, or integrin (33 subunit, is expressed in a variety of cells, including fibroblasts (13), endothelial cells (14), macrophages (15,16), and megakaryocytes (17), expression of GPIIb is restricted to megakaryocyte and is detected at an early stage of the development of this lineage (18-21). Therefore, GPIIb is a good candidate as a marker for megakaryocytopoeisis. For this reason, the GPIIb gene was isolated (22) and its 5"flanking region was examined (23). We found that a genomic fragment extending la80 bp upstream from the transcription initiation start site is able to drive the expression of a reporter gene in a tissue-specific manner. Different binding sites for nuclear proteins were identified within this domain, including consensus sequences for the binding of the erythroid factor GATAl and sites for proteins present in megakaryocytic cells. In the present study, the promoter region of the GPIIb gene was further investigated using 5'deletion analysis. We found that it contains a domain that exhibits all the characteristics of a tissue-specific enhancer active in both megakaryocytic and erythroid cells.

MATERIALS AND METHODS
Plasmid Construction-The basic CAT plasmids used were pBLCAT3 (24), pRSVCAT (25), and pCAT promoter plasmid (Promega Biotec). The pBLCAT3 series of constructs was obtained after unidirectional deletions using the exonuclease III/mungbean system (Stratagene), as described elsewhere (23). The pRSV-luciferase plasmid was used as internal standard expressing firefly luciferase under the control of the Rous sarcoma virus promoter (26). All plasmids used for transfection were purified by double banding on CsCl gradients.
Cell Culture and DNA Transfection-HEL, K562, and HeLa cells were grown in RPMI 1640 medium (GIBCO) and 10% fetal calf serum (Boehringer Mannheim). HEL and K562 were transfected by electroporation, using a gene pulser (Bio-Rad) set at 960 microfarads, 400 V. Each assay was done with 10 pg of one of the CAT constructs, 10 pg of pRSV-luciferase, and 50 pg of salmon sperm DNA/1O7 cells in a total volume of 800 p1. HeLa cells were transfected at 960 microfarads and 200 V, in a total volume of 180 11.
Luciferase and CATAssays-Cells were harvested 48 h after transfection, and cell extracts were obtained by three cycles of freeze and thaw lysis. Luciferase activity of the extracts was measured using the luciferase assay system (Promega Biotec). CAT assays were performed essentially as described by Gorman et al. (25). The amount of protein in the extract tested was normalized in function of luciferase activity.
Gel Retardation Assays-The gel retardation assays were performed as already described (23) by a combination of the procedures of Halligan and Desiderio (27) and Singh et al. (28). The oligonucleotides D and E corresponding to the footprinted areas D and E have already been described (23). Their sequences were 5"CCTCAGGTTT-TATCGGGGGCAGCAGCT for oligo D and 5"TCCTAGAAGGAG-GAAGTGGGTAAATG for oligo E, respectively. These oligonucleotides were synthesized on an Applied Biosystem DNA synthesizer, 5' end-labeled and incubated with HEL, K562, and HeLa nuclear protein extracts, as previously described (23).
Methylation Interference Assay-The oligonucleotide probes, "Plabeled at one 5' end, were partially methylated with dimethyl sulfate (Maxam and Gilbert sequencing kit, Du Pont-New England Nuclear). The methylated probes were incubated with nuclear extracts of HEL cells, and the DNA-protein complexes were used in gel retardation assays, under conditions previously described (23). After autoradiography, the protein-bound and free oligonucleotide probes were cut out and recovered from the gel. DNA was then purified by electroelution, cleaved in piperidine at 90 "C for 30 min, electrophoresed on 15% acrylamide, 8 M urea gels, and autoradiographed.
Site-directed Mutagenesis-Site-directed mutagenesis was performed with the "oligonucleotide-directed in vitro mutagenesis system, version 2" (Amersham Corp.) using the instructions of the manufacturer. The sequence of the oligonucleotides used were CCTCAGGTTTTAMGGGGGCAGCAGCT for the mutant oligo D and TCCTAGAAGGAEAAGTGGGTAAATG for the mutant oligo E.

Analysis of 5"Deletion Mutants of the GPIIb Promoter-
To delineate the functional DNA sequences within the 5' region of the GPIIb gene, a genomic fragment containing the domain of the gene extending from +33 to -1050 and including the initiation start site was inserted into the XbaI site of the pBLCAT3 vector containing the CAT gene. We have already reported that the fusion mRNA produced by this construct is initiated at the correct transcription start site (23). Deletion mutants of this domain were generated by 5' digestion with exonuclease I11 and blunted with the mung bean nuclease. Each resulting fragment was controlled by sequencing and cotransfected in HEL cells with the pRSVL plasmid containing the luciferase gene used as internal standard for calibration and quantification of the CAT assays.
Extracts of HEL cells transfected with plasmid 1050 yielded 15% of the CAT activity obtained with the pRSVCAT positive control. Since deletions of the region between -1050 and -813 did not change this figure, the activity obtained with plasmid -813 was arbitrary given the value 100. In a control experiment, the +33/-1050 fragment was fused to the CAT gene in an opposite direction. As shown in Fig. 1, the CAT activity obtained with this plasmid was comparable with that obtained with the pBLCAT3 plasmid alone, indicating that the CATIIb promoter activity is orientation-dependent. Further digestion of the 813 fragment down to -554 had no effect on the transient expression of the enzyme. In contrast, deletions from -554 to -414 resulted in a 70% decrease of the CAT activity obtained in the extracts of transfected HEL cells (Fig. 1). Further deletions at -170 had no detectable additional effect. Finally, a greater decrease of the CAT activity was noted upon deletion to -29. A residual activity was still detected upon deletion to -13.
These data suggested that the region between -554 and -414 contains a positive acting DNA element necessary for an optimal promoter activity in HEL cells. Consequently, this region was examined for its potential enhancer properties.
Characterization of the Enhancer Domain-A 192-bp DNA fragment, corresponding to the sequence between -598 and -406, including the -5541-414 region, was produced by digestion of the 1050 plasmid with Hind111 and PuuII, purified, and fused to the CAT gene, driven by the SV40 promoter in the pCAT promoter plasmid. Three different constructs were made. In two of them, the 192-bp fragment was inserted in the direct or opposite direction upstream of the SV40 promoter, whereas the third construct contained the fragment in direct orientation downstream from the CAT gene (Fig. 2). These plasmids were introduced in HEL, K562, and HeLa cells to evaluate the enhancer activity and the tissue specificity of the IIb promoter DNA fragment. The CAT activity obtained with the enhancerless vector containing the CAT gene and the SV40 promoter alone was arbitrary given the value 1. Addition of the 192-bp fragment in either position produced a 4.4-5.9-fold increase of this basal activity when the plasmids were introduced into HEL or K562 cells (Fig.  2). In contrast, the CAT activity measured in extracts of transfected HeLa cells remained close to the basal level (0.7-1.4). These results indicated that the 192-bp DNA fragment, containing the GPIIb nucleotide sequence between -406 and -598, is able to increase the activity of an heterologous promoter in megakaryocytic or erythroid cells but not in HeLa cells, in a position-and orientation-independent manner.
Analysis of the Enhancer Region-In our previous studies of the GPIIb promoter, we have shown that the nucleotide sequence corresponding to the region -554/-414 is footprinted by nuclear proteins of HEL cells but not by nuclear proteins from HeLa cells (23). These protected areas, corresponding to cell-specific interactions were designated domains D and E and found to form a DNA cluster essential for the activity of the GPIIb promoter. To verify if the DNA-protein complexes were directly involved in the activity of the enhancer, the contact sites between the DNA and the proteins were first identified using a methylation interference procedure. Synthetic oligonucleotides corresponding to domain D and domain E were incubated with nuclear extracts of HEL cells. The fragments were then partially methylated and analyzed in gel shift assays. The DNA-protein complexes were eluted, submitted to Maxam and Gilbert G>A reaction, and analyzed on polyacrylamide gels (Fig. 3). The DNA-binding protein that interacted with domain D contacted at nucleotides -453 and -454, whereas the contact site of domain E were at nucleotides -515 and -516. These D and E contact sites were mutated either individually or together in the 192bp enhancer fragment. The different mutants of the enhancer region were incorporated in the pCAT promoter vector, and the effect of these mutations on the CAT activity was examined in extracts of transfected HEL cells. As shown in Fig. 4, each mutation, either alone or in combination, eliminated the enhancing effect of the fragment, indicating that domains D and E are implicated in the enhancer function.
To verify if this conclusion was still valid when the enhancer is located within the sequence of the GPIIb promoter, the same mutations were introduced together or individually in the 813 plasmid. Mutations at the D site or at the E site both produced a 60% inhibition of the activity (Fig. 5). This indicated that a single mutation of either binding site was sufficient to switch off the enhancer. Again, the double mutation did not produce a greater inhibition, indicating that the effect of the D and E cis-acting elements are cooperative.
Tissue-specific Activity of the Enhancer-Since the enhancer region was found to be active in both erythrocytic and megakaryocytic cell lines (Fig. 2), the presence of the nuclear proteins interacting with domains D and E was verified in K562 by mobility shift assays. The pattern of retarded bands observed with D and E oligonucleotides was similar when nuclear extracts from HEL or K562 cells were used (Fig. 6).
NO retarded bands were detected with extracts from HeLa cells. To verify that these DNA-protein complexes were also Platelet GPIIb Promoter bp upstream and 33 nucleotides downstream from the transcription start site, was inserted at the XbaI site of the pBLCAT3 vector in the direct or the reverse orientation. 5' deletion mutants were generated by digestion of this fragment with exonuclease I11 and blunted with mung bean nuclease as previously described (23). Each plasmid was transfected into HEL cells, and the respective CAT activities were measured 48 h after transfection. In each assay, the 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 -813 plasmid, which was taken as the 100% value. Each value is an average of at least three independent experiments. The pRSVCAT plasmid was used as a positive control. and the promoterless plasmid pBLCAT3 was used to measure the background level.

ment. This fragment was produced by digestion with Hind111 and
PuuII of the 1050-bp GPIIb genomic fragment. It was inserted in direct or reverse orientation, upstream or downstream from the CAT gene driven by the SV40 promoter in the pCAT promoter plasmid. The different constructs were introduced into HEL, K562, and HeLa cells for transient transfection assays. The volumes of protein extract for each CAT assay were normalized according to the luciferase activity. CAT activity was expressed relative to the basal activity obtained with the enhancerless pCAT promoter plasmid. The values reported represent the mean value of a t least three different experiments. active in K562 cells, the CAT activity was measured after transfection of K562 cells, with the different constructs containing mutations at the contact sites. As shown in Fig. 4, the mutations produced an inhibition of the enhancer activity similar to that observed in HEL cells.
Taken together, these results indicate that the enhancer domain interacts with similar positive transcription factors present in erythrocytic and megakaryocytic cells and that the enhancer exhibits an identical activity in both cell lines.

DISCUSSION
The production of circulating platelets is the ultimate reaction of a complex multistep differentiation process. It involves the commitment of an hematopoietic early pluripotent stem cell to the megakaryocytic lineage, the mitotic develop- thrombogenesis. In our effort to search for elements that are transcriptionally active in megakaryocytes, we have analyzed the promoter region of the platelet-specific GPIIb gene. In the present study, we describe the presence of a tissue-specific enhancer within the promoter of this gene.
We have previously shown that the genomic region extending from nucleotide -1080 to nucleotide +33, within the 5'flanking region of the GPIIb gene was able to confer cell specificity to the CAT reporter gene (23). One possible explanation for this observation is that the GPIIb gene contains a regulatory element whose activity is restricted to its promoter. Therefore, this DNA region was further analyzed to delineate the DNA sequences that play an active role in the transcription. The presence of a positive regulatory element within this promoter, centered at position -484, was suggested by the observation that deletion of the region between -554 and -414 produces a 70% decrease of the CAT activity in HEL cells but not in HeLa cells. A DNA segment corresponding to this sequence was able to activate the enhancerless ubiquitous SV40 promoter in a position-and orientation-independent manner in HEL but not in HeLa cells. Additional 5' deletions down to the nucleotide -113 had a minimal effect on the residual activity of the promoter. When the sequence between -113 and -29 was deleted, however, an additional 50% inhibition of the residual GPIIb promoter activity was observed. This loss of activity may be partly due to the deletion of a consensus binding site for the erythroid factor GATAl centered at position -54. We have already shown that this binding site interacts with a nuclear protein similar to this erythroid factor and is functional when fused to the CAT gene and transfected into HEL cells (29). Interestingly enough, the construct that contains the sequence between +33 and -13 exhibited a significant residual activity, as compared with the promoterless CAT construct. This suggests that the GPIIb promoter that does not have an obvious TATA motif may contain sequences that are critical for the positioning of the transcriptional protein complex close to the transcription start site.
These experiments indicate that the GPIIb promoter contains at least four distinct functional domains. One of these domains, centered at -484, exhibits an enhancing activity. This DNA region is composed of mult,iple binding sites for nuclear proteins that are present in megakaryocytic cells but not in HeLa cells. One of these sites, designated domain D, interacts with only one factor, whereas the other site, called domain E, can form two DNA-protein complexes, as detected by mobility shift assays. When these sites were mutagenized either in the isolated fragment or within the complete promoter sequence, the activity of the enhancer was lost, establishing a functional implication of these DNA-protein complexes. Alteration of these sites, either alone or in combination, produced the same effect, indicating that these DNAprotein complexes do not function separately, and are cooperative rather than synergistic.
To verify whether this enhancer domain was megakaryocyte-specific or not, the K562 cell line was used as a model for the erythroid lineage. This cell line does not express the platelet GPIIb protein, although the enhancer domain was active at a level similar to that observed with HEL cells. Furthermore, the nuclear proteins that interact with this region are also present in nuclear extracts of K562 cells, but not in myelomonocytic or fibroblastic cells (23). Two series of independent results suggest that the K562 nuclear proteins are similar to the factors present in megakaryocytic cells. 1) The electrophoretic pattern of the DNA complexes were similar. 2) Mutagenesis of the contact sites produced a comparable inhibition of the enhancer activity in both HEL and K562 cells. One of the binding sites, domain D, contains a TTATC motif, which is equivalent to the binding site of the erythroid factor GATA1. This factor has been shown to be implicated in both promoter and enhancer activity of erythroid gene. The present study suggests that it may also be implicated in the enhancer of a megakaryocytic gene. Further studies, however, are needed to determine if the observed positive effect of this IIb enhancer is mediated by GATAl in association with an as yet nonidentified factor.
A number of independent observations indicate that megakaryocytic and erythroid lineages share a number of phenotypic features. Both cells express the receptor for erythropoietin, a major regulator of the production of erythrocytes, and chromosomal markers have been used to demonstrate the existence of a bipotent progenitor cell (30, 31). Thus, understanding the switch between the two lineages is an interesting challenge. The complete GPIIb promoter is inactive in K562 cells, but our results indicate that the GPIIb enhancer domain is tissue-specific but is not lineage-specific. Therefore, this enhancing activity is not solely responsible for the megakaryocyte-specific expression of the gene. Then, why is the GPIIb gene not expressed in erythroid cells? Selectivity of expression can be achieved by negative, as well as positive regulatory elements. Negative regulation has been shown, for instance, to repress the a-globin gene in nonerythroid cells (32). A similar mechanism may control gene expression in the megakaryocyte. Though from the results presented in this study we can conclude that the GPIIb gene contains a classical enhancer element, this element must function in synergy with another positive or negative element. In support of this hypothesis, is a recent observation that the promoter region of another megakaryocyte marker, the PF4 gene, also contains positive and negative regulatory domains (33). More recently, we were able to identify a silencing domain within the GPIIb promoter that turns off the transcription of the gene in nonmegakaryocytic cells (34). Thus, it is likely that the establishment of the megakaryocyte lineage is mediated by a specific mechanism that controls the transcription of megakaryocytic genes.