The Rous Sarcoma Virus Long Terminal Repeat Promoter Is Regulated by TFII-I

Abstract

Many viral genes contain core promoters with two basal control elements, the TATA box and the pyrimidine-rich initiator (Inr). However, the molecular mechanisms involved in transcription initiation from composite core promoters (TATA⁺ Inr⁺) containing Inr elements are unclear. The Rous sarcoma virus (RSV) long terminal repeat (LTR) contains a transcriptionally potent enhancer and core promoter composed of a TATA box and an Inr-like sequence, termed the transcription start site core (TSSC). Previously we demonstrated that the TSSC binds the multifunctional Inr-binding protein YY1. Here we present evidence that the TSSC also binds the multifunctional transcription factor TFII-I and that both TFII-I and YY1 are required for RSV LTR transcriptional activity. Gel shift assays using anti-TFII-I antibody show that TFII-I is present in a protein complex that specifically binds to the TSSC. Mutations in the TSSC that reduce TFII-I binding also reduce RSV LTR enhancer and promoter activity. Transient-transfection assays demonstrate that TFII-I transactivates the RSV LTR from ca. fourfold (basal) to ca. sevenfold (enhanced) in both human and natural host cell lines. Importantly, the activity of the TSSC element can be attributed to the binding activity of TFII-I and the YY1 protein, since mutation of each of these binding sites within the TSSC element abolishes all viral expression as demonstrated by transient-transfection assays. Taken together, these data demonstrate that expression of RSV viral mRNA is dependent on both TFII-I and YY1.

Transcription initiation of protein-coding genes is the cornerstone for understanding the complexity of gene expression. The initiation of mRNA synthesis is directed by core promoter elements, most commonly the TATA box, the pyrimidine-rich initiator (Inr) element (38, 41, 56), and/or the recently described downstream promoter element, which is located ∼30 bp distal to the transcription initiation site (4, 5). Each of these core promoter elements can function independently or synergistically to nucleate the formation of a stable preinitiation complex competent for transcription (4, 5, 10, 22, 41). Composite core promoters (TATA⁺ Inr⁺) are typically found in viral genes (36). Indeed, some of the most well-characterized composite promoters are viral promoters, such as the human immunodeficiency virus type 1 (HIV-1) promoter (31, 42–45), adenovirus major late promoter (AdMLP) (26, 42, 50), and adeno-associated virus P5 promoter (25, 47, 48, 52, 53). The presence of two core promoter elements may allow the virus to form a more stable preinitiation complex and/or integrate a variety of cellular signals to function optimally under various conditions.

The first step in TATA-directed transcription is recognition of the AT-rich TATA box sequence located ca. 25 to 30 bp upstream of the transcription initiation site by the TATA box-binding protein (TBP) component of the TFIID complex (38, 41). Following template recognition by TBP, this interaction is stabilized by TFIIA, and then the remaining general transcription factors (TFIIB, TFIIF, TFIIE, TFIIH, and TFIIJ) and RNA polymerase II are recruited to the DNA template, completing the formation of the preinitiation complex (39, 41). Despite the extensive knowledge of TATA-directed transcription, the mechanism for Inr-mediated transcription initiation is not as well defined (41, 51). The pyrimidine-rich Inr has a loose consensus sequence of YYA₊₁N(T/A)YY, derived by extensive mutational analyses (15, 19, 25). The factors required for Inr-mediated transcription include those required for TATA-directed basal transcription, and, at least in promoters containing both a TATA box and an Inr, one other factor called TAF150 in Drosophila and CIF150 in humans (18, 20) is required. TAF150/CIF150 stabilizes the TFIID-DNA interaction but does not appear to be responsible for direct recognition of the Inr element (18, 20). However, several other proteins have been proposed as candidates for Inr recognition and shown to contribute to Inr function, including USF (9, 44), RNA polymerase II (6, 41), TBP-associated factors (14, 19, 40, 41, 54), YY1 (44, 52, 53), and TFII-I (7, 16, 28, 31, 34, 35, 37, 42, 45, 57).

TFII-I is a multifunctional transcription factor originally identified as a factor that could bind to the Inr elements present in the AdMLP, HIV-1, and terminal deoxynucleotidyltransferase (TdT) gene promoters (44). The TFII-I protein is a 957-amino-acid phosphoprotein with an apparent molecular mass of 120 kDa (21, 35). The primary structure of TFII-I revealed several novel features of the protein, including six highly conserved direct repeats each approximately 90 amino acids long, a hydrophobic zipper region that is not flanked by a basic region, three clusters of acidic amino acids, and six domains reminiscent of helix-loop-helix motifs present in each direct repeat (13). Currently, only one other protein, MusTRD1, which is highly expressed in skeletal muscle and required for slow muscle fiber-specific gene expression, is known to have homology to TFII-I (37). The role of TFII-I in Inr-mediated transcription was established by experiments showing that TFII-I is required for transcription from both Inr-containing TATA-less promoters and TATA- and Inr-containing promoters (16, 28, 31, 34, 37, 45, 57). However, TFII-I has recently been shown to bind to sites which bear no obvious homology to the pyrimidine-rich Inr, including the E-box Myc site (E-box) (44), the c-sis/platelet-derived growth factor-inducible element (SIE), and the serum response element (SRE) (13). Other studies have implicated a role for TFII-I in cell cycle-regulated gene expression (16) and in serum-inducible transcription from the c-fos promoter (13). Furthermore, TFII-I physically interacts with c-Myc (42), USF (44), serum response factor (SRF), Phox1 (13), Btk (Bruton's tyrosine kinase) (17, 58), and possibly NF-κB (31). Interestingly, TFII-I is identical to BAP-135, a protein involved in X-linked immune deficiency (58), and has been identified in the breakpoint regions of the 7q11.23 Williams-Beuren syndrome deletion (17). The disparate functions of TFII-I probably reflect its multifunctional potential as a transcription factor involved in a broad spectrum of biological activities.

To study the molecular events involved in gene expression, we have employed the Rous sarcoma virus (RSV) long terminal repeat (LTR), which contains a potent enhancer and basal promoter active in multiple cell types (11). We have previously characterized a basal control element, the transcription start site core (TSSC), present in the core promoter of the RSV LTR, that is required for efficient viral expression and is regulated in part by the multifunctional Inr-binding protein, YY1. Three specific protein complexes (A, B, and C) form with the TSSC in gel shift assays (30), and we have previously shown that YY1 is a component of TSSC(C). Here we report the characterization of the protein complex TSSC(A), which binds to the TSSC region overlapping the initiating nucleotide in the RSV LTR promoter. We show that TFII-I is a component of the TSSC(A) complex by gel shift analyses with anti-TFII-I antibody. Mutational analysis of the RSV LTR demonstrates that the TFII-I binding site is necessary for full enhancer and promoter activity. Furthermore, transient-transfection assays overexpressing TFII-I demonstrated that TFII-I can transactivate the RSV LTR basal promoter more than fourfold and the RSV LTR enhancer as much as sevenfold. In addition, point mutations in both the TFII-I and YY1 sites in the RSV LTR core promoter recapitulate the effect of a deletion of the TSSC element. Taken together, these data suggest that RSV LTR transcription initiation is regulated through the TSSC by both YY1 and TFII-I.

MATERIALS AND METHODS

Plasmid construction.

The chloramphenicol acetyltransferase (CAT) reporter gene plasmids p(B)SRA and p(B)e⁻ have been described previously (30). Briefly, p(B)SRA contains RSV LTR sequences from −489 to +103 of the provirus and p(B)e⁻ contains the RSV minimal promoter sequences from −54 to +103 linked to the CAT gene. Plasmid p(B)e⁻_mII-I was created from p(B)e⁻ by using standard site-directed mutagenesis techniques to create specific mutations within the putative TFII-I site (consisting of the sequences +11 to +18 relative to the transcription start site) in the RSV LTR promoter, as described below. Plasmid p(B)SRA_mII-I was prepared as follows. The CAT plasmid p(B)SRA was digested with ClaI, and the digested products were size fractionated on a 1% agarose gel. The appropriate DNA fragment containing the RSV enhancer sequences from −489 to −54 of the provirus was excised and purified as previously described (30). This fragment was then inserted into the ClaI site of p(B)e⁻_mII-I, resulting in plasmid p(B)SRA_mII-I.

Preparation of the luciferase plasmids pSRA-Luc and pe⁻-Luc has been described previously (30). The luciferase plasmids pSRA-Luc_mII-I and pe⁻-Luc_mII-I were derived from the CAT plasmids p(B)SRA_mII-I and p(B)e⁻_mII-I, respectively. The 787-bp HindIII restriction fragment of plasmid p(B)SRA_mII-I (containing the RSV enhancer and minimal promoter) was inserted into the HindIII site of the pGL3-basic luciferase vector (Promega) to create pSRA-Luc_mII-I. Plasmid pe⁻-Luc_mII-I was prepared by inserting the XhoI-HindIII restriction fragment from p(B)e⁻_mII-I into the XhoI-HindIII restriction sites of the pGL3-basic luciferase vector.

The CAT reporter construct, p(B)e⁻_mYY1/mII-I, which contains the 5′YY1 TFII-I double mutation, was derived from p(B)e⁻_mII-I using standard mutagenesis techniques as described below (the pe⁻-Luc_mII-I plasmid was not prepared at this time). Plasmid p(B)e⁻_mYY1/mII-I was then used to produce the corresponding luciferase reporter plasmid pe⁻-Luc_mYY1/mII-I, which was used in the subsequent transient-transfection assays. This was achieved by inserting the 184-bp XhoI-HindIII restriction fragment from p(B)e⁻_mYY1/mII-I into the XhoI-HindIII restriction sites of the pGL3-basic luciferase vector.

The mammalian expression vector for TFII-I was prepared from the bacterial expression vector pET11-d-II-I (the kind gift of A. L. Roy, Tufts University) (45). pET11-d-II-I was digested with NcoI and BamHI to remove the full-length TFII-I cDNA. The digested products were size fractionated on a 0.8% agarose gel. The appropriate DNA fragment was excised and purified described above. The purified TFII-I cDNA fragment was then inserted into the SmaI site of the expression vector pSVK3.

Site-directed mutagenesis.

Standard mutagenesis techniques (23, 24, 55) were used to create specific mutations within the YY1 and TFII-I sites of the RSV LTR as previously described (30). The presence of the specific mutation was confirmed by chain termination DNA sequencing of the plasmids, as described by Sanger et al. (46), using the following oligonucleotides (mutated nucleotides are shown in lowercase bold type): mII-I (−1 to +30), 5′-GCGGTAAACTGGTccccccTGTAACCACACG-3′; and mYY1/mII-I (−8 to +22), 5′-ATGTccccccTGGTCAAAcaaCGTTTATTG-3′.

Cell culture.

HeLa (human cervical carcinoma) cells were maintained in a 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F-12 nutrient mixture (DMEM/F-12) enriched with 10% calf serum and ferric sulfate complex at 6 mg/ml. In addition, the following were added to the medium: penicillin G sodium (25 U/ml), streptomycin sulfate in 0.85% saline (25 mg/ml), and sodium bicarbonate (2.44 g/liter).

QT6 (quail fibrosarcoma) cells were obtained from American Type Culture Collection. QT6 cells were maintained in medium 199 (M199) supplemented with 10% tryptose phosphate broth, 5% fetal bovine serum, and 1% chicken serum. In addition, penicillin G sodium (25 U/ml), streptomycin sulfate in 0.85% saline (25 mg/ml), and sodium bicarbonate (1.7 g/liter) were added to the medium.

Chicken embryo fibroblasts (CEF) were prepared from 10-day-old embryos (SPAFAS) as described by Boulden and Sealy (2, 3). The prepared cells were resuspended and maintained in medium 199 (M199) containing penicillin G sodium (25 U/ml), streptomycin sulfate in 0.85% saline (25 mg/ml), and sodium bicarbonate (1.7 g/liter) and supplemented with 10% tryptose phosphate broth, 5% fetal bovine serum, and 1% chicken serum.

Transfection and enzymatic assays.

Transfections were performed by the calcium phosphate coprecipitation technique as described by Graham and van der Eb (12). HeLa cells were plated in a 60-mm-diameter dish 1 day prior to transfection at a density of 5 × 10⁵ cells/dish. Transfections were performed with 10 μg of plasmids p(B)SRA, p(B)SRA_mII-I, p(B)e⁻, or p(B)e⁻_mII-I and 2.5 μg of pe⁻-Luc_mYY1/mII-I per 5 ml of DMEM/F-12 medium and corresponding amounts of the empty reporter vector p(B)CAT or pGL3-basic in control assays. Cells were exposed to the CaPO₄-DNA precipitate for 24 h, at which time the medium was removed and replaced with fresh supplemented medium and the cells were allowed to grow for 36 to 48 h.

QT6 cells were also transfected by the calcium phosphate coprecipitation technique (12). QT6 cells were plated in a 60-mm dish 1 day prior to transfection at a density of 5 × 10⁵ cells/dish. Transfections were performed with 50 ng of pSRA-Luc or pSRA-Luc_mII-I or with 2.5 μg of pe⁻-Luc, pe⁻-Luc_mII-I, or pe⁻-Luc_mYY1/mII-I and corresponding amounts of the empty reporter vector pGL3-basic in control assays. Cells were exposed to the CaPO₄-DNA precipitate for 6 to 8 h, at which time the medium was removed and replaced with fresh supplemented medium and the cells were allowed to grow for 36 to 48 h.

CEF cells were transfected by the calcium phosphate coprecipitation technique (12) by plating cells into a 60-mm dish 1 day prior to transfection at an approximate density of 1 × 10⁶ cells/dish. Cells were refed with fresh supplemented medium 1 to 2 h prior to transfection either with 50 ng of pSRA-Luc or pSRA-Luc_mII-I, or with 2.5 μg pe⁻-Luc or pe⁻-Luc_mII-I and corresponding amounts of the empty reporter vector pGL3-basic in control assays. Cells were exposed to the CaPO₄-DNA precipitate for 6 to 8 h, at which time the medium was removed and replaced with fresh supplemented medium and the cells were allowed to grow for 36 to 48 h.

Transfections of cells overexpressing TFII-I were performed in 5 ml of DMEM/F-12 medium (HeLa) or in 5 ml of M199 (QT6). Each transfection mixture contained a total of 20.05 μg of DNA for transfections with pSRA-Luc or 22.5 μg of DNA for transfections with pe⁻-Luc, consisting of reporter plus variable amounts of pSVK3/II-I expression vector supplemented with the parental expression plasmid pSVK3 to adjust for equal amounts of input DNA as indicated in the relevant figure legend. Cells were exposed to the CaPO₄-DNA precipitate for 24 h (HeLa) or 6 to 8 h (QT6), at which time the medium was removed and replaced with fresh supplemented medium and the cells were allowed to grow for 36 to 48 h.

All cells were harvested by being washed once in cold (4°C) phosphate-buffered saline, incubated for 15 min at room temperature in PBS buffer containing 5 mM EDTA and 5 mM EGTA, and then scraped from the plates. The cells were collected at 4°C by centrifugation for 5 min at 630 × g. The cell pellet was resuspended in 250 μl of 0.25 M Tris (pH 8) containing 1 mM phenylmethylsulfonyl fluoride (PMSF), lysed by sonication, and clarified by centrifugation for 15 min at 10,000 × g as described by Boulden and Sealy (2).

CAT assays were performed by the method of Nordeen et al. (33), and CAT activity was quantitated by liquid scintillation counting. The Promega luciferase assay system with reporter lysis buffer was used for luciferase assays as specified by the manufacturer. Luciferase activity was assayed by using an Analytical Luminescence Laboratory Monolight 2010 luminometer with 10 μl of cell lysate. The total protein in each sample was determined by Bradford assay, and the reporter activity was normalized to the total protein present in each sample.

Radiolabeled and competitor oligonucleotides.

Oligonucleotides for radiolabeled probes and competitor DNAs were prepared by automated DNA synthesis in the Diabetes Research and Training Center DNA Core (Vanderbilt University) and purified over desalting columns followed by n-butanol extraction. The amount of DNA was quantitated by measurement of the absorbance at 260 nm. Oligonucleotides were labeled using T4 polynucleotide kinase, electrophoresed on a 12% native polyacrylamide gel in Tris-borate-EDTA (TBE), and purified by electroelution as previously described (3). Nonradiolabeled competitor DNAs were prepared, purified, and quantitated as described above. Oligonucleotides with non-wild-type sequence are indicated by lowercase bold type. The oligonucleotides are shown in Table 1.

TABLE 1.

Oligonucleotides used in this study

Oligonucleotide (positions)	Sequence (5′ to 3′)
TSSC (−5 to +26)	TAAACGCCATTTTACCATTCACCACATTGGT
	ATTTGCGGTAAAATGGTAAGTGGTGTAACCA
TSSCmII-I (−5 to +26)	TAAACGCCATTTTACCAggggggACATTGGT
	ATTTGCGGTAAAATGGTccccccTGTAACCA
TSSCm3′YY1 (−5 to +26)	TAAACGCCATTTTAttgTTCACCACATTGGT
	ATTTGCGGTAAAATaacAAGTGGTGTAACCA
TSSCmYY1/mII-I (−5 to +26)	TAAACGttgTTTTACCAggggggACATTGGT
	ATTTGCaacAAAATGGTccccccTGTAACCA

Antibodies.

Antibody to TFII-I was a generous gift from A. L. Roy. The anti-TFII-I antibody was raised in rabbits against a synthetic polypeptide corresponding to the putative DNA-binding domain of TFII-I (28, 44). The polyclonal serum was obtained from a 10-week-old bleed and specifically recognizes the 120-kDa polypeptide in Jurkat nuclear extract and purified TFII-I (28).

Electrophoretic mobility shift assays.

Electrophoretic mobility shift assays were performed with HeLa cell nuclear extracts previously prepared by the method of Shapiro et al. (49). Electrophoretic mobility shift assay mixtures contained 1 μl of HeLa nuclear extract diluted 1:10 with nuclear dialysis buffer (10 mM morpholineethanesulfonic acid [MES], 0.1 mM EDTA, 50 mM NaCl, 50% glycerol), 1.25 μg of poly(dI-dC)-poly(dI-dC), and 0.5 ng of TSSC ³²P-labeled DNA in a final volume of 20 μl of binding buffer containing 10 mM HEPES (pH 8), 5 mM Tris (pH 7.9), 2 mM dithiothreitol, 1 mM EDTA, 50 mM NaCl, and 20% glycerol. Nonradiolabeled competitor DNAs and probe were added to the reaction mixture simultaneously, and the HeLa nuclear extract was always added last. Gel shift assays using anti-TFII-I antibody were performed by preincubating 1 μl of HeLa nuclear extract, diluted as described above, with 1 μl of antibody diluted 1:20 with 1% bovine serum albumin for 10 min at 4°C with occasional mixing. The reaction was carried out in the presence of a protease inhibitor mix containing 0.1 mM pepstatin, 10 mM β-glycerol phosphate, 0.1 mM sodium vanadate, 1 mM phenylmethylsulfonyl fluoride, 5 μg of leupeptin per ml, and 1 μg of aprotinin per ml. The antibody-extract mixture was then processed as described above. All binding reactions were carried out for 30 min at room temperature, and the products were electrophoresed on a native 6% polyacrylamide gel containing 25 mM Tris base, 190 mM glycine, and 1 mM EDTA (pH 8). Gels were subsequently dried for autoradiography.

RESULTS

TFII-I is a component of the TSSC(A) complex.

TFII-I was originally shown to bind to the initiator elements of the AdMLP, HIV-1, and terminal deoxynucleotidyl transferase (TdT) promoters (44). Footprinting analyses and binding competition studies with the Inr elements from these promoters suggested a consensus TFII-I Inr binding sequence of 5′-YAYTCYYY-3′ (1, 44). TFII-I has been proposed as a key regulator in Inr-mediated transcription (7, 28, 34, 42, 43, 57). Since the TSSC exhibits some Inr-like functions and is regulated by the Inr-binding protein YY1 (30), we inspected it for sequence similarities to the consensus TFII-I Inr binding motif. Figure 1B shows that an almost perfect match (7 out of 8 residues) to the TFII-I motif is present in the TSSC element. The TFII-I site within the TSSC overlaps a weak YY1 consensus site (3′ YY1 site) (5′-TACCATTCAC-3′); however, in our previous study (30) the YY1 protein was shown to bind exclusively to another, higher-affinity YY1 consensus site just upstream (5′ YY1 site) (5′-CGCCATTTT-3′) (30). To determine whether the TFII-I motif was important for factor binding to the TSSC element, we performed gel shift assays with HeLa nuclear extract, TSSC DNA, and either wild-type, mII-I, or m3′YY1 TSSC oligonucleotides as competitor (Fig. 1A). Although the specific residues required for TFII-I binding to DNA have not been determined, the TSSC_mII-I oligonucleotide replaces the last 6 residues of the TFII-I consensus sequence with guanine residues, presumably altering the binding of TFII-I to its consensus site. TSSC_m3′YY1 oligonucleotide was also used as competitor, since the TFII-I consensus site overlaps this sequence. The mutation within the 3′YY1 site is consistent with data showing that such mutations prohibit YY1 binding (15). Binding of both TSSC(C) (or YY1) and TSSC(B) complexes was diminished equivalently by addition of a 100-fold molar excess of wild-type TSSC or either of the two mutant oligonucleotides as competitor (Fig. 1A, lanes 2 to 4). However, TSSC(A) binding was not affected by the addition of mII-I DNA (lane 3). This suggests that only TSSC(A) requires a TFII-I-binding site for complex formation. Impaired competition of TSSC(A) with the m3′YY1 oligonucleotide as competitor suggests that nucleotides within the 3′YY1 site may also be necessary for TSSC(A) complex formation. These data are consistent with previous results which also indicated that TSSC(A) binding was dependent on sequences within the 3′YY1 site (30). We also performed gel shift experiments using the TSSC_mII-I oligonucleotide as labeled probe and the oligonucleotides in Table 1 as competitors. As anticipated, the TFII-I mutation prohibits binding of the TSSC(A) factor to the DNA (lane 5). Also, consistent with the competition analyses with wild-type TSSC, the TFII-I mutation had no effect on the binding activities of TSSC(B) and TSSC(C), since each complex was detected (lane 5).

FIG. 1.

TFII-I is a component of the TSSC(A) complex. Gel shift assays were performed with HeLa nuclear extract, 1.25 μg of poly(dI-dC):poly (dI-dC), and 0.5 ng of ³²P-labeled DNA as described in Materials and Methods. (A) Either no competitor (lanes 1 and 5) or a 100-fold molar excess of the indicated competitor oligonucleotide was added to the binding-reaction mixtures (lanes 2 to 4 and lanes 6 to 8). (B) Sequences of the oligonucleotides used in the gel shift analysis are shown. The TFII-I-binding site is boxed. Mutations are shown in lowercase bold type. The TFII-I consensus sequence is also shown, where Y is a pyrimidine residue. (C) HeLa nuclear extract was mixed at 4°C for 10 min with 1 μl of preimmune serum (lane 2) or 1 μl of anti-TFII-I peptide antibody diluted 1:20 with 1% bovine serum albumin (lane 3) in the presence of a protease inhibitor mix. The protein samples were then mixed with 0.5 ng of ³²P-labeled TSSC as described in Materials and Methods. Specific TSSC protein-DNA complexes are indicated.

Having established that the TSSC TFII-I site is necessary for TSSC(A)-binding activity, we tested whether the TSSC(A) complex is immunologically related to the TFII-I protein. To do this, gel shift assays were performed with HeLa nuclear extract preincubated with an anti-TFII-I peptide antibody that recognizes the putative DNA-binding domain of the TFII-I protein (28). The results of these assays are shown in Fig. 1C. The TSSC(A) band was abrogated by the anti-TFII-I antibody (Fig. 1C, lane 3) but was unaffected by the control preimmune serum (lane 2). This is consistent with previous reports showing that the TFII-I antibody blocks TFII-I binding without producing a supershifted band (28). Taken together, these data indicate that the TSSC(A) complex contains the TFII-I transcription factor and requires both the TFII-I consensus site and 3′YY1 core motifs to bind DNA.

RSV LTR transcription is decreased by mutation of the TFII-I-binding site.

To correlate the DNA-binding activity of TFII-I at the TSSC element with RSV transcriptional activity, we assessed the effect of the TFII-I mutation in vivo by transient-transfection assays. The transcriptional activity of RSV enhancer and promoter constructs containing the TFII-I mutation that disrupted TSSC(A) complex formation was examined in three different cell lines, HeLa, QT6, and CEF. Transfections in HeLa cells were performed with the CAT reporter constructs p(B)SRA, p(B)SRA_mII-I, p(B)e⁻, and p(B)e⁻_mII-I. The activity of the TFII-I mutants was compared to the activity of their wild-type counterpart (set at 100%). The luciferase reporter constructs, pSRA-Luc, pSRA-Luc_mII-I, pe⁻-Luc, and pe⁻-Luc_mII-I, were used to transfect QT6 and CEF cells. The RSV luciferase constructs are analogous to their corresponding CAT plasmids, except that they drive the expression of the luciferase reporter gene. pSRA-Luc_mII-I and pe⁻-Luc_mII-I also contained the TFII-I mutation, which prohibited TSSC(A) binding in gel shift analyses. The results of the transfections performed with HeLa cells are shown in Fig. 2A. Mutation of the TFII-I site reduced the transcriptional activity of the enhancer-containing promoter to 37% and that of the enhancerless promoter to 50% of wild-type activity. The effect of the enhancer was not eliminated by mutation of the TFII-I site, since the activity of the pSRA-Luc_mII-I remained ∼72-fold higher than the activity of pe⁻-Luc_mII-I (data not shown). This indicates that in HeLa cells, TFII-I is not essential for enhancer activity, which is consistent with the role of TFII-I in basal transcription. Parallel studies performed with natural host cell lines for RSV produced similar results for the enhancer-containing constructs. As shown in Fig. 2B, mutation of the TFII-I site reduced transcription from the enhancer-containing plasmids to 45% in QT6 cells and 47% in CEF cells. The twofold reduction in transcriptional activity observed in QT6 and CEF cells with the enhancer-containing constructs is similar to the results obtained with HeLa cells. Interestingly, the effect of the TFII-I mutation on RSV transcription from the minimal promoter was much more dramatic in the natural host cell lines than in HeLa cells. Mutation of the TFII-I site reduced activity to 24% in QT6 cells and to 8.0% in CEF cells. The dramatic decrease of transcriptional activity from the minimal promoter TFII-I mutants is similar to the effect observed in QT6 cells from the enhancerless promoter when the 5′YY1 site was mutated and is consistent with the effect observed when the TSSC element was deleted (30). Apparently, the minimal promoter is much more dependent on basal factors for transcriptional activity in a natural host cell line than in HeLa cells. This may reflect cell-type-specific differences in the ability of other factors to compensate for the loss of TFII-I when present in a specific cellular environment. In any case, the TFII-I-binding site is required to maintain wild-type levels of RSV transcriptional activity.

FIG. 2.

Mutation of the RSV LTR TFII-I site impairs transcription in HeLa cells and avian cells. (A) HeLa cells were transfected with 10 μg of either CAT reporter plasmid p(B)SRA, p(B)SRA_mII-I, p(B)e⁻, p(B)e⁻_mII-I, or the empty CAT reporter vector p(B)CAT as described in Materials and Methods. (B) Transient-transfection assays were performed by introducing either 50 ng of the luciferase reporter constructs pSRA-Luc or pSRA-Luc_mII-I or 2.5 μg of the luciferase reporter constructs pe⁻-Luc or pe⁻-Luc_mII-I into QT6 and CEF cells. For both experiments, protein extracts were prepared from cells harvested 36 to 48 h posttransfection as described in Materials and Methods. The reporter activity was calculated and plotted relative to the corresponding wild-type construct, which was set at 100%. The graph represents data from at least three separate experiments. The relative luciferase units (RLU) and standard error for each experiment are shown. The star indicates the construct used for normalization, which therefore does not contain standard error.

Overexpression of TFII-I exerts a positive effect on RSV transcription.

We determined that TFII-I is a component of the TSSC(A) complex and correlated binding to transcriptional activity by using functional assays. To directly examine the role of TFII-I in RSV transcription, we tested whether overexpression of the protein could transactivate the RSV LTR. Transient-transfection assays were performed with HeLa and QT6 cells by using increasing amounts of the TFII-I expression plasmid pSVK3/II-I and either the enhancer-containing reporter construct pSRA-Luc or the minimal reporter construct pe⁻-Luc. The results of the experiments performed with the enhancerless luciferase reporter plasmid pe⁻-Luc are shown in Fig. 3A. We found that overexpression of TFII-I stimulated transcription from the minimal promoter in a dose-dependent manner. In HeLa cells, overexpression of TFII-I activated transcription, with the maximal response producing nearly a fivefold activation with respect to the pe⁻-Luc construct plus the maximal amount of parental or empty expression plasmid. A similar level of stimulation (fourfold) was observed in QT6 cells with the minimal promoter. Parallel experiments were performed with the enhancer-containing reporter construct pSRA-Luc, and the results of these assays are shown in Fig. 3B. Compared to control experiments with the pSRA-Luc reporter plus parental expression vector, we observed a dose-dependent response in HeLa cells ranging from a three- to a sevenfold activation. TFII-I also trans-activated the RSV enhancer and promoter in QT6 cells, although not to the same extent as observed in HeLa cells. A three- to fourfold stimulation was observed in QT6 cells at the maximal amount of TFII-I expression vector. This is in contrast to the effect observed with overexpression of TFII-I using the enhancerless construct in HeLa cells and QT6 cells, where the stimulation in each of these cell lines was comparable. The significance of this observation is not clear; however, it may reflect cell-type-specific differences in enhancer function. The overexpression studies do not directly assay the specific nucleotides required for TFII-I activity, and therefore we cannot completely rule out the possibility that TFII-I is acting indirectly. However, taken together with the mutational analyses of the TFII-I motif in the TSSC, these data provide strong evidence that TFII-I mediates its activity through the TFII-I-binding site. Furthermore, it should be noted that the observed inductions may underestimate the functional contribution of TFII-I, since the RSV LTR is very active in each of these cell lines and, at least in HeLa cells, TFII-I is abundantly expressed.

FIG. 3.

Overexpression of TFII-I transactivates the RSV LTR enhancer and promoter. TFII-I was tested for its ability to transactivate the RSV LTR enhancer and promoter. (A) HeLa cells or QT6 cells were cotransfected with 2.5 μg of the reporter construct pe⁻-Luc or the pGL3-basic empty reporter vector and the indicated amounts of the TFII-I expression vector pSVK3/II-I as described in Materials and Methods. (B) HeLa cells or QT6 cells were cotransfected with 50 ng of the reporter construct pSRA-Luc or the pGL3-basic empty reporter vector and the indicated amounts of the TFII-I expression vector pSVK3/II-I as described in Materials and Methods. In each experiment, the total amount of DNA was kept constant by addition of the empty expression vector pSVK3. Protein extracts were prepared from cells harvested 36 to 48 h posttransfection, and luciferase activity was measured and then normalized to total protein for each sample as described in Materials and Methods. The graph represents the fold activation of at least three independent experiments (measured in relative luciferase units [RLU]). The fold activations were calculated relative to the experiment indicated by a star (whose fold activation was set at 1). The standard error for each experiment is indicated.

YY1/TFII-I double mutants severely impair transcription from the RSV LTR promoter.

We have previously demonstrated that deletion of the TSSC results in a complete loss in RSV LTR activity. Mutation of the 5′YY1-binding site in the TSSC did not completely eliminate TSSC function but indicated that at least 50% of TSSC activity is attributable to the transcription factor YY1 (30). Therefore, we tested whether a YY1/TFII-I double mutant could duplicate the effect observed by deletion of the TSSC element. First, we examined the DNA-binding properties of a TSSC mutant oligonucleotide, TSSC_mYY1/mII-I, that contained a mutation in both the 5′YY1 site and the TFII-I site (Fig. 4A). The mutations in these sites are identical to those used in gel shift assays described earlier for YY1 (30) and TFII-I (Fig. 1), where binding for each protein was individually abolished by the factor-specific mutation. As shown in Fig. 4A, TSSC(B) bound to the TSSC_mYY1/mII-I mutant but the TSSC(A) and TSSC(C) complexes did not (lane 4). Hence, neither YY1 nor TFII-I can bind to the TSSC element in the presence of these mutations. Furthermore, addition of a 100-fold molar excess of TSSC_mYY1/mII-I did not significantly compete the TSSC(A) or TSSC(C) complexes (compare lanes 1 and 2 to lane 3). Next, we tested the functional consequence of the loss of YY1- and TFII-I-binding activity on RSV transcription. Site-directed mutagenesis was used to generate the luciferase reporter construct pe⁻-Luc_mYY1/mII-I. This plasmid contains the RSV minimal promoter with the 5′YY1 and TFII-I mutations that prohibited binding driving the expression of the luciferase reporter gene. The effect of the double mutant on RSV transcriptional activity was evaluated in vivo by transient transfection of the wild-type plasmid pe⁻-Luc and the double-mutant plasmid pe⁻-Luc_mYY1/mII-I into HeLa and QT6 cells (Fig. 4B). As a control for background activity, the empty reporter vector pGL3-basic was also transfected into each cell line. The activity of the empty reporter and the double-mutant plasmids was normalized to wild-type activity, which was set at 100%. Transcription from pe⁻-Luc_mYY1/mII-I in both HeLa and QT6 cells was reduced to levels comparable to those in the control reactions with the empty reporter plasmid pGL3-basic (HeLa, 13.5% ± 1.6%; QT6, 17.4% ± 1.2%). These results are comparable to the results observed when the TSSC was deleted (30), suggesting that YY1 and TFII-I together mediate TSSC function.

FIG. 4.

Specific mutation of the YY1- and TFII-I-binding sites duplicates the effect of the TSSC deletion on RSV LTR transcription. (A) Gel shift assays were performed with HeLa nuclear extract and either TSSC wild-type or TSSCm_YY1/mII-I DNA as probe. Lanes 1 and 4 contain no competitor DNA. Lanes 2 and 3 contain 100-fold molar excess amounts of TSSC and TSSC_mYY1/mII-I oligonucleotide, respectively. The positions of the TSSC complexes are indicated by arrows. The complex indicated by a star has not been identified but could arise from a proteolytic fragment of TFII-I or another component(s) of the TSSC(A) complex. The sequence of the TSSC element (TSSC) and the YY1 and TFII-I mutations within the TSSC element (mYY1/mII-I) are shown below. The 5′- and 3′-YY1 sites in the RSV LTR TSSC element are indicated by black bars, and the TFII-I site is boxed. The mutated sequence is shown in lowercase bold type. (B) Transient-transfection assays were performed with HeLa and QT6 cells by introducing 2.5 μg of the reporter plasmids pe⁻-Luc or pe⁻-Luc_mYY1/mII-I into each cell type. Luciferase activity was calculated from protein extracts prepared from cells harvested 36 to 48 h posttransfection and normalized to total protein as described in Materials and Methods. The luciferase activity is plotted relative to the activity of the corresponding wild-type, which was set at 100%. The graph represents data from at least three separate experiments. The normalized luciferase activity (measured in relative luciferase units [RLU]) and standard error for each experiment is indicated.

DISCUSSION

Much of our understanding of Inr-mediated transcription has come from the characterization of a variety of genes regulated by various Inr-binding proteins through different Inr elements. Thus, the specific molecular events that mediate Inr-directed transcription remain elusive, due in part to the heterogeneity of protein factors binding to Inr sites and its loose consensus sequence. In our characterization of the RSV LTR transcriptional control mechanisms, we have previously identified an Inr-like sequence termed the TSSC, which functions similarly to a bona fide initiator (30).

To identify factors that may regulate RSV expression through the TSSC element, we inspected the TSSC sequence for homology to well-characterized Inr-binding proteins. The TSSC contains two copies of the core sequence 5′-CCAT-3′, which is recognized by the YY1 transcription factor. We previously demonstrated that YY1 binds the 5′ CCAT motif. This binding is necessary for RSV promoter function in vitro and in vivo, although YY1 is not by itself responsible for TSSC function. A nearly perfect match to the consensus sequence for TFII-I (YAYTCYYY (1, 44) is located within the TSSC element juxtaposed to the 5′YY1 site and overlapping the 3′YY1 site (30). Competition binding assays and immunological techniques verified that TFII-I binds specifically to the TSSC and that it is a component of the TSSC(A) complex. Disruption of TFII-I binding impaired both enhancer-driven and enhancerless RSV promoter activity, although the magnitude of the effect was found to be cell type dependent. In our analysis of the RSV LTR in HeLa cells, we observed a reduction in promoter strength by more than 50% upon mutation of the TFII-I-binding site. The reduction in transcriptional activity was similar with both the enhancer-containing construct and the minimal promoter construct. Interestingly, in QT6 and CEF cells, mutation of the TFII-I site in the enhancerless construct reduced promoter strength to near background levels. TFII-I is widely expressed, but the expression level of the protein varies depending on cell type (13). Thus, the greater dependence of the minimal promoter on an intact TFII-I-binding site in some cells may reflect different requirements for certain basal factors under some cellular conditions. In any case, the dual mutation of both 5′ YY1- and TFII-I-binding sites in the TSSC reduced RSV promoter function to background levels in both HeLa and QT6 cells, recapitulating the effect of a TSSC deletion (30). Therefore, we conclude that TSSC function in the RSV promoter requires both YY1 and TFII-I, although the relative contributions of each factor can vary depending on the cell type and on whether the upstream enhancer is present.

The TFII-I binding site in the TSSC exhibits extensive homology to the Inr elements found in the AdMLP, TdT, and T-cell receptor variable region-derived (Vβ) promoters (TSSC, CATTCACC; AdMLP, CACTCTCT; TdT, CATTCTGG; Vβ, CACTTTCT) (28). The initiating nucleotide in the AdMLP, TdT, and Vβ Inr elements is the conserved adenine residue contained within the TFII-I binding sequence (underlined), but in RSV the TFII-I site is located downstream of the start site (nucleotides +11 to +18). Thus, even though the TSSC TFII-I sequence is nearly identical to the TFII-I sites in these Inr elements, its placement within the promoter is different. This may reflect a unique role for TFII-I in RSV activity versus the activity of TFII-I in a prototypical Inr, since the RSV TSSC does not function identically to well-characterized Inr elements (30). Classical Inr elements are capable of initiating accurate basal levels of transcription independent of a TATA box (28, 50), but the TSSC element, at least in the context of the viral enhancer and promoter, cannot direct specific basal level transcription without the intact TATA box (30).

Interestingly, at least two other gene promoters that do not contain a classical Inr element are directly regulated by TFII-I. Similar to RSV, an atypical Inr element termed the transcription start site region (SSR) in the HIV-1 promoter has been shown to bind TFII-I (59). The HIV-1 core promoter lacks a prototypical Inr but does contain an SSR that influences promoter strength, similar to the RSV TSSC. In the HIV-1 promoter, the SSR is crucial for transcriptional activity but is dependent on the presence of at least a weak TATA box for activity (59). TFII-I is also essential for the transcriptional activity of the KDR/flk-1 promoter (57). TFII-I mediates this activity by binding to a functional Inr within the KDR/flk-1 promoter, although it retains little homology to the classical pyrimidinde-rich initiator sequence or the consensus TFII-I binding site (57). Likewise, the TSSC and the SSR exhibit only partial sequence similarity to the consensus Inr sequence (59). Apparently, TFII-I can mediate transcriptional activity through atypical initiator elements with variable sequence similarity to the consensus Inr or even to the consensus TFII-I-binding site. Little is known about the mechanism of transcription initiation through Inr-like elements that function only in conjunction with a TATA box (41). However, TFII-I is known to play a key role in recruitment of the preinitiation complex to promoters of class II genes that utilize an Inr (42–44). Therefore, further characterization of the role of TFII-I in TSSC function may help to clarify the functional heterogeneity between classical Inr elements and elements like the TSSC that are TATA dependent.

Although TFII-I was originally characterized as an Inr-binding protein (44, 45), it has proven to be a diverse transcriptional factor that regulates genes from upstream as well as basal elements. TFII-I binds to several different promoter elements with no obvious sequence similarity, namely, initiators, E-boxes, and SRE and SIE sites (13, 21, 44, 45). In the HIV-1 and AdMLP promoters, protein interactions between TFII-I and other factors binding to the same region of DNA are important for modulating TFII-I activity (13, 31, 45). Indeed, TFII-I interacts with factors as diverse as Phox1 (13), NF-κB (31), Myc (42), USF (44), SRF (13, 21), STAT1 and STAT3 (21), and potentially YY1 at the RSV LTR promoter. In the AdMLP promoter, TFII-I acts independently and synergistically with USF1 to activate transcription in vivo through E-box elements present in the promoter (45). TFII-I promotes the formation of a Phox1/SRF complex that mediates serum-inducible transcription through the SRE (13). TFII-I activity is also regulated by phosphorylation (35). Tyrosine phosphorylation of TFII-I is dispensable for DNA binding but is required for its Inr-mediated transcriptional activity (35). Presumably, the phosphorylation status of TFII-I is necessary for protein-protein interactions with the basal machinery and/or its translocation to the nucleus (35). In addition, induced tyrosine phosphorylation of TFII-I by epidermal growth factor correlates with activation of the c-fos promoter through upstream elements (21). These observations suggest that TFII-I plays a broader role in regulating gene transcription potentially by integrating regulatory signals from upstream components to the basal machinery and/or by interacting with other transcription factors. Whether TFII-I plays such a role within the RSV LTR is not yet known. However, the RSV LTR enhancer contains SRF-binding sites, and TFII-I can form protein-protein complexes with SRF. An interaction between TFII-I and SRF could be one pathway by which enhancer function is coupled to the promoter in the RSV LTR. This is supported by a sixfold reduction in RSV enhancer activity when the TFII-I site is mutated (data not shown).

Composite core promoters are found primarily in viral genes (36). The presence of both TATA and Inr elements in a promoter could serve to (i) increase the efficiency of TBP recruitment to the promoter by utilizing multiple pathways, (ii) increase the stability of the preinitiation complex on the promoter through multiple contacts at the TATA box and the Inr, (iii) integrate a variety of extracellular signals, or (iv) counteract the effect of repressors by utilizing a separate pathway. Rather than being redundant, the initiator region has been demonstrated to be important in several viral systems. This is evident from studies with the HIV-1 SSR, which demonstrated that the YY1-binding site within this region is important for virion production (29). Studies with the AdMLP initiator element showed that mutations in the Inr and TATA box produced viruses with growth defects. This phenotype was linked to a decrease in transcriptional efficiency (26). Similar findings demonstrating the importance of the initiation site in viral transcription have been observed with the Epstein-Barr virus EBNA-1 initiator (32) and the human cytomegalovirus initiation site (27). For the RSV LTR, the exceptionally potent transactivation potential of the enhancer has been well characterized (8, 11) and the TSSC element has been shown to be essential for both basal and enhanced transcription (30). The activity of the TSSC is mediated through both YY1 and TFII-I. Thus, an understanding of such multifunctional factors as TFII-I in basal and activated transcription from the RSV LTR may help to elucidate the relevancy of Inr-mediated transcription in viral and eukaryotic systems.