TEF-1 Transrepression in BeWo Cells Is Mediated through Interactions with the TATA-binding Protein, TBP*

Transcription enhancer factor-1 (TEF-1) has been im- plicated in transactivating a placental enhancer (CSEn) that regulates human chorionic somatomammotropin (hCS) gene activity. We demonstrated that TEF-1 represses hCS promoter activity in choriocarcinoma (BeWo) cells S. and N. L. J. Biol. Chem. suggesting that TEF-1 in- teracts with basal transcription factors. Here we demonstrate that hTEF-1 overexpression inhibits minimal hCS promoters containing TATA and/or initiator elements, Rous sarcoma virus and thymidine kinase pro- moters in BeWo cells. Cotransfection of TEF-1 antisense oligonucleotides alleviated exogenous TEF-1-mediated repression and increased basal hCS promoter activity, indicating that endogenous TEF-1 exerts repressor ac- tivity. GST-TEF-1 fusion peptides fixed to glutathione-Sepharose beads retained in vitro -generated human TATA-binding protein, hTBP. The TEF-1 proline-rich domain was essential for TBP binding, but polypeptides also containing the zinc finger domain bound TBP with higher apparent affinity. TBP supershifted hTEF-GT- IIC DNA complexes, but TEF-1 inhibited in vitro binding of TBP to the TATA motif. Coexpression of TBP and TEF-1 in BeWo cells alleviated TEF-1-mediated transrepression, indicating that the TBP-TEF-1 interaction is functional in vivo . The data indicate

Transcription enhancer factor-1 (TEF-1) has been implicated in transactivating a placental enhancer (CSEn) that regulates human chorionic somatomammotropin (hCS) gene activity. We demonstrated that TEF-1 represses hCS promoter activity in choriocarcinoma (BeWo) cells (Jiang, S. W., and Eberhardt, N. L. (1995) J. Biol. Chem. 270, 13609 -13915), suggesting that TEF-1 interacts with basal transcription factors. Here we demonstrate that hTEF-1 overexpression inhibits minimal hCS promoters containing TATA and/or initiator elements, Rous sarcoma virus and thymidine kinase promoters in BeWo cells. Cotransfection of TEF-1 antisense oligonucleotides alleviated exogenous TEF-1-mediated repression and increased basal hCS promoter activity, indicating that endogenous TEF-1 exerts repressor activity. GST-TEF-1 fusion peptides fixed to glutathione-Sepharose beads retained in vitro-generated human TATA-binding protein, hTBP. The TEF-1 proline-rich domain was essential for TBP binding, but polypeptides also containing the zinc finger domain bound TBP with higher apparent affinity. TBP supershifted hTEF-GT-IIC DNA complexes, but TEF-1 inhibited in vitro binding of TBP to the TATA motif. Coexpression of TBP and TEF-1 in BeWo cells alleviated TEF-1-mediated transrepression, indicating that the TBP-TEF-1 interaction is functional in vivo. The data indicate that TEF-1 transrepression is mediated by direct interactions with TBP, possibly by inhibiting preinitiation complex formation.
The mechanism of TEF-1 action has not been elucidated but appears to be complex. TEF-1 binds to sequences related to the general "enhancer core consensus," 5Ј-TGTGG(T/A)(T/A)(T/ A)G-3Ј (Weiher et al., 1983;Kariya et al., 1993), and mutation of the binding sites for TEF-1 results in loss of transcriptional activation (Xiao et al., 1991). Nevertheless, when cells are cotransfected with TEF-1 expression vectors and promoter constructs containing the TEF-1 binding site, inhibition or squelching of promoter activity instead of activation is typically observed (Xiao et al., 1991;Ishiji et al., 1992;Hwang et al., 1993;. This has led to the concept that TEF-1 requires a limiting transcription factor for its transactivation function. In the case of the placenta-specific enhancer CSEn, we recently provided evidence that a factor, designated CSEF-1, which is distinct from TEF-1 and is abundant in choriocarcinoma and COS cells, bound to the GT-IIC and Sph-I/Sph-II enhansons with binding specificity identical to that of TEF-1 . CSEn transactivation in COS and BeWo cells was correlated with the presence of CSEF-1, but not TEF-1, since COS cells contain very low levels, if any, of TEF-1. This raises the question of whether TEF-1 mediates transrepression but not transactivation in some cell types. To address these issues, we reexamined the effects of overexpression of hTEF-1 on hCS promoter and enhancer function in choriocarcinoma cells (BeWo). Exogenous TEF-1 expression led to marked repression of basal hCS promoter activity without affecting relative enhancer-mediated stimulation of promoter activity. Interestingly, this repression appears to be independent of a GT-IIC binding site and was observed with several promoters, suggesting that basal transcriptional factors might be involved. Cotransfection of hTEF-1 antisense oligonucleotides not only reversed these effects, but also stimulated basal hCS promoter activity, indicating that endogenous TEF-1 was acting as a repressor in BeWo cells. Based on these studies we undertook an analysis of the ability of TEF-1 to bind to known components of the basal transcription apparatus. We found that TEF-1 bound to TATA-binding protein (TBP) and that the TEF-1 proline-rich and zinc finger domains were involved in this interaction. Since Hwang et al. (1993) demonstrated that these domains were required for both autoinhibitory and stimulatory responses, our data suggest that TEF-1-TBP interactions were important for TEF-1 function in vivo. Interestingly, TEF-1 inhibited the binding of TBP to the hCS proximal promoter region containing the TATA box, whereas TBP supershifted TEF-1-GT-IIC DNA complexes. Importantly, overexpression of TBP in BeWo cells alleviated the TEF-1mediated inhibition of hCS promoter activity, indicating that TBP-TEF-1 interaction is functional in vivo. Our data are consistent with the hypothesis that TEF-1 transrepression may be the result of interference with the binding of TBP to the TATA element and that TEF-1-mediated transactivation requires additional factors. Alternatively, transactivation formally attributed to TEF-1 may be mediated by different factors. The latter possibility is strongly supported in the case of the placental enhancer, since transactivation of CSEn is highly correlated with the distinct GT-IIC-binding factor, CSEF-1  Cell Culture-BeWo cells (American Type Culture Collection) were maintained in RPMI 1640 (Life Technologies, Inc.), supplemented with 10% fetal bovine serum (Whittaker), 100 units/ml penicillin (Life Technologies, Inc.), and 2 mM L-glutamine (Life Technologies, Inc.). Cells were grown in monolayer at 37°C in an atmosphere containing 5% CO 2 and 100% humidity.
Cell Transfection and Luciferase Assay-Cell transfection by electroporation, preparation of cell lysates, and measurement of luciferase activity (light units/g of protein) were performed as described previously . In cotransfection experiments with CMVp.TEF, CMVp.␤GAL and TKp.CAT were used as controls to assess the potential effects of excess promoter. Plasmid DNA was isolated by large scale alkaline lysis and purified by double CsCl gradient ultracentrifugation. For antisense and sense oligonucleotide transfections, cells were mixed with 20 l of phosphorothioate oligonucleotide and plasmids previously diluted at different concentrations in TE (1 mM Tris⅐HCl, 1 mM EDTA) before transfection.
Data Analysis-Data groups were subjected to analysis of variance (ANOVA). For those data that indicated significant effects (p Ͻ 0.05), multiple comparison analyses were performed with Student's t tests employing a Bonferronni inequality (Snedecor and Cochran, 1980). ANOVA values are designated in the figure legends.
Northern Blot Analysis-Triplicate transfections using 5 ϫ 10 6 BeWo cells and 15 g of hCSp.LUC and either 500 ng of CMVp.␤GAL or CMVp.TEF-1 were performed as described . After various periods of incubation at 37°C, cellular RNA was isolated by the method of Xie and Rothblum (1991). RNA samples (50 g) were subjected to standard conditions for Northern gel electrophoresis and transfer (Sambrook et al., 1989). Random primed DNA probes (1 ϫ 10 6 cpm/ml) generated from the luciferase coding region were hybridized at 45°C for 15 h according to Sambrook et al. (1989). The washed filters were exposed to Kodak x-ray film with intensifying screens at Ϫ70°C for 5 days.
Expression of TEF-1 in Escherichia coli-HB101 bacteria transformed with the various TEF expression plasmids were grown in 2 ϫ YT-G medium (Sambrook et al., 1989) in the presence of 100 g/ml ampicillin. The bacteria were induced by 100 mM isopropyl-1-thio-␤-Dgalactopyranoside treatment at 30°C for 4 h and then collected by centrifugation and resuspended in 1 ϫ phosphate-buffered saline. After sonication on ice, Triton X-100 was added to a final concentration of 1% to aid the solubilization of proteins.

35
S Labeling of hTBP, TFIIB, and TR␤ by in Vitro Translation-The GST fusion proteins were bound to glutathione-Sepharose 4B beads and eluted with reduced glutathione according to the manufacturer's instructions. Protein concentration was determined by Coomassie dye binding assay. The purity and size of the fusion proteins were confirmed on 10% SDS-PAGE by silver or Coomassie staining. The hTBP, TFIIB, and human thyroid hormone receptor TR␤ were labeled by in vitro translation using TNT reticulocyte lysate (Promega) according to the manufacturer's directions. Aliquots of the translation products were resolved by 10% SDS-PAGE and visualized by overnight exposure to Kodak x-ray film.
Protein Retention Assays-Protein-protein interaction experiments were performed with minor modifications of the procedure of Seto et al. (1992). Beads coated with GST-TEF fusion protein (20 l) obtained from the batch purification step (manufacturer's protocol) were washed three times with 400 l of incubation buffer (50 mM KCl, 40 mM HEPES⅐HCl (pH 7.5), 2 mM MgCl 2 , 0.5% nonfat milk, 0.5% Nonidet P-40, 1 mM dithiothreitol and 1 mM phenylmethylsulfonyl fluoride). Aliquots (10 l) of in vitro translated 35 S-labeled TBP, TFIIB, or TR␤ proteins were mixed with the beads in 100 l of incubation buffer. The beads were suspended by mechanical agitation at 4°C for 2 h, collected by centrifugation, and washed five times with cold incubation buffer. Bound proteins were recovered by brief boiling in 40 l of SDS loading buffer (50 mM Tris⅐HCl (pH 6.8), 100 mM dithiothreitol, 2% SDS, 0.2% bromphenol blue, and 20% glycerol) and analyzed by 10% SDS-PAGE along with prestained protein standards (Bio-Rad). After electrophoresis, the gels were dried and exposed to Kodak x-ray film at Ϫ70°C for 2-5 days with intensifying screens.
Gel Shift Assays-Double-stranded SV40 GT-IIC-and AdML TATA box-containing oligonucleotides (Table I) were used as probes. Probe preparation, incubation conditions, and electrophoretic analysis were carried out as described previously . Gels were dried in vacuo and exposed to Kodak x-ray film at Ϫ80°C with intensifying screens for 1-3 days.

TEF-1 Inhibits Basal but Not Relative CSEn-stimulated hCS
Promoter Activity-We reported previously that cotransfection of a TEF-1 expression vector inhibited the basal hCS promoter  . This suggested that TEF-1-mediated inhibition of promoter activity was independent of a TEF-1 DNA binding site and might result from interactions of TEF-1 with basal transcription factors. To test this hypothesis, we cotransfected the CMVp.TEF-1 expression plasmid with wild-type and mutated or CSEn-linked versions of the hCS promoter to ascertain whether TEF-1 function depended on a previously identified promoter element. Confirming our previous results , cotransfection of increasing amounts (40 -1,080 ng) of the CMVp.TEF DNA with the hCSp.LUC gene resulted in progressive repression of basal CS promoter activity (Fig. 1A). Significant repression of hCS promoter activity was observed with 120 ng of cotransfected CMVp.TEF. This result mirrors the behavior of TEF-1 overexpression with the HPV-16 P97 promoter in keratinocytes (Ishiji et al., 1992) and with its ability to repress endogenous HeLa TEF-1 activity both in vivo and in vitro (Xiao et al., 1991). Importantly, although the total enhancer-stimulated activity decreased significantly (Fig. 1A), TEF-1 overexpression did not affect significantly the relative enhancer-stimulated activity (Fig. 1B). Therefore, TEF-1-mediated transrepression results from interactions with the promoter and does not depend on GT-IIC sequences present in CSEn.
The hCS promoter is dominantly controlled by Sp1, TATA box, and initiator elements in BeWo cells . We therefore analyzed CS promoter mutants to determine which of these sequences might be involved in mediating TEF-1 transrepression. Substitution mutation of the InrE binding site (sequences hCS Ϫ20/ϩ1, 493CS_⌬APA), deletion of the TATA element (493CS_⌬TATA) in the context of the otherwise intact hCS promoter, or deletion of sequences upstream of the TATA box, including the Sp1 site that contributes to basal hCS promoter activity (36CS) did not alleviate the inhibitory activity of TEF-1 ( Fig. 2A). Similarly, TEF-1 overexpression inhibited the activities of the heterologous RSV and TK promoters (Fig. 2 and C) in a manner indistinguishable from the hCS promoter. This repression was not due to the presence of excess promoter DNA, since cotransfection of large amounts of DNAs containing the CMV or TK promoters was without effect on hCS, RSV, or TK promoter activity (Fig. 2, A-C). These data further support the concept that overexpression of TEF-1 interferes with a common, basal transcription mechanism.
Overexpression of TEF-1 Decreases hCSp.LUC Transcripts-TEF-1 transrepression in HeLa cells has been shown to be mediated at the level of transcription by in vitro transcription studies (Xiao et al., 1991;Hwang et al., 1993). We found that cotransfection of CMVp.TEF-1 inhibited hCSp.LUC mRNA transcripts in BeWo cells by analysis of Northern blots (Fig. 3). Low levels of hCSp.LUC transcripts were readily detected 3 h after transfection. No difference between control and TEF-1overexpressing cells was observed up to 7 h after transfection. However, 18 h after transfection significantly lower levels of hCSp.LUC transcripts were observed in TEF-1-transfected cells. The time-dependent repression probably reflects the time required for TEF-1 to accumulate to sufficient levels to affect hCSp.LUC expression. These data are consistent with the concept that TEF-1 acts by a transcriptional mechanism in BeWo cells.
Cotransfection of TEF-1 Antisense Oligonucleotides Stimulates Basal hCS Promoter Activity-We demonstrated previously that TEF-1 levels in BeWo cells could be up-and downregulated by transfection of a TEF-1 expression vector or TEF-1 antisense oligonucleotides . We utilized this approach to assess the physiological role of endogenous TEF-1 on transcription in BeWo cells. First, varying amounts of the sense and antisense oligonucleotides were cotransfected along with the fixed amounts of hCSp.LUC and CMVp.TEF genes. The low level of hCS promoter activity in the presence of the CMVp.TEF gene was derepressed with increasing concentrations of the antisense oligonucleotide, but not the control sense oligonucleotide (Fig. 4A), confirming that the antisense oligonucleotide was fully functional. Second, the function of endogenous TEF-1 was studied in the absence of TEF-1 overexpression. Increasing concentrations of TEF-1 antisense, but not sense, oligonucleotide resulted in a 2-fold increase in hCS promoter activity (Fig. 4B). The fact that TEF-1 antisense oligonucleotides almost completely reversed the effects of TEF-1 overexpression indicates that the oligonucleotide acts by inhibiting TEF-1 expression and not by another mechanism. It is noteworthy that the concentrations of antisense oligonucleotides at which the half-maximal effect occurs is left-shifted in the data in Fig. 4B (22.6 M) versus that in Fig.  4A (28.5 M). This dose-response sensitivity probably reflects the relatively higher concentration of TEF-1 mRNA (compare Fig. 3) in cells containing exogenous TEF-1 (panel A) than in cells that only contain endogenous TEF-1 (panel B). The inhibition of promoter activity observed at oligonucleotide concentrations greater than 40 M with both the sense and antisense oligonucleotides appears to be due to nonspecific toxicity. These data indicate that endogenous levels of TEF-1 in BeWo cells inhibit hCS promoter activity and provide additional evidence that TEF-1 function in BeWo cells is dominantly inhibitory.
TEF-1 Binds to the TATA-binding Protein, TBP-The TATA box binding factor TBP and TFIIB are frequently involved in interaction with other transcriptional factors (Seto et al., 1992;Lin et al., 1991;Ing et al., 1992), and previous studies (Brou et al., 1993;Gruda et al., 1993) have suggested that components of the basal TFIID transcription complex might be involved with TEF-1 action. Since our studies suggested the possible TEF-1 involvement with basal transcriptional factor(s), we sought to determine whether TEF-1 interacts directly with hTBP or TFIIB. Initially a GST-TEF fusion protein was expressed in E. coli and the purified 26-kDa GST peptide, 74-kDa GST-TEF-1, and 48-and 26-kDa thrombin cleavage products were characterized by SDS-PAGE (Fig. 5A). The GST and GST-TEF fusion proteins were fixed to glutathione Sepharose beads and in vitro-generated 35 S-labeled hTBP, hTFIIB, and hTR␤ were incubated with the beads (Fig. 5B). Although no 35 S-labeled protein was recovered from GST-coated beads (data not shown), the GST-TEF-1 beads retained 35 S-labeled hTBP, but not 35 S-labeled hTFIIB or hTR␤, indicating that TEF bound hTBP specifically (Fig. 5C).
We next analyzed gel shift experiments with TEF and the GT-IIC enhanson in the presence or absence of TBP. When purified TBP was added to gel shift assays using different amounts of in vitro-generated TEF-1, a weak but visible supershift was observed (Fig. 6). Since neither GST protein nor TBP binds to the GT-IIC probe, the supershift represents the TBP-TEF-GT-IIC complex. This result confirms the TEF-TBP interaction and indicates that this protein-protein interaction does not affect the ability of TEF-1 to bind DNA.
TEF-1 Inhibits in Vitro Binding of TBP to the TATA Element-Since TBP also has specific DNA binding activity, we analyzed the effects of TEF-1 on the ability of TBP to bind to the TATA element. Surprisingly, the addition of increasing amounts of GST-TEF inhibited the binding of TBP to the proximal (nucleotides Ϫ40/Ϫ10) hCS promoter region that contains the TATA element (Fig. 7). The inhibition was not due to the presence of GST on the fusion protein, since GST had no observable effect on TBP binding to the TATA element (Fig. 7).

FIG. 5. TEF-1 interacts specifically with TBP but not TFIIB.
Panel A, expression of GST-TEF fusion protein in E. coli. GST-TEF fusion protein was expressed in HB101 and purified using glutathione-Sepharose 4B beads. Aliquots (30 ng) of thrombin-cleaved, intact GST-TEF fusion protein and GST peptide were resolved by SDS-PAGE and visualized by silver staining. Panel B, 35 S-labeled hTBP, hTFIIB, and hTR␤ from in vitro translation. No 35 S-labeled product was observed from control reactions lacking plasmid DNA (TNT). Panel C, recovery of 35 S-labeled proteins after binding to GST-TEF-coated glutathione-Sepharose 4B beads.

FIG. 6. TBP binds to TEF-GT-IIC complex resulting in supershifted complexes.
Gel shift assays were performed with increasing amounts of GST-TEF and an oligonucleotide containing the GT-IIC enhanson. The addition of 2 footprinting units (fpu) of hTBP to the binding reaction resulted in a supershifted complex. No protein-DNA complex was observed when 500 ng of GST peptide or 2 fpu of TBP was incubated with the GT-IIC-containing oligonucleotide. The molar ratio of TBP to GST-TEF-1 was estimated to be about 2.1, assuming that the TEF-1 and TBP were 100% active, GST-TEF-1 was 50% pure (Fig. 5A), and using Promega quality control data on the recombinant TBP (95% pure as estimated from Coomassie-stained SDS-PAGE gels and 20 ng of protein/fpu).
FIG. 7. TEF-1 inhibits the binding of TBP to DNA containing the TATA motif. Purified hTBP (4 footprinting units, fpu) and oligonucleotides containing the AdML-TATA box motif were used in gel shift assays in the presence or absence of varying amounts (100 -300 ng) of in vitro-generated GST-TEF. Inhibition of TBP binding to the TATA motif was due to TEF-1, since GST peptide (500 ng) had no inhibitory effect. The molar ratios of TBP to GST-TEF-1 at 100, 200, and 300 ng input of GST-TEF-1 were estimated to be about 3.3, 1.7, and 1.1, respectively, as detailed in the legend to Fig. 6.

TEF-1-TBP Interactions Mediate Transrepression
These data suggest that TEF-1 might inhibit promoter action by inhibiting preinitiation complex formation. Also, if TEF-1 has any transactivation function in BeWo cells, the data support the concept that additional factors would be required for such activity.
The TEF-1 Proline-rich and Zinc Finger Domains Are Required for Binding TBP-Our data suggest that the TEF-TBP interaction may be essential for TEF function. Three TEF-1 domains have been identified which are required for its transcriptional activation and autointerference activities in HeLa cells (Hwang et al., 1993). These domains include a COOHterminal zinc finger, STY-, and proline-rich domains (Fig. 8).
To ascertain whether any of these domains might be important for the TEF-1-TBP interaction, we analyzed the TEF-1 domains that were required for TBP binding. Several in-frame GST-TEF-1 deletion mutants ( Fig. 8 and Fig. 9, A and B) were expressed in E. coli (Fig. 9B), and glutathione-Sepharose beads coated with truncated GST-TEF-1 were incubated with in vitro-generated 35 S-labeled TBP and washed, and the bound TBP was eluted and analyzed by gel electrophoresis (Fig. 9B). All of the fusion polypeptides were expressed at similar levels except GST-TEF 1/133 and GST-TEF 340/426 , which were more abundantly expressed (Fig. 9B) and which did not bind TBP appreciably (Fig. 9C). All of the polypeptides that contained the proline-rich domain (amino acids 143-204) bound TBP, indicating that this region was essential for TBP binding. Those polypeptides that included the region downstream of the proline-rich domain, particularly the zinc finger domain, bound TBP with the highest apparent affinity. However, the zinc finger domain alone did not bind TBP. These data suggest that cooperative interactions between the proline-rich region and downstream amino acid sequences, including the zinc finger and possibly the STY-rich domain, are required for optimal TBP interaction. Thus, previously identified regions required for transactivation and squelching in HeLa cells (Hwang et al., 1993) are required to bind TBP.
Overexpression of TBP Inhibits TEF-1-mediated Transrepression of hCS Promoter Activity-To determine whether the TBP-TEF-1 interaction that was established by the in vitro experiments (above) represented a physiologically relevant interaction, we performed cotransfection experiments with TBP and TEF-1 expression vectors. We reasoned that if TEF-1 inhibited hCS promoter function by interactions with TBP, then overexpression of TBP might be able to sequester TEF-1 away from functional TBP-containing holo-TFIID complexes. As illustrated in Fig. 10, in the absence of exogenous TBP, transfection of 5,000 ng of pXJ40-TEF-1 plasmid inhibited hCS promoter activity by 95%. Inclusion of 200 and 1000 ng of the TBP expression plasmid hIID-CX reduced basal activity of the hCS promoter activity by 24.3 and 50.5%, respectively, indicating that TBP overexpression by itself may cause some "squelching" activity. However, the transrepression elicited by 5,000 ng of the pXJ40-TEF-1 plasmid was reduced dramatically to 64.8 and 30.2% (not significantly different from control), respectively, by the inclusion of 200 and 1,000 ng of the hIID-CX plasmid. Thus overexpression of TBP can reverse the repressive actions of TEF-1 overexpression, indicating that TBP and TEF-1 interact functionally in vivo. DISCUSSION TEF-1 is proposed to transactivate a variety of eukaryotic enhancers and promoters, including the SV40 (Davidson, 1988;Xiao et al., 1991;Hwang et al., 1993) and hCS (Walker et al., 1990;Jacquemin et al., 1994;Eberhardt, 1994, 1995) enhancers, muscle-specific genes (Shimizu et al., 1993;Kariya et al., 1993;Stewart et al., 1994), human papillomavirus-16 E6 and E7 oncogenes (Ishiji et al., 1992), and mouse early developmental genes (Melin et al., 1993). Since overexpression of TEF-1 in various cells results in transcriptional squelching (Xiao et al., 1991;Ishiji et al., 1992;Hwang et al., 1993;, it has been proposed that other limiting transcription factors along with TEF-1 are required for transactivation. Recently we found that another factor, CSEF-1, possibly unrelated to TEF-1, is correlated with CSEn transactivation in choriocarcinoma cells (BeWo). In these cells TEF-1 overexpression failed to activate the hCS enhancer and repressed hCS promoter activity in the absence of the enhancer . These results mirror those found with the myosin heavy chain ␤ gene, in which two distinct factors, a ubiquitous mouse TEF-1 homolog and an unrelated musclespecific factor, bind to the GT-IIC-related element, and the ubiquitous TEF-1 homolog failed to transactivate myosin heavy chain ␤ gene expression in mouse skeletal muscle cells (Shimizu et al., 1993). Also, Stewart et al. (1993) have shown that GAL4 chimeras containing a novel isoform of chicken TEF-1 which has 13 additional COOH-terminal amino acids can transactivate GAL4-dependent reporter genes, whereas chimeras corresponding to the ubiquitous hTEF-1 isoform only exhibit squelching activity. These studies raise the possibility that the dominant function of the ubiquitous form of TEF-1 is a repressor.
The present studies expand our previous observation that TEF-1 represses hCS promoter activity independently of the presence of GT-IIC motifs . Consistent with the concept that TEF-1 acts at the level of transcription, steady-state levels of luciferase transcripts were diminished in BeWo cells cotransfected with the hCSp.LUC and CMVp.TEF-1 genes (Fig. 3). Repression was not restricted to the hCS promoter, since it occurs with the heterologous TK and RSV promoters (Fig. 2, B and C), which have not been previously known to respond to TEF-1. Furthermore, TEF-1 inhibits transcription from the minimal hCS promoter (Fig. 2 A), which contains a TATA box and an initiator element . Equivalent repression was observed with constructs lacking either the TATA or initiator element ( Fig. 2A). Thus TEF-1-mediated repression may involve interactions with basal transcriptional factors that can function with both TATA and initiator elements. The TEF-1-mediated repression could be overcome by cotransfecting antisense oligonucleotides (Fig.  4). More importantly, the data demonstrate that endogenous levels of TEF-1 negatively regulate hCS promoter activity in BeWo cells, since antisense oligonucleotides increased hCS promoter activity in the absence of cotransfected CMVp.TEF-1 (Fig. 4). This finding suggests that TEF-1 does not function as a transactivator in these cells.
Previous studies suggested that TEF-1 might interact with TBP. First, in studies primarily directed at SV40 T antigen-TEF-1 interactions, it was observed that TBP could bind to TEF-1 (Gruda et al., 1993). Second, Brou et al. (1993) had shown that the activity of GAL-TEF-1 chimeras appeared to be mediated by at least two distinct classes of TFIID complexes. Using GST-TEF-1 pull-down assays we confirmed the observation of Gruda et al. (1993), demonstrating that TEF-1 specifically interacts with hTBP, but not hTFIIB or as a control the

TEF-1-TBP Interactions Mediate Transrepression
human TR␤ receptor (Fig. 5C). Interestingly, TBP was able to form supershifted complexes with TEF-1 and DNA containing the GT-IIC enhanson (Fig. 6); however, TEF-1 inhibited the ability of TBP to bind to the TATA element (Fig. 7), suggesting that TEF-1 binding to TBP might inhibit preinitiation complex formation. We localized an essential region for binding TBP to the proline-rich sequences located between amino acids 143 and 204 (Fig. 9C). In addition, polypeptides containing the zinc finger domain bound TBP with higher affinity (Fig. 9C). Thus two of the three TEF-1 domains shown to be required for its transactivation and transrepression in HeLa cells (Hwang et al., 1993) are required for optimal TBP binding. Our studies indicate that the domains containing the proline-rich and zinc finger sequences link TEF-1 to the basal transcriptional machinery through its ability to bind TBP.
On both the TATA-containing and non-TATA-containing promoters, TBP as part of the TFIID complex is the first component to enter the preinitiation complex. The association of TBP with the promoter may be rate-limiting (Chatterjee and Struhl, 1995;Klages and Strubin, 1995). Accordingly, it is not surprising that a large number of viral and cellular transcriptional factors have been found to interact with TBP (Table II). These factors represent various classes of DNA-binding and non-DNA-binding proteins, and their interaction with TBP produces a variety of responses. For example, NC1, NC2, and DBF4 (Meisterernst et al., 1991) and Dr1 (Inostroza et al., 1992) do not affect the binding of TBP with DNA, but they do prevent the association of TBP with TFIIA and/or TFIIB. TFIIA increases the stability of the TBP-TATA complex (Yokomori et al., 1994) and in conjunction with TBP may be essential for committing a complex to transcription (Auble and Hahn, 1993). A yeast transcription factor, ADI, disrupts the TBP-TATA complex in an ATP-dependent manner, but this effect is blocked by prior association of TFIIA (Auble and Hahn, 1993). The HBV protein X-TBP interaction is also inhibited by nonhydrolyzable ATP analogs, suggesting that protein interactions with TBP which are ATP-dependent may be common. The Drosophila p230 subunit of TFIID inhibits the TBP TATA binding activity and represses transcription (Kokubo et al., 1994), activities that are shared by TEF-1.
Importantly, the TEF-1-mediated transrepression of the hCS promoter is alleviated by the cotransfection of increasing amounts of TBP expression plasmids (Fig. 10), indicating that TBP and TEF-1 functionally interact in vivo. This result provides strong support for the concept that the in vitro TBP-TEF-1 interactions observed in these studies represents a physiologically relevant interaction and suggests, but does not prove, that TEF-1 can interact with holo-TFIID. It is noteworthy that when overexpressed in eukaryotic cells, like TEF-1, many of the TBP-binding factors (Table II) cause strong squelching effects on heterologous promoters. Tax1 transactivates several viral and cellular promoters and enhancers through specific DNA elements. Tax1 interacts specifically with TBP and exerts strong squelching effects on transcription (Caron et al., 1993). Like the TBP-mediated inhibition of TEF-1 repression (Fig. 10), overexpression of TBP was able to stimulate the transactivation of GAL4-Tax1 chimeras and partially alleviates Tax1-mediated squelching (Caron et al., 1993). Also, c-Fos-and FosB-mediated inhibition of transcription can be partially relieved by overexpression of TBP (Metz et al., 1994a(Metz et al., , 1994b. These studies suggest that TBP levels may be limiting, in which case squelching mechanisms involving additional factors are considered unlikely to explain the transcription inhibition. However, although TBP overexpression increased p65/ NFB-mediated transactivation, neither TBP nor TFIIB overexpression was able to alleviate p65/NFB-mediated squelching (Schmitz et al., 1995), suggesting that additional proteins were required for transactivation in this case. Also, in BeWo cells, overexpression of TBP itself results in reduced promoter activity, suggesting that other limiting factors may FIG. 11. Comparison of identified TBP binding domains among transcription factors, whose interaction with TBP is essential for transcriptional activity. The protein sequences (Swissprotein data base) were initially aligned with the program Pileup (Wisconsin Sequence Analysis Package), and additional gaps were introduced manually to obtain maximal alignment. Underlined sequences in p230 have been shown to be important for TBP binding (Kokubo et al., 1994), and the arrows identify sites that have been demonstrated to be important for TBP binding in c-Fos (Metz et al., 1994b). Uppercase letters indicate homology between two sequences; uppercase bold letters indicate homology among three or more sequences. be involved in mediating hCS promoter function. Metz et al. (1994b) have provided evidence for the existence of a TBP binding motif that is shared by VP16, E1A, and c-Fos. We therefore compared the essential proline-rich TEF-1 domain and a number of other transcription factors that were known to bind to TBP and whose TBP binding domain had been narrowly mapped. Alignment of the sequences around the TBP binding motif, (F/L)V(F/L)D, indicates that TEF-1 shares some homology with these TBP-binding proteins (Fig. 11). Interestingly, TEF-1, p53, and VP16 show a region of homology just downstream of the putative TBP binding motif in a proline-rich domain, suggesting that these factors are more closely related, and the proline-rich region might constitute part of a conserved TBP binding domain.
Although TEF-1 blocks the formation of TBP-TATA complexes, this does not exclude the possibility that TEF-1 can serve as a transactivator in certain cell types. In such cells, additional cofactors may prevent TEF-1 inhibition of TBP-TATA complex formation in a manner similar to TFIIA inhibition of ADI-mediated disruption of the TBP-TATA complex (Auble and Hahn, 1993). However, the data do support the concept that TEF-1 may be an exclusive repressor in some cell types. This conclusion is strengthened by our recent findings that a 30-kDa GT-IIC-binding factor, apparently unrelated to TEF-1, is correlated with hCS enhancer activity in COS-1 cells, which express very low levels of TEF-1 . Accordingly, it is possible that TEF-1 provides a counterregulatory stimulus to the actions of other factors that mediate transactivation through the GT-IIC enhanson.