Cooperative and Competitive Protein Interactions at the Hsp70 Promoter*

Drosophila heat shockfactor (HSF) binds to specific sequence elements of heat shock genes and can activate their transcription 200-fold. Though HSF has an acidic activation domain, the mechanistic details of heat shock gene activation remain undefined. Here we report that HSF interacts directly with the general transcription factor TBP (TATA-box binding protein), and these two factors bind cooperatively to heat shock promoters. A third factor that binds heat shock promoters, GAGA factor, also interacts with HSF and further stabilizes HSF binding to heat shock elements (HSEs). The interaction of HSF and TBP is explored in some detail here and is shown to be mediated by residues in both the amino- and carboxyl-terminal portions of HSF. This HSF/TBP interaction can be specifically disrupted by competition with the potent acidic transcriptional activator VP16. We further show that the acidic domain of the largest subunit of Drosophila RNA polymerase II (Pol II) associates with TBP in vitro and is specifically displaced from TBP upon addition of HSF. The region of TBP that mediates both HSF and Pol II acidic domain binding maps to the conserved carboxyl-terminal repeats and depends on at least one of the TBP residues known to be contacted by VP16 and to be critical for transcription activation. We discuss these findings in the context of a model in which HSF triggers hsp70 transcription by freeing the hsp70 promoter-paused Pol II from the constraints on elongation caused by the affinity of Pol II for general transcription factors.

Heat shock triggers formation of heat shock factor (HSF) 1 protein trimers (1,2) that bind tightly to heat shock elements upstream of the hsp70 promoter. HSF binding is concomitant with a rapid and vigorous (200-fold) increase in the rate of transcription. The uninduced heat shock promoter is primed for rapid activation. This promoter is contained in a chromatin structure that is open and easily accessible and contains one paused Pol II per promoter (3). A rate-limiting step in transcription appears to be the escape of this promoter-paused Pol II into productive elongation. Even after heat shock, when Pol II fires from the hsp70 promoter every 6 s, transient Pol II pausing can still be detected on hsp70 (4,5) such that Pol II progression through this specific region at the 5Ј end of the transcription unit remains rate-limiting.
The relative generality of transcriptional control at the level of paused polymerase (6) indicates that many upstream activators can act to stimulate escape of the paused polymerase into a productive elongation mode. How might this happen? We favor a model in which the paused polymerase is restrained via its strong association with the promoter and general transcription factors. In this scenario, polymerase recruitment is vigorous while escape (beyond the pause) is limiting. The activator could act to increase the rate of Pol II escape by modifying the polymerase complex to produce an elongationally competent form or perhaps more simply by competing with Pol II for one or more binding sites on the general transcription apparatus. We have demonstrated previously that Pol II can bind TBP in vitro and can be displaced from TBP by competition with specific transcriptional activator proteins (VP16 and CTF1) (7,8).
TBP is a good candidate for a heat shock factor target given its constitutive presence on hsp70 (4) and given the close proximity of the TATA element to the proximal HSF binding sites. Also, the potent acidic activator VP16 has been shown to associate with TBP (9), and it is known that acidic activators like GAL4 can activate an hsp70 promoter in transgenic fly lines (10). Finally, the carboxyl-terminal domain (CTD) heptad repeats and the acidic domain (the so-called "H" domain) of RNA polymerase have been shown to interact genetically with each other (8,11), and both have been shown to bind to TBP in vitro (8,12). In our hands, the H-domain/TBP interaction is stronger than the CTD/TBP interaction.
TBP-binding is only one of a variety of activities displayed by transcriptional activators. VP16 has been implicated in the recruitment of TFIIH (13). Since TFIIH has both DNA helicase activity and CTD-specific kinase activity, this suggests a role for activators in promoter melting and/or CTD phosphorylation. Such a modification of the polymerase complex might also play a role in the progression of the paused polymerase into productive elongation. VP16 has also been shown to associate with TFIIB in vitro (14). The multiple interactions of activators with basal factors are consistent with multiple layers of activator-mediated regulation and the synergistic effect of activators (15). A fraction of TBP is complexed in vivo, as TFIID, with at least eight TBP-associated factors (TAFs), which also have been implicated as promoter-specific activator targets (reviewed in Ref. 16). The TBP core of TFIID also serves as the foundation for assembly of the basal transcription apparatus (17). TBP is, therefore, capable of many interactions, some of which must occur simultaneously. This may be possible if these interactions are specific for small portions of the TBP surface (as has been shown for several basal factor-TBP interactions, see Ref. 17), allowing TBP to support additional activator contacts. Additionally, any of these protein-protein interactions may be quite dynamic, such that multiple factors could bind to the same site on the TBP surface.
Here we present first tests of a simple "competition" hypoth-esis for heat shock gene regulation. Affinity chromatography assays demonstrate that HSF can bind to TBP in vitro and that this binding is competitive with the acidic H-domain of RNA polymerase. Portions of HSF and TBP that are required for this association are mapped. Competitive binding assays also suggest that HSF associates with TBP in a fashion similar to the potent acidic activator VP16. Additionally, it is shown that HSF and TBP bind cooperatively to heat shock promoters and that GAGA factor further stabilizes the HSF-HSE complex.

MATERIALS AND METHODS
Plasmid Constructions-All DNA manipulations were carried out using standard procedures (18). Plasmid construction information will be made available from the authors upon request.
Protein Production-All recombinant proteins were produced in BL21 cells at A 600 ϭ 0.5 by induction using 1 mM isopropyl-1-thio-␤-Dgalactopyranoside (Sigma), and cells were harvested by centrifugation 3 h later. Bacterial extracts for purification of recombinant proteins were generated by sonication of harvested cells in 100 mM NaCl, 20 mM Tris, pH 7.5, 1 mM DTT, and 0.1 mM phenylmethylsulfonyl fluoride, followed by centrifugation for 20 min at 10,000 rpm using an SA-600 rotor. In the case of polyhistidine-tagged proteins, extracts were produced in the manufacturer (Novagen) recommended buffer, supplemented with 0.05% Nonidet P-40. GST and GST-derivative proteins were purified using GSH-agarose beads (Sigma) according to the manufacturer (Pharmacia Biotech Inc.) recommendations. MBP and MBPderivative proteins were purified with amylose resin (NEB) according to the manufacturer recommendations using 100 mM NaCl throughout. Polyhistidine-tagged proteins were purified using His-bind resin (Novagen) according to the manufacturer recommendations, followed by dialysis into buffer containing 20 mM Tris, pH 7.5, 1 mM DTT, 0.1 mM EDTA, 100 mM NaCl, and 10% glycerol. Protein concentrations were estimated by comparison to known amounts of BSA standard on SDS-PAGE gels stained with Coomassie Blue.
Binding Assays-Buffer composition for binding assays was typically 10 mM Tris, pH 7.4, 10 mM HEPES, pH 8.0, 0.5 mM DTT, 0.1 mM EDTA, 0.05% Nonidet P-40, 10% glycerol, 3 mM MgCl 2 , 0.5 mg/ml BSA, and NaCl at either 100 mM (for binding and wash buffers) or 1 M (for salt elutions). Wash buffer was the same as the binding buffer, but lacking BSA. In instances where columns were used, flow rates were approximately 1 drop/min.
Approximately 100 ng of the various recombinant proteins used were equilibrated with 10 -20 l of GST or GST-derivative resins (for batch chromatography) or 250 l of resin (for column chromatography). Beads were then washed with 20 -60-bead volumes of wash buffer (typically 6 ϫ 10 volumes), followed by treatments as described. Where indicated, approximately 10 -100 g of competitor protein (titrated at 3-fold dilutions) was included in the equilibration mixture to assay for competitive binding.
Drosophila nuclear extracts were prepared as described previously (19), and the extract protein concentration was approximately 20 mg/ ml. The final buffer composition was 25 mM HEPES, pH 7.6, 15% glycerol, 0.1 mM EDTA, 1 mM DTT, 0.1% phenylmethylsulfonyl fluoride, and 100 mM KCl. After addition of 50 l of nuclear extract to GST or GST-HSF resins, the beads were washed with 10 column volumes of wash buffer and eluted with 1 M NaCl.
Bacculovirus-produced HSF was a gift of M. Fernandes, and approximately 200 ng of this HSF was applied to GST and GST-TBP columns, which were then treated as above. Final 1 M elution fractions were precipitated with TCA prior to analysis.
Western blotting analysis was done using standard approaches (18), and input standard dilutions were performed as an indicator of signal linearity.
Dissociation constant determination experiments were done as follows. Increasing concentrations of GST-HSF coupled to approximately 10 l of GSH beads (also containing a constant excess of GST) were equilibrated with approximately 5 ng of His 6 -dTBP in 200 l of binding buffer also containing 0.5 mg/ml BSA. Beads were then collected by centrifugation and washed once with 0.6 ml of binding buffer, and the bound material was analyzed by SDS-PAGE. GST-HSF concentration was measured by comparison to known amounts of BSA standard by staining with Coomassie Blue, and percent TBP recovery was determined by comparison to known amounts of TBP by Western blotting with an antibody against TBP.
Bandshift and Footprinting Assays-Radioactive DNA probes were generated by polymerase chain reaction using an end-labeled primer.
Buffer composition was the same as that described for the proteinbinding assays, with the addition of 100 ng of poly-dGdC/25-l reaction. Approximately 2 fmol of end-labeled probe was incubated for 30 -60 min at room temperature with approximately 50 ng of the indicated proteins (titrated at 2-fold dilutions where indicated), followed by electrophoresis on 1% agarose gels in 0.5 ϫ TBE (Tris-borate-EDTA) or treatment for 30 s with approximately 0.01 unit of DNaseI (Worthington) followed by 25 l of stop buffer (1% SDS, 50 mM EDTA, 28 g/ml glycogen). DNase-treated samples were then treated with phenol and were ethanol precipitated, followed by electrophoresis on 6% sequencing gels.

Direct
Interactions of HSF and TBP-To evaluate possible targets with which HSF may interact when bound to heat shock promoters, we measured the relative binding of Drosophila HSF to two general transcription factors, TBP and TFIIB. Both of these general factors have been previously demonstrated to bind activating domains of regulatory proteins (9,20). As shown in Fig. 1A, when nuclear extract was chromatographed on agarose beads containing GST or GST-HSF in 100 mM NaCl, 5-10% of the TBP remained bound to the HSF beads, and substantially less TFIIB remained bound. A similar assay was done using bacterially produced purified Drosophila TBP and TFIIB, as shown in Fig. 1B, yielding a similar result and indicating that HSF can interact with TBP in the absence of other Drosophila proteins. Immunoblotting of quantitative dilutions of these samples indicated that under these conditions, HSF binds to TBP with approximately 30-fold greater affinity than to TFIIB (data not shown). Fig. 1C shows the complemen- tary assay of chromatographing HSF over immobilized TBP. Incubation of bacculovirus-produced HSF with GST-TBP beads resulted in 5-10% of the input HSF being bound, but no detectable HSF bound to control GST-containing beads. Therefore, we conclude that HSF can bind directly to TBP in vitro.
The binding affinity of HSF to TBP is comparable with that of the well documented VP16/TBP interaction, which has a dissociation constant of approximately 10 Ϫ7 or 10 Ϫ8 M Ϫ1 (9, 21) (see below, also data not shown). To measure the dissociation constant of the HSF/TBP interaction, we performed co-precipitation experiments using a limiting concentration of TBP and various GST-HSF concentrations, followed by a single wash. Fig. 1D summarizes the results of four such experiments. The concentration of GST-dHSF required for recovery of 50% of the TBP is approximately 10 Ϫ8 M Ϫ1 .
It is noteworthy, however, that despite the relatively tight binding displayed by purified factors, we were unable to detect this interaction by co-immunoprecipitation using strictly endogenous factors, and antibodies to either TBP or HSF, in our standard nuclear extract (data not shown). This may simply reflect technical difficulties associated with the assay, such as the fact that chromatin removal (from the extract) is necessary to ensure that any co-precipitated signal seen is not DNAmediated. Using an extract from heat-shocked cells (where the HSF is primarily nuclear and in its active conformation), it is likely that the bulk of the active (HSE-bound) HSF is removed with the chromatin, thus rendering these extracts inherently deficient in active HSF. Solubilization of such chromatin-committed HSF requires salt concentrations above that at which many protein-protein interactions could be expected to be stable. Additionally or alternatively, our inability to detect an interaction between endogenous HSF and TBP in extracts may be an indicator that the prime candidate scenario for a bona fide HSF-TBP interaction in vivo is when the two factors are maintained in proximity to one another via HSEs and the TATA element and that the two factors, not surprisingly, do not exist in a stable complex free in solution.
Cooperative-binding of HSF and TBP to the Heat Shock Promoter-To assess the possibility that a direct interaction between HSF and TBP may stabilize the binding of the two factors to their respective promoter elements, we performed bandshift and DNase-footprinting assays using combinations of the two factors, and a DNA probe containing hsp70 87A sequences from Ϫ230 to ϩ85, which includes the first three HSEs, the TATA box, and the transcription start site. A titration of HSF was performed in the absence or presence of TBP, and the resulting complexes were subjected to electrophoresis on a nondenaturing agarose gel. As shown in Fig. 2, a greater fraction of the probe was shifted at lower HSF concentrations in the presence of TBP than in its absence. Also, TBP alone at this concentration does not produce a substantial shifted complex. This indicates that HSF and TBP can bind DNA cooperatively.
Footprinting assays on protein-DNA mixtures similar to those used for the band-shift experiment provide further support for the cooperative binding of HSF and TBP to the hsp70 promoter. As shown in Fig. 3, the DNA-binding of each factor seems to be aided by the presence of the other factor. In Fig. 3A, HSF is able to more stably occupy HSEs 1, 2, and 3 in the presence of the TBP-TFIIA complex (compare lanes 4 and 7). In Fig. 3B, yTBP is seen to more stably occupy the TATA element in the presence of HSF (compare lanes 3 and 6). Similar effects were seen using an hsp26 promoter probe (Fig. 3C), suggesting that the HSF/TBP cooperative DNA-binding is at least partially independent of promoter geometry. Comparison of lanes 3 and 5 reveals that TBP more stably occupys the TATA ele-ment in the presence of HSF, and comparison of lanes 7 and 8, or lanes 9 and 10, reveals that HSF more stably occupys the proximal hsp26 HSEs in the presence of TBP. We hypothesize that cooperative DNA-binding of HSF and TBP is the result of the direct interaction of these proteins as shown in Fig. 1.
GAGA Factor Interacts with HSF-GAGA factor also binds to the hsp70 and hsp26 regulatory regions and is critical for the full regulation of these promoters (22)(23)(24). GAGA factor has been implicated in maintaining heat-shock promoters in a nucleosome-free configuration (25) and is present on heat shock promoters prior to and during heat shock (26). As shown in Fig.  4A, bacterially produced, purified GAGA factor binds GST-HSF beads. Similarly, HSF binding to an hsp70 promoter fragment is increased modestly in the presence of GAGA factor (compare lanes 3 and 5). Therefore, some portion of the role of GAGA factor in heat shock promoter function may be a consequence of direct interaction with HSF and stabilization of HSF binding.
In contrast, GAGA factor showed less binding to GST-TBP beads than to GST-HSF beads, and no cooperative DNA-binding was detected between GAGA factor and TBP. As shown in Fig. 4B, the GAGA factor footprint partially overlaps that of the TBP-TFIIA complex, and GAGA factor appears to compete with TBP for binding to DNA. The two bands at the 3Ј end of the TATA region, indicated by an arrow in Fig. 4B, serve as an indicator of TBP binding in the presence of GAGA factor. GAGA factor appears to destabilize the TBP-TATA complex, presumably via steric competition (compare lanes 2-4). Thus, in the case where a GAGA site overlaps the TATA box region, the binding of the two factors to DNA appears to be competitive.
Localization of Domains Critical for HSF/TBP Binding-HSF, like a variety of transcriptional activators that interact with TBP, has an acidic activation domain. To assess whether TBP-binding is mediated exclusively by the acidic activation domain (27) located at the carboxyl terminus of HSF, we tested the ability of truncated (GST-fused) HSF proteins to bind TBP. As shown in Fig. 5A, both the amino-and carboxyl-terminal regions of HSF were sufficient to bind to TBP in this assay. This suggests that there are at least two distinct HSF surfaces that can mediate the HSF/TBP interaction.
The carboxyl-terminal repeats of TBP are highly conserved across species, and are critical for cell viability (28,29). To determine if the conserved region of TBP is necessary for HSFbinding, we used two truncated Drosophila TBP constructs, each fused to GST. As shown in Fig. 5B, a TBP derivative containing the carboxyl-terminal conserved repeats (and lacking nonconserved amino-terminal sequences) bound HSF efficiently in a standard bead-binding assay. Conversely, a TBP derivative lacking an intact conserved domain failed to associate strongly with HSF.
HSF Competes with VP16 for TBP-binding-VP16 is a wellcharacterized potent acidic activator that binds to TBP in vitro (30). The carboxyl-terminal sequences of HSF display similarities to sequences within the first activation domain of VP16, and we reasoned that HSF may associate with TBP in a fashion similar to that of VP16. Fig. 6A shows that 10 -20% of input TBP remains bound to GST-VP16 beads after washing. If HSF is included in the binding mixture, however, a reduced fraction of the input material remains bound to the VP16 beads after washing. These results imply that HSF and VP16 can compete for binding to a common surface of TBP.
To more precisely probe the relationship of the specificities of HSF and VP16 binding to TBP, we measured HSF binding to both wild-type yeast TBP and the TBP point mutant L114K, which has been shown to be deficient in both in vitro VP16 binding and response to acidic transcriptional activators (31). This point mutation resides in the first of two highly conserved direct repeats of yeast TBP. As shown in Fig. 6B, L114K mu-FIG. 3. HSF and TBP display co-operative DNA-binding by DNaseI footprinting. A, increasing concentrations of His 6 -HSF were assayed by DNaseI footprinting for binding to an end-labeled hsp70 promoter fragment in the presence or absence of His 6 -TBP/TFIIA (T/ A). B, increasing concentrations of His 6 -yTBP were assayed by DNaseI footprinting for binding to an end-labeled hsp70 promoter fragment in the presence or absence of His 6 -HSF. C, increasing concentrations of His 6 -yTBP were assayed by DNaseI footprinting for binding to an end-labeled hsp26 promoter fragment in the presence or absence of His 6 -HSF, and decreasing concentrations of His 6 -HSF were assayed for binding to the same DNA fragment in the presence or absence of His6-yTBP.
FIG . 4. GAGA factor mediates the DNA-complex stabilization of HSF but not TBP. Protein binding assay: His 6 -GAGA was chromatographed on beads containing GST, GST/HSF, or GST/TBP. Retained fractions were analyzed by SDS-PAGE and visualized by Western blotting using an antibody against GAGA factor. DNase footprinting: Increasing concentrations of HSF (A) or TBP/TFIIA (B) were assayed by DNase footprinting in the absence or presence of His 6 -GAGA. tant TBP exhibited reduced binding to GST-HSF beads, relative to wild-type TBP. These experiments further indicate that the HSF/TBP interaction is mediated by the conserved region of TBP, and that a TBP residue that is critical for transcription activation and VP16 binding is also critical for HSF binding.
HSF Competes with Pol II for TBP-Binding-The carboxyl terminus of the largest subunit of Drosophila RNA polymerase II has sequence similarity to acidic transcription activators (8). The pattern of hydrophobic and acidic residues in this region resembles the activation domains of VP16 and GAL4. We have previously described the ability of the homologous region (the H-domain) of yeast RNA polymerase to bind to yeast TBP (8). This domain of the yeast polymerase functions as a potent transcriptional activator, similar in strength to the activation domain of VP16, when fused to a heterologous DNA-binding domain (8). This activating property of a domain of Pol II led us to hypothesize a mechanism of transcriptional activation that is centered on the competition between the activator and RNA polymerase for binding to one or more sites on the basal transcription apparatus (8).
To examine the ability of the H-domain of Drosophila PolII to bind to TBP or TFIIB, we incubated these factors with GST or GST-H-domain beads. As shown in Fig. 7A, TBP bound effectively to the H-domain beads, but TFIIB showed weaker binding. To test the ability of HSF to disrupt the TBP-H-domain complex, we exposed TBP to beads containing GST or GST-Hdomain in the absence or presence of HSF. As shown in Fig. 7B, in the presence of HSF, less TBP is retained by the H-domain beads, indicating that the TBP/polymerase interaction can be compromised by HSF.
If HSF and RNA polymerase compete for a specific binding site on TBP, it seems likely that a TBP mutation that reduces HSF binding would also reduce polymerase binding. To test this, we passed yTBP and the yeast TBP point mutant L114K over beads containing GST-H-domain and GST only. Fig. 7C shows that, like HSF, polymerase H-domain TBP-binding is reduced by a point mutation in this hydrophobic TBP residue, which has been shown to be critical for response to acidic transcriptional activators in vitro (31) and in vivo (32). DISCUSSION We have shown here that the heat shock gene-specific activator, HSF, binds efficiently to the general transcription factor TBP in vitro. In these experiments, comparable fractions of input TBP were recovered by HSF affinity chromatography using either Drosophila nuclear extracts or purified recombinant TBP. A second general factor TFIIB, which also has been reported to bind acidic activators (20), shows only weak affinity for HSF. The HSF/TBP interaction appears to influence the association of these factors with their DNA targets, in that we observe that purified HSF and TBP bind cooperatively to heat shock promoters in vitro. Likewise, GAGA factor, another component of the hsp70 and hsp26 promoters, also aids the binding of HSF to the hsp70 promoter in vitro. Both TBP and GAGA factor occupy these heat shock promoters prior to induction by heat shock and are thus positioned to facilitate HSF recruitment. These interactions, coupled with the open chromatin configuration of heat shock promoters (33), may help to explain the fact that HSF binding to HSEs in vivo is dependent on the presence of intact TFIID and GAGA binding sites (34). In addition, these interactions of HSF with TBP and GAGA factor may stabilize promoter associations of these factors during multiple rounds of activated transcription when proposed contacts of these factors with RNA polymerase II and other components of the basal machinery are likely to be disrupted during each cycle of transcription.
The binding of HSF to TBP is mediated by residues in both the DNA-binding/trimerization domain and in the acidic carboxyl-terminal domain of HSF. This binding is targeted to the conserved carboxyl-terminal repeats of TBP. The binding of HSF to TBP is similar in both avidity and character to the FIG. 5. Protein domains required for HSF/TBP binding. A, His6-TBP was equilibrated with beads containing GST or GST-HSF truncated derivatives as indicated. After washing, samples were electrophoresed and visualized by Western blotting using an antibody against TBP. B, His 6 -HSF was equilibrated with beads containing GST or the indicated GST-TBP truncated derivatives. After washing, retained fractions were subjected to SDS-PAGE followed by visualization using an antibody against dHSF.
FIG. 6. HSF competes with VP16 for binding to TBP. A, His 6 -TBP was equilibrated with equimolar amounts of MBP or various concentrations of MBP-HSF (equal protein concentration was maintained by supplementation with MBP) followed by addition of beads containing GST or GST-VP16. After further incubation, beads were washed, and bound fractions were analyzed by SDS-PAGE and visualized by immunoblotting using an antibody against TBP. B, His6-yTBP or His 6 -yTBP/L114K were incubated with beads containing GST or GST/HSF. After washing, bound fractions were electrophoresed and visualized using an antibody against yTBP.
binding of the acidic transcription activator VP16 to TBP. Both interactions are affected by a specific mutation in TBP (L114K). Moreover, VP16 and HSF compete for binding to TBP.
We have also shown that an acidic domain (H) of Drosophila RNA polymerase II binds to TBP in vitro in a manner similar to the polymerase/TBP interaction previously reported in yeast (8). This interaction is disrupted upon addition of HSF, suggesting that polymerase and HSF can compete for the same site on TBP. This site on TBP also maps to the conserved TBP carboxyl-terminal repeats and is specifically reduced by the L114K mutation. We suggest that some of the same polymerase-general factor affinities that facilitate recruitment of Pol II to the promoter, like the TBP interaction seen here, also act as a "tether" that hinders polymerase escape into functional elongation and thus contribute to the formation of paused polymerase. These results provide the basis for a simple competition model for hsp70 transcriptional activation in which HSF frees the hsp70 paused polymerase from one component of the constraint on elongation caused by affinity of polymerase for general transcription factors at the core promoter.
A model for activated transcription must, of course, account for multiple rounds of transcription. If HSF displaces a critical Pol II contact by binding to the core promoter complex, HSF may then occupy a site that is important for the next round of polymerase recruitment. How does the next Pol II molecule enter? Pol II recruitment appears to be accomplished via multiple known general factor contacts, including interactions involving TFIIF (35), CTD with TBP (11), and polymerase with the promoter DNA itself. This recruitment rate would have to be much faster than Pol II escape to account for the observed full occupancy of the uninduced promoter by paused Pol II (4). We propose that the TBP-binding activity of HSF increases the efficiency of the rate-limiting step (polymerase escape) while having a negligible inhibitory effect on the inherently fast polymerase recruitment provided by the multiple polymerase contacts of this strong promoter. While this competition model is attractive in its simplicity, it does not exclude other mechanisms that might act alternatively or additionally to increase the rate of escape of Pol II from the pause site into productive elongation. For example, HSF may facilitate recruitment of other general factors, which could modify the promoter-paused Pol II and thereby affect its escape to productive elongation. Furthermore, we have examined here only one of what may be several common contacts of HSF and Pol II with general factors.
Eukaryotic transcription has many steps that can be fine tuned to the needs of the thousands of differentially regulated promoters. Many distinct regulatory steps have been documented, including TFIID-recruitment (36 -38), Pol II recruitment (39), promoter melting (40), and elongational control after promoter escape (41,42). In each case, the slow step in transcription must be the target of regulatory factors that either enhance or inhibit one of many specific molecular interactions required for establishing a productive transcription complex. The fact that regulatory factors and RNA polymerase interact with multiple general transcription factors provides the potential for modulation at any of multiple distinct steps. FIG. 7. HSF competes with RNA polymerase for binding to TBP, and polymerase-TBP binding is mediated by TBP conserved repeats. A, purified His 6 -dTBP or TFIIB were equilibrated with GSH beads containing GST or GST-dH-domain. After washing, bound proteins were analyzed by SDS-PAGE followed by immunoblotting with antibodies against TBP or TFIIB. B, His 6 -dTBP was equilibrated with equimolar amounts of MBP or various concentrations of MBP-HSF (equal protein concentration was maintained by supplementation with MBP) followed by addition of beads containing GST or GST-dH-domain. After further incubation, beads were washed, and bound fractions were analyzed by SDS-PAGE and immunoblotting with an antibody against dTBP. C, His 6 -yTBP or His 6 -yTBP/L114K were equilibrated with beads containing GST or GST-dH-domain. After washing, bound fractions were resolved by SDS-PAGE and visualized using an antibody against yTBP.