Dual Roles for Transcription Factor IIF in Promoter Escape by RNA Polymerase II*

Transcription factor (TF) IIF is a multifunctional RNA polymerase II transcription factor that has well established roles in both transcription initiation, where it functions as a component of the preinitiation complex and is required for formation of the open complex and synthesis of the first phosphodiester bond of nascent transcripts, and in transcription elongation, where it is capable of interacting directly with the ternary elongation complex and stimulating the rate of transcription. In this report, we present evidence that TFIIF is also required for efficient promoter escape by RNA polymerase II. Our findings argue that TFIIF performs dual roles in this process. We observe (i) that TFIIF suppresses the frequency of abortive transcription by very early RNA polymerase II elongation intermediates by increasing their processivity and (ii) that TFIIF cooperates with TFIIH to prevent premature arrest of early elongation intermediates. In addition, our findings argue that two TFIIF functional domains mediate TFIIF action in promoter escape. First, we observe that a TFIIF mutant selectively lacking elongation activity supports TFIIH action in promoter escape, but is defective in suppressing the frequency of abortive transcription by very early RNA polymerase II elongation intermediates. Second, a TFIIF mutant selectively lacking initiation activity is more active than wild type TFIIF in increasing the processivity of very early elongation intermediates, but is defective in supporting TFIIH action in promoter escape. Taken together, our findings bring to light a function for TFIIF in promoter escape and support a role for TFIIF elongation activity in this process.


Transcription factor (TF) IIF is a multifunctional RNA polymerase II transcription factor that has well established roles in both transcription initiation, where it functions as a component of the preinitiation complex and is required for formation of the open complex and synthesis of the first phosphodiester bond of nascent transcripts, and in transcription elongation, where it is capable of interacting directly with the ternary elongation complex and stimulating the rate of transcription.
In this report, we present evidence that TFIIF is also required for efficient promoter escape by RNA polymerase II. Our findings argue that TFIIF performs dual roles in this process. We observe (i) that TFIIF suppresses the frequency of abortive transcription by very early RNA polymerase II elongation intermediates by increasing their processivity and (ii) that TFIIF cooperates with TFIIH to prevent premature arrest of early elongation intermediates. In addition, our findings argue that two TFIIF functional domains mediate TFIIF action in promoter escape. First, we observe that a TFIIF mutant selectively lacking elongation activity supports TFIIH action in promoter escape, but is defective in suppressing the frequency of abortive transcription by very early RNA polymerase II elongation intermediates. Second, a TFIIF mutant selectively lacking initiation activity is more active than wild type TFIIF in increasing the processivity of very early elongation intermediates, but is defective in supporting TFIIH action in promoter escape. Taken together, our findings bring to light a function for TFIIF in promoter escape and support a role for TFIIF elongation activity in this process.
TFIIF 1 was originally identified by its requirement in promoter-specific transcription initiation by RNA polymerase II. Mammalian TFIIF is a heterodimer composed of ϳ30 kDa (RAP30) and ϳ74 kDa (RAP74) subunits (1). Substantial evidence suggests that TFIIF performs multiple functions in transcription initiation. Although TFIIF is not essential for selec-tive binding of RNA polymerase II to promoters, TFIIF functions as an integral component of the preinitiation complex and strongly stabilizes binding of polymerase to TFIID and TFIIB at the promoter (2)(3)(4). In addition, TFIIF is required for entry of TFIIE and TFIIH into the preinitiation complex (2, 4 -6), for subsequent open complex formation catalyzed by the TFIIH DNA helicase, and for synthesis of the first phosphodiester bond of nascent transcripts (7)(8)(9). Although it is presently not known whether TFIIF merely functions as a scaffold for binding of TFIIH to the preinitiation complex or whether TFIIF actively participates in formation of the open complex, evidence suggests that TFIIF interacts with promoter DNA in the preinitiation complex and induces a dramatic conformational change that results in wrapping of DNA for nearly a full turn around RNA polymerase II and may facilitate formation of the open complex (10 -12).
In addition to its role in transcription initiation, TFIIF is also capable of potently activating the rate of elongation by RNA polymerase II through a direct interaction with the ternary elongation complex. Evidence suggests that TFIIF activates elongation by suppressing transient pausing by polymerase at many sites along the DNA (13,14). In this respect, TFIIF functions similarly to general elongation factor Elongin (originally known as SIII) (15,16), ELL (17), ELL2 (18), Tat-SF1 (19), and the Cockayne syndrome complementation group B protein (20).
Although a role for TFIIF elongation activity in promoterspecific transcription has not been established, we and others have shown that TFIIF initiation and elongation activities are carried out by distinct TFIIF functional domains (14,21,22). We previously identified RAP30 mutations that selectively block TFIIF initiation and elongation activities. We observed that TFIIF initiation activity is mediated at least in part by RAP30 C-terminal sequences that contain a cryptic DNA-binding domain, which Werner and co-workers (23) recently demonstrated by NMR is similar in structure to the "winged" helixturn-helix DNA-binding domains of linker histone H5 and hepatocyte nuclear transcription factor HNF3/forkhead (23). In addition, we observed that TFIIF elongation activity is mediated in part by RAP30 sequences located upstream of the C terminus in a region proposed to bind RNA polymerase II.
We recently identified post-initiation roles for TFIIE, TFIIH, and an ATP cofactor in efficient promoter escape by RNA polymerase II. Our findings led to the model that TFIIE and TFIIH function together to increase the efficiency of promoter escape by suppressing premature arrest of early RNA polymerase II elongation intermediates in an ATP-dependent process that depends on TFIIH DNA helicase activity. Because TFIIF is required for interaction of TFIIH with the preinitiation complex and for open complex formation catalyzed by the TFIIH DNA helicase, we investigated the possibility that TFIIF might also be required for TFIIH action in promoter escape. In this report, we present evidence that TFIIF functions not only in promoter-specific transcription initiation, but also in promoter escape. By analyzing the activities of TFIIF mutants selectively deficient in promoter-specific transcription initiation or in stimulation of elongation by RNA polymerase II, we have obtained evidence that TFIIF performs dual roles in promoter escape, by cooperating with TFIIH to modulate premature arrest of early elongation intermediates, and, in a reaction dependent on TFIIF elongation activity, by increasing the processivity of very early elongation intermediates. Here we present these findings, which bring to light a novel function for TFIIF in promoter escape and support a role for TFIIF elongation activity in this process.
Preparation of DNA Templates for Transcription-The EcoRI to NdeI fragment containing the AdML promoter from pDN-AdML was prepared as described (24). DNA fragments containing the double-stranded AdML and the premelted Ad(Ϫ9/Ϫ1) promoters were prepared from single-stranded M13mp19-AdML DNA as described (25), except that KpnI digestion was performed before extension of oligonucleotide primers by DNA polymerase. The premelted Ad(Ϫ9/ϩ9) promoter fragment was prepared using a similar procedure, with an oligonucleotide primer with a sequence corresponding to positions Ϫ58 to Ϫ10 and ϩ10 to ϩ22 of the AdML promoter on the plasmid pDN-AdML, with an 18-base mismatch sequence, AAGTAGAAGCGAGAGACA, which extends from Ϫ9 to ϩ9.
Preparation of RNA Polymerase II and Transcription Factors-RNA polymerase II (26) and TFIIH (rat ␦, DEAE 5-PW fraction or SP 5-PW fraction (27)) were purified from rat liver nuclear extracts as described previously. Recombinant yeast TBP (AcA 44 fraction (28)) and rat TFIIB (29) were expressed in Escherichia coli and purified as described.
Preparation of the Elongin ABC Complex-Individual recombinant histidine-tagged Elongin A, B, and C subunits were expressed in E. coli, purified from guanidine hydrochloride-solubilized inclusion bodies by Ni 2ϩ -nitrilotriacetic acid-agarose chromatography, and refolded together as described previously (31). The reconstituted Elongin ABC complex was further purified as described (15) by chromatography on consecutive TSK SP 5-PW and TSK phenyl-5-PW high pressure liquid chromatography columns.
Assay of Transcription-Transcription reactions were performed essentially as described (25). Preinitiation complexes were assembled at the AdML promoter at 28°C by a 45-min incubation of 30-l reaction mixtures containing 10 mM Hepes-NaOH (pH 7.9), 20 mM Tris-HCl (pH 7.9), 55 mM KCl, 7 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM dithiothreitol, 0.5 mg/ml bovine serum albumin, 1.5% (w/v) polyvinyl alcohol, 6% (v/v) glycerol, 5 units of recombinant placental ribonuclease inhibitor, ϳ10 ng of DNA template, ϳ5 ng of recombinant yeast TBP, ϳ10 ng of recombinant TFIIB, ϳ4 ng of TFIIE, 0.01 units of RNA polymerase II, and the amounts of TFIIF, TFIIH, and Elongin ABC complex indicated in the figure legends. Transcription reactions were performed at 28°C with the concentrations of nucleotides indicated in the figure legends. Transcription reactions measuring synthesis of trinucleotide or 3Ј-O-MeG-terminated transcripts were stopped by addition of 15 l of a solution of 100 mM EDTA and 0.5 mg/ml proteinase K. After a 10-min incubation at room temperature, 55 l of a solution of 9 M urea, 0.025% bromphenol blue, and 0.025% xylene cyanol FF were added to reaction mixtures. The samples were vortexed for 10 s, heated at 90°C for 2 min, and analyzed by electrophoresis through 25% (w/v) acrylamide, 3% (w/v) bisacrylamide, 6 M urea gels as described (32). Transcription reactions measuring full-length run-off transcripts were stopped by addition of an equal volume of a solution of 200 mM Tris-HCl (pH 7.6), 300 mM NaCl, 25 mM EDTA, 2% SDS, 0.5 mg/ml yeast tRNA, and 0.1 mg/ml proteinase K. After a 15-min incubation at room temperature, transcripts were precipitated with ethanol, resuspended in 30 l of a solution of 9 M urea, 0.025% bromphenol blue, and 0.025% xylene cyanol FF, heated at 90°C for 5 min, and analyzed by electrophoresis through 6% (w/v) acrylamide, 0.3% (w/v) bisacrylamide, 7 M urea gels. Transcript synthesis was quantitated using a Molecular Dynamics PhosphorImager.

TFIIF Can Increase the Processivity of Very Early RNA Polymerase II Elongation
Intermediates-To begin to investigate the role of TFIIF in promoter escape, we took advantage of the artificial AdML promoter derivative Ad(Ϫ9/Ϫ1), which contains a premelted transcriptional start site (Fig. 1A). Ad(Ϫ9/ Ϫ1) can support initiation by RNA polymerase II in the absence of TFIIF, TFIIE, and TFIIH (8, 25, 33-36) and is therefore a useful model for investigating post-initiation roles of these Under our reaction conditions, transcription from the Ad(Ϫ9/Ϫ1) promoter, with or without TFIIF, was inhibited completely by 1 g/ml ␣-amanitin and reduced by at least 90% by omission of either TBP or TFIIB from reactions (data not shown), arguing that it was dependent on RNA polymerase II and primarily promoter-specific.
Consistent with results of previous studies indicating that efficient promoter escape depends upon the presence of TFIIE and TFIIH (25, 36 -38), the rate of synthesis of full-length 3Ј-O-MeG-terminated transcripts was maximal when transcription reactions contained all five general initiation factors (Fig. 1B, lanes 29 -32). Omission of TFIIE, TFIIH, or both factors from transcription reactions resulted in a significantly reduced rate of synthesis of 3Ј-O-MeG-terminated transcripts, and a substantially increased rate of appearance of abortive trinucleotide transcripts and abortive or arrested transcripts shorter than ϳ9 nucleotides (lanes 5-8, [13][14][15][16][21][22][23][24]. Omission of TFIIF from transcription reactions resulted in a further reduction in the rate of synthesis of abortive trinucleotide transcripts. In addition, at the low ribonucleoside triphosphate concentrations used in our assays, omission of TFIIF from transcription reactions resulted in a drastic reduction in synthesis of transcripts longer than 3 nucleotides (lanes 1-4, 9 -12, 17-20, and 25-28), even at times when the levels of trinucleotide synthesized was equal to or greater than the levels of trinucleotide synthesized in the presence of TFIIF (compare lanes 4 and 5, 12 and 13, 20 and 21, and 28 and 29), arguing that, under the conditions of our assays, TFIIF can increase the processivity of very early RNA polymerase II elongation intermediates.
A TFIIF Mutant Selectively Lacking Elongation Activity Is Defective in Suppressing the Frequency of Abortive Transcription by Very Early RNA Polymerase II Elongation Intermediates-In a previous study, we identified RAP30 mutations that differentially affect TFIIF initiation and elongation activities (21) (Fig. 2A). Our findings revealed that TFIIF elongation activity is most strongly sensitive to RAP30 mutations that fall between residues 91 and 135, in a region proposed to bind RNA polymerase II. In contrast, TFIIF initiation activity is most strongly sensitive to RAP30 mutations that fall between residues 136 and 210, in a C-terminal region that includes a cryptic DNA-binding domain. The RAP30 C-terminal DNA-binding domain exhibits limited sequence similarity to conserved region 4 of bacterial factors (39,40), and results of NMR studies indicate that it has a winged helix turn helix structure similar to the DNA-binding domains of linker histone H5 and the transcriptional activator HNF-3/forkhead (23).
To explore further the role of TFIIF in promoter escape, we compared the activities of two TFIIF mutants, F-RAP30(⌬181-195) and F-RAP30(⌬91-105) ( Fig. 2A), which we previously showed were most severely defective in initiation and elongation, respectively (21). The TFIIF mutants were purified from the soluble fraction of E. coli coinfected with M13 vectors encoding RAP74 and either histidine-tagged wild type RAP30, RAP30(⌬181-195), or RAP30(⌬91-105). As shown in Fig. 2B, following chromatography on consecutive Ni 2ϩ -agarose, phosphocellulose, TSK DEAE-NPR, and TSK SP-NPR columns, wild type TFIIF and the two TFIIF mutants were more than 95% pure and contained approximately stoichiometric amounts of RAP30 and RAP74.
The initiation activities of purified wild type TFIIF and the TFIIF mutants were compared in dinucleotide-primed abortive initiation assays, which have been widely used to measure synthesis of the first phosphodiester bond of nascent transcripts by both prokaryotic and eukaryotic RNA polymerases (41)(42)(43)(44)(45). The elongation activities of wild type TFIIF and TFIIF mutants were compared in oligo(dC)-tailed template assays, which allow measurement of the rate of RNA chain elongation by RNA polymerase II in the absence of transcription factors needed for promoter-specific initiation (46 -48). Consistent with our previous results (21), TFIIF mutant F-RAP30(⌬181-195) supported little or no abortive initiation by RNA polymerase II from the AdML promoter (Fig. 2C), but stimulated elongation by polymerase in oligo(dC)-tailed template assays with a specific activity 30 -50% that of wild type TFIIF. In addition, consistent with our previous results (21), TFIIF mutant F-RAP30(⌬91-105) did not detectably stimulate elongation by RNA polymerase II, but could support abortive initiation by polymerase from the AdML promoter, although at reduced levels compared with wild type TFIIF. Although the specific activity of F-RAP30(⌬91-105) in abortive initiation was not as great as that of wild type TFIIF, F-RAP30(⌬91-105) was capable of supporting abortive initiation ϳ50% as well as wild type TFIIF when present at saturating concentrations.
Because F-RAP30(⌬91-105) could support initiation by RNA polymerase II on a duplex DNA template, we tested its activity in promoter escape from a duplex AdML promoter derivative (Fig. 2C)  the results of the experiment of Fig. 3, the ratio of 16-mer to transcriptional starts was ϳ3-fold greater in the presence of wild type TFIIF than in the presence of F-RAP30(⌬91-105), in both the presence and absence of TFIIH (Fig. 4, A and B). In addition, TFIIH stimulated the efficiency of promoter escape similarly (15-20-fold) in the presence of either wild type TFIIF or F-RAP30(⌬91-105) (Fig. 4C), suggesting that the relative inactivity of this mutant was not due to defective interaction with TFIIH.
Surprisingly, in the absence of TFIIH, the initiation-defective TFIIF mutant F-RAP30(⌬181-195) supported more effi-cient promoter escape than wild type TFIIF (Fig. 5). In this experiment, reactions were performed in the presence of levels of wild type and mutant TFIIF that supported synthesis of similar levels of transcripts. A larger fraction of early RNA polymerase II elongation intermediates suffered abortion or arrest after synthesis of 5-8 nucleotide transcripts in the presence of wild type TFIIF than in the presence of F-RAP30( ⌬181-195). At all TFIIF concentrations tested, the ratio of 16-mer to transcriptional starts was approximately 4 -5-fold greater in the presence of F-RAP30(⌬181-195) than in the presence of wild type TFIIF. In contrast, in the presence of TFIIH, 16-mers were synthesized more efficiently in the presence of wild type TFIIF than in the presence of F-RAP30(⌬181-195). Although TFIIH stimulated promoter escape in the presence of both wild type TFIIF and F-RAP30(⌬181-195), the ratio of 16-mer to transcriptional starts was increased ϳ15-30-fold by TFIIH in the presence of wild type TFIIF but only 1-4.5-fold in the presence of F-RAP30(⌬181-195), depending on the F-RAP30-(⌬181-195) concentration. Considered together, these results raise the possibilities (i) that interactions important for transcription initiation between the RAP30 C terminus and other components of the initiation complex may actually present an impediment to promoter escape by RNA polymerase II and (ii) that TFIIH may function during promoter escape to help overcome this impediment.
The elongation factor Elongin cannot replace TFIIF in reconstitution of promoter-specific transcription initiation by RNA polymerase II from the duplex AdML promoter; however, when TFIIF is absent from transcription reactions, Elongin can strongly stimulate the rate of abortive initiation and synthesis of 16-nucleotide 3Ј-O-MeG-terminated transcripts by polymerase from the premelted Ad(Ϫ9/Ϫ1) promoter in the presence of TBP and TFIIB. Consistent with the model that RAP30 Cterminal sequences important for transcription initiation present an impediment to efficient promoter escape by RNA polymerase II, we observe that, in the presence of TBP, TFIIB, and TFIIE, Elongin supports 20 -30-fold more efficient promoter escape than wild type TFIIF in the absence of TFIIH and at least as efficient promoter escape as wild type TFIIF in the presence of TFIIH (Fig. 5A, lanes 8 -10 and 18 -20). In addition, the activity of Elongin in promoter escape is independent of TFIIH. We note that Elongin does not support promoter escape more efficiently than TFIIF simply because it is intrinsically more active in elongation, since these experiments were performed in the presence of amounts of Elongin, TFIIF, and F-RAP30(⌬181-195) with similar activities in elongation assays.
Downstream Extension of the Premelted Region of the AdML Promoter Is Sufficient to Relieve the TFIIF-dependent Impediment to Efficient Promoter Escape-To investigate further the possibility that TFIIF initiation activity might provide an impediment to promoter escape by RNA polymerase II, we attempted to identify transcription reaction conditions that could relieve the impediment and increase the efficiency of promoter escape in the presence of wild type TFIIF. Because we have recently shown that stimulation of promoter escape by TFIIH is strongly dependent on its ATP-dependent DNA helicase activity (49), we asked whether downstream extension of the premelted region of the Ad(Ϫ9/Ϫ1) promoter might cause promoter escape to be largely independent of TFIIH and to be similarly efficient in the presence of either wild type TFIIF or the initiation defective TFIIF mutant F-RAP30(⌬181-195). In these experiments, we took advantage of the Ad(Ϫ9/ϩ9) promoter derivative, which contains a premelted region extending from Ϫ9 to ϩ9 relative to the normal AdML transcriptional start site (Fig. 6A). Transcription was reconstituted, in the presence and absence of TFIIE and TFIIH, with RNA polymerase II, TBP, TFIIB, wild type TFIIF, and the duplex AdML As expected, transcription by RNA polymerase II from the duplex AdML promoter was strongly dependent on TFIIE and TFIIH, and the efficiency of promoter escape from the Ad(Ϫ9/ Ϫ1) promoter was stimulated ϳ20-fold in the presence of TFIIE and TFIIH (Fig. 6, B, lanes 1, 2, 4, and 5, and C). TFIIE-and TFIIH-dependent stimulation of promoter escape from the Ad(Ϫ9/Ϫ1) promoter is strongly dependent on ATP and is inhibited by addition of ATP␥S (data not shown and Refs. 25 and 49). In contrast, promoter escape by RNA polymerase II from the Ad(Ϫ9/ϩ9) promoter was very efficient (with a ratio of 16-mer to total starts of ϳ12%) even in the absence of TFIIE and TFIIH and was only slightly stimulated by TFIIE and TFIIH (Fig. 6, B, lanes 3 and 6, and C). In addition, significantly fewer 4 -15-nucleotide abortive or arrested transcripts were synthesized from the Ad(Ϫ9/ϩ9) promoter than from either the duplex AdML promoter or the Ad(Ϫ9/Ϫ1) promoter. Synthesis of both abortive trinucleotide and 16-nucleotide 3Ј-O-MeG terminated transcripts from the Ad(Ϫ9/ϩ9) promoter was strongly dependent on both TBP and TFIIF (lanes 7-10), arguing that transcription on this template was primarily promoter-specific.
In the experiment of Fig. 6D, transcription from the Ad(Ϫ9/ ϩ9) promoter was reconstituted, in the presence and absence of TFIIE and TFIIH, with RNA polymerase II, TBP, TFIIB, and wild type TFIIF, F-RAP30(⌬181-195), or F-RAP30(⌬91-105). As shown in lanes 1, 2, 4, and 5, wild type TFIIF and F-RAP30(⌬181-195) both supported efficient promoter escape, independent of the presence of TFIIE and TFIIH. In contrast to the results seen when reactions were performed with the Ad(Ϫ9/Ϫ1) template, similar efficiencies of promoter escape were observed in the presence of F-RAP30(⌬181-195) and in the presence of wild type TFIIF, consistent with the model that extending the pre-melted promoter region to ϩ9 relieves the impediment to escape normally imposed by RAP30 C-terminal sequences important for initiation on duplex templates. Finally, promoter escape on the Ad(Ϫ9/ϩ9) template was substantially less efficient than in the presence of wild type TFIIF when reactions contained the elongation-defective mutant F-RAP30(⌬91-105) (Fig. 6D, lanes 3 and 6), suggesting that TFIIF elongation activity contributes to promoter escape by a mechanism that is insensitive to template topology. Interestingly, the efficiency of both initiation and promoter escape was stimulated by TFIIE and TFIIH in reactions containing F-RAP30(⌬91-105). This stimulation was largely independent of ATP (data not shown), however, suggesting that TFIIE-and TFIIH-dependent stimulation of initiation and promoter escape from the Ad(Ϫ9/ϩ9) promoter is not due to unwinding of the template by the TFIIH DNA helicase but rather to stabilization of transcription complexes containing F-RAP30(⌬91-105). DISCUSSION In previous studies investigating the mechanism of promoter-specific transcription by RNA polymerase II, we identified post-initiation roles for TFIIE, TFIIH, and an ATP cofactor in efficient escape of polymerase from the promoter (25,49,50). Our findings from these studies led to the model that TFIIE and TFIIH function together to increase the efficiency of promoter escape by suppressing premature arrest of early RNA polymerase II elongation intermediates in an ATP-dependent process that depends strongly on the TFIIH XPB DNA helicase. In this report, we present evidence that TFIIF also affects the efficiency of promoter escape by RNA polymerase II.
First, we observe that TFIIF is capable of suppressing abortive transcription by very early RNA polymerase II elongation intermediates by increasing their processivity. Under the conditions of our assays, nearly all transcription complexes initiated without TFIIF aborted transcription after synthesizing only a single phosphodiester bond. In the presence of TFIIF, however, a much larger fraction of transcription complexes were able to synthesize longer (4 -10 nucleotides) abortive transcripts or to escape the promoter without aborting transcription (Fig. 1, see also Ref. 49).
Consistent with the possibility that TFIIF elongation activity plays a role in this process, we observe that a TFIIF mutant, F-RAP30(⌬91-105), which supports transcription initiation by RNA polymerase II but selectively lacks elongation activity, is defective in suppressing abortive transcription. Although it is presently not clear how TFIIF elongation activity might increase the processivity of RNA polymerase II at this early stage of elongation, ternary transcription complexes containing transcripts less than ϳ10 nucleotides in length are much more unstable than complexes containing longer transcripts and are therefore prone to abort transcription (for example, see Refs. 32, 51, and 52). As a consequence, there is a competition between dissociation of the transcription complex and formation of the next phosphodiester bond at each step of nucleotide addition prior to establishment of the stably elongating complex. Thus, TFIIF elongation activity may decrease the frequency of abortive transcription simply by increasing the rate of nucleotide addition. In this regard, it is noteworthy that Pan and Greenblatt (8) observed less dependence on TFIIF for escape from premelted promoters in transcription reactions containing higher concentrations of ribonucleoside triphosphates, where the rate of TFIIF-independent elongation is likely to have been greater than in our assays. In addition, although initiation from a premelted promoter has been found to be dependent on the presence of at least TBP and TFIIB in most previous studies, Keene and Luse (52) recently reported that RNA polymerase II can initiate transcription and escape from a premelted AdML promoter in the absence of any initiation factors, including TFIIF. Taken together, these findings suggest that the relative contribution of TFIIF elongation activity to promoter escape by RNA polymerase II in vitro is likely to depend at least in part on the experimental conditions used.
It is also conceivable that TFIIF elongation activity may play a more active role in preventing abortive transcription. We have observed (i) that, like Elongin (53), TFIIF has the capacity to promote template-directed extension by RNA polymerase II of the 3Ј-OH termini of DNA molecules and (ii) that mutations of RAP30 that interfere with TFIIF elongation activity also interfere with this reaction. 2 As suggested by Salzman and co-workers (54), the template directed addition of ribonucleotides to 3Ј-OH termini of DNA by RNA polymerase II may occur in a reaction that mimics formation of the ternary elongation complex. In this case, TFIIF and Elongin might promote extension of DNA molecules by RNA polymerase II by facilitating proper positioning of DNA 3Ј-OH termini with respect to the polymerase active site and, similarly, might stimulate elongation by helping to maintain proper positioning of the 3Ј-end of nascent transcripts in the polymerase active site. In light of these observations, it is possible that TFIIF elongation activity increases the processivity of very early elongation complexes by preventing release of the nascent transcript before it has grown long enough to remain stably associated with elongating polymerase.
Second, by taking advantage of premelted AdML promoter derivatives that do not require TFIIF for initiation, we observe that a TFIIF mutant F-RAP30(⌬181-195), which is capable of strongly stimulating elongation by RNA polymerase II but selectively lacks initiation activity, is partially defective in supporting TFIIH activity in promoter escape. Thus, sequences within the RAP30 C-terminal domain contribute to functional interactions between TFIIF and TFIIH. In addition, we have made the surprising finding that F-RAP30(⌬181-195) can support more efficent promoter escape than wild type TFIIF in the absence of TFIIH, raising the possibilities (i) that a TFIIF function important for transcription initiation may actually present an impediment to promoter escape by RNA polymerase II and (ii) that TFIIH may function during promoter escape to help overcome this impediment.
How might a TFIIF function important for initiation block promoter escape? Based on results of cross-linking experiments and on analysis of electron micrographs of the preinitiation complex (10 -12, 55, 56), it has been proposed that promoter DNA is tightly wrapped around RNA polymerase II and the general initiation factors in the preinitiation complex (12) (see Fig. 7). Notably, the TFIIF subunits RAP30 and RAP74 are cross-linked to promoter DNA both upstream and downstream of the TATA box in the preinitiation complex, and it is proposed that these contacts are in large part responsible for formation and maintenance of the tightly wrapped structure. TFIIF-dependent bending of the DNA within the preinitiation complex may facilitate unwinding of promoter DNA during formation of the open complex.
The TFIIF mutant F-RAP30(⌬181-195) contains a deletion in RAP30 C-terminal sequences that encode a cryptic DNAbinding domain exhibiting sequence similarity to the highly conserved promoter-binding domains present in region 4 of bacterial factors (39,40). Recent NMR studies have revealed that this RAP30 C-terminal region is remarkably similar in structure to the winged helix turn helix DNA-binding domains of linker histone H5 and hepatocyte nuclear transcription factor HNF3/forkhead (23). It is therefore possible that TFIIF could promote wrapping of DNA around RNA polymerase II in the preinitiation complex by binding directly to promoter DNA, in part through its RAP30 C-terminal DNA-binding domain. Binding of TFIIF to promoter DNA downstream of the transcriptional start site could provide an impediment to elongation by early RNA polymerase II elongation intermediates, and the TFIIH DNA helicase could function at least in part to increase the efficiency of promoter escape by disrupting TFIIF interactions with downstream DNA (Fig. 7). In light of evidence (i) that premelting promoter DNA downstream of the transcriptional start site largely removes the TFIIF-induced impediment to early elongation and (ii) that the RAP30 C-terminal DNA-binding domain interacts most tightly with doublestranded DNA (39), it is possible that the TFIIH DNA helicase could disrupt TFIIF interactions with the promoter simply by unwinding downstream DNA.