Structural and Functional Interactions of Transcription Factor (TF) IIA with TFIIE and TFIIF in Transcription Initiation by RNA Polymerase II*

A topological model for transcription initiation by RNA polymerase II (RNAPII) has recently been proposed. This model stipulates that wrapping of the promoter DNA around RNAPII and the general initiation factors TBP, TFIIB, TFIIE, TFIIF and TFIIH induces a torsional strain in the DNA double helix that facilitates strand separation and open complex formation. In this report, we show that TFIIA, a factor previously shown to both stimulate basal transcription and have co-activator functions, is located near the cross-point of the DNA loop where it can interact with TBP, TFIIE56, TFIIE34, and the RNAPII-associated protein (RAP) 74. In addition, we demonstrate that TFIIA can stimulate basal transcription by stimulating the functions of both TFIIE34 and RAP74 during the initiation step of the transcription reaction. These results provide novel insights into mechanisms of TFIIA function.

Initiation of transcription by RNA polymerase II (RNAPII) 1 proceeds through the formation of a preinitiation complex containing RNAPII and the general transcription factors (TF) TBP (the TATA box-binding protein of TFIID), TFIIB, TFIIE, TFIIF, and TFIIH on promoter DNA (reviewed in Refs. 1 and 2). The first step in preinitiation complex assembly is the recognition of the TATA element of the promoter by TBP. The binding of TBP to the TATA box induces a DNA bend of ϳ90 o (3,4). TFIIB can associate with the TBP-promoter complex (5). Mammalian TFIIF, which is composed of the subunits RAP74 and RAP30, directly binds to RNAPII and has been shown to participate in recruitment of the enzyme to the preinitiation complex (6,7). TFIIE, which is also composed of two subunits called TFIIE56 and TFIIE34, is involved in the melting of promoter DNA at the transcription initiation site through a mechanism that is ATPindependent (8,9). Finally, TFIIH, which has kinase and helicase activities, mediates the ATP-dependent melting of the promoter DNA in the region of the initiation site and is involved in the transition between the initiation and elongation states of the complex (10 -15).
Recent results describing both the structure of the basal transcription machinery and the topological organization of the preinitiation complex have considerably improved our understanding of transcription initiation mechanisms. Determination of the atomic structure of yeast RNAPII at 2.8 angstroms resolution and that of elongating yeast RNAPII at 3.3 angstroms by Kornberg and co-workers (16,17) has revealed key features of both the interaction between the enzyme and template DNA and the basis of its catalytic activity. Analysis of the molecular organization of the preinitiation complex using sitespecific protein-DNA photo-cross-linking has provided insights on the topology of the preinitiation complex containing RNAPII and the general transcription factors (18 -26). Recently, we have proposed a topological model, the DNA wrapping model, which describes transcription initiation by RNAPII (23,26,27). This model accounts for our photo-cross-linking data and several additional data obtained in various laboratories. The DNA wrapping model stipulates that a role for the general transcription factors is to help in the wrapping of the promoter DNA in the preinitiation complex in such a way that a torsional strain is progressively developed upstream of the transcription start site and results in the partial unwinding of the DNA helix. This region of unwound DNA is used as a substrate by the singlestranded DNA helicases of TFIIH that catalyze open complex formation (26).
First isolated as a general transcription factor, TFIIA has a rather controversial role in transcription initiation. Human TFIIA is composed of three subunits: ␣ (35 kDa), ␤ (19 kDa), and ␥ (12 kDa) (28 -31). The ␣ and ␤ subunits are encoded by the same gene and are produced by posttranslational cleavage of a precursor (29,30). In yeast, TFIIA is composed of only two subunits encoded by two different genes, TOA1 and TOA2 (32). The N-terminal part of the polypeptide produced from the TOA1 gene is homologous to the human ␣ subunit, and the C-terminal part is homologous to the ␤ subunit (29,30). TOA2 encodes a polypeptide homologous to ␥ (28,31). The posttranslational cleavage of human ␣/␤ has been demonstrated to be non-essential because wild type activity can be recovered with uncleaved recombinant ␣/␤ and ␥ renatured together (33). TFIIA is not essential for basal transcription in vitro, but it has been shown to stimulate basal transcription in a variety of systems (28,(33)(34)(35)(36)(37). TFIIA binds TBP and increases the affinity of TBP for the TATA box (36 -38). TFIIA can displace certain repressors, including Dr1-DRAP1/NC2, topoisomerase 1, HMG1, and Mot-1, from the TFIID complex, indicating that TFIIA is involved in antirepression (39 -43). Human TFIIA also plays a role in activated transcription, being required for the functioning of some activators (28, 29, 31, 33, 34, 44 -47). For example, TFIIA binds to the activator Zta and mediates its stimulation of TFIID binding to the TATA box (28). Similarly, TFIIA enhances the activation of transcription by the activators Sp1, VP16, and NTF1 (34). The activators VP16 and Zta, which bind TFIIA, stimulate the assembly of a TFIIA-TFIIDpromoter complex, consistent with the roles in vivo of these factors in activated transcription (46). The function of TFIIA in transcriptional antirepression and activation have been separated and is associated with distinct subunits of the factor (35). Subunits ␤ and ␥ are essential for antirepression, whereas ␣ is not. Conversely all three subunits are required for activation.
Previous photo-cross-linking experiments performed with a TBP-TFIIA-promoter complex have revealed that TFIIA makes promoter contacts both in the region of the TATA box and upstream of it (18,20). We have now determined the position of TFIIA in a preinitiation complex assembled in the presence of TBP, TFIIB, TFIIE, TFIIF, and RNAPII. Our results indicate that TFIIA makes promoter contacts not only on the TATA box and upstream of it in the Ϫ40 region, as is observed in the TBP-TFIIA-promoter complex, but also in the ϩ26 region. Given the two extreme promoter positions approached by TFIIA and the small size of TFIIA (67 kDa), these results suggest that TFIIA is located near the cross-point of the wrapped DNA structure, where it can simultaneously contact nucleotides Ϫ40 and ϩ26. TFIIF, TFIIE, and RNAPII also cross-link to the Ϫ40 and ϩ26 positions (23,26), suggesting that TFIIA may directly interact with these factors. We report here that TFIIA directly interacts with RAP74, TFIIE56, and TFIIE34 in addition to the previously determined interaction with TBP. Furthermore, we use an abortive initiation assay to provide evidence that the stimulatory effect of TFIIA on basal transcription is exerted through a stimulation of the activity of RAP74 and TFIIE34 at the initiation stage of transcript formation.
Protein-DNA Photo-cross-linking-The synthesis of the photoreactive nucleotide N 3 R-dUMP, the preparation of the probes, and the conditions for binding reactions were as described (55). Two photoprobes containing the modified nucleotide at positions Ϫ39/Ϫ40 and ϩ26 were used. For each probe, the concentration of poly(dI-dC) in the binding reactions was optimized to favor specific over nonspecific binding. A typical reaction with all the factors contained 200 ng each of TBP, TFIIB, RAP30, RAP74, TFIIE56, TFIIE34, rTFIIA, and RNAPII. UV irradiation, nuclease treatment, and SDS-PAGE analysis of radiolabeled photo-cross-linking products were performed as described previously (55).
Protein-Protein Interactions-Protein-protein interactions were analyzed essentially as we previously described (22). RAP74 wt, RAP74 deletion mutants, RAP30, TFIIE34, TFIIE56, RNAPII, TFIIB, TBP, and bovine serum albumin (BSA) were immobilized on Affi-gel 10 (Bio-Rad) at a concentration of 1 to 5 mg/ml resin. Microcolumns were made with ϳ20 l of this resin. A volume of 50 l of nTFIIA (Fig. 2), recombinant TFIIA␣/␤, recombinant TFIIA␥ (Fig. 3), or rTFIIA (Fig. 4), which contained 100 ng of nTFIIA or 200 ng of TFIIA␣/␤, TFIIA␥ or rTFIIA, was then loaded on the different columns. The flowthrough was collected and the columns successively eluted with 50 l of ACB buffer (10 mM Hepes, pH 7.9, 0.2 mM EDTA, 20% glycerol and 1 mM dithiothreitol) containing 0.1 M, 0.3 M, and 0.5 M NaCl. An aliquot of the input, the flowthrough, and the various salt elutions were analyzed on SDS-PAGE and revealed by silver ( Fig. 2 and Fig. 4) or zinc staining (Fig. 3). The intensity of the bands was evaluated using the UN-SCAN-IT software. It was considered that TFIIA, or one of its subunits, was binding to a particular column when the intensity of the 0.3 M salt band was higher than the intensity of both the 0.1 M band and the flowthrough band. In contrast, when the intensity of the 0.3 M band was lower than the 0.1 M or the flowthrough band, we considered that TFIIA was not binding to the column. The specificity of TFIIA binding was assessed by comparing the binding of the TFIIA subunits to the binding of a contaminant polypeptide of the nTFIIA preparation (Fig. 2).
Gel Mobility Shift Assay-Plasmid DNA containing the adenovirus major late promoter (AdMLP) was digested with the restriction enzymes BamHI and DraI, and the 110-base pair fragment containing the promoter was filled-in using the Klenow fragment of DNA polymerase in the presence of [␣-32 P]dGTP. Gel mobility shift assays were performed as described previously (56). Complexes were assembled using highly purified TBP (20 ng), TFIIA (120 ng), and, when Transcription Assay-Transcription assays were performed as described previously (57). TBP (120 ng), TFIIB (120 ng), RAP30 (120 ng), RAP74 (260 ng), TFIIE34 (160 ng), TFIIE56 (240 ng), RNAPII (660 ng), and various amounts of TFIIA were incubated with 500 ng of the supercoiled DNA template containing the AdMLP from nucleotides Ϫ50 to ϩ10 fused to a G-less cassette. Under these conditions a 391-nt run-off transcript is produced.
Abortive Initiation Assay-Templates were prepared by annealing two 80-base pair DNA oligonucleotides carrying the strands of the

TFIIA Is Located Near the Cross-point of the Wrapped DNA Structure in the Initiation Complex-We have previously
shown that TFIIA assembled with TBP on the AdMLP (e.g. TBP-TFIIA-promoter complex) cross-linked to positions Ϫ31/ Ϫ29, Ϫ25/Ϫ30, Ϫ39/Ϫ40, and Ϫ42 (18). We have now analyzed the position of TFIIA in a preinitiation complex composed of TBP, TFIIB, RAP74, RAP30, TFIIE56, TFIIE34, and RNAPII using site-specific protein-DNA photo-cross-linking. Both TFIIA␣/␤ and TFIIA␥ cross-linked in the region of the TATA box to photoprobe Ϫ31/Ϫ29 (data not shown), upstream of it to photoprobe Ϫ39/Ϫ40 (Fig. 1A, left panel) and downstream of it to photoprobe ϩ26 (Fig. 1A, right panel). The cross-linking of TFIIA to positions Ϫ39/Ϫ40 required the presence of TBP but not that of RAP30 (Fig. 1A), TFIIB, or RNAPII (data not shown). The cross-linking of TFIIA to position ϩ26, however, required the presence of TBP and RAP30 (Fig. 1A, right panel), as well as that of TFIIB and RNAPII (data not shown). These results indicate that promoter contacts by TFIIA in the Ϫ39/ Ϫ40 region do not necessitate assembly of a preinitiation complex containing TFIIB, TFIIF, and RNAPII (e.g. a TBP-TFIIApromoter complex is sufficient), whereas the promoter contact by TFIIA in the ϩ26 region requires the assembly of a preinitiation complex (e.g. a TBP-TFIIA-TFIIB-TFIIF-RNAPII-TFIIE-promoter complex) in which promoter DNA adopts a wrapped structure (see Fig. 1B for a schematic representation).
The cross-linking of TFIIA to promoter regions downstream of the transcription initiation site (ϩ10 to ϩ30) was not unexpected. According to the DNA wrapping model, the DNA helices upstream of TATA (Ϫ40/Ϫ60 region) and downstream the initiation site (ϩ10/ϩ30 region) are juxtaposed in space. Our results provide additional support for the notion that promoter DNA is wrapped in the initiation complex and indicate that TFIIA is localized near the cross-point of the wrapped DNA structure.
TFIIA Directly Interacts with TBP, RAP74, TFIIE56, and TFIIE34 -In the context of a preinitiation complex assembled with TFIIA, TBP, TFIIB, TFIIE, TFIIF, and RNAPII, we obtained cross-linking of TFIIA to photoprobes that are also cross-linked by other components of the complex. More specifically TFIIE34, RAP74, and RAP30 cross-link to photoprobe ϩ26, while Rpb2, TFIIE34, RAP74, and RAP30 cross-link to photoprobe Ϫ39/Ϫ40 (23,26). These observations indicate that TFIIA is in close proximity to these factors in the preinitiation complex, suggesting that TFIIA could directly interact with TFIIE, TFIIF, and RNAPII. To test this hypothesis, nTFIIA was chromatographed over different affinity columns containing immobilized RAP74, RAP30, TFIIE56, TFIIE34, TFIIB, and RNAPII. Columns containing TBP and BSA were used as positive and negative controls, respectively because TBP has been shown to interact with TFIIA (36,37). The flowthrough was collected in each case and the columns were successively eluted with buffer containing 0. (␣, ␤, and ␥ subunits) was retained on the TFIIE56, TFIIE34, RAP74, and TBP columns but not on the TFIIB, RAP30, RNA-PII, and BSA columns (Fig. 2). Contaminant bands of the TFIIA fraction were visible in the flowthrough of all the columns. The binding of TFIIA to the affinity columns is most easily visualized by examining the elution of the ␤ and ␥ subunits.
To further characterize these interactions, we next used the individual subunits of TFIIA in our affinity chromatography experiments. TFIIA␣/␤ and TFIIA␥ were individually chromatographed on affinity columns containing immobilized RAP74, TFIIE34 and TFIIE56. BSA was added to the input as an internal negative control. Fig. 3 shows that TFIIA␣/␤, but not TFIIA␥, interacts with RAP74 and TFIIE34. TFIIA␣/␤ did not bind to the TFIIE56 column, whereas TFIIA␥ was retarded on the TFIIE56 column, suggesting a weak interaction. This finding is surprising in view of the observation that nTFIIA binds strongly to the TFIIE56 column. Perhaps this is due to the association of TFIIA␣/␤ and TFIIA␥ resulting in a conformational change that favors binding to TFIIE56.
Two groups have previously reported interactions between TFIIA and TFIIE. Both used recombinant TFIIA, not the natural protein. In agreement with our results, Yamamoto et al. RAP74 Contains Two Distinct TFIIA-binding Domains-To determine the domain or domains of RAP74 responsible for the interaction with TFIIA, a series of RAP74 deletion mutants were immobilized on different affinity columns and rTFIIA (␣/␤ and ␥ renatured together) chromatographed through each column. BSA was added to the input as an internal control, and a BSA column served as a negative control. Fig. 4A shows some representative data, and a summary is presented in Fig. 4B. All the N-terminal fragments of RAP74, except for RAP74-(1-75), which contains only the first 75 amino acids of the polypeptide, were bound by rTFIIA (Fig. 4A). Because rTFIIA did bind to RAP74-(1-136) but not to RAP74-(1-75), our results define a first domain of interaction between these two proteins that encompasses amino acids 76 -136. We next used the C-terminal fragments of RAP74 in our affinity chromatography experiments. RAP74-(207-517), RAP74-(358 -517), and RAP74-(407-517) were all observed to bind to RAP74, indicating the existence of a second interacting domain. To delineate this domain more precisely, two additional mutants, RAP74-(363-444) and RAP74-(363-409), were used. TFIIA bound well to RAP74-(363-444) but not to RAP74-(363-409) (Fig. 4A). These results define a second TFIIA-interacting domain of RAP74 located between amino acids 410 and 444. The existence of an additional putative TFIIA-binding domain in the central RAP74 region was ruled out by passing rTFIIA on columns containing RAP74 -(136-258) and RAP74-(258 -356). In each case, rTFIIA was not retained on the column. The two TFIIA-interacting domains of RAP74 that we identified are both localized in conserved regions of the protein, domain 76 -136 being localized in conserved region I and domain 410 -444 in conserved region III (see Fig. 4B for a schematic representation).
TFIIA Stimulates the Activity of TFIIE34 and RAP74 in Transcription Initiation-Several reports have shown that TFIIA can stimulate basal transcription by RNAPII in vitro, but is not essential for the basal transcription reaction (28,(33)(34)(35)(36)(37). For example, recombinant TFIIA increased the forma-FIG. 6. Stimulation of basal transcription by TFIIA. A, run-off transcription assays were performed on a supercoiled template carrying the AdMLP using TBP, TFIIB, TFIIE, TFIIF, and RNAPII in either the absence or the presence of increasing amounts of TFIIA (100, 200, and 400 ng). The position of the accurately initiated transcript (391 nt) is indicated. B, abortive initiation assays were performed on a synthetic double-stranded oligonucleotide carrying the AdMLP using TBP, TFIIB, RAP30, and RNAPII in either the absence or the presence of RAP74 alone, RAP74 and TFIIE34, and RAP74 and TFIIE56. Increasing amounts of TFIIA were added to the reactions. The positions of the abortive transcripts (4 -10 nt) are indicated. C, quantification of the stimulatory effect of TFIIA on abortive initiation. The intensity of the bands corresponding to the abortive transcripts from 4 -6 experiments were quantitated using a PhosphorImager. The measured intensities were used to calculate the ratios of amount of transcript produced in the presence of TFIIA to that produced in its absence in each case (Fold Stimulation). tion of a 391-nt run-off transcript from a supercoiled template carrying the AdMLP fused to a G-less cassette in the presence of TBP, TFIIB, TFIIE, TFIIF, and core RNAPII (Fig. 6A). To test whether or not TFIIA acts on the initiation step of the transcription reaction, we developed an abortive initiation assay in which an 80-base pair double-stranded oligonucleotide carrying the AdMLP was used to drive transcription initiation in the presence of RNAPII and the general transcription factors. In this assay only abortive transcripts of 2 to 10 nt in length can be synthesized because the transcription reaction is performed in the absence of GTP on a template that harbors a G at position ϩ11 (see Fig. 6B). Abortive initiation was shown to be promoter-specific because the use of DNA templates with mutations in the AdMLP that impair transcription in run-off assays did not support synthesis of abortive transcripts. 2 Under our reaction conditions, abortive initiation minimally requires the presence of TBP, TFIIB, RAP30, and RNAPII (Fig.  6). With this minimal set of factors, the addition of TFIIA does not stimulate the formation of abortive transcripts (Fig. 6, B and C). When either RAP74 alone, or RAP74 and TFIIE34 are added to the system, the formation of abortive transcripts is increased, indicating that these two factors are involved in transcriptional initiation (Fig. 6, data not shown, and Refs. 58,61,62). The addition of TFIIA to reactions containing either RAP74 alone, or RAP74 and TFIIE34, in addition to TBP, TFIIB, RAP30, and RNAPII, had a stimulatory effect (Fig. 6, B and C), indicating that TFIIA can stimulate the initiation stage of the transcription reaction. Because TFIIA stimulates transcription initiation only when RAP74 and TFIIE34 are present in the reaction, our results suggest that TFIIA enhances the functions of TFIIF and TFIIE to stimulate the basal transcription reaction.
The role of TFIIA in RNAPII transcription is not completely resolved. However, several reports suggest that an important function of this factor is to act as a co-factor in transcriptional activation (28, 29, 33, 34, 44 -47). In this paper, we establish the existence of structural and functional interactions between TFIIA and both TFIIE and TFIIF within the initiation complex. TFIIE and TFIIF have been shown to be involved in the melting of promoter DNA near the initiation site during open complex formation (9,58). Considering that an activator such as Gal4-VP16, whose full activity requires the presence of TFIIA, can stimulate promoter melting near the initiation site (63), it is possible to envision an activation mechanism in which TFIIA functions as a bridge between the activator proteins and the general transcription factors TFIIE and TFIIF. This connection may help to explain how upstream activators influence and stimulate the biochemical events occurring at the transcription start site.