Transcriptional Activation of the Type II Transforming Growth Factor-β Receptor Gene upon Differentiation of Embryonal Carcinoma Cells*

Previously, it has been shown that differentiation of embryonal carcinoma (EC) cells turns on the expression of functional transforming growth factor type-β receptors. Here, we show that the type II receptor (TβR-II) gene is activated at the transcriptional level when EC cells differentiate. We show that the differentiated cells, but not the parental EC cells, express transcripts for TβR-II. In addition, the expression of TβR-II promoter/reporter gene constructs are elevated dramatically when EC cells differentiate and we identify at least two positive and two negative regulatory regions in the 5′ flanking region of the TβR-II gene. Moreover, we identify a cAMP response element/activating transcription factor site that acts as a positive cis-regulatory element in the TβR-IIpromoter, and we demonstrate that the transcription factor ATF-1 binds to this site and strongly stimulates the expression of theTβR-II promoter/reporter gene constructs when ATF-1 is overexpressed in EC-derived differentiated cells. Equally important, we identify a negative regulatory element in a 53-base pair region that had previously been shown to inhibit strongly the expression ofTβR-II promoter/reporter gene constructs. Specifically, we demonstrate that this region, which contains an inverted CCAAT box motif, binds the transcription factor complex NF-Y (also referred to as CBF) in vitro. Furthermore, expression of a dominant-negative NF-YA mutant protein, which prevents DNA binding by NF-Y, enhances TβR-II promoter expression. Together, these studies suggest that the transcription factors ATF-1 and NF-Y play important roles in the regulation of the TβR-IIgene.

Previously, it has been shown that differentiation of embryonal carcinoma (EC) cells turns on the expression of functional transforming growth factor type-␤ receptors. Here, we show that the type II receptor (T␤R-II) gene is activated at the transcriptional level when EC cells differentiate. We show that the differentiated cells, but not the parental EC cells, express transcripts for T␤R-II. In addition, the expression of T␤R-II promoter/ reporter gene constructs are elevated dramatically when EC cells differentiate and we identify at least two positive and two negative regulatory regions in the 5 flanking region of the T␤R-II gene. Moreover, we identify a cAMP response element/activating transcription factor site that acts as a positive cis-regulatory element in the T␤R-II promoter, and we demonstrate that the transcription factor ATF-1 binds to this site and strongly stimulates the expression of the T␤R-II promoter/reporter gene constructs when ATF-1 is overexpressed in EC-derived differentiated cells. Equally important, we identify a negative regulatory element in a 53-base pair region that had previously been shown to inhibit strongly the expression of T␤R-II promoter/reporter gene constructs. Specifically, we demonstrate that this region, which contains an inverted CCAAT box motif, binds the transcription factor complex NF-Y (also referred to as CBF) in vitro. Furthermore, expression of a dominant-negative NF-YA mutant protein, which prevents DNA binding by NF-Y, enhances T␤R-II promoter expression. Together, these studies suggest that the transcription factors ATF-1 and NF-Y play important roles in the regulation of the T␤R-II gene.
Transforming growth factor type-␤ (TGF-␤) 1 refers to a com-plex family of genetically distinct polypeptides that are secreted by virtually all cells and which exert potent effects on cell proliferation, differentiation, extracellular matrix production, and immunoregulation (1)(2)(3)(4). Based on these activities and the defined spatial and temporal pattern of expression of the three mammalian isoforms of TGF-␤ (TGF-␤1, TGF-␤2, and TGF-␤3) during mouse embryogenesis, it has been argued that the TGF-␤s play important roles in the generation and modification of extracellular signals that direct critical processes during mammalian development (5)(6)(7)(8)(9)(10)(11)(12). This is borne out by the induction of fetal defects in embryos that are unable to produce the different isoforms of TGF-␤ (13)(14)(15)(16)(17), in particular TGF-␤2, and in embryos that cannot produce functional TGF-␤ receptors due to inactivation of the T␤R-II gene by gene targeting (18). Given the importance of TGF-␤ during development, it is not surprising that defects in TGF-␤ signal transduction have also been implicated in the pathological processes of many diseases, including arthritis, ulcerative diseases, atherosclerosis, and glomerulonephritis (2,19). Equally important, cells that lose the ability to respond to TGF-␤ are more likely to exhibit uncontrolled growth and to undergo neoplastic transformation (20 -27).
TGF-␤s primarily exert their biological effects through interactions with three distinct high affinity TGF-␤ cell surface receptors (designated types I, II, and III). All three receptors have been cloned and characterized (28 -33). Both 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. The type III (T␤R-III) receptor (also referred to as betaglycan) is a transmembrane proteoglycan devoid of intrinsic signaling ability, but which may act to present ligand to other signaling receptors (34). The most commonly held model for the activation of the TGF-␤ signal transduction cascade proposes the selective binding of TGF-␤ to the type II receptor, a constitutively active kinase (35). Ligand binding to T␤R-II induces recruitment of T␤R-I into a stable complex. Once this complex is formed, T␤R-II transphosphorylates T␤R-I at serine and threonine residues, resulting in signal transduction to downstream substrates. Thus, loss of responsiveness to TGF-␤ could result from changes in the expression of functional TGF-␤ receptors rather than defects in the expression or activation of TGF-␤ ligand.
Efforts to define the roles of TGF-␤ and their receptors have involved the study of embryonal carcinoma (EC) cells, as they are a model system used frequently for studying early mammalian development (36). EC cells resemble cells of the early mouse embryo morphologically and biochemically. Moreover, they can be induced to differentiate into many of the cell types formed during mammalian embryogenesis (37), making them particularly well suited for the investigation of the signal transduction events involved in cellular differentiation. Furthermore, they provide a useful tool for the identification of mechanisms involved in carcinogenesis, as EC cells are tumorigenic whereas their differentiated cells are largely non-tumorigenic. Using this model system, we demonstrated that EC cells do not express detectable cell surface receptors for TGF-␤ until after they are induced to differentiate (38). Equally important, the up-regulation in the expression of TGF-␤ receptors by the EC-derived differentiated cells coincides with the ability of exogenous TGF-␤ to inhibit their proliferation as well as to their loss of tumorigenic potential (38,39). In the present study, we examined the expression of the T␤R-II gene both in F9 EC cells and their differentiated counterparts. Our findings demonstrate that few, if any, T␤R-II transcripts are expressed by F9 EC cells, whereas there is a dramatic increase in the expression of T␤R-II mRNA when F9 EC cells are induced to differentiate. This observation is supported by the differences in expression of various T␤R-II promoter/reporter gene constructs in EC cells and their differentiated counterparts, arguing strongly that the large increase in the steady-state levels of T␤R-II mRNA that accompany differentiation is due, at least in part, to an increase in the transcription of the T␤R-II gene promoter. In addition, our results identify both positive and negative regulatory regions in the T␤R-II promoter that appear to contribute significantly to the transcriptional activity of the T␤R-II gene in both EC and EC-derived differentiated cells. In this regard, mutation of a CRE/ATF motif (Ϫ196 to Ϫ185) within one of the positive regulatory regions reduces substantially T␤R-II promoter activity in the EC-derived differentiated cells. Gel mobility shift analyses demonstrates that the transcription factor ATF-1 is able to bind to this CRE/ATF motif in vitro. We demonstrate further that expression of ATF-1 in vivo up-regulates T␤R-II promoter activity in the differentiated cells, most likely through the CRE/ATF motif. Conversely, we have identified an inverted CCAAT box motif (Ϫ83 to Ϫ74) that appears to negatively influence the transcriptional activity of the T␤R-II promoter in both EC cells and their differentiated counterparts. The T␤R-II CCAAT box motif is bound in vitro by the transcription factor complex, NF-Y (also referred to as CBF) in both cell types. Importantly, we demonstrate that expression of a dominant-negative NF-YA mutant protein, which prevents DNA binding of the NF-Y transcription factor complex, specifically enhances the expression of T␤R-II promoter/reporter gene constructs in both F9 EC cells and their differentiated cells. Together, these studies suggest that ATF-1 and NF-Y play important roles in the regulation of the T␤R-II gene. Filters were prehybridized for 3 h at 42°C in a solution consisting of 5ϫ SSPE, 5ϫ Denhardt's solution, 50% deionized formamide, 1% SDS, and 100 g/ml denatured salmon testis DNA. Hybridization was performed in the same buffer with 2 ϫ 10 6 cpm/ml of 32 P-labeled probe for 16 -18 h at 42°C. The 32 P-labeled probe was obtained by SacI digestion of the human T␤R-II cDNA clone, H2-3FF (30). The approximately 1.5-kilobase pair fragment was radioactively labeled with [␥-32 P]dCTP using a random primed DNA labeling kit (Boehringer Mannheim GmbH, Mannheim, Germany) to a specific activity of 1-2 ϫ 10 9 cpm/g. Following hybridization, the membranes were washed twice in 2ϫ SSPE, 0.1% SDS for 10 min at room temperature followed by a single wash in 0.5ϫ SSPE, 0.1% SDS for 15 min at 50°C. Filters were autoradiographed at Ϫ80°C with Kodak X-Omat AR film. All prehybridization, hybridization and wash conditions were the same for hybridizations with the hGAPDH normalization probe. The 780-base pair probe was obtained by PstI and XbaI digestion of the plasmid clone, HcGAP (ATCC, Rockville, MD) and radioactively labeled as described above.

Cells and
Expression Plasmids-T␤R-II promoter/chloramphenicol acetyltransferase (CAT) reporter gene expression plasmids were generated and amplified by polymerase chain reaction using genomic DNA containing the 5Ј-untranslated region of the human T␤R-II gene as a template and cloned into pGEM4-SVOCAT (42). The constructs were named pT␤RII-n, where n is the distance in nucleotides from the transcription initiation site identified by Humphries et al. (43) and Bae et al. (42). The expression plasmids pECEATF-1 and pECEATF-2 were provided by Dr. Michael O'Reilly (University of Rochester, Rochester, NY). These plasmids contained the human ATF-1 and ATF-2 cDNAs (44) inserted into the expression plasmid pECE (45). The eukaryotic expression plasmid pNFYA29 was obtained with permission of Dr. R. Mantovani from Dr. Peter Edwards (UCLA School of Medicine). This plasmid uses an SV40 promoter to drive the expression of the NFYA29 mutant protein (46). The normalization plasmid, pCMV-␤ (CLON-TECH) contains the ␤-galactosidase reporter gene under the control of the CMV immediate early promoter (47). The plasmid pDOL-CMV-CAT contains the CAT reporter gene under the control of the CMV immediate early promoter (48) and was used as a positive control to monitor the general transcriptional activity of F9 EC and F9-differentiated cells. All plasmids were verified by sequencing, and purified by Qiagen tip-500 columns.
Transient Transfection Assay-F9 EC cells, F9-differentiated cells (day 3), PYS-2 cells, and PSA-5E cells were transfected by the calcium phosphate precipitation method as described previously (49). Typically, 20 g of each T␤R-II promoter/CAT plasmid was co-transfected with 2 g of the internal standard, pCMV-␤. After an overnight incubation with the DNA-CaPO 4 precipitate, the cells were washed twice with serum-free medium and refed with Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. CAT activities were determined 48 h after transfection by the method of Seed and Sheen (50) and normalized to ␤-galactosidase activity by the method of Rosenthal (51) to adjust for differences in transfection efficiency (52). In some experiments, 10 g of pT␤RII-n constructs were co-transfected with optimal concentrations of the expression plasmids containing human ATF-1 or ATF-2 cDNAs or the dominant negative mutant, NFYA29.
For gel mobility supershift analyses with the wild-type T␤R-II Ϫ210 to Ϫ185 probe, the binding reactions were incubated for 4 h at 4°C with 4 g of the antibodies indicated, prior to the addition of the radiolabeled probe. The ATF-1 specific monoclonal antibody was generated against recombinant ATF-1 and reacts only with complexes containing ATF-1. This antibody was kindly provided by Dr. Steven Hinrichs (University of Nebraska Medical Center, Omaha NE). Polyclonal antibodies specific to ATF-2, c-Jun, CREB, and CREM were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). For gel mobility supershift analyses with the wild-type T␤R-II Ϫ104 to Ϫ67 probe, the binding reactions were incubated for 1 h at 4°C with 1 g of the antibodies indicated, prior to the addition of the radiolabeled probe. The NF-YA antibody and nonspecific IgG antibodies were purchased from Rockland Inc. (Gilbertsville, PA). The nondenaturing 4% polyacrylamide gels (30:1, acrylamide:bisacrylamide) were electrophoresed at 120 V for 3-5 h at 4°C in high ionic strength buffer containing 50 mM Tris, 100 mM glycine, and 2 mM EDTA, then dried, and subjected to autoradiography.

Expression of T␤R-II mRNA in F9 EC and F9-differentiated
Cells-T␤R-II expression is normally associated with ligand binding and growth responsiveness of cells to TGF-␤ (21-23, 58). To identify possible mechanisms for the significant increase in TGF-␤ ligand binding and growth responsiveness that is observed when EC cells are induced to differentiate (38), we initially examined mRNA expression of the T␤R-II gene in F9 EC cells and their differentiated counterparts. Northern blot analysis of poly(A) ϩ RNA from undifferentiated F9 EC cells detected little or no transcript of approximately 5.2 kilobases that corresponds to the size predicted for the T␤R-II gene ( Fig. 1) (30). However, the intensity of this transcript increased substantially (Ͼ15-fold) after F9 EC cells were induced to differentiate with RA ( Fig. 1). Thus, it appears that there is a strong correlation between the expression of T␤R-II mRNA in F9 EC cells and F9-differentiated cells and the ability of each cell type to bind and respond to TGF-␤.
Transcriptional Regulation of the T␤R-II Gene-Due to the large increase in the steady-state levels of T␤R-II mRNA when F9 EC cells are induced to differentiate, we examined the transcriptional activity of the T␤R-II gene promoter in F9 EC cells and their differentiated counterparts. These studies employed chimeric gene constructs in which various amounts of the 5Ј-flanking region of the human T␤R-II gene were inserted upstream of the CAT reporter gene in the plasmid pGEM-SVOCAT (42). Transient transfection of HepG2 cells with these constructs had previously identified several distinct regulatory regions in the T␤R-II promoter including: two positive regulatory regions (Ϫ219 to Ϫ172 and ϩ1 to ϩ35), two negative regulatory regions (Ϫ1240 to Ϫ504 and Ϫ100 to Ϫ67), and the core promoter region (Ϫ47 to Ϫ1) (42). In the current study, the T␤R-II promoter-CAT constructs, pT␤RII-1883/ϩ50, pT␤RII-274/ϩ50, and pT␤RII-137/ϩ50 were transiently transfected into F9 EC cells. (These constructs contain a common 3Ј end located at ϩ50 in relationship to the primary T␤R-II transcription start site (42,43), and increasing amounts of the T␤R-II 5Ј-flanking sequence ranging from nucleotide Ϫ1883 to Ϫ137.) Consistent with the virtual absence of T␤R-II transcripts in EC cells (Fig. 1), virtually no CAT activity was detected over background (Fig. 2). This result is unlikely to be explained by low transfection efficiency, as both the normalizing plasmid, pCMV-␤, and positive control plasmid, pDOL-CMV-CAT, are expressed strongly (data not shown). In stark contrast to our observations in F9 EC cells, all of the constructs expressed substantially greater levels of CAT activity in the F9-differentiated cells (Fig. 2). Equally important, the 9-fold increase in transcriptional activity of pT␤RII-274/ϩ50 when compared with pT␤RII-137/ϩ50 suggests the presence of a strong positive regulatory element(s) located in the region between Ϫ137 and Ϫ274 of the T␤R-II gene promoter. Furthermore, the 2-fold decrease in transcription of the pT␤RII-1883/ϩ50 construct relative to pT␤RII-274/ϩ50 points to a weak negative regulatory element(s) located in the region between Ϫ274 and Ϫ1883 (Fig. 2).
To ensure our observations were not unique to F9-differentiated cells, we also examined the expression of these constructs in two stable EC-derived differentiated cell lines, PYS-2 (parietal endoderm-like) and PSA-5E (visceral endoderm-like). For the most part, these cell lines demonstrated a pattern of expression for each of the constructs that is similar to that observed for the F9-differentiated cells (Fig. 2). One notable difference is that the negative regulatory region between Ϫ274 and Ϫ1883 appears to have a stronger influence on the expression of the reporter gene in both the PYS-2 and PSA-5E cells than in the F9-differentiated cells.
Identification of a Cis-regulatory Element That Elevates T␤R-II Promoter Activity in F9-differentiated Cells-The large difference between the level of CAT activity expressed in the differentiated cells by constructs pT␤RII-274/ϩ50 and pT␤RII-137/ϩ50 suggests the presence of a positive regulatory element(s) located in the region between Ϫ274 and Ϫ137. Studies by Bae et al. (42) identified a putative CRE/ATF site located between Ϫ196 and Ϫ190 that contributed significantly to the basal transcriptional activity of the T␤R-II gene in HepG2 cells. This observation along with our previous findings dem- (F9-diff d5) was performed as described under "Materials and Methods." A 1.5-kilobase pair SacI fragment of the human T␤R-II cDNA clone, H2-3FF, was used as a hybridization probe. Following autoradiography, the T␤R-II probe was removed and the same blot was rehybridized with a human GAPDH cDNA probe for normalization.
onstrating that a CRE/ATF element located in the TGF-␤2 gene promoter was essential for its expression in EC-differentiated cells (59) as well as other cell types (60,61), led us to examine the effect of the putative CRE/ATF site on T␤R-II promoter expression in F9-differentiated cells. For these studies, we utilized a set of shorter constructs that eliminated a second positive regulatory region (ϩ1 to ϩ35) identified by Bae et al. (42) in HepG2 cells. Similar to our observations with the pT␤RII-274/ϩ50 and pT␤RII-137/ϩ50 constructs, there was about a 9-fold difference in the expression of pT␤RII-219/ϩ2 when compared with pT␤RII-137/ϩ2 (compare Figs. 2 and 3), suggesting that in F9-differentiated cells the function of the region between Ϫ219 and Ϫ137 is not dependent on the putative downstream positive regulatory element(s).
To determine whether the putative CRE/ATF site influenced basal transcriptional activity, the sequence at Ϫ196 to Ϫ190 in pT␤RII-219/ϩ2 was modified from TTAGTCA to TGCTGCA. CAT activity was reduced approximately 50% when F9-differentiated cells were transfected with the mutant pT␤RII-219 M/ϩ2 construct instead of pT␤RII-219/ϩ2 (Fig. 3). This finding argues that the CRE/ATF site located at Ϫ196 to Ϫ190 exerts a significant influence on the basal transcriptional activity of the T␤R-II promoter in F9-differentiated cells. However, as CAT activity was reduced by only 50%, which is similar to the results observed in HepG2 cells (42), other cis-regulatory elements within the region Ϫ219 to Ϫ137 are likely to contribute to the activity of the T␤R-II promoter.
Analysis of the Binding of Nuclear Proteins from F9 EC Cells and Their Differentiated Cells to the CRE/ATF Site in the T␤R-II Promoter-Previous studies in HepG2 cells (42), as well as studies in F9-differentiated cells (Fig. 3), demonstrate the importance of the CRE/ATF site for expression of the T␤R-II gene. However, the factor(s) that binds to this site has not been identified. This led us to initially examine the in vitro binding of nuclear proteins to the CRE/ATF site. Gel mobility shift analysis of radiolabeled dsODNs containing the T␤R-II CRE/ ATF motif with nuclear extracts prepared from F9 EC and F9-differentiated cells resulted in the formation of a single prominent DNA-protein complex that migrated with similar mobility in each extract (Fig. 4). The protein(s) in each complex appears to bind specifically to the CRE/ATF site of the probe, as a 25-fold excess of both unlabeled wild-type probe and unlabeled probe (M1) mutated slightly upstream (Ϫ203 to Ϫ200) of the CRE/ATF site competed effectively for the formation of DNA-protein complex, whereas a 25-fold excess of unlabeled probe (M2) mutated within (Ϫ195 to Ϫ192) the CRE/ATF site competed only very weakly (Fig. 4). Moreover, a 25-fold excess of unlabeled probe containing the essential CRE/ATF element (Ϫ74 and Ϫ67) from the human TGF-␤2 gene promoter (53,60) also competed effectively for the formation of the DNA-protein complex binding to the T␤R-II CRE/ATF (Fig. 4). Similarly, the unlabeled wild-type T␤R-II probe and not the CRE/ATF mutant counterpart (M2) was able to compete effectively for the factors that bind to the CRE/ATF motif of the TGF-␤2 gene (data not shown).

ATF-1 Is Present in the DNA-Protein Complexes Formed between the T␤R-II CRE/ATF Site and Nuclear Extracts from F9 EC and F9-differentiated Cells-
To identify the transcription factor(s) present in the DNA-protein complexes formed between the T␤R-II CRE/ATF site and nuclear extracts prepared from F9 EC and F9-differentiated cells, we used a battery of antibodies that individually recognize transcription factors that bind to CRE/ATF motifs, including antibodies that recognize ATF-1, ATF-2, c-Jun, CREB, and CREM. Each of these antibodies were incubated with nuclear extracts prepared from F9 EC and F9-differentiated cells and analyzed by gel mobility shift assay with the T␤R-II CRE/ATF specific probe. Only ATF-1, or a closely related transcription factor, appears to be present in the DNA-protein complexes formed with the F9 EC (Fig. 5) and F9-differentiated ( Fig. 6) cell nuclear extracts, as determined by both the change in migration and intensity of the prominent DNA-protein complex. However, neither of the heterodimeric partners of ATF-1 identified to date, CREB and CREM, were detected in the DNAprotein complex, suggesting that ATF-1 is binding either as a homodimer or as a heterodimer with an as yet unidentified transcription factor.
The Transcription Factor ATF-1 Up-regulates the T␤R-II Promoter in Vivo-Despite the finding that ATF-1 binds to the T␤R-II CRE/ATF site in vitro, it was possible that other members of the CREB/ATF family of transcription factors are responsible for the basal transcriptional activity mediated through the CRE/ATF site in vivo. Therefore, to examine the ability of ATF-1 to influence the expression of the T␤R-II gene, we employed eukaryotic expression vectors in transient transfection assays that express either ATF-1 (pECEATF-1) or ATF-2 (pECEATF-2) proteins under the control of the SV40 promoter. When F9-differentiated cells were co-transfected with pECEATF-1 and the pT␤RII-219/ϩ2 promoter/reporter construct, T␤R-II promoter activity was up-regulated by over 40-fold compared with the expression of the pT␤RII-219/ϩ2 construct co-transfected with pSV⌬ (an SV40 promoter vector without ATF-1 or ATF-2) (Fig. 7). In contrast, co-transfection of the pECEATF-2 plasmid with pT␤RII-219/ϩ2 resulted in only a modest (4-fold) induction of T␤R-II promoter activity (Fig. 7). The increase in pT␤RII-219/ϩ2 expression by ATF-1 does not appear to be due to general effects on cellular transcription, as overexpression of ATF-1 had only a minor effect (Ͻ3-fold induction) on two other T␤R-II promoter constructs (pT␤RII-137/ϩ2 and pT␤RII-47/ϩ2) that do not contain the CRE/ATF site (Fig. 8). Similarly, overexpression of ATF-1 did not have a significant effect on the CMV promoter (Fig. 8). Together, our findings argue that ATF-1 contributes to the expression of the T␤R-II gene in EC-differentiated cells.

Identification of a Cis-regulatory Element Located between Ϫ83 and Ϫ74 of the T␤R-II Promoter-Previous studies by Bae et al. (42) identified a regulatory region located between Ϫ137
and Ϫ47 in the T␤R-II promoter that had a strong negative effect on the basal transcriptional activity of the T␤R-II gene in HepG2 cells. However, the location of the negative regulatory element and, more important, the transcription factor that binds to this site were not identified. Utilizing a set of T␤R-II promoter/CAT deletion constructs that ranged from nucleotide Ϫ219 to Ϫ47, and each ending at ϩ2, we determined that the region located between Ϫ100 and Ϫ47 also suppresses T␤R-II promoter activity in F9-differentiated cells (data not shown, also see Fig. 10). Sequence analysis of this region identified a 10-base pair putative cis-regulatory element (TGATTGGCAG) located between Ϫ83 and Ϫ74 that contains an inverted CCAAT box motif. Moreover, expression of T␤R-II promoter/ reporter constructs with this site mutagenized is elevated when transfected into HepG2 cells. 2 Interestingly, this sequence is identical to the core sequence (Ϫ789 to Ϫ780) of a negative regulatory element identified in the human CYP1A1 gene using HepG2 cells (62,63). Moreover, this same sequence has been demonstrated to be an essential cis-regulatory element of the fibroblast growth factor-4 (FGF-4) gene in F9 EC cells (64 -66). In both the CYP1A1 gene and the FGF-4 gene, it was determined that the transcription factor complex NF-Y was able to bind in vitro to the core CCAAT box motif.
These observations led us to examine the binding of nuclear protein(s) from F9 EC cells and their differentiated counterparts to the Ϫ104 to Ϫ67 region of the T␤R-II promoter. Gel mobility shift analysis of a radiolabeled dsODN containing the T␤R-II Ϫ104 to Ϫ67 region with nuclear extracts prepared from F9 EC and F9-differentiated cells, resulted in the formation of at least two distinct DNA-protein complexes (Fig. 9,  compare A and B). It is important to note that a third less distinct and slower migrating DNA-protein complex was con-2 D. Kim and S.-J. Kim, unpublished results.

FIG. 4. Binding of nuclear proteins from F9 EC and F9-differentiated cells to the CRE/ATF site in the T␤R-II promoter.
Gel mobility shift assay of the 32 P-labeled wild-type T␤R-II CRE/ATF dsODN was performed with 12 g of crude nuclear extract prepared from either F9 EC or F9-differentiated cells as described under "Materials and Methods." Competition analysis of the DNA binding activity was performed by the addition of 25-fold molar excess of unlabeled dsODNs containing either the wild-type (WT) T␤R-II CRE/ATF site (indicated by the line above the sequence), a mutation 5Ј of the T␤R-II CRE/ATF site (M1), or a mutation within the T␤R-II CRE/ATF site (M2). In addition, competition with a 25-fold molar excess of unlabeled dsODN containing the TGF-␤2 CRE/ATF site (underlined; ␤2) was also performed. This, and all gel shift studies described in this report, were repeated at least once with similar results. In addition, the same results were observed with different preparations of nuclear extracts.

FIG. 5. Identification of nuclear proteins from F9 EC cells that bind to the T␤R-II CRE/ATF site.
Gel mobility supershift assay of the 32 P-labeled wild-type T␤R-II CRE/ATF dsODN was performed with 12 g of crude nuclear extract prepared from F9 EC cells as described under "Materials and Methods." Reaction mixtures containing nuclear extract were left untreated (lane 2) or were preincubated with ATF-1specific monoclonal antibody (an IgA-type antibody) (lane 3) or polyclonal antibodies (IgG-type antibodies) specific for ATF-2, c-Jun, CREB, and CREM in lanes 4, 5, 6, and 7, respectively. Non-specific mouse IgA and IgG were used as negative controls in lanes 8 and 9. sistently formed in multiple nuclear extract preparations of F9 EC cells, but not in their differentiated counterparts. It is also important to note that the two distinct DNA-protein complexes formed with each nuclear extract migrate with very similar mobilities when electrophoresed on the same gel under identical conditions (data not shown). In this regard, the DNA-protein complexes shown in Fig. 9 (A and B) were run on different gels for different lengths of time.
Nuclear proteins in each complex from both the F9 EC and F9-differentiated cells appear to bind specifically to the Ϫ83 to Ϫ74 CCAAT box-containing motif in the T␤R-II promoter. This is supported by the finding that specific protein binding is abolished completely by a 50-fold molar excess of dsODNs containing either the Ϫ104 to Ϫ67 or the Ϫ91 to Ϫ67 region of the T␤R-II promoter, whereas dsODNs containing either the region between Ϫ104 to Ϫ87 or a mutated CCAAT box consensus sequence were unable to compete for the binding of any of the DNA-protein complexes (Fig. 9, A and B, data not shown). Equally important, dsODNs containing the Ϫ125 to Ϫ97 region of the murine FGF-4 promoter (which contains the T␤R-II TGATTGGCAG motif) competed as effectively for the binding of all of the DNA-protein complexes as that observed by both the T␤R-II Ϫ104/Ϫ67 and Ϫ91/Ϫ67 competitor dsODNs (Fig.  9A). Similarly, a dsODN containing the human CYP1A1 NRE also was able to compete for the binding of all of the DNAprotein complexes (data not shown).
The Transcription Factor Complex NF-Y Is Present in the DNA-Protein Complexes Formed between the T␤R-II Putative CCAAT Box Motif and Nuclear Extracts from F9 EC and F9differentiated Cells-Previous studies using gel mobility super-shift analysis determined that the transcription factor complex, NF-Y is one of the transcription factors that binds to the TGATTGGCAG sequence in both the FGF-4 gene (64 -66) and human CYP1A1 gene (63). The ability of oligonucleotides containing the FGF-4 CCAAT box and the CYP1A1 NRE to compete effectively for the binding of proteins to the T␤R-II CCAAT

FIG. 6. Identification of nuclear proteins from F9-differentiated cells that bind to the T␤R-II CRE/ATF site.
Gel mobility supershift assay of the 32 P-labeled wild-type T␤R-II CRE/ATF dsODN was performed with 12 g of crude nuclear extract prepared from F9-differentiated cells as described under "Materials and Methods." Reaction mixtures containing nuclear extract were left untreated (lane 2) or were preincubated with ATF-1-specific monoclonal antibody (an IgA-type antibody) (lane 3) or polyclonal antibodies (IgG-type antibodies) specific for ATF-2, c-Jun, CREB, and CREM in lanes 4, 5, 6, and 7, respectively. Non-specific mouse IgA and IgG were used as negative controls in lanes 8 and 9.

FIG. 7. Effects of ATF-1 and ATF-2 transcription factors on
T␤R-II promoter activity. F9-differentiated cells were co-transfected in monolayer with 10 g of the T␤R-II promoter/CAT construct pT␤RII-219/ϩ2 and with optimal concentrations (7.5 g) of either the ATF-1 or ATF-2 expression plasmids, pECEATF-1 and pECEATF-2, respectively. SV⌬ plasmid (7.5 g), which lacks ATF-1 and ATF-2 expression genes, was used as a control. All transfections included the ␤-galactosidase normalizing plasmid, pCMV-␤. The bars represent CAT activities relative to the expression of pT␤RII-219/ϩ2 when transfected with the SV⌬ control (1,036 cpm). The experiment was repeated three times with similar results.
FIG. 8. Effect of ATF-1 on T␤R-II promoter activity appears to be specific to the CRE/ATF site. F9-differentiated cells were cotransfected in monolayer with 10 g of the T␤R-II promoter/CAT constructs pT␤RII-47/ϩ2, pT␤RII-137/ϩ2, and pT␤RII-219/ϩ2 along with 7.5 g of the pECEATF-1 or pSV⌬ plasmids. All transfections included the ␤-galactosidase normalizing plasmid, pCMV-␤. The pCMV-CAT (5 g) was also co-transfected with 7.5 g of either the pECEATF-1 or pSV⌬ to monitor effects on general transcription. The bars represent relative CAT activities of each of the reporter plasmids in the presence or absence of ATF-1. CAT activities of pT␤RII-47/ϩ2, pT␤RII-137/ϩ2, pT␤RII-219/ϩ2, and pCMV-CAT when transfected with the pSV⌬ control were 4,688, 1,003, 6,189, and 58,933 cpm, respectively. The experiment was repeated twice with similar results. box motif raised the possibility that NF-Y may also bind to the T␤R-II CCAAT box motif. As NF-Y is reported to be a trimeric complex containing NF-YA, NF-YB, and NF-YC (also known as CBF-B, CBF-A, and CBF-C, respectively), we used an antibody that specifically recognizes the NF-YA subunit to characterize the DNA-protein complexes that form between the T␤R-II CCAAT box motif and nuclear proteins from both F9 EC and F9-differentiated cells. In both cell types, only the slower of the two closely migrating distinct DNA-protein complexes was supershifted by the addition of the NF-YA antibody, whereas the nonspecific IgG antibody had no effect on the mobility of any of the DNA-protein complexes (Fig. 9, A and B). In addition, the NF-YA antibody also appears to recognize the less distinct slowest migrating DNA-protein complex formed with nuclear extract from F9 EC cells (Fig. 9A), suggesting that NF-Y may be interacting with another factor(s). Thus, it appears that the NF-Y transcription factor complex binds the T␤R-II CCAAT box motif in vitro and differentiation does not overtly affect the ability of NF-Y to bind to this site. In contrast to the slower migrating DNA-protein complex, the faster migrating complex observed in both F9 EC cells and their differentiated counterparts was not supershifted by the NF-YA antibody (Fig. 9, A  and B). Hence, the factor(s) in this DNA-protein complex appears to be distinct from NF-Y and thus far it has not been identified.

NF-Y Influences T␤R-II Expression in Vivo-
The results of our transient expression studies, combined with our in vitro binding analyses to the T␤R-II CCAAT box motif, implies a role for NF-Y in T␤R-II transcription. However, additional studies are required to address the question of whether NF-Y affects T␤R-II expression in vivo. To this end, Mantovani et al. (46) described a mutant NF-YA protein, NFYA29, which contains mutations in three amino acids of the DNA binding domain of NF-YA. The NFYA29 protein continues to bind to the YB subunit of NF-Y, but not to DNA, thereby functioning as a dominant negative repressor of NF-Y-mediated effects on transcription (46). Results from studies co-expressing this dominantnegative NF-YA demonstrated an in vivo role of NF-Y in the sterol-dependent expression of the farnesyl diphosphate (FPP) synthase gene, the 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) synthase gene (67), and the FGF-4 gene (66).
To examine whether NF-Y influences the expression of the T␤R-II gene, we co-transfected both F9-differentiated cells (where the endogenous T␤R-II gene is expressed) with various T␤R-II promoter-CAT constructs and the expression plasmid for the NFYA29 mutant protein. In these cells, the T␤R-II promoter-CAT constructs, pT␤RII-219/ϩ2, pT␤RII-100/ϩ2 and pT␤RII-47/ϩ2 are expressed as described previously. Specifically, the deletion of the region between Ϫ219 and Ϫ100 containing the positive CRE/ATF site resulted in a dramatic reduction (approx. 11-fold) in the transcription of the T␤R-II promoter (Fig. 10, also see Fig. 3). However, when the region between Ϫ100 and Ϫ47 is deleted, pT␤RII-47/ϩ2 activity returned to levels higher than the pT␤RII-219/ϩ2 construct, indicating the presence of a strong negative regulatory element in the Ϫ100/Ϫ47 region (Fig. 10). (The promoter fragment Ϫ47/ϩ2 contains a Sp1 site that when mutated, significantly diminishes the activity of the pT␤RII-47/ϩ2 construct (42), implicating a potential role for Sp1 in the transcriptional activity of this region.) Co-transfection of pNFYA29 with the pT␤R-II promoter-CAT constructs resulted in a dramatic increase (approximately 17-fold) in the expression of the pT␤RII-100/ϩ2 construct, which contains the T␤R-II CCAAT box motif, but had little or no effect on the expression of the pT␤RII-47/ϩ2 construct, which does not contain the CCAAT box motif (Fig.  10). In addition, NFYA29 also increased the expression of the pT␤RII-219/ϩ2 construct to a level similar to the expression of the pT␤RII-100/ϩ2 construct (Fig. 10). Importantly, expression of pNFYA29 had little or no effect on the expression of a CMV-CAT control vector or a plasmid containing the ␤-galactosidase reporter gene under the control of a SV40 promoter suggesting that the NFYA29 mutant protein does not exert general effects on transcription or transfection efficiency (data not shown). The increase in T␤R-II promoter activity caused by the expression of mutant NFYA29 in these experiments argues strongly that binding by NF-Y to the T␤R-II CCAAT box motif can act to inhibit transcription of this gene in vivo.
As a result of the above observations, we also transfected F9 EC cells, where the T␤R-II gene is not expressed, with the same set of deletion constructs. As might be expected, the pT␤RII-219/ϩ2 and pT␤RII-100/ϩ2 constructs were expressed weakly in F9 EC cells; however, removal of the region between Ϫ100 and Ϫ47 resulted in a significant increase in the basal transcription of the pT␤RII-47/ϩ2 construct (Fig. 11). As in the case of EC-differentiated cells, co-transfection of EC cells with pNFYA29 and pT␤RII-100/ϩ2 or pT␤RII-219/ϩ2 constructs resulted in a substantial increase (6.3-and 3-fold, respectively) in the overall expression of these constructs while having little or no effect on the expression of the pT␤RII-47/ϩ2 construct (Fig. 11). Thus, the CCAAT box motif may also play a role in limiting the transcription of the T␤R-II gene in F9 EC cells. DISCUSSION TGF-␤ ligand binding and cellular responsiveness to TGF-␤ increase dramatically when EC cells are induced to differentiate, which suggests strongly that TGF-␤ receptors are differentially expressed by EC cells and their EC-derived differentiated counterparts (38). In this report, we examined the expression of the T␤R-II gene in F9 EC and F9-differentiated cells, as T␤R-II expression is normally associated with ligand binding and growth responsiveness of cells to TGF-␤ (21)(22)(23)58). Our studies demonstrate that the expression of T␤R-II is closely associated with the induction of differentiation in F9 EC cells. Few, if any, transcripts for the T␤R-II gene are detected in F9 EC cells, whereas there is a dramatic increase in the mRNA steady-state levels of this gene in F9-differentiated cells. Equally important, the high levels of expression of T␤R-II promoter/CAT reporter chimeric gene constructs in F9-differentiated cells, but not in their undifferentiated parental cells, argues strongly that the increase in the steady state levels of T␤R-II mRNA results, at least in part, from increased transcription of the T␤R-II gene. Furthermore, this study identifies two different transcription factors, ATF-1 and NF-Y, that bind in vitro and in vivo to a positive regulatory element and negative regulatory element, respectively.
The up-regulation of the T␤R-II gene during differentiation coincides with the ability of TGF-␤ to both bind and inhibit the growth of the EC-derived differentiated cells as well as to their loss of tumorigenic potential. Therefore, EC cells provide a powerful model system for studying the mechanisms that control T␤R-II expression during normal development as well as during the processes of malignant transformation. In regard to the latter, several lines of evidence suggest that T␤R-II may act as a tumor suppressor gene. A number of tumors, including small cell lung carcinoma (26), thyroid tumors (27), and prostate adenocarcinoma (25) show a loss of or reduced expression of functional TGF-␤ type II receptors. Additionally, transfection of wild-type T␤R-II constructs into hepatoma cells and MCF-7 breast carcinoma cells lacking functional type II receptors restores sensitivity to TGF-␤ and suppresses their tumorigenic phenotype (21,22). Moreover, gross structural mutations of both T␤R-II alleles has been observed in 71-90% of colorectal tumors and cell lines and in 71% of gastric cancer cell lines with microsatellite instability (23,24,68). However, in a number of instances in which cells fail to express the T␤R-II gene at the RNA or protein level, no apparent structural mutations within the coding region of the gene were observed. This raises the possibility that defects in the promoter region of the T␤R-II gene and/or in the mechanisms regulating the transcription of the T␤R-II gene may play important roles in the aberrant expression of T␤R-II in certain neoplasms. In this regard, the low levels of T␤R-II mRNA expressed by A431 tumor cells is thought to be a result of a point mutation located in the 5Ј flanking region of the T␤R-II promoter (69).
To understand further the mechanism(s) that control the regulation and expression of the T␤R-II promoter during differentiation, our studies also examined the expression of T␤R-II promoter/reporter gene constructs in the differentiated cells derived from various EC cell lines to identify putative DNA regulatory elements that are required for normal T␤R-II promoter expression. The data presented demonstrates the existence of at least four distinct regulatory regions in the T␤R-II promoter including, two positive regulatory regions located between Ϫ274/Ϫ100 and Ϫ47/ϩ2, and two negative regulatory regions located between Ϫ1883/Ϫ274 and Ϫ100/Ϫ47. In addition, we determined that a putative CRE/ATF site located at Ϫ196 to Ϫ190 exerts a significant positive influence on T␤R-II promoter activity in the differentiated cells, whereas a putative inverted CCAAT box motif located at Ϫ83 to Ϫ74 exerts a significant negative influence on T␤R-II promoter activity in both F9 EC cells and their differentiated counterparts.
In regard to the CRE/ATF motif, mutation of this site reduces transcription of wild-type T␤R-II promoter/reporter constructs in the differentiated cells by approximately 50%, which is similar to what was observed in HepG2 cells (42). Gel mobility shift analysis further demonstrated that protein complexes containing the transcription factor ATF-1 specifically recognize and bind the T␤R-II CRE/ATF site. Overexpression of ATF-1 in F9-differentiated cells resulted in a dramatic increase in T␤R-II promoter activity most likely through the CRE/ATF site, suggesting that ATF-1 not only interacts with this site in vitro but also in vivo. Interestingly, the expression of the TGF-␤2 gene in EC cells and their differentiated counterparts is also regulated by mechanisms involving the binding of ATF-1 to an essential CRE/ATF element (7, 53, 59, 60). FIG. 11. Effect of the dominant-negative NFYA29 mutant protein on T␤R-II promoter activity in F9 EC cells. F9 EC cells were co-transfected in monolayer with 10 g of the T␤R-II promoter/CAT constructs pT␤RII-219/ϩ2, pT␤RII-100/ϩ2, and pT␤RII-47/ϩ2 with 5 g of the pNFYA29 expression plasmid (solid bars) as described under "Materials and Methods." As a control, the parent vector lacking the NFYA29 insert was used to equalize the amount of DNA transfected into the cells (open bars). All transfections included the ␤-galactosidase normalizing plasmid, pCMV-␤. The bars represent CAT activities relative to the expression of the pT␤RII-47/ϩ2 construct when transfected with the control vector (6,750 cpm). The experiment was repeated three times with similar results.
Another important aspect of the work reported in this study is the demonstration that a 53-base pair negative regulatory region located between Ϫ100 and Ϫ47 contains an inverted CCAAT box motif that negatively regulates T␤R-II promoter activity. We demonstrate further that the transcription factor complex, NF-Y, binds to the T␤R-II CCAAT box motif in vitro when nuclear extracts from both F9 EC cells and their differentiated counterparts are used in gel mobility supershift analyses. Importantly, we also demonstrate that expression of the dominant-negative NFYA29 mutant protein in both F9 EC and F9-differentiated cells specifically increases the expression of T␤R-II promoter/reporter constructs that contain the CCAAT box motif. In further support of these findings, we have shown that NF-Y in nuclear extracts prepared from HepG2 cells binds to the CCAAT box motif (data not shown) and expression of T␤R-II promoter/reporter constructs with this site mutagenized is elevated when transfected into HepG2 cells. 2 Thus, our findings argue strongly that in vivo binding of NF-Y to the CCAAT box motif represses the activity of the T␤R-II promoter.
CCAAT box motifs are found in the promoter regions of many genes, and a survey of over 500 unrelated promoter sequences determined that most proximally located CCAAT boxes reflect the target sequence for NF-Y rather than C/EBP or NF1 (70). NF-Y was originally identified as a ubiquitously expressed protein that binds to the Y box motif, originally defined as an inverted CCAAT box motif (CTGATTGGYY) in all MHC class II genes that is critical for tissue specific gene expression (71). Although the exact mechanism(s) by which NF-Y regulates transcription is unclear, it has been demonstrated to act as both a positive (66,72,73) and a negative (74) transcription factor. Several studies suggest that NF-Y acts by stabilizing or recruiting the binding of additional factors to adjacent promoter elements (75,76). Whether there are coordinate interactions between ATF-1 and NF-Y that act to regulate the T␤R-II promoter in F9 EC and F9-differentiated cells is unknown, however these two factors appear to act in concert to regulate a number of other gene promoters, including the human cyclin A gene and the rat hexokinase II gene (72,77).
In conclusion, our studies demonstrate clearly that transcription of the T␤R-II gene is influenced significantly by both positive and negative cis-regulatory elements and trans-acting factors. However, the exact mechanism(s) by which differentiation turns on the expression of the T␤R-II gene remains to be determined. Given the important role of the TGF-␤ type II receptor during normal development and in its aberrant expression in certain neoplasms, further study of this gene in this model system is clearly warranted. In addition, it will be interesting to determine whether differentiation also leads to the transcriptional activation of the genes for the type I and type III TGF-␤ receptors.