Thrombopoietin-induced Expression of the Glycoprotein IIb Gene Involves the Transcription Factor PU.1/Spi-1 in UT7-Mpl Cells*

Thrombopoietin (TPO) is the major regulator of proliferation and differentiation of megakaryocytes and their progenitors. These actions can be reproduced in the human megakaryoblastic cell line UT7 into which the murine TPO receptor, c-Mpl, was introduced. In these cells, TPO enhanced the expression of the specific megakaryocytic marker integrin glycoprotein (GP) IIb-IIIa while decreasing the expression of erythroid genes (Porteu, F., Rouyez, M.-C., Cocault, L., Benit, L., Charon, M., Picard, F., Gisselbrecht, S., Souyri, M., and Dusanter-Fourt, I. (1996) Mol. Cell. Biol.16, 2473–2482). We have now analyzed the effect of TPO on the transcriptional activity of the GPIIb promoter in these cells. Using transient transfection assays of a series of human GPIIb promoter fragments, we delineated a TPO-responsive element within the previously reported enhancer region of the promoter. Although this enhancer included GATA- and Ets-binding sites (EBSs), we found that only EBS −514 was important for TPO response. We identified PU.1/Spi-1 as the endogenous Ets transcription factor that strongly and preferentially interacted with this enhancer EBS. This factor did not interact with other proximal EBSs in the GPIIb promoter. We next showed that TPO induced a strong and selective increase of PU.1/Spi-1 expression and DNA binding activity in UT7-Mpl cells. In contrast, TPO did not affect the expression of Ets-1/2 while weakly increasing the levels of Fli-1. Overexpression of PU.1/Spi-1 was further shown to enhance GPIIb promoter activity in the absence and presence of TPO. Overall, our data indicated that, in UT7-Mpl cells, TPO increased the transcriptional activity of a GPIIb gene in part due to an enhanced expression of an unexpected transcription factor, the Ets family PU.1/Spi-1 factor. To our knowledge, this is the first evidence of a role for the PU.1/Spi-1 factor in the regulation of megakaryocytic genes.

Megakaryocytic differentiation is characterized by the increase of DNA content in the cell, which involves an endomitotic process. This will lead to an increase in megakaryocyte ploidy and size and to the ultimate production of platelets from the fragmentation of megakaryocyte cytoplasm. Megakaryocytic differentiation is also characterized by the synthesis of a number of platelet proteins. Among these, the integrin ␣ IIb ⅐␤ 3 complex (glycoprotein (GP) 1 IIb-IIIa); the glycoproteins GPIb, GPIX, and GPV; platelet factor 4; and the ␤-thromboglobulin are exclusively expressed on platelets and megakaryocytes. In the past few years, the 5Ј-regulatory regions of a number of megakaryocytic genes were cloned and sequenced (1-7). These promoter regions were shown to share many characteristic features. In particular, they all contain associated binding sites for GATA and Ets factors (2,5,8,9). The ␣ IIb (GPIIb) gene is one of the most studied megakaryocytic genes. This TATA-less gene encodes a protein, the ␣-subunit of integrin ␣ IIb ⅐␤ 3 , which is expressed at a very early stage of megakaryocytic differentiation. We (10,11) and others (12) previously showed that an ϳ800-bp DNA fragment located upstream of the initiation start site of the GPIIb gene contains the cis-acting elements necessary for lineage-restricted expression of GPIIb in vitro and in vivo. Sequences crucial for tissue-specific expression of the GPIIb gene were identified using serial deletions of promoter constructs introduced into megakaryocytic cells. Studies showed that the two associated GATA/Ets-binding sites present in the human or rat GPIIb promoter are necessary for full promoter activity, with a dominant participation of the distal GATA/Ets motif (4,8,9). This distal GATA/Ets motif is located in an enhancer region that is active in both erythroid and megakaryocytic cell lines (11). A short enhancer-less GPIIb promoter fragment was also shown to be active in megakaryocytic cells in vitro only. This short 100-bp-long promoter was transactivated in non-megakaryoblastic HeLa cells when cotransfected with the GATA-1, Ets-1, or Fli-1 expression vector, suggesting the positive role of these factors in regulating the expression of the GPIIb gene in megakaryocytes (8,13). However, although a number of Ets factors including Ets-1, Ets-2, Fli-1, and PU.1 were reported to be present in megakaryocytes, it is still unclear which Ets factors bind to the various GPIIb Ets-binding sites (EBSs) and which of these factors are essential for GPIIb promoter activation in vivo. Tissue-specific expression of the GPIIb gene was also shown to depend on as yet unidentified negative factors that bind to a repressor domain in the proximal part of the promoter. These factors specifically suppress the expression of the GPIIb gene in non-megakaryocytic cells (4,14).
While GPIIb gene expression is restricted to megakaryocytic cells, its expression is also regulated along with megakaryocytic differentiation. Only a limited number of studies analyzed the promoter elements that participate in this differentiationdependent expression of the GPIIb gene. These studies always used erythromegakaryocytic cell lines induced to differentiate in the presence of phorbol esters. Interestingly, in these chemically induced differentiation models, regulatory elements that participated in the temporal expression of GPIIb were identified and located at the same positions as the tissue-specific positive and negative sequences (15). However, the identity of the factors physiologically important in the regulated expression of GPIIb in vivo awaits further investigation.
The cloning of thrombopoietin (TPO) and its receptor, c-Mpl, has provided a means toward understanding the molecular basis of megakaryocytic development. TPO was found to control the proliferation and differentiation of megakaryocytes and their progenitors specifically (16,17) and to participate in the activation of platelets (18). The TPO receptor is a member of the cytokine receptor superfamily and is believed to be activated through homodimerization (19,20). The human erythromegakaryoblastic cell line UT7 is dependent on the presence of the granulocyte-monocyte colony-stimulating factor (GM-CSF) or erythropoietin (EPO) for its growth and survival (21). We had previously shown that, upon introduction of the TPO receptor, these cells acquired the ability to proliferate and differentiate along the megakaryocytic lineage in the presence of TPO (22). In these cells, TPO-dependent differentiation was associated with an increase in the levels of GPIIb proteins and mRNAs (22). In the present study, we analyze the regulation of the human GPIIb promoter in UT7-Mpl cells in response to the natural megakaryocytic regulator TPO. Using transient transfection assays of a series of human GPIIb promoter-reporter constructs, we first delineate a TPO-inducible enhancer. We identify PU.1/Spi-1 as the major endogenous Ets factor bound to the TPO-responsive enhancer and provide evidence that TPO up-regulates the expression of active PU.1/Spi-1 proteins. This is the first identification of PU.1/Spi-1 as a transcription factor responsible for the regulated expression of GPIIb in a megakaryocytic cell line.
Cell Culture and Transfections-Human erythromegakaryoblastic UT7 cells (21) or the stable transfectant cells expressing the murine TPO receptor, UT7-Mpl (22), were cultured in ␣-minimal essential medium supplemented with 10% fetal calf serum (Life Technologies, Inc.) and EPO (2 units/ml; Boehringer Mannheim). The cells (5 ϫ 10 6 /transfection) were transfected by electroporation (250 V, 960 microfarads) in the presence of 20 g of the indicated luciferase reporter plasmid, 10 g of the pCMV-gal reference plasmid, and, where indicated, 2-10 g of expression vector. Unrelated DNAs were added to a final 50-g amount of total foreign DNA added to the cells. The transfected cells were immediately divided into two aliquots and incubated in the basal EPO-containing culture medium at 37°C for 48 h. The culture medium of one of the aliquots was supplemented with murine TPO or other tested cytokines. Murine TPO was obtained from baby hamster kidney cells engineered to stably express murine TPO (Zymogenetics) (23). This murine TPO at 1000 units/ml resulted in the same proliferation and differentiation of UT7-Mpl cells as recombinant human TPO at 20 ng/ml. Where indicated, recombinant human GM-CSF (2.5 ng/ml), interleukin-3 (10 ng/ml), interleukin-6 (10 ng/ml), and interferon-␥ (1000 units/ml) were used. 48 h after transfection, total cell extracts were prepared by three cycles of freeze-thawing lysis (Ϫ196°C to ϩ37°C).
COS cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and transfected by the DEAEdextran method. 48 h after transfection, the cells were resuspended in lysis buffer (0.1 M Tris-HCl (pH 7.5), 1% Triton X-100, 15% glycerol, and 0.5 mM phenylmethylsulfonyl fluoride) and incubated for 1 h at 4°C. Insolubilized material was removed by centrifugation (20,000 ϫ g, 10 min). HEL cells were maintained in ␣-minimal essential medium-conditioned medium supplemented with 10% fetal calf serum Luciferase and ␤-Galactosidase Assays-10 -20 l of cell extracts were mixed with 100 l of luciferase assay buffer (Promega) at room temperature and immediately scored for luciferase activity using a Lumac-3M luminometer (Berthold, Wildbad, Germany). 1-5 l of the same cell extracts were assayed for ␤-galactosidase activity according to the manufacturer's instructions using a Tropix kit and a Berthold luminometer. The galactosidase activity was used to evaluate transfection efficiency and to normalize luciferase activity in all tested extracts. Normalized luciferase activity in TPO-treated cells was compared with normalized luciferase activity in untreated cells and expressed as -fold activation.
Plasmid Constructions-Fragments of the human GPIIb promoter had been previously prepared and tested for their activity in the context of GPIIb-pBLCAT3 constructs (11,13). The SalI-BglII or HindIII-BglII fragment of the hGPIIb promoter was subcloned into the pGL2 vector containing the luciferase reporter gene (Promega) and checked by sequence analysis. Promoter fragments from ␤-actin (bp Ϫ340 to ϩ10) and the EPO receptor (bp Ϫ659 to Ϫ59) were also inserted into the pGL2 vector as controls. Expression vectors coding for PU.1/Spi-1, Fli-1, and the sole Ets domain (DNEts) under the control of the SV40 early gene promoter have been previously described (24 -26).
Western Blot Analysis-Whole cell extracts (100 g) or nuclear extracts (50 g) were analyzed by electrophoresis on 10% polyacrylamide gels followed by Western blotting and immunodetection using peroxidase-coupled antibodies and chemiluminescence detection (ECL, Amersham Corp.) as described (29).

Thrombopoietin Activates the GPIIb Promoter in UT7-Mpl
Cells-The human UT7-Mpl cell line expressing the murine TPO receptor, c-Mpl, can be maintained in culture in the presence of GM-CSF or EPO with similar proliferative rates (22). If UT7-Mpl cells were initially maintained in the presence of EPO, the megakaryocytic differentiation induced by TPO was far greater than if cells were initially maintained in the presence of GM-CSF, with a stronger induction of the CD41 marker and the appearance of polyploid cells, as previously reported (22). The induced expression of GPIIb protein was correlated with an accumulation of GPIIb mRNA (22), suggesting that TPO modulated the expression of the GPIIb gene at the transcriptional level. To test this hypothesis, we introduced into UT7-Mpl cells a reporter construct composed of a fragment of the human GPIIb gene promoter (bp Ϫ787 to ϩ33) fused to the luciferase reporter gene, and we analyzed its expression in the presence of TPO using transient transfection assays. Efficient and reproducible transient transfection of UT7-Mpl cells required the maintenance of EPO in the medium of the freshly transfected cells. Therefore, following transfection, UT7-Mpl cells were cultivated in the presence of EPO alone or EPO plus TPO.
TPO treatment stimulated luciferase activity in a dose-dependent manner, with a linear increase in the luciferase signal at TPO concentrations from 100 to 1000 units/ml (Fig. 1A). This last concentration was chosen for all subsequent experiments. The expression of phGPIIb-luc was also a function of time of TPO treatment (Fig. 1B): a moderate induction of luciferase activity (2-3-fold) was observed after the first 6 -18 h of treatment, followed by a strong increase thereafter (24 -48 h, 20fold induction over untreated cells). This time course of expression of phGPIIb-luc correlated well with the kinetics of accumulation of endogenous GPIIb mRNA in TPO-treated UT7-Mpl cells (data not shown). Some increased expression of our reference plasmid (pCMV-gal) or luciferase reporter constructs under the control of the ubiquitous ␤-actin promoter was also detected (Fig. 1B, black bars). These findings suggest that TPO may also activate basal transcription or may possess some general stabilizing post-transcriptional effects in addition to a specific action on GPIIb promoter activity.
To test the specificity of TPO action on phGPIIb-luc expression, we also incubated transiently transfected UT7-Mpl cells with GM-CSF (Fig. 1B, white bars), interleukin-3, interleukin-6, or interferon-␥ instead of or in combination with TPO. None of these cytokines affected the expression of phGPIIb-luc, and none modulated the action of TPO during the time of the transient transfection assay (data not shown), despite the expression of functional receptors for these factors in the cells (21,29). These data indicate that the region from bp Ϫ787 to ϩ33 of the hGPIIb gene contains a functional promoter and regulatory elements that mediate TPO-dependent transcriptional activation.
The hGPIIb Promoter Contains a TPO-responsive Enhancer That Includes an Important Ets-binding Site-To further delineate the specific DNA elements required for TPO induction, we tested deletion constructs in the region from bp Ϫ787 to ϩ33 of the hGPIIb promoter as depicted in Fig. 2. The basal expression of this series of hGPIIb promoter-luciferase deletion constructs was first tested in the absence of TPO. In contrast to our data obtained with HEL cells (14), progressive 5Ј-deletions of the hGPIIb promoter down to position Ϫ307 did not result in any significant change in luciferase expression in TPO-untreated cells (Fig. 2, left column). The next further truncations (to positions Ϫ170, Ϫ126, and Ϫ109) generated promoter fragments with the same change in transcriptional activity as we (14) and others (15) obtained in HEL and K562 cells.
Upon TPO treatment (Fig. 2, right column), we observed that deletion of the distal promoter region from bp Ϫ787 to Ϫ553 of the hGPIIb promoter did not affect the levels of TPO-induced luciferase activity. However, an additional deletion from bp Ϫ553 to Ϫ414 reproducibly decreased (ϳ2.5-fold) the cytokinemediated expression of the reporter gene, indicating that this 130-bp region contains a TPO-responsive element. No further deletions significantly impaired TPO-induced luciferase activity. The same results were obtained in the parental UT7 cells, where hGPIIb-reporter constructs were introduced along with a c-Mpl expression vector in similar transient transfection assays (data not shown), further strengthening our data.
The TPO-responsive region (bp Ϫ553 to Ϫ414) overlaps with a previously characterized tissue-specific megakaryocytic enhancer whose activity depends on the integrity of two motifs, namely a GATA-binding site (GATA Ϫ463) and an Ets-binding site (EBS Ϫ514) (11). To determine which of these two elements is necessary for TPO action, we tested reporter constructs containing individual mutations of these two sites in the context of the fully responsive Ϫ597/ϩ33 hGPIIb-reporter construct. As shown in Fig. 2 (lower constructs), mutation of GATA Ϫ463 did not affect the TPO responsiveness of the hGPIIb promoter. In contrast, mutation of the sole EBS Ϫ514 (changing the Ets-binding core sequence GGAA to TTAA) resulted in a significant decrease in the TPO response, of the same magnitude as the one observed with the enhancer-deleted Ϫ414 construct (Fig. 2, upper constructs).
The hGPIIb promoter fragment (bp Ϫ787 to ϩ33) also contains two other proximal EBSs at positions Ϫ39 and Ϫ139 from the transcriptional start site (see scheme in Fig. 2), which exhibits the known consensus binding site (G/C)(A/C)GGAAGT of Ets family proteins (30). Surprisingly, mutations of either of these two proximal EBSs did not affect the TPO-mediated activation of the hGPIIb promoter (Fig. 2, lower constructs). However, as expected from previous studies with HEL cells (8) the hGPIIb-luc construct in TPO-untreated UT7 cells. Thus, EBS Ϫ514 is the only EBS necessary for TPO-induced expression of the hGPIIb promoter.
TPO Induces the Binding of PU.1/Spi-1 to the Enhancer EBS-To determine if specific regulatory proteins bind to the TPO-inducible enhancer, EMSAs were performed using oligonucleotides spanning the different EBSs of the hGPIIb promoter or an optimized EBS from the Drosophila E74 promoter (30). As shown in Fig. 3, the E74 oligonucleotide formed four major DNA-protein complexes (complexes C1-C4) in the presence of nuclear extracts from untreated cells (lane 1). Three of these complexes (C1, C2, and C4) were highly specific as they were inhibited by a 100-fold excess of unlabeled E74 DNA (lane 2), but not by a mutant oligonucleotide (E74mut) in which the GGAA core sequence required for Ets binding was replaced by CCAA (lane 3). Similarly, using an oligonucleotide spanning EBS Ϫ514 as a probe (the G514 oligonucleotide), four DNAprotein complexes were detected, in the presence of untreated cell extracts, that migrated as those detected in the presence of the E74 probe. However, complex C1 was often poorly detected. These complexes were all specific for Ets-binding consensus sequence as they were inhibited by an excess of unlabeled E74 DNA, but not by the E74mut oligonucleotide (lanes 7 and 8). TPO treatment of UT7-Mpl cells led to a strong increase in the quantity of the faster migrating DNA-protein complex (C4), with no detectable change in the other complexes (lanes 4 -6).
This increase was initially detected after 18 h of TPO treatment (data not shown). Similar data were obtained when TPOtreated cell extracts were incubated in the presence of the E74 probe instead of G514 (Fig. 4, lanes 5 and 6). The identity of the Ets-binding protein present in this TPO-induced complex was next studied. Using a series of antibodies directed against specific Ets family members, we found that polyclonal anti-PU.1/Spi-1 antibodies totally eliminated complex C4, but failed to affect any of the other G514-protein complexes in supershift assays (Fig. 3, lane 9). The anti-Ets-1/2, anti-Fli-1, anti-GABP␣ (data not shown), and anti-Elf-1 (Fig. 3, lane 10) antibodies did not suppress the formation of complex C4.
We next focused our attention on the two other GPIIb EBSs (see Fig. 2). Using the same extracts as above and the oligonucleotide spanning EBS Ϫ39 or EBS Ϫ139, we never detected DNA-protein complexes migrating with mobility close to that of complex C4 or interacting with anti-PU.1/Spi-1 antibodies ( Fig.  4 and data not shown). Taken together, these data demonstrate that the Ets transcription factor PU.1/Spi-1 specifically binds to EBS Ϫ514 of the GPIIb promoter (as well as to the optimized E74 EBS) and exhibits an inducible DNA binding activity following TPO treatment.
The TPO-responsive Enhancer EBS Interacts Preferentially with PU.1/Spi-1-To better understand the function of the TPO-responsive EBS Ϫ514, we wondered whether this EBS

FIG. 2. Analysis of GPIIb promoter activity in TPO-treated or -untreated UT7-Mpl cells.
A schematic representation of a series of GPIIb promoter-luciferase constructs that were introduced into UT7-Mpl cells is shown. GATA-and Ets-binding sites are represented as hatched and black motifs, respectively. When mutated, the EBS (GGAA) and GATA (GATA) core sequences were changed to TTAA and TCTA, respectively. Each construct was introduced along with the reference plasmid into UT7-Mpl cells, and the cells were maintained for 48 h in the absence (Basal activity) or presence of TPO. The activities of promoter fragments were evaluated through luciferase activities. Left column, normalized luciferase activities were measured in TPO-untreated cells and are expressed relative to the full-length GPIIb promoter construct (bp Ϫ787 to ϩ33), which was arbitrarily set as 100%. Right column, normalized luciferase activities were measured in TPO-treated cells and are expressed relative to the activities measured in the untreated cells expressing the same pGPIIb construct (-fold induction). Results are the means of four independent experiments. could bind Ets family members equally. We therefore incubated this EBS with a large excess of various Ets factors. Because it was previously reported that the expression of two Ets transcription factors, namely Ets-1 and Fli-1, allowed the transcriptional activation of a minimal enhancer-less hGPIIb promoter (bp Ϫ75 to ϩ26) in non-megakaryocytic HeLa cells (8,13), we overexpressed the Ets-1, Fli-1, or PU.1/Spi-1 transcription factor in COS cells, and we prepared extracts from transfected COS cells as a source of a given Ets factor (Fig. 5). We tested the binding selectivity of the different GPIIb EBSs for these Ets factors in EMSAs. As indicated in Fig. 5, EBS Ϫ514 strongly bound PU.1/Spi-1 (lane 4) and weakly bound Ets-1 (lanes 2 and 5) and nearly failed to interact with Fli-1 (lane 3; one complex was still detectable after overexposure of the autoradiograph).
In contrast, the optimized E74 EBS interacted strongly with all three Ets factors tested (lanes 9 -11).
Using the same COS cell extracts, we found that neither EBS Ϫ39 nor EBS Ϫ139 bound the PU.1/Spi-1 factor. However, both EBS Ϫ39 and EBS Ϫ139 interacted with Ets-1, although with a rather weak affinity as compared with the optimized E74 EBS (data not shown). Therefore, among the three GPIIb EBSs present in the active GPIIb promoter fragment tested, only EBS Ϫ514 specifically bound PU.1/Spi-1. These data strengthen the relevance and specificity of PU.1/Spi-1 binding to the enhancer EBS Ϫ514.
TPO Induces the Expression of PU.1/Spi-1 Protein-We further studied the expression of a number of Ets family proteins by Western blot analysis using the same UT7-Mpl extracts as those used in EMSAs (Fig. 6). We detected low levels of PU.1/ Spi-1 and Fli-1, but no Ets-1, in untreated cells. TPO treatment greatly increased the level of PU.1/Spi-1 protein, which reached the same levels as those expressed in erythroleukemic HEL cells (Fig. 6A, upper panel). TPO treatment also resulted in a weak increase in the amount of Fli-1 proteins as compared with the amount in HEL cells (Fig. 6A, center panel). In contrast, Ets-1 protein was still undetectable in TPO-treated UT7-Mpl cells, although it was clearly detected in HEL cells (Fig. 6A,  lower panel). Elf-1 and GABP␣ Ets family proteins were also expressed in UT7-Mpl extracts, but their levels were not affected by TPO treatment (data not shown).
UT7-Mpl cells were also treated with GM-CSF instead of TPO, and the expression of Ets factors was similarly assayed by Western blot analysis. As shown in Fig. 6B, even after 4 days of treatment, GM-CSF did not significantly enhance the expression of any Ets factors tested in UT7-Mpl cells. All together, our data indicate that TPO selectively up-regulated the expression of PU.1/Spi-1.
We also introduced a dominant-negative variant of Ets proteins carrying only the DNA-binding Ets domain of Ets-1 (pD-NEts) into UT7-Mpl cells. This sole Ets domain was previously reported to behave as a general dominant-negative form for all Ets family factors (26). Coexpression of the Ets domain (pD-NEts) with pGPIIb-luc profoundly inhibited the TPO-dependent activation of the hGPIIb promoter while affecting only slightly the basal expression of GPIIb-luc in TPO-untreated cells (Fig. 7). The effect of this dominant-negative form of Ets proteins was specific for the expression of the GPIIb construct and was not related to a toxic side effect since it did not affect the expression of an EPO receptor-luciferase construct (data not shown). These data strengthen the importance of Ets factors in the TPO-induced expression of the GPIIb gene.

DISCUSSION
In this study, we provided evidence that, in the UT7-Mpl cell line, the transcription factor PU.1/Spi-1 regulates the expression of the megakaryocytic GPIIb gene through binding to a specific TPO-responsive element. To our knowledge, this is the first evidence of a direct involvement of the PU.1/Spi-1 transcription factor in the control of megakaryocytic gene expression. We further showed that TPO regulates the expression of active PU.1/Spi-1 protein in UT7-Mpl cells. This is the first evidence of the regulation of a specific Ets family transcription factor by a cytokine.
The PU.1/Spi-1 factor is an Ets-related transcription factor that was first discovered as a target of insertional activation at the Spi site by the Friend virus (31). This insertion leads to an overexpression of normal PU.1/Spi-1 proteins in infected erythroid progenitors and consequently to immortalization of erythroblasts via as yet uncharacterized mechanisms (32). Independently, PU.1/Spi-1 was identified as a macrophage and B lymphoid transcription factor (33) that regulated the expres-sion of a number of genes in this two lineages (reviewed in Ref. 35). The expression of PU.1/Spi-1 was shown to be restricted to hematopoietic organs and was detected in myeloid, erythroid, and B (but not T) lymphoid cell lines. In bone marrow, in situ immunohistochemical analysis showed that PU.1/Spi-1 was present in early (but not late) erythroid and granulocytic precursors. In the same studies, PU.1/Spi-1 was shown to be surprisingly highly expressed in mature megakaryocytes (34). A number of other studies reported the expression of PU.1/ Spi-1 in cultures of early erythroid progenitors (32,35,36), which suggested a requirement for the PU.1/Spi-1 factor during very primitive erythroid development (37). Interestingly, Friend murine erythroleukemia cells, which were derived from Friend virus-infected erythroid progenitors and overexpressed PU.1/Spi-1, express quite high levels of a number of megakaryocytic markers in addition to the erythroid ones. Because of the expression of this megakaryocyte-like program, these cells were considered as a bipotential model of differentiation rather than being purely erythroid in nature (42). In light of our present data, it is now possible to suggest that overexpression of PU.1/Spi-1 in murine erythroleukemia cells is responsible for the observed expression of megakaryocytic markers. Therefore, although PU.1/Spi-1 was long believed to be a master switch in hematopoietic development in programming hematopoietic cell commitment and differentiation along the myeloid lineage, other observations also suggested a role for this transcription factor in early erythroid and late megakaryocytic development (32, 34 -37, 42).
Insertional inactivation of the PU.1 gene causes multiple hematopoietic abnormalities, with a lack of mature macrophages, neutrophils, and B and T cells leading to perinatal death of PU.1/Spi-1 Ϫ/Ϫ mice (38,39). Overall development of the erythroid and megakaryocytic lineages seemed not to be strongly affected in PU.1/Spi-1 null neonates, with platelets being detected in peripheral blood smears and erythroid progenitors being present in bone marrow. However, the precise number and functional capacities of the erythroid and megakaryocytic progenitors in PU.1/Spi-1 null mice have not yet been determined. Development of erythroid and megakaryocytic progenitors during embryonic development of PU.1/Spi-1 null mice might be due to other Ets family protein(s) compensating for the absence of PU.1/Spi-1 and allows some megakaryocytic differentiation in these animals. In agreement with this hypothesis, we (Fig. 7) and others (8,13) found a strong induction of GPIIb expression in cells overexpressing Ets-1 or Fli-1, indicating a possible functional complementation/redundancy among various Ets members for the activation of megakaryocytic genes.
Analyses of the mechanisms involved in megakaryocytic differentiation and the expression of megakaryocytic genes were performed with phorbol ester-induced differentiation models. In cells such as DAMI, Meg01, HEL, and K562, phorbol esters were shown to induce a strong expression of GPIIb along with the expression of Ets-1 and Fli-1 (but not Ets-2 and PU.1/Spi-1) mRNAs (8,43). Neither of these studies had evaluated the expression of Ets proteins in addition to Ets transcripts. We did not find any Ets-1 proteins expressed in TPO-treated or -untreated UT7-Mpl cells, indicating that endogenous Ets-1 proteins should not play any role in the basal and TPO-dependent expression of GPIIb in our cells. Furthermore, although phorbol esters induced a strong expression of megakaryocytic markers in UT7-Mpl cells (22), these agents did not induce the expression of PU.1/Spi-1 proteins after 2-4 days of treatment (data not shown). Thus, the mechanisms involved in the TPOdependent induction of megakaryocytic differentiation are at least partially unrelated to those triggered by phorbol esters. It is worth noting that, although we delineated a TPO-responsive enhancer in the GPIIb promoter, TPO still induced up to an 8-fold activation of the enhancer-less GPIIb promoter construct (see Fig. 2). This indicates the existence of PU.1/Spi-1-independent signal(s) able to activate the GPIIb proximal promoter upon TPO treatment. The identities of these signals are unknown. Whether these signals involve phorbol ester-responsive pathway(s) is currently under investigation.
The mechanism(s) by which PU.1/Spi-1 might regulate the transcriptional activity of the GPIIb promoter is still unknown. However, the GPIIb gene is a TATA-less gene, and such genes have often been shown to depend on Sp1 and Ets factors to tether the transcriptional initiation complex at the transcriptional start site. Recent studies with primary rat megakaryocytes suggested that factors bound to the proximal EBS and Sp1 sites of the rat GPIIb promoter could interact with as yet uncharacterized enhancer factor(s) to form a transcription complex with the TFIID factor, joining the basal transcriptional machinery to upstream enhancer elements (40). PU.1/Spi-1 has been previously reported to directly interact via its N-terminal transactivation domain with TFIID in vitro (41). According to these observations and our present data, PU.1/Spi-1 bound to EBS Ϫ514 may provide a physical link between the basal transcription complex (TFIID) at the transcriptional start site and the enhancer region, which will regulate the magnitude of GPIIb transcription. Since a number of other known megakaryocytic genes (PF4, GPV, etc.) possess multiple Ets-binding sites in their regulatory region (3,44), it remains to be established whether the PU.1/Spi-1 factor will also participate in the regulation of other megakaryocytic gene expression.