Unique and selective effects of five Ets family members, Elf3, Ets1, Ets2, PEA3, and PU.1, on the promoter of the type II transforming growth factor-beta receptor gene.

Previous studies have shown that the promoter of the type II TGF-beta receptor gene (TbetaR-II) is strongly stimulated by Elf3, a member of the Ets transcription factor family. The TbetaR-II gene behaves as a tumor suppressor and it is expressed in nearly all cell types, whereas Elf3 is expressed primarily in epithelial cells. Hence, the TbetaR-II gene is likely to be regulated by other Ets proteins in nonepithelial cells. In this study, we examined the effects of four other Ets family members (Ets1, Ets2, PEA3, and PU.1) on TbetaR-II promoter/reporter constructs that contain the two essential ets sites of this gene. These studies employed F9 embryonal carcinoma cells and their differentiated cells, because transcription of the TbetaR-II gene increases after F9 cells differentiate. Here we demonstrate that Ets2, which is expressed in F9-differentiated cells along with Elf3, does not stimulate or bind to the TbetaR-II promoter in these cells. In contrast, PEA3 stimulates the TbetaR-II promoter in F9-differentiated cells, but it inhibits this promoter in F9 cells. Thus, the effects of PEA3 on the TbetaR-II promoter are cell context-dependent. We also show that the effects of Elf3 are cell context-dependent. Elf3 strongly stimulates the TbetaR-II promoter in F9-differentiated cells, but not in F9 cells. In contrast to Elf3 and PEA3, Ets1 strongly stimulates this promoter in both F9 cells and F9-differentiated cells. Finally, we show that PU.1 exerts little or no effect on the activity of the TbetaR-II promoter. Together, our findings indicate that Elf3 is not the only Ets protein capable of stimulating the TbetaR-II promoter. Importantly, our findings also indicate that each of the five Ets proteins influences the TbetaR-II promoter in a unique manner because of important differences in their biochemical properties or their patterns of cellular expression.

TGF-␤ 1 consists of three isoforms that exert potent, pleiotropic effects during growth and differentiation (1,2). TGF-␤ also plays critical roles in many diseases. In the case of cancer, TGF-␤ has been shown to exert complex effects on tumor growth and metastasis (3,4). TGF-␤ has also been implicated in the suppression of the immune system (5)(6)(7)(8)(9), and decreased expression of functional TGF-␤ receptors by cells of the immune system in an animal model has been shown to suppress tumor metastasis and enhance host survival (8,9). TGF-␤s exert their biological effects through interactions with three distinct high affinity TGF-␤ cell surface receptors, which are designated types I, II, and III (1,2). The type III receptor, also known as betaglycan, is a transmembrane proteoglycan devoid of intrinsic kinase activity. The type I (T␤R-I) and type II (T␤R-II) receptors are transmembrane serine/threonine kinases that act in concert to mediate intracellular signaling. Ablation of any of the three TGF-␤ receptors by gene targeting causes lethal defects during development (10 -12), and defects in these receptors, in particular T␤R-I and T␤R-II, are strongly associated with tumorigenicity (1,2,13). Although T␤R-I and T␤R-II clearly behave as tumor suppressor genes, inactivation of these genes exerts different effects on tumor growth and tumor metastasis (3,4).
During the past several years, transcriptional regulation of the T␤R-II gene has been studied extensively. The T␤R-II promoter lacks a consensus TATA box, but it has been reported to be regulated by a negative regulatory element and by at least six positive cis-regulatory elements located within 400 base pairs of its major transcription start site (14 -25). The positive cis-regulatory elements include a CRE/ATF site located ϳ200 base pairs upstream of the major transcription start site. More than one transcription factor has been reported to activate the T␤R-II promoter through this cis-regulatory element (16,24). A CCAAT box, which binds the transcription factor complex NF-Y, has been identified ϳ80 base pairs upstream of the major transcription start site (16,22). Other studies have identified at least two conserved GC boxes that are located close to the CCAAT box. Sp1 appears to be the prime candidate for activating the T␤R-II promoter via these GC boxes (19 -22). In addition, the murine T␤R-II 5Ј-flanking region contains a functional Egr-1 binding site (25). The activity of the T␤R-II promoter is also influenced heavily by a region containing two overlapping ets sites located just downstream of the major transcription start site (17,23). Several studies have demonstrated that the transcription factor Elf3 (also known as ESX, ESE-1, Jen, and ERT) can strongly activate the T␤R-II promoter via these ets sites in epithelial cells (17,23).
Elf3 is a member of the Ets family of transcription factors, which is composed of over 25 members (26). This family of transcription factors is characterized primarily by a highly conserved DNA binding domain, known as an ETS domain (26). Elf3 not only activates T␤R-II promoter/reporter constructs in epithelial cells, it also has been shown by chromatin immunoprecipitation studies to bind to the T␤R-II promoter in vivo (23). Like many other Ets family members, Elf3 appears to contain a potent autoinhibitory domain that limits its binding to DNA in vitro (23). Although the position of this domain within Elf3 has not been determined, it appears to be located N-terminal of its DNA binding domain. Other studies using athymic nude mice and the breast cancer cell line Hs578t, which exhibits little or no endogenous Elf3, have shown that forced expression of Elf3 elevates expression of T␤R-II and diminishes tumor growth (27). Moreover, Elf3 knockout mice, which exhibit severe defects in the developing gut, express lower levels of T␤R-II in their enterocytes (absorptive cells of the gut) (28).
Elf3 is expressed by all epithelial cell types examined thus far, but not by other cell types (29,30). Given that T␤R-II is expressed by nearly all cell types, it is likely that other members of the Ets family of transcription factors stimulate the transcription of the T␤R-II gene in cell types that do not express significant levels of Elf3. In this regard, Ets proteins have been found in all cells, and several Ets proteins are expressed ubiquitously (26,(31)(32)(33). Furthermore, the DNA binding domains of Ets proteins appear to be structurally very similar. X-ray crystallographic structures of six Ets proteins demonstrate that their DNA binding domains share a common structural fold, which contains three ␣-helices and four ␤-strands (34 -38). This raises the possibility that the T␤R-II gene is stimulated indiscriminately by other members of the Ets family. However, this is unlikely to be the case. Work with other promoters argues that Ets proteins do not promiscuously stimulate promoters that contain ets sites. In fact, a high degree of specificity is observed, such that only one, or at most a few, Ets family members activate any given promoter with ets sites. Currently, the mechanisms responsible for this specificity are only partially understood. However, autoinhibitory domains, which are present in at least eight Ets family members and which limit the binding of these Ets proteins to DNA in vitro (23, 39 -44), appear to play critical roles in controlling access to ets sites in vivo. In addition, in several well documented studies, cooperative binding with other transcription factors that bind to adjacent DNA regulatory sequences has been shown to overcome the negative effects of the autoinhibition in vitro (45)(46)(47).
Several tumor model systems, including embryonal carcinoma (EC) cells and their differentiated counterparts, have been used to investigate the transcriptional regulation of the T␤R-II gene. EC cells are used widely, because their differentiation in culture mimics important stages of mammalian embryogenesis (48). EC cells also provide an excellent model system for studying the regulation of TGF-␤ receptors, because differentiation of EC cells suppresses their tumorigenicity (49 -51) and leads to the appearance of cell surface TGF-␤ receptors (52). Accordingly, EC cells do not respond to TGF-␤, whereas, TGF-␤ inhibits the growth of their differentiated cells (52). It is also significant that the differential expression of TGF-␤ receptors by EC cells and their differentiated counterparts parallels the onset of TGF-␤ receptor expression by similar cell types in the early embryo (53). Although T␤R-I mRNA does not appear to change significantly with the differentiation of EC cells (23), the transcription of the T␤R-II gene increases when EC cells differentiate (16,23). Moreover, the mRNA for T␤R-II and the mRNA for Elf3 increase in parallel, and forced overexpression of Elf3 in EC-differentiated cells strongly increases T␤R-II promoter activity (23). Together, these findings suggest that the up-regulation of Elf3 plays a key role in transcriptional activation of the T␤R-II gene when EC cells differentiate into nontumorigenic cells.
To understand the transcription of the T␤R-II gene more fully, this study compared the abilities of five Ets family members to regulate T␤R-II promoter activity. Specifically, the effects of Elf3, PEA3, Ets2, Ets1, and PU.1 on the T␤R-II promoter were compared in F9 EC cells and F9-differentiated cells. The effects of Ets2 were examined, because Ets2 (54) and Elf3 (23) are both expressed in F9-differentiated cells. PEA3 was examined, because it is expressed in EC cells, but its expression is down-regulated strongly when F9 EC cells differentiate (54). Thus, PEA3 expression decreases as the transcription of the T␤R-II gene increases (16,54). Ets1 and PU.1 were included in this study as candidate Ets proteins that may be involved in regulating the transcription of the T␤R-II gene in nonepithelial cells. In this regard, Ets1 and PU.1 are expressed in the cells of the immune system, which express T␤R-II but not Elf3 (31,32,(55)(56)(57). Together, our findings indicate that each of these Ets family members influences the T␤R-II promoter in a unique manner because of important differences in their biochemical properties or their patterns of cellular expression. This study also provides further characterization of the ets sites of the T␤R-II promoter and demonstrates that these two overlapping ets sites are both required for the synergistic stimulation of the T␤R-II promoter by Ets proteins.

EXPERIMENTAL PROCEDURES
Materials-Dulbecco's modified Eagle's medium (DMEM) was obtained from Invitrogen. Fetal bovine serum (FBS) was purchased from HyClone Laboratories (Logan, UT). Unless indicated otherwise, all chemicals were obtained from Sigma.
Cell Culture-F9 EC cells were cultured on gelatin-coated tissue culture plastic in DMEM containing 10% FBS, as described previously (16,23). F9 cells were induced to differentiate by a 72-h treatment with 5 M retinoic acid, and 293T cells were cultured in DMEM containing 10% FBS, as described previously (23). All cells were cultured at 37°C in a humidified atmosphere of 5% CO 2 in air.
Ets Protein Expression Constructs-Expression constructs for Elf3 (pcDNA3.0-mElf3), Elf3⌬N233 (Elf3(ATHϩDBD)), and Elf3⌬N270 (Elf3(DBD)) have been described previously (23). The expression construct GFP-Elf3 was generated by insertion of the coding sequence for Elf3, from the Elf3 expression vector pcDNA3.0-mElf3, into the GFP (green fluorescent protein) vector, pEGFP-C1 (Clontech Laboratories, Palo Alto, CA). Expression vectors Ets2 and Ets2⌬N331 were obtained from Craig Hauser (Birnham Institute, La Jolla, CA) (58). The expression vector Ets2⌬C453 was generated from Ets2 by replacing the codon for amino acid 453 with a stop codon using the Quick Change method. This was accomplished using the primer pairs 5Ј-TGCTGGGCTTCACT-TGAGAGGAACTGC-3Ј and 5Ј-GTTCCTCTCAAGTGAAGCCCAG-CAAGT-3Ј (the underlined sequences refer to the mutation generated to insert a stop codon and eliminate 16 amino acids from the C-terminal tail of Ets2). The absence of a BspEI site, which was deleted to generate the mutation, was used to screen for mutant clones. The mutant sequence was verified by sequence analysis at the Genomic Core Research Facility (University of Nebraska, Lincoln, NE). The expression vector for Ets1, pCMV-Tag2a-FlagEts1, was obtained from Barbara Graves (University of Utah, Salt Lake City, UT) (59). YFP-Ets1 was generated by inserting the coding sequence for Ets1, from pCMV-Tag2a-FlagEts1, into the yellow fluorescent protein (YFP) vector, pEYFP-C1 (Clontech Laboratories, Palo Alto, CA). The expression vector for PEA3 was obtained from John Hassell (McMaster University, Hamilton, Ontario, Canada) (44), and the expression vector PU.1 was obtained from R. Kawahara (University of Nebraska Medical Center) (61).
Transfection and Activities of Promoter/Reporter Gene Constructs-F9 EC cells and F9-differentiated cells were transiently transfected by the calcium phosphate precipitation method, as described previously (16,23). In addition to 15 g of T␤R-IIϪ108/ϩ56 promoter/ reporter gene constructs, the cells were transfected with 1 g of the internal control plasmid, pCMV-␤-gal, to normalize for any differences in transfection efficiency or general effects on transcription. In this regard, Ets1, Elf3, and PU.1, but not PEA3, Ets2, Elf3⌬N233, or Elf3⌬N270, appeared to decrease the activity of pCMV-␤-gal more than 3-fold when their expression vectors were used at concentrations greater than 3 g. This was also true for other normalizing plasmids tested, which were driven by the SV40 promoter. Hence, expression vectors for Ets1, Elf3, and PU.1 were not used at concentrations above 3 g. At 3 g or less, the effects of these Ets proteins on pCMV-␤-gal were generally less than 3-fold. Chloramphenicol acetyltransferase (CAT) and ␤-galactosidase activities were determined 48 h after transfection of F9 EC cells and 72 h after transfection of F9-differentiated cells, as described previously (16).
Transfection and Expression of GFP and YFP Fusion Proteins-F9 EC cells or F9-differentiated cells were transiently transfected with expression vectors for Ets-fluorescent fusion proteins using Lipo-fectAMINE in conjunction with the LipofectAMINE PLUS reagent (Invitrogen) as described previously (62). This transfection protocol was used, because previous studies determined that only small differences in transfection efficiency are observed when these reagents are used (62). Expression of fluorescent proteins was determined 24 h after transfection by flow cytometry using a Becton Dickinson FACSCalibur flow cytometer. The data were collected and analyzed as described previously (62). Examination of YFP-Ets1 and GFP-Elf3 expression by fluorescence microscopy demonstrated that these proteins were localized to the nuclear compartment.
Electrophoretic Mobility Shift Assays (EMSA)-Nuclear extracts were prepared from 293T cells that had been transiently transfected with 20 g of one of the flag-tagged Ets protein expression vectors, as described previously (23). Double-stranded oligonucleotide probes were radiolabeled, as described previously (16,23). The amounts of flagtagged Ets proteins in the nuclear extracts were determined by Western blot analysis using the monoclonal antibody (M2) that recognizes the flag epitope, as described previously (23). For each EMSA, the amount of nuclear extract added to the binding buffer (23) was adjusted so that equal amounts of flag-tagged Ets proteins were compared. In cases where supershift assays were performed, the monoclonal antibody M2 was preincubated with the nuclear extract for 40 min at room temperature prior to adding the radiolabeled DNA probe. In all experiments shown, the DNA-protein complexes were shown to be specific by virtue of competition with an unlabeled double-stranded DNA oligonucleotide with the same sequence as the radiolabeled DNA probe and by the lack of competition with an unlabeled double-stranded DNA oligonucleotide with a different sequence. The DNA-protein complexes were resolved on 6% nondenaturing polyacrylamide gels, as described previously (16,23). The gels were dried and visualized by a PhosphorImager (Amersham Biosciences). The DNA probes used corresponded to nucleotides 3-32 of the mT␤R-II gene (23): B/A probe, 5Ј-TGGCGAGGAGTTTCCTGTTTC-CCTCGGCGC-3Ј and its complement; X/A probe, 5Ј-TGGCGAGGAGA-GCGCTGTTTCCCTCGGCGC-3Ј and its complement; B/X probe, 5Ј-T-GGCGAGGAGTTTCCTGTTCTAGACGGCGC-3Ј and its complement; B/B probe, 5Ј-TGGCGAGGAGTTTCCTGTTTCCTGTGGCGC-3Ј and its complement. In each case, the mutant sequence is underlined. For each probe, the sequence, TAGC, was added to the 5Ј end of each oligonucleotide for use in radiolabeling.

Elf3 Synergistically Increases the Activity of the T␤R-II Promoter via Two Overlapping Ets Sites-Although
Ets proteins bind to an essential core sequence of 5Ј-GGA(A/T)-3Ј (63), their binding to DNA is affected by flanking sequences. Based on the DNA binding preferences of eight Ets proteins (64 -68), Ets proteins appear to bind to a 9-base pair consensus ets site that consists of 5Ј-(A/G)(C/G/T)(C/A)GGA(A/T)(G/A)(C/T)-3Ј. Analysis of the region 12-27 relative to the major transcription start site of the mouse T␤R-II gene indicates that it contains two potential ets sites that overlap by 2 base pairs (15,23). These sites are arranged in the same 5Ј to 3Ј direction. The sequence of the upstream ets site (5Ј-ACAGGAAAC-3Ј; designated the B-site) matches the sequence of a consensus site, whereas the sequence of the downstream ets site (5Ј-GAGGGAAAC-3Ј; designated the A-site) differs from the consensus sequence above by 2 base pairs, which are shown in bold (Fig. 1A). Except for 1 base pair in the putative A-site of the human gene (5Ј-GGGG-GAAAC-3Ј), which is shown in bold, the two overlapping ets sites in the mouse and human T␤R-II genes are conserved fully.
Previous studies have examined the effects of mutating both sites or just the B-site in promoter/reporter gene constructs that contain either the human or the mouse 5Ј flanking region of the T␤R-II gene (17,23). To determine the contribution of the A-site relative to the B-site, we disrupted each site individually by site-directed mutagenesis. For this purpose, we used the promoter/reporter gene construct mT␤R-IIϪ108/ϩ56(B/A) that contains the mouse T␤R-II promoter region and the wild type sequence for each ets site (B/A). The activities of the mutant constructs, mT␤R-IIϪ108/ϩ56(B/X) and mT␤R-IIϪ108/ϩ56(X/ A), were compared with the activities of the parent construct mT␤R-IIϪ108/ϩ56(B/A) and the double mutant construct mT␤R-IIϪ108/ϩ56(X/X) by transfecting these constructs individually into F9-differentiated cells (Fig. 1B). Disruption of the A-site or the B-site reduced the activities of the promoter/ reporter gene constructs ϳ40 and 60%, respectively. Disruption of both sites reduced the activity of the promoter/reporter construct by at least 80% (Fig. 1B).
To further characterize the contributions of the B-site and the A-site, we compared the ability of Elf3 to stimulate the activities of the parent T␤R-II promoter/reporter construct, mT␤R-IIϪ108/ϩ56(B/A), and the two mutant constructs, mT␤R-IIϪ108/ϩ56(B/X) and mT␤R-IIϪ108/ϩ56(X/A). For this purpose, F9-differentiated cells were transiently transfected with two concentrations of the Elf3 expression vector and one of the T␤R-II promoter/reporter gene constructs (Fig. 1C). At the highest concentration tested, Elf3 stimulated the activity of the parent construct, mT␤R-IIϪ108/ϩ56(B/A), 23-fold. In contrast, Elf3 at the same concentration stimulated the activity of the mT␤R-IIϪ108/ϩ56(B/X) construct and the mT␤R-IIϪ108/ ϩ56(X/A) construct only 3.7-and 4.7-fold, respectively. This demonstrated that both sites are essential to respond fully to Elf3. More importantly, it argues that the two ets sites function cooperatively to enhance synergistically the expression of the parent mT␤R-IIϪ108/ϩ56(B/A) construct.
Ets1, but Not PU.1, Strongly Stimulates the T␤R-II Promoter-Given the impact of Elf3 on the activity of the T␤R-II promoter, we examined whether other Ets proteins could stimulate the activity of this promoter in F9-differentiated cells. Because the T␤R-II gene is transcribed in cells that do not express Elf3, we initially examined the ability of Ets1 to stimulate T␤R-II promoter activity. Ets1, the founding member of the Ets family, is not expressed in F9 EC cells or F9-differen-FIG. 1. Both ets sites of the T␤R-II promoter are required for maximal response to Elf3. A, the sequences of the wild-type and mutated ets sites in mT␤R-IIϪ108/ϩ56(B/A), mT␤R-IIϪ108/ϩ56(X/A), mT␤R-IIϪ108/ϩ56(B/X), and mT␤R-IIϪ108/ϩ56(X/X) are shown. These sequences and the relative positions of the two ets sites are underlined. The mutated sequences of ets sites are boxed and shaded. B, F9-differentiated cells were transiently transfected with 15 g of each construct along with 1 g of an internal control plasmid, pCMV-␤-gal. CAT and ␤-galactosidase activities were assayed and normalized, as described under "Experimental Procedures." The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ϩ56(B/A), which was set to 1. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed three times, and similar results were obtained in each case. C, F9-differentiated cells were transiently transfected with 15 g of the mT␤R-II promoter/reporter construct shown, an Elf3 expression vector (0, 1, or 3 g), and 1 g of an internal control plasmid, pCMV-␤-gal. The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ϩ56(B/A), which was set to 1. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed three times, and similar results were obtained in each case. tiated cells (54). It is expressed primarily in lymphoid tissues, which express T␤R-II, but not Elf3 (32,56). Like Elf3, Ets1 strongly stimulated the parent mT␤R-IIϪ108/ϩ56(B/A) construct in F9-differentiated cells ( Fig. 2A). Furthermore, the strong response to Ets1 required both ets sites, because Ets1 barely stimulated either mT␤R-IIϪ108/ϩ56(B/X) or mT␤R-IIϪ108/ϩ56(X/A). These findings led us to examine the ability of a third Ets family member, PU.1, to stimulate the activity of the T␤R-II promoter. PU.1 is expressed in B-cells and myeloid cells, which express T␤R-II, but not Elf3 (57). Unlike Elf3 and Ets1, PU.1 exerted only a modest effect on the activity of mT␤R-IIϪ108/ϩ56(B/A) (Fig. 2B). This modest effect of PU.1 does not appear to be the result of its inability to bind to the T␤R-II promoter, because PU.1 reduced the response of mT␤R-IIϪ108/ϩ56(B/A) to Elf3 (Fig. 2B). As a control, we determined that the inhibitory effect of PU.1 was not the result of an effect FIG. 2. Ets1 can strongly stimulate the T␤R-II promoter, but PU.1 does not. A, F9-differentiated cells were transiently transfected with 15 g of the mT␤R-II promoter/reporter gene construct shown and 1 g of an internal control plasmid, pCMV-␤-gal. Where indicated, the cells were also transfected with an Ets1 expression vector at the amounts shown. CAT and ␤-galactosidase activities were assayed and normalized, as described under "Experimental Procedures." The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ϩ56(B/A), which was set to 1. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed three times, and similar results were obtained in each case. B, F9-differentiated cells were transiently transfected with 15 g of mT␤R-IIϪ108/ϩ56(B/A) and 1 g of an internal control plasmid, pCMV-␤-gal. Where indicated, the cells were also transfected with an expression vector for Elf3 or PU.1 at the amounts shown. The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ϩ56(B/A), which was set to 1. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed three times, and similar results were obtained in each case. C, F9-differentiated cells were transiently transfected with 2 g of an expression vector for either GFP-Elf3 or GFP as indicated using LipofectAMINE and the PLUS reagent, as described under "Experimental Procedures." Where indicated, the cells were also transfected with 2 g of an expression vector for PU.1. GFP fluorescence was measured by flow cytometry, as described under "Experimental Procedures." Positive cells were determined as the percentage of cells expressing fluorescence above that of the mock transfection control. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed twice, and similar results were obtained. on Elf3 expression. Using a GFP-Elf3 fusion protein, which we have shown to be fully active in the stimulation of the T␤R-II promoter, we determined that PU.1 does not reduce the expression of GFP-Elf3 in F9-differentiated cells (Fig. 2C). If anything, PU.1 appeared to enhance the level of GFP-Elf3. However, this enhancement appears to be nonspecific, because it also enhanced the level of GFP itself. Together, these findings argue that Elf3 and Ets1 can strongly stimulate the T␤R-II promoter, whereas PU.1 exerts little if any effect on the T␤R-II promoter in F9-differentiated cells.
PEA3 Strongly Stimulates the T␤R-II Promoter in F9-differentiated Cells, but Not in the Parental F9 EC Cells-Given the differential effects of Ets1 and Pu.1, we examined the effect of PEA3 on the T␤R-II promoter. Initially, we examined the ability of PEA3 to stimulate mT␤R-IIϪ108/ϩ56(B/A) in F9-differentiated cells, even though it is expressed in F9 EC cells and is strongly down-regulated in F9-differentiated cells (54). Like Elf3 and Ets1, PEA3 exerted a strong effect on the activity of the T␤R-II promoter (Fig. 3). However, we consistently observed a smaller nonspecific effect of PEA3 on mT␤R-IIϪ108/ ϩ56(X/X). In addition, we determined that the responses of mT␤R-IIϪ108/ϩ56(B/X) and mT␤R-IIϪ108/ϩ56(X/A) to PEA3 were similar to that observed with mT␤R-IIϪ108/ϩ56(X/X) (data not shown). Thus, the full response of the T␤R-II promoter to PEA3 appears to require both ets sites.
The ability of PEA3 to stimulate the T␤R-II promoter in F9-differentiated cells raised an important question. Does PEA3 stimulate the T␤R-II promoter in F9 EC cells, in which PEA3 is expressed (54), but where the expression of the T␤R-II gene is very low (16,23)? This question was addressed by co-transfecting F9 EC cells with mT␤R-IIϪ108/ϩ56(B/A) and an expression vector for PEA3 (Fig. 4A). PEA3 exerted little effect on the activity of this construct in these cells. In fact, at the higher levels of PEA3, the activity of mT␤R-IIϪ108/ ϩ56(B/A) was reduced below the basal activity of this construct observed in the absence of the PEA3 expression vector. This suggested that PEA3 may in fact be inhibitory to the T␤R-II promoter in F9 EC cells. This possibility was tested by taking advantage of our finding that Ets1 stimulates the activity of mT␤R-IIϪ108/ϩ56(B/A) construct in F9 EC cells (Fig. 4A). When F9 EC cells were co-transfected with expression vectors for both Ets1 and PEA3, the activity of mT␤R-IIϪ108/ϩ56(B/A) decreased in a dose-dependent manner (Fig. 4B). As a control, we demonstrated that the expression of YFP-Ets1 fusion protein, which we have shown strongly stimulates the T␤R-II promoter (data not shown), is not reduced by overexpression of PEA3 in F9 EC cells (Fig. 4C). In view of the diametrically different responses of the T␤R-II promoter to PEA3 in F9 EC cells and in F9-differentiated cells, it is evident that the response to this member of the Ets family is cell context-dependent.
Elf3 Does Not Stimulate the T␤R-II Promoter in F9 EC Cells-The differential effects of PEA3 in F9 EC cells and F9-differentiated cells prompted us to examine whether Elf3, which is not expressed at significant levels in F9 EC cells (23), can stimulate the T␤R-II promoter when ectopically expressed in undifferentiated F9 EC cells. This possibility was examined by transiently transfecting F9 EC cells with mT␤R-IIϪ108/ ϩ56(B/A) and an Elf3 expression vector. Unlike the strong response to Elf3 in F9-differentiated cells (Fig. 1C), the response in F9 EC cells was very modest, ϳ2-fold, even at the highest amount tested (Fig. 5A). In contrast, Ets1 at a lower level of expression increased the activity of the reporter gene ϳ6-fold. As a control, we demonstrated that Elf3 was readily expressed in F9 EC cells. Moreover, we determined that the expression of Elf3 and Ets1 relative to one another in F9 EC cells was similar to their relative expression in F9-differentiated cells (data not shown). Thus, Ets1 can strongly stimulate the T␤R-II promoter in both F9 EC cells and their differentiated counterparts, whereas Elf3 only exerts a strong effect on this promoter in F9-differentiated cells. Therefore, like the effect of PEA3, the effect of Elf3 on the T␤R-II promoter appears to be cell context-dependent.
The minimal effect of Elf3 in these cells raised the question of whether Elf3 binds effectively to the T␤R-II promoter in F9 EC cells. In this connection, an autoinhibitory domain of Elf3 (23) could interfere with its binding to the T␤R-II promoter in F9 EC cells. This question was addressed by testing whether Elf3 would interfere with the ability of Ets1 to stimulate the activity of mT␤R-IIϪ108/ϩ56(B/A) in F9 EC cells. If Elf3 is able to bind to the T␤R-II promoter, but it is unable to stimulate its activity in these cells, it should reduce the response of mT␤R-IIϪ108/ϩ56(B/A) to Ets1. However, unlike PEA3 (Fig. 4B), Elf3 exerted little or no effect on the response to Ets1. In contrast, the truncated form of Elf3, Elf3⌬N233, which behaves as a dominant-negative (Ref. 23; also see Fig. 9B below), reduced the response to Ets1 (Fig. 5B) without reducing Ets1 expression (data not shown). Thus, it appears that Elf3 does not bind readily to the T␤R-II promoter when expressed ectopically in F9 EC cells.
Ets2 Does Not Stimulate the T␤R-II Promoter in F9-differentiated Cells-Given the potent effects of Elf3, Ets1, and PEA3 on the T␤R-II promoter in F9-differentiated cells, we examined the ability of Ets2 to stimulate the activity of mT␤R-IIϪ108/ ϩ56(B/A). Ets2 was of particular interest, because Ets2 (54), like Elf3 (23), is expressed in F9-differentiated cells. In addition, previous studies have shown that Ets2 is transcriptionally   FIG. 3. PEA3 can strongly stimulate the T␤R-II promoter. F9-differentiated cells were transiently transfected with 15 g of mT␤R-IIϪ108/ ϩ56(B/A) or mT␤R-IIϪ108/ϩ56(X/X) and 1 g of an internal control plasmid, pCMV-␤-gal. Where indicated, the cells were also transfected with an expression plasmid for PEA3 at the amounts shown. CAT and ␤-galactosidase activities were assayed and normalized, as described under "Experimental Procedures." The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ϩ56(B/ A), which was set to 1. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed three times, and similar results were obtained in each case. active in F9-differentiated cells (69) and we determined that the activity of the heterologous promoter/reporter gene construct Py2-Luc, which contains two ets sites (70), is stimulated when F9-differentiated cells are cotransfected with Ets2 (data not shown). However, overexpression of Ets2 in F9-differentiated cells exerted only a minimal effect on the activity of mT␤R-IIϪ108/ϩ56(B/A) (Fig. 6A). Moreover, this minimal effect appears to be nonspecific, because comparable increases were observed with the construct mT␤R-IIϪ108/ϩ56(X/X), in which both ets sites were disrupted. In addition, the lack of effect by Ets2 is not the result of its inability to be overexpressed in F9-differentiated cells. We have determined that GFP-Elf3 and GFP-Ets2 are expressed at comparable levels in F9-differentiated cells (data not shown). To examine the effect of Ets2 on the T␤R-II promoter more closely, we examined whether Ets2 would influence the ability of Elf3 to stimulate FIG. 4. PEA3 does not stimulate the mT␤R-II promoter in F9 EC cells. A, F9 EC cells were transiently transfected with 15 g of mT␤R-IIϪ108/ϩ56(B/A) and 1 g of an internal control plasmid pCMV-␤-gal. Where indicated, the cells were also transfected with an expression vector for Ets1 or PEA3 at the amounts shown. CAT and ␤-galactosidase activities were assayed and normalized, as described under "Experimental Procedures." The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ ϩ56(B/A), which was set to 1. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed three times, and similar results were obtained in each case. B, F9 EC cells were transiently transfected with 15 g of mT␤R-IIϪ108/ ϩ56(B/A) and 1 g of an internal control plasmid, pCMV-␤-gal. Where indicated, the cells were also transfected with an expression vector for Ets1 or PEA3 at the amounts shown. The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ϩ56(B/A), which was set to 1. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed three times, and similar results were obtained in each case. C, F9 EC cells were transiently transfected with 2.5 g of an expression vector for either YFP-Ets1 or YFP using LipofectAMINE and the PLUS reagent, as described under "Experimental Procedures." Where indicated, the cells were also transfected with 2.5 g of an expression vector for PEA3 or Elf3⌬N270. YFP fluorescence was measured by flow cytometry, as described under "Experimental Procedures." Positive cells were determined as the percentage of cells expressing fluorescence above that of the mock transfection control. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed twice, and similar results were obtained. the activity of mT␤R-IIϪ108/ϩ56(B/A). If Ets2 is able to bind to the ets sites in vivo, but it is unable to stimulate the T␤R-II promoter, then it should behave as a dominant-negative and block the response to Elf3. However, Ets2 did not reduce the response to Elf3 at any Ets2 concentration tested (Fig. 6B). This suggested that Ets2 is unable to bind to the ets sites of the T␤R-II promoter in vivo. This finding was interesting, because Ets2 is reported to contain an autoinhibitory domain, which strongly influences its binding to DNA in vitro (39). This prompted us to test whether a truncated form of Ets2, Ets2⌬N331, which contains primarily the DNA binding domain and the 26-amino acid C-terminal tail of Ets2, could interfere with the effect of Elf3 on the T␤R-II promoter. When Ets2⌬N331was overexpressed in conjunction with Elf3 in F9differentiated cells, we observed a dose-dependent inhibition of reporter gene activity (Fig. 6B). At the highest amount tested, Ets2⌬N331 reduced the activity of the mT␤R-IIϪ108/ϩ56(B/A) in the presence of Elf3 by ϳ80%. As a control, we verified that Ets2⌬N331 did not reduce the expression of Elf3 (Fig. 6C). Thus, these findings argue that full-length Ets2 does not stimulate the T␤R-II promoter in F9-differentiated cells, nor does it bind to the ets sites of this promoter in these cells.
To examine why Ets2 is unable to stimulate the T␤R-II promoter, we generated another truncated form of Ets2, Ets2⌬C453, which lacks 16 amino acids of its C-terminal tail. Ets2⌬C453 was tested, because previous studies reported that the C terminus of Ets2 contained an autoinhibitory domain and that removal of this region enabled recombinant Ets2 to bind in vitro to an ets site from the mb-1 gene (39). Using a heterologous promoter/reporter construct, which contains four ets sites from the stromelysin promoter, we determined that Ets2⌬C453, like its fulllength counterpart, is transcriptionally active (data not shown). However, Ets2⌬C453, like its parent protein, Ets2, exerted little effect on the activity of mT␤R-IIϪ108/ϩ56(B/A) in F9-differentiated cells (Fig. 7). Importantly, it also failed to reduce the response to Elf3. If Ets2⌬C453 was able to bind to the T␤R-II promoter in these cells, it would have reduced the response to Elf3. Thus, Ets2⌬C453 does not appear to bind to the ets sites of the T␤R-II promoter in F9-differentiated cells.
Ets2 Only Binds Effectively in Vitro to One of the Ets Sites in the T␤R-II Promoter-In view of the fact that Ets2⌬N331 blocked the response to Elf3 (Fig. 6B), but Ets2⌬C453 did not (Fig. 7), we examined the in vitro binding of different forms of Ets2 to DNA. For this purpose, Ets2, Ets2⌬C453, and Ets2⌬N331 were expressed separately in 293T cells, and nuclear extracts from these cells were used in EMSA with the 30-base pair DNA probe (B/A) that contains the two overlapping ets sites and flanking sequence from the promoter of the mouse T␤R-II gene. As reported earlier with a DNA probe that contained an ets site from the mb-1 gene (39), Ets2 exhibited no binding to the B/A probe (Fig. 8A). Furthermore, Ets2⌬C453 did not form a DNA-protein complex with the B/A probe (data not shown). In direct contrast, Ets2⌬N331 formed a prominent DNA-protein complex with the B/A probe, when tested at the same concentration used for Ets2 (Fig. 8A).
Surprisingly, Ets2⌬N331 only formed a single DNA-protein complex with the B/A DNA probe, which contains the two ets sites of the T␤R-II promoter. This differs markedly from the binding of Elf3⌬N270 to the B/A DNA probe. As shown previously, Elf3⌬N270, which lacks the autoinhibitory domain of Elf3, forms both a binary and a ternary DNA-protein complex with the B/A DNA probe (Fig. 8B), and both of these DNAprotein complexes are shifted with the M2 antibody that recognizes the N-terminal flag tag of Elf3⌬N270 (23). To better understand the reason behind the difference in binding between Ets2⌬N331 and Elf3⌬N270, two additional DNA probes, B/X and X/A, were used. The B/X probe contains the sequence of the B-site with the sequence of the A-site disrupted. The X/A probe contains the sequence of the A-site with the sequence of the B-site disrupted. (The bases changed in these two probes were identical to those used to create the corresponding promoter/reporter gene constructs, mT␤R-IIϪ108/ϩ56(B/X) and mT␤R-IIϪ108/ϩ56(X/A).) When the binding of equal amounts of Ets2⌬N331 and Elf3⌬N270 were compared using these probes and the B/A probe, Elf3⌬N270 formed DNA-protein complexes with each of the probes (Fig. 8B). However, Ets2⌬N331 formed a single DNA-protein complex with the B/A and B/X probes, but exhibited little if any binding to the X/A probe (Fig. 8B). This argues that Ets2⌬N331 exhibits a significantly lower affinity for the A-site than it does for the B-site of the T␤R-II promoter. Given that both ets sites are required for the synergistic effects of Elf3 on the T␤R-II promoter (Fig. 1C), FIG. 5. Elf3 does not strongly stimulate the mT␤R-II promoter in F9 EC cells. A, F9 EC cells were transiently transfected with 15 g of mT␤R-IIϪ108/ϩ56(B/A) and 1 g of an internal control plasmid, pCMV-␤-gal. Where indicated, the cells were also transfected with an expression vector for Elf3 or Ets1 at the amounts shown. CAT and ␤-galactosidase activities were assayed and normalized, as described under "Experimental Procedures." The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ϩ56(B/A), which was set to 1. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed four times, and similar results were obtained in each case. B, F9 EC cells were transiently transfected with 15 g of mT␤R-IIϪ108/ϩ56(B/A) and 1 g of an internal control plasmid, pCMV-␤-gal. Where indicated the cells were also transfected with an expression vector for Ets1, Elf3, or Elf3⌬N233 at the amounts shown. The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ϩ56(B/A), which was set to 1. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed three times, and similar results were obtained in each case.

FIG. 6. Ets2 does not stimulate T␤R-II promoter in F9-differentiated cells.
A, F9-differentiated cells were transiently transfected with 15 g of the mT␤R-II promoter/reporter construct shown and 1 g of an internal control plasmid, pCMV-␤-gal. Where indicated the cells were also transfected with an expression vector for Elf3 or Ets2 at the amounts shown. Activities were assayed and normalized as described under "Experimental Procedures." Data shown are means and S.D. for duplicate samples. This experiment was performed three times, and similar results were obtained. B, F9-differentiated cells were transiently transfected with 15 g of mT␤R-IIϪ108/ϩ56(B/A) and 1 g of an internal control plasmid, pCMV-␤-gal. Where indicated, the cells were also transfected with 1 g of Elf3 and Ets2 or Ets2⌬N331 at the amounts shown. CAT and ␤-galactosidase activities were assayed and normalized, as described under "Experimental Procedures." The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ϩ56(B/A), which was set to 1. Data shown are means and S.D. for duplicate samples from a representative experiment. This experiment was performed twice, and similar results were obtained. C, F9-differentiated cells were transiently transfected with 2 g of the expression vector for GFP-Elf3 or GFP as shown using LipofectAMINE and the PLUS reagent, as described under "Experimental Procedures." Where indicated, the cells were also transfected with 2 g of the Ets2⌬N331 expression vector. GFP fluorescence was measured by flow cytometry, as described under "Experimental Procedures." Positive cells were determined as the percentage of cells expressing fluorescence above that of the mock transfection control. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed twice, and similar results were obtained. this raised the possibility that the reason Ets2 is unable to stimulate the T␤R-II promoter in F9-differentiated cells is because it cannot bind effectively to the ets sites of this promoter.
Ets2 Does Not Activate a T␤R-II Promoter/Reporter Gene Construct Containing Two Consensus Ets Sites-To determine whether the inability of Ets2 to activate the T␤R-II promoter was because of low affinity for the A-site, we performed two additional studies. First, we examined the in vitro binding of Ets2⌬N331 to a DNA probe, referred to as the B/B probe, in which the sequence of the A-site in the B/A DNA probe was changed to a second B-site. Ets2⌬N331 formed both a binary and a ternary complex with this probe, whereas Ets2⌬C453 and Ets2 failed to form any detectable DNA-protein complexes (Fig. 8C). As expected, Elf3⌬N270 also formed both a binary and a ternary complex with the B/B probe in an EMSA (data not shown).
The finding that Ets2⌬N331 formed a ternary complex with the B/B probe formed the basis for the design of the second study, in which we examined the ability of Ets2 to stimulate the activity of the mT␤R-IIϪ108/ϩ56(B/B) construct in F9differentiated cells. This promoter/reporter gene construct contains a second B-site in place of the A-site. Not surprisingly, the basal activity of mT␤R-IIϪ108/ϩ56(B/B), in the absence of exogenously expressed Ets proteins, was elevated ϳ8-fold in comparison to the parent mT␤R-IIϪ108/ϩ56(B/A) construct (Fig. 9A). This elevated activity is apparently because of activation of the T␤R-II promoter by endogenous proteins, because overexpression of Elf3⌬N270 or Elf3⌬N233, which we previously demonstrated behave as a dominant-negative (23), reduced the basal activity of the mT␤R-IIϪ108/ϩ56(B/B) construct (Fig. 9B). Interestingly, overexpression of Elf3 in F9differentiated cells increased the activity of the mT␤R-IIϪ108/ ϩ56(B/B) construct ϳ5-fold, whereas overexpression of Ets2 exerted little or no effect on mT␤R-IIϪ108/ϩ56(B/B) in F9differentiated cells (Fig. 9A). Thus, the inability of Ets2 to stimulate the T␤R-II promoter in F9-differentiated cells does not appear to be the result of its inability to bind to the A-site in vivo.

DISCUSSION
The findings described in this study demonstrate that the T␤R-II promoter is not only strongly stimulated by Elf3, it is also highly responsive to Ets1 and PEA3 in F9-differentiated cells. Each of these Ets proteins mediates its effects via the two ets sites, which are conserved in the mouse and the human T␤R-II gene. It is evident that both ets sites are functional and that both sites are required for the synergistic stimulation of the T␤R-II promoter by Elf3, Ets1, and PEA3. Moreover, we demonstrate that five Ets family members, Elf3, Ets1, PU.1, Ets2, and PEA3, influence the activity of the T␤R-II promoter in distinctly different ways. Differential expression of these Ets proteins in F9 EC cells and their differentiated counterparts is one of the important mechanisms for determining which Ets protein regulates the T␤R-II promoter. Importantly, we demonstrate that the effects of at least two of the five Ets proteins, Elf3 and PEA3, are cell type-dependent. We also demonstrate that there are important differences in the ability of two Ets proteins, Elf3 and Ets2, which are both expressed in F9-differentiated cells, to bind to the two ets sites of the T␤R-II promoter, both in vitro and in vivo. Moreover, our studies argue that differential binding of Ets proteins to the two ets sites can contribute significantly to the differential activation of the T␤R-II promoter (see below). Hence, nature uses a wide variety of mechanisms to control which members of the Ets family of transcription factors regulate the T␤R-II promoter.
Two previous studies demonstrated that the activity of the T␤R-II promoter was reduced significantly when both ets sites were disrupted in combination (17,23). Neither study specifically examined the effect of disrupting each site individually. In the case of the human T␤R-II promoter/reporter gene constructs, mutations introduced to disrupt the A-site also introduced changes in the B-site (17). As a consequence, the earlier data appear to indicate that a larger reduction in promoter activity occurs when the A-site is disrupted. In the work presented in this report, care was taken to disrupt each site separately. Given that the B-site matches a 9-base pair consensus sequence for an ets site, but that the sequence of the A-site differs by 2 base pairs of the 9, it was not surprising that a larger decrease in promoter activity was observed with the mT␤R-IIϪ108/ϩ56(X/A) construct than with the mT␤R-IIϪ108/ϩ56(B/X) construct, 60% versus 40% (Fig. 1B). Furthermore, disruption of both sites in mT␤R-IIϪ108/ϩ56(X/X) reduced promoter activity by at least 80%, which is similar to what had been observed in mouse T␤R-II promoter/reporter gene constructs when the region 3-56 was deleted (23). What is particularly interesting is the effect of overexpressing Elf3 on the activity of these constructs. Although the T␤R-II promoter is stimulated more than 20-fold by Elf3 when both sites are intact, disruption of either site reduces the response of the promoter to less than 5-fold. This argues that the two ets sites work cooperatively to respond synergistically to Elf3, as well as Ets1 and PEA3. The mechanism(s) responsible for this cooperative effect has not been determined, but it may involve cooperative binding of these Ets proteins to the T␤R-II promoter. In this regard, Ets1 has been shown to bind cooperatively to two adjacent ets sites in the stromelysin promoter (71).
It is evident from the work presented in this study that the two overlapping ets sites differ in at least two important respects. First, as discussed above, the B-site matches the sequence of a consensus ets site, whereas the A-site differs by 2 base pairs from the consensus sequence. Not surprisingly, the 2 base pairs in the A-site that differ from the consensus sequence are outside the essential core 5Ј-GGA(A/T)-3Ј sequence (Fig. 1A). Interestingly, conversion of the A-site to a second B-site elevates the expression of the activity of our mT␤R-IIϪ108/ϩ56 promoter/reporter construct 8-fold (Fig. 9A). Thus, we suspect that the presence of a nonconsensus ets site limits the activity of the endogenous T␤R-II promoter and helps to FIG. 7. A C-terminal tail truncated form of Ets2 is unable to stimulate the T␤R-II promoter. F9-differentiated cells were transiently transfected with 15 g of mT␤R-IIϪ108/ϩ56(B/A) or mT␤R-IIϪ108/ϩ56(X/X) and 1 g of an internal control plasmid, pCMV-␤-gal. Where indicated the cells were also transfected with an Elf3 expression and/or an Ets2⌬C453 expression vector at the amounts shown. CAT and ␤-galactosidase activities were assayed and normalized, as described under "Experimental Procedures." The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ϩ56(B/A), which was set to 1. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed three times, and similar results were obtained in each case.
achieve the appropriate level of T␤R-II transcription in vivo. Given the potent and pleiotropic effects of TGF-␤, elevating T␤R-II expression would likely be detrimental to normal cellular physiology.
A second potentially important difference between the B-site and A-site is the differential binding of Ets2. Although it is evident that Ets2⌬N331 can bind to the B-site, it exhibits significantly lower affinity for the A-site. In this regard, when Elf3⌬N270 and Ets2⌬N331 were tested at the same concentration, Elf3⌬N270 formed a prominent DNA-protein complex with the X/A probe, whereas the DNA-protein complex formed between Ets2⌬N331 and the X/A probe was barely detectable (Fig. 8B). In addition, Elf3⌬N270 formed a ternary DNA-protein complex in an EMSA with a DNA probe that contained both ets sites, but Ets2⌬N331 formed only a binary DNAprotein complex. Given that both ets sites in the T␤R-II pro-  1 or B-2), and ternary complex (T) are shown. Nuclear extract from mock-transfected 293T cells was added to the control lane. C, EMSA was performed with nuclear extracts prepared from 293T cells transfected with an expression plasmid for Ets2, Ets2⌬C453, or Ets2⌬N331. Nuclear extract from mock-transfected 293T cells was added to lane 1. Nuclear extract from 293T cells transfected with Ets2, Ets2⌬N331, and Ets2⌬C453 were added to lanes 2, 3, and 4, respectively. Equal amounts of the flag-tagged Ets2 proteins were added to the reaction mixtures. The DNA probe used contained the sequence for two B-sites (B/B). The positions of the free probe (P), binary complex (B), and ternary complex (T) are shown. moter are needed for maximal response to Elf3, PEA3, and Ets1, this raised the possibility that the low affinity of Ets2 for the A-site may be responsible for its inability to stimulate T␤R-II promoter activity.
To test whether the A-site in the T␤R-II promoter contributed significantly to the lack of response to Ets2, we generated the T␤R-II promoter/reporter gene construct, mT␤R-IIϪ108/ ϩ56(B/B), which contains a second B-site in place of the A-site. This substitution enhanced the activity of the promoter ϳ8fold. Elf3 further stimulated the activity of this construct another 5-fold. In contrast, Ets2 exerted little, if any, effect on the modified B/B construct. Moreover, our findings argue that Ets2 cannot bind in vivo to the ets sites of the T␤R-II gene in F9-differentiated cells. Earlier studies reported that removal of 16 amino acids from the C-terminal tail of Ets2 eliminates an autoinhibitory domain of Ets2 and enhances its binding to DNA in vitro (39). These studies used Ets2 proteins produced by in vitro transcription/translation. However, the same truncated form of Ets2 produced in 293T cells exhibited virtually no binding to the ets sites of the T␤R-II promoter in vitro. In contrast, Ets2⌬N331, which lacks a large region that is Nterminal of the DNA binding domain, bound readily to the B-site of the T␤R-II promoter. Thus, these findings parallel the apparent binding of Ets2 proteins when overexpressed in F9differentiated cells. More specifically, Ets2 and Ets2⌬C453 did not activate the T␤R-II promoter, nor did they interfere with the ability of Elf3 to stimulate the T␤R-II promoter (Figs. 6B and 7). In direct contrast, Ets2⌬N331 behaved as a dominantnegative and blocked the response to Elf3 (Fig. 6B). This suggests that the failure of Ets2 to strongly stimulate the activity of the T␤R-II promoter in F9-differentiated cells (Fig. 6A) is the result primarily of its inability to bind to the ets sites of T␤R-II promoter both in vitro and in vivo and not of its low affinity for the A-site.
The failure of Ets2 to bind to the ets sites of the T␤R-II promoter in F9-differentiated cells raises an important question. Why does Ets1 bind to the ets sites of the T␤R-II promoter, but Ets2 does not? Ets1 and Ets2 are closely related evolutionarily (31). Their DNA binding domains differ by only 3 amino acids, and these Ets proteins are ϳ57% identical overall. One possible explanation for the differential binding of Ets1 and Ets2 could be that there are subtle, but important, differences in the tertiary structures of their DNA binding domains. Such a difference could be a result of the fact that their DNA binding domains differ by 3 amino acids or more likely could result from interactions with other regions of the protein. An alternative explanation for the differential effects of Ets1 and Ets2 is the availability of factors that interact with Ets proteins to promote their binding to DNA in vivo. Thus far, at least eight Ets proteins, including Ets1 and Ets2, have been reported to possess an autoinhibitory domain, which influence their binding to DNA (23, 39 -44). In the case of Ets1, CBF-␣2 (also known as FIG. 9. Ets2 is unable to stimulate a T␤R-II promoter/reporter gene construct containing two B-sites. A, F9differentiated cells were transiently transfected with 15 g of mT␤R-IIϪ108/ ϩ56(B/A) or mT␤R-IIϪ108/ϩ56(B/B) and 1 g of an internal control plasmid, pCMV-␤-gal. Where indicated, the cells were also transfected with an expression vector for Elf3 or Ets2 at the amounts shown. CAT and ␤-galactosidase activities were assayed and normalized, as described under "Experimental Procedures." The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ ϩ56(B/A), which was set to 1. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed three times, and similar results were obtained in each case. B, F9-differentiated cells were transiently transfected with 15 g of mT␤R-IIϪ108/ϩ56(B/B) and 1 g of an internal control plasmid pCMV-␤-gal. Where indicated, the cells were also transfected with an expression vector for Elf3⌬N270 or Elf3⌬N233 at amounts shown. The promoter activity of each construct is calculated relative to the CAT activity observed with mT␤R-IIϪ108/ϩ56(B/B), which was set to 1. Data shown are means and S.D. for duplicate samples from one representative experiment. This experiment was performed twice, and similar results were obtained in each case. AML1 or PEBP2␣B) has been shown to interact directly with the autoinhibitory domain of Ets1 and to enhance Ets1 binding to DNA (45). Furthermore, there is a reciprocal effect of Ets1 on the binding of CBF-␣2 to DNA. Thus, an explanation for the differential effects of Ets1 and Ets2 on the T␤R-II promoter may be the absence of a transcription factor in F9-differentiated cells that is able to cooperate with Ets2 and promote its binding to T␤R-II promoter. If this is the case, it is also reasonable to speculate that F9-differentiated cells express transcription factors that, although unable to interact with Ets2, do interact effectively with Ets1 and Elf3 to promote their binding to the T␤R-II promoter. Moreover, the failure of Elf3 to strongly stimulate the T␤R-II promoter in F9 EC cells (Fig. 5A) or to interfere with the ability of Ets1 to stimulate the T␤R-II promoter in these cells (Fig. 5B) may be because of the lack of an Elf3-cooperating factor in F9 EC cells. In contrast, one or more factors that interact with Ets1 may be present in F9 EC cells that enable it to stimulate the T␤R-II promoter. Thus far, only a few factors have been shown to interact with Ets1 and promote its binding to DNA (45,47). To date, no factors have been identified that promote Elf3 binding to DNA. Further study will be needed to address this issue.
Our findings with PEA3, Elf3, and PU.1 illustrate that other mechanisms are also likely to be used by nature to limit the ability of Ets proteins to stimulate T␤R-II promoter activity. Although PEA3 can stimulate the activity of the T␤R-II promoter in F9-differentiated cells, it is not expressed in these cells. In direct contrast, PEA3 does not stimulate the T␤R-II promoter in F9 EC cells where it is expressed. In fact, our findings suggest that PEA3 may help repress the transcription of the endogenous T␤R-II gene in the parental F9 EC cells (Fig.  4A). By analogy with nuclear receptors in the presence and absence of hormone (60), when PEA3 binds to the T␤R-II promoter in F9-differentiated cells, it may associate with co-activators that increase the activity of the T␤R-II promoter, whereas in F9 EC cells, PEA3 may inhibit promoter activity by interacting with co-repressors rather than co-activators. Regardless of the mechanism, our findings indicate that the effects of PEA3 are cell context-dependent. Interestingly, this is also true for Elf3. However, the mechanisms involved appear to differ for these two Ets proteins. PEA3 can bind to the T␤R-II promoter in F9 EC cells (Fig. 4B). In contrast, we propose that Elf3 exhibits little or no binding to the T␤R-II promoter in these cells, because Elf3 did not interfere with the ability of Ets1 to stimulate this promoter in F9 EC cells (Fig. 5B). Finally, the failure of PU.1 to stimulate the T␤R-II promoter (Fig.  2B) may result from its inability to interact effectively with appropriate co-activators in F9-differentiated cells. Alternatively, our findings with PU.1 may reflect promoter-specific regulation of the T␤R-II promoter by different Ets proteins. Further study will be required to resolve this issue.
In conclusion, the work reported in this study demonstrates that more than one Ets protein can stimulate the activity of the T␤R-II promoter. Given the essential roles played by TGF-␤ in normal and diseased tissues, it will be important to examine whether different Ets proteins, in particular Ets1, are able to stimulate the expression of the T␤R-II gene in nonepithelial cells, including the cells of the immune system, which are often suppressed by TGF-␤ in diseases, such as cancer. Our work also demonstrates that the five Ets proteins tested in this study exhibit important biochemical and cellular differences in the way they influence the activity of the T␤R-II promoter. The insights gained in this study are no doubt applicable to the mechanisms used to generate specificity in the activation of other genes by this large family of transcription factors. Given that the DNA binding domains of Ets proteins are structurally very similar, and that several Ets proteins are often expressed in the same cell, it is not surprising that nature has developed multiple mechanisms to prevent promiscuous access of Ets proteins to promoters of specific genes.