Interactions of a DNA-bound Transcriptional Activator with the TBP-TFIIA-TFIIB-Promoter Quaternary Complex*

Site-specific protein-DNA photo-cross-linking was used to show that, when bound to its cognate site at various distances upstream of the TATA element, the chimeric transcriptional activator GAL4-VP16 can physically interact with a TATA box-binding protein (TBP)- transcription factor IIA (TFIIA)-TFIIB complex assembled on the TATA element. This result implies DNA bending and looping of promoter DNA as a result of the physical interaction between GAL4-VP16 and an interface of the TBP-TFIIA-TFIIB complex. This protein-protein interaction on promoter DNA minimally requires the presence of one GAL4 binding site and the formation of a quaternary complex containing TBP, TFIIB, and TFIIA on the TATA element. Notably, the topology of the TBP-TFIIA-TFIIB-promoter complex is not altered significantly by the interaction with DNA-bound activators. We also show that the ability of GAL4-VP16 to activate transcription through a single GAL4 binding site varies according to its precise location and orientation relative to the TATA element and that it can approach the efficiency obtained with multiple binding sites. Taken together, our results indicate that the spatial positioning of the DNA-bound activation domain is important for efficient activation, possibly by maximizing its interactions with the transcriptional machinery including the TBP-TFIIA-TFIIB-promoter quaternary complex.

Most models of transcriptional activation imply a physical interaction of DNA-bound transcriptional activators and the transcription machinery assembled on core promoters (1)(2)(3). This contact between the activation domain and components of the transcription machinery has been attributed diverse functions including: (i) the recruitment of key transcription factors (e.g. general transcription factors or co-activators) at the promoter, (ii) the stimulation of enzymatic activities involved in the transcription reaction (e.g. promoter melting, phosphorylation, initiation of RNA chain synthesis), and (iii) the relief of transcriptional blockades induced by various types of repressors including nucleosomes. In support of this view of transcriptional activation, a number of protein-protein interactions between various activation domains and members of the RNA polymerase II (RNAPII) 1 transcription machinery have been characterized in solution and found to be important in mediating transcriptional activation (1)(2)(3). However, little is known about the formation of these protein-protein interactions when the interacting partners are bound to promoter DNA.
Ultimately, transcriptional activators regulate the activity of the basal RNAPII transcription machinery. The transcription reaction is a multi-step process in which a preinitiation complex containing TBP, TFIIB, TFIIE, TFIIF, TFIIH, and RNA-PII is first assembled onto promoter DNA (4,5). TBP recognizes and binds the TATA element, inducing a bend of about 80°in the DNA helix (6,7). TFIIB binds to and stabilizes the TBP-promoter complex (8). TFIIF, a factor composed of two subunits called RAP74 and RAP30, binds tightly to RNAPII (9,10), recruits the enzyme to the TBP-TFIIB-promoter complex, and induces an isomerization of the preinitiation complex that includes wrapping of promoter DNA around RNAPII (11,12). TFIIE, a factor also composed of two subunits called TFIIE␣ and TFIIE␤, stabilizes the preinitiation complex (13) and is involved in the melting of promoter DNA at the transcription initiation site through an ATP-independent mechanism (14,15). TFIIH, a nine-subunit factor that has both kinase and helicase activities, mediates the ATP-dependent melting of promoter DNA and is involved in the transition between the initiation and elongation states of the complex (16 -21) (also see Ref. 22 for a review). The initiation of chain synthesis proceeds through a cycle of abortive initiation events during which RNA-PII synthesizes short 2-10-nucleotide transcripts before escaping the promoter and entering a productive elongation mode (23). The activity of elongating RNAPII is modulated through the action of a number of elongation factors (24,25).
The acidic activator GAL4-VP16 is a chimeric polypeptide composed of the DNA-binding domain of the yeast GAL4 protein (amino acids 1-147) and the transcriptional acidic activation domain of the viral protein VP16 (amino acids 412-490) (26). When GAL4 DNA-binding sites are located upstream of the TATA box of a promoter, this activator has been shown to stimulate transcription in a synergistic manner both in vivo and in vitro (see Ref. 27 for example). GAL4-VP16 can act on various steps of the transcription reaction, probably via its many interactions with components of the basal transcriptional machinery. These include interactions with TBP (28,29), TFIIB (30,31), TFIIA (32), TFIIH (33), and hTAF II 32 (34). GAL4-VP16 was shown to stimulate three distinct events during the formation of the preinitiation complex: (i) formation of the TFIID-TFIIA-DNA complex (34); (ii) TFIIB recruitment (35); and, (iii) RAP30, TFIIE␣, and RNAPII recruitment (36). A * This work was supported by a grant from the Canadian Institutes of Health Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 (37), transcript elongation (38), and reinitiation (39,40).
Using site-specific protein-DNA photo-cross-linking, we have analyzed the physical interaction between GAL4-VP16 molecules bound to promoter sites various distances from the TATA element and a complex containing TBP, TFIIB, and TFIIA assembled onto the TATA element. Using templates with either one or five GAL4 DNA-binding sites, we show that DNA-bound GAL4-VP16 does indeed cross-link to many positions in the vicinity of the TATA element where a TBP-TFIIA-TFIIB complex is assembled. Interaction with DNAbound GAL4-VP16 does not significantly modify the conformation of the TBP-TFIIA-TFIIB-promoter complex. We also show that a GAL4-VP16 dimer bound to one GAL4 binding site can stimulate transcription with an efficiency approaching that of GAL4-VP16 dimers bound to five sites if the single dimer is properly positioned upstream of the TATA box.

MATERIALS AND METHODS
Protein Factors-Recombinant yeast TBP (29), human TFIIB (41), human TFIIA (42), and GAL4-VP16 (43) were purified as described previously. The activity of all purified proteins was tested in various DNA binding assays and in in vitro transcription reactions.
In Vitro Transcription-In vitro transcriptions were performed as described previously (44). Promoters containing the adenovirus-2 major late (Ad2ML) promoter from nucleotides Ϫ50 to ϩ10 fused to a G-less cassette and a varying number of GAL4 DNA-binding sites were used. Supercoiled template DNA (300 ng) and competitor DNA (300 ng, empty vector) were incubated with 1-5 l of HeLa nuclear extract in both the absence and presence of 5-40 ng of GAL4-VP16. Transcript formation was monitored by primer extension and quantified using a PhosphorImager (Amersham Biosciences).
Primer Extension Analysis-Primer extension analyses were performed as previously described (45). Transcription reactions were incu-bated with a radiolabeled 30-mer complementary to nucleotides ϩ37 to ϩ67 of the coding strand of the Ad2ML promoter. This primer was labeled with [␥-32 P]ATP using T4 polynucleotide kinase (United States Biochemical).
Protein-DNA Photo-cross-linking-The synthesis of the photoreactive nucleotide N 3 R-dUMP, the preparation of the probes, and the conditions for the binding reactions were performed as previously described (46). Photoprobes containing the photoreactive nucleotide at positions Ϫ45/Ϫ48, Ϫ39/Ϫ40, Ϫ34, Ϫ20, Ϫ19, and Ϫ15 were used. A typical reaction contained 100 ng of TBP, 100 ng of TFIIB, 200 ng TFIIA, and various amounts of GAL4-VP16 (as indicated in the figures). UV irradiation, nuclease treatment, and SDS-PAGE analysis of radiolabeled photo-cross-linking products were performed as described previously (46).

DNA-bound GAL4-VP16 Specifically Cross-links to the Region of the TATA Box in the Presence of TBP, TFIIA, and
TFIIB-Most models of transcriptional activation stipulate that an activator bound to its specific binding site upstream of the basal promoter could make contacts with the transcriptional machinery at the basal promoter through the formation of a loop in the promoter DNA (reviewed in Refs. [1][2][3]. To obtain direct evidence of the interaction of activators with the transcriptional machinery on promoter DNA (as opposed to interactions between free molecules in solution), we asked whether or not GAL4-VP16 bound to its cognate sites upstream of the TATA box could cross-link to the region of TATA in the presence of TBP, TFIIB, and TFIIA. All three of these factors previously have been shown to bind to the activation domain of VP16 in solution (28 -32). To perform the cross-linking experiments with a template that supports transcriptional activation, we tested the ability of templates carrying between two and five GAL4 binding sites, placed in both possible orientations upstream of the TATA box, to support GAL4-VP16 activation in vitro from the Ad2ML promoter. As summarized in FIG. 1. Transcriptional activation using templates with multiple GAL4 binding sites. Various numbers of tandemly repeated GAL4 binding sites (two sites, GML 272 and GML 274; three sites, GML 276 and GML 278; four sites, GML 280 and GML 282; five sites, GML 33) were placed in both possible orientations 37 bp upstream of the TATA box of the Ad2ML basal promoter fused to a G-less cassette. Maximal activation levels, expressed as the ratio of the number of transcripts produced in the presence GAL4-VP16 to that produced in its absence (fold activation), represent an average of 3-5 independent experiments. Fig. 1, the template with five GAL4 sites most efficiently supported transcriptional activation by GAL4-VP16 in a HeLa nuclear extract (11.9-fold). Notably, we were unable to obtain synergistic activation of this promoter under our experimental conditions. We then decided to use the template carrying five GAL4 sites (GML 33) in our photo-cross-linking experiments.
As shown in Fig. 2A, GAL4-VP16 does cross-link to photoprobe Ϫ39/Ϫ40 in the presence of TBP, TFIIB, and TFIIA (left panel). As expected, the p55 subunit of both recombinant TFIIA and TBP also cross-linked to photoprobe Ϫ39/Ϫ40 (Refs. 47-49; see below). The intensity of the GAL4-VP16 cross-linking signal in the presence of TBP, TFIIA, and TFIIB is about 30-fold higher than that obtained in the absence of TBP, TFIIA, and TFIIB (right panel). To ensure that the cross-linking of GAL4-VP16 to positions Ϫ39/Ϫ40 in the presence of TBP, TFIIA, and TFIIB requires binding of Gal-VP16 to the GAL4 sites located 37 bp upstream of TATA, we compared cross-linking signals obtained with photoprobes possessing or lacking GAL4 binding sites. Fig. 2B shows that the cross-linking of GAL4-VP16 to photoprobe Ϫ39/Ϫ40 is dependent upon the presence of the GAL4 binding sites. Together, these results indicate that a TBP-TFIIA-TFIIB-GAL4-VP16 complex can form on promoter DNA and that its formation requires the presence of both the TATA element and the GAL4 binding sites, thereby implying DNA bending and looping.
DNA-bound GAL4-VP16 Does Not Significantly Modify the Conformation of a TBP-TFIIA-TFIIB Complex on Promoter DNA-We and others have previously shown that TBP and TFIIA cross-link to a number of positions in the region of the TATA element of a promoter (47)(48)(49). We next compared the cross-linking of TBP and TFIIA at positions Ϫ45/Ϫ48, Ϫ34, Ϫ20, Ϫ19, and Ϫ15 in both the presence and the absence of GAL4-VP16 (Fig. 3). GAL4-VP16 specifically cross-linked to all of the photoprobes except Ϫ19. In the context of DNA looping, this finding is not surprising because five GAL4-VP16 dimers bound to a stretch of 100 bp of DNA (e.g. the five GAL4 binding sites) are expected to make multiple contacts with the TBP-TFIIA-TFIIB complex associated with the TATA box region.
Our results also show that the cross-linking of TBP and TFIIA to positions Ϫ39/Ϫ40 (Fig. 2), Ϫ34, Ϫ20, Ϫ19, and Ϫ15 (Fig. 3) is not significantly affected by the presence of GAL4-VP16. However, the cross-linking of TFIIA to photoprobe Ϫ45/ Ϫ48 is weakly, but reproducibly, increased by increasing amounts of GAL4-VP16. Because this stimulation of the TFIIA cross-links by GAL4-VP16 is observed only at positions Ϫ45/ Ϫ48, it is unlikely to be the result of either a general stabilization or a more efficient assembly of the TBP-TFIIA-TFIIBpromoter complex. We favor the alternative conclusion that DNA looping, induced by the contact of DNA-bound GAL4-VP16 with the TBP-TFIIA-TFIIB complex on the TATA box, tethers the DNA helix at Ϫ45/Ϫ48 to the bulk of TFIIA (see Fig.  6B for a schematic representation).
We have been unable to obtain cross-linking of DNA-bound GAL4-VP16 in the region of the TATA box in the presence of TBP alone, TBP and TFIIA, or TBP and TFIIB (data not shown), indicating that a TBP-TFIIA-TFIIB-promoter quaternary complex is minimally required to produce DNA looping and stable binding to DNA-bound GAL4-VP16. We did obtain cross-linking of DNA-bound GAL4-VP16 to the TATA box when we used photoprobes carrying a single GAL4 binding site (Fig. 4). This cross-linking signal is specific because most of it was lost in experiments performed in the absence of TBP, TFIIA, and TFIIB or when we used a template lacking GAL4 binding sites (data not shown). This result indicates that a single GAL4-VP16 dimer bound to DNA can interact with a TBP-TFIIB-TFIIA complex on the TATA box through DNA looping.

The Location and Orientation of a Single GAL4 Binding Site Relative to the TATA Element Influences Activation Levels in
Vitro-Our conclusion that GAL4-VP16 molecules bound to a single GAL4 binding site can efficiently contact a TBP-TFIIA-TFIIB-promoter complex suggests that a single GAL4 site could support efficient transcriptional activation if it is positioned properly upstream of the core promoter. To test the idea two series of templates with the Ad2ML basal promoter placed under the control of one GAL4 binding site (in both possible orientations) were constructed (Fig. 5A). In each series, the distance between the GAL4 site and the TATA box is different, being 37 (GML 248 and  Because of the helical structure of DNA, GAL4-VP16 dimers bound to the various templates are localized on different faces of the DNA helix. Each 2-bp increment will move the activator molecule by 72°(a full helical turn being 360°). We used in vitro transcription assays in either the presence or absence of GAL4-VP16 to test activation with each of the constructs. The results are shown in Fig. 5B and are summarized in Fig. 5C. The different templates supported activation levels ranging from 2.3-to 7.5-fold. This result indicates that the precise location of the GAL4 binding site relative to the TATA element affects activation. We believe that activation depends on the precise face of the helix where the VP16 activation domain is positioned (see Fig. 6A for a schematic representation). As the site moves away from the TATA box, making one full turn around the DNA helix (in 72°i ncrements), transcription activation increases up until the center of the site is located at nucleotide Ϫ80 and then decreases thereafter. This finding argues in favor of a preferred orientation, or side, of the helix for the GAL4-VP16 activation domain such that the interactions with the transcriptional machinery are maximized. In support of this conclusion, the templates that most efficiently support activation are those that place the GAL4-VP16 dimer on the same face of the DNA helix, whereas those that are least efficient place the GAL4-VP16 dimer on the opposite face of the helix (Fig. 6A).
As shown in Fig. 5, the activation level obtained with a template containing five GAL4 binding sites (11.7-fold) is only slightly higher than that obtained using a template with one GAL4 site at position Ϫ71 (7.5-fold). Templates containing two, 3, and four GAL4 binding sites produced equivalent or lower activation levels (Fig. 1). In contrast to the results of Lin et al. ; depicted in red) that previously has been shown to be essential for transcriptional activation (62). However, our results indicate that this TFIIB domain alone is not sufficient for the interaction of DNA-bound GAL4-VP16 with the transcriptional machinery bound to core promoter. As shown, DNA looping increases the contact of TFIIA with the DNA helix around positions Ϫ45/Ϫ48. under our conditions, that is using non-chromatin templates, the adequate positioning of a single GAL4 binding site relative to TATA has an effect comparable with the use of multiple binding sites.
Together, our results support the notion that DNA-bound GAL4-VP16 can contact the transcription machinery around the TATA box, induce a looping of the DNA double helix, and stimulate transcription in a nonsynergistic manner. Although our results do not allow for the determination of the exact cause-and-effect relationship between these events, they indicate that the positioning of DNA-bound activation domains upstream of TATA is important in activation mechanisms and fully support the concept of a crucial role for direct physical contact between an activator and the RNAPII machinery through promoter DNA looping. An increasing body of evidence indicates that promoter DNA is wrapped around the transcription machinery during transcription initiation. By making contacts with the transcription machinery, a process that induces DNA bending and looping, transcriptional activators possibly stimulate the formation of a DNA wrap around RNAPII.
In recent years, many reports have indicated that transcriptional activators such as VP16 interact not only with the basal transcription machinery but also with the Mediator (TRAP-SMCC-ARC) complex (50 -52). In yeast, the VP16 activation domain can also modify the transcriptional status of chromatin by (i) relieving nucleosome-mediated repression through the recruitment of histone acetyltransferase complexes such as SAGA and NuA4 (53,54) and by (ii) targeting the Swi/Snf complex, thus facilitating chromatin remodeling (55). In mammalian cells, VP16 interacts functionally with p300 (56,57). The multiple interactions of a given activation domain with various transcriptional targets are required to produce both the high levels of activation obtained in vivo and in vitro on chromatin templates and the synergistic response obtained when iterative binding sites for a transcriptional activator drive the expression of a reporter gene on nucleosome-containing templates (1)(2)(3)(4)(5). Although previous reports have established that VP16 binds directly to TFIIA, TBP, and TFIIB, our results indicate that, separately, these interactions are not stable when the proteins are bound to promoter DNA and suggest that the TBP-TFIIA-TFIIB complex as a whole entity is the target of DNA-bound GAL4-VP16 dimers. More recently, a number of reports have shown that the ordered recruitment (and concomitant ordered release) of many transcription factors, including the general initiation factors, RNAPII, the Mediator complex, and diverse chromatin-modifying complexes, is required for gene activation in vivo (58). Here, we have provided biochemical evidence showing that specific proteinprotein interactions between factors that are bound to distant positions on a DNA fragment can induce the formation of a loop that is important for transcriptional activation. The loop structure may be a key topological requirement for the association of some transcription complexes to promoter DNA. For example, and as we have suggested previously (22), the loop structure formed within promoter DNA as a result of the assembly of a TFIID-TFIIA-promoter complex (59) can provide the architectural requirement for the subsequent binding of the RNAPII complex. Rapid, tightly regulated association-dissociation of transcription complexes on promoter DNA could be facilitated by the formation of the loop structure. In support of this notion, topological constraints imposed to promoter DNA such as tight bending in the TATA box and initiation site regions were suggested to be involved in the displacement of a key repressive nucleosome during activation of the interferon-␤ gene (60,61).