Intermediates in formation and activity of the RNA polymerase II preinitiation complex: holoenzyme recruitment and a postrecruitment role for the TATA box and TFIIB

A special role of the TATA box

To examine the role of the TATA box, we compared PIC formation and activity on the HIS4 and HIS4mTATA templates (Fig. ​(Fig.1).1). In basal and Gal4–AH-stimulated PIC assembly assays with HIS4mTATA, the levels of all factors probed for were reduced compared with the HIS4 template (Fig. ​(Fig.2a,2a, lanes 2 and 3 vs. 5 and 6, also see Fig. ​Fig.8b,8b, below). As mentioned above, the TBP signal was reduced at least twofold and was not stimulated by the activator. The levels of all other factors tested were reduced two- to fourfold in basal and activated reactions, but still showed an approximately twofold stimulation by Gal4–AH. As described above for the HIS4 template, complexes formed on the HIS4mTATA template were localized to the 93-bp region between the Gal4 site and the DNA 45-bp downstream of the deleted TATA box (not shown). Thus, the TATA box mutation does result in a reduction in the formation or stability of PICs. Importantly however, PIC formation is still seen in the absence of the TATA box.We next measured the transcription activity of PICs on the HIS4mTATA template. No products were detected in the primer extension assay that detects transcripts of ∼100 nucleotides (Fig. ​(Fig.2c).2c). To test whether these complexes were defective in synthesis of short transcripts, an S1-nuclease protection assay was used with an oligonucleotide designed to detect transcripts at least 30 nucleotides in length. In the presence of Gal4–AH, synthesis of short transcripts was reduced 13-fold by the TATA mutation (Fig. ​(Fig.2d).2d). Assays to directly detect open complex formation by use of permanganate sensitivity or abortive initiation have not been successful in the yeast system (unpublished results). To test if the Pol II in these PICs was transcribing in the opposite direction, we performed primer extension reactions with a primer that recognized a sequence upstream of the Gal4 site on the nontranscribed strand. No α-amanitin-sensitive RNAs were detected (data not shown). Thus, even though some PICs were formed on the TATA mutant template, they were severely compromised in synthesis of transcripts at least 30 nucleotides in length. These results are consistent with a role for the TATA box in the extent of PIC formation, as well as in an essential function that affects the activity of the PIC. It is possible that the destruction of the TATA sequence affects the ability of TBP to distort the DNA in a way that promotes formation of functional PICs.

Use of mutant extracts to examine mechanisms of PIC formation and function

To examine the roles of several subunits of the transcription machinery in PIC formation and function, we generated nuclear extracts from a number of strains carrying conditional alleles in specific factors. We reasoned that by using an unfractionated extract containing a mutation in a specific factor, we could ask how this defect affects PIC formation and function in the context of the remaining transcription machinery.

A TBP mutant is defective in stable PIC formation

It is well established that TBP is required for transcription of genes transcribed by Pol II as it nucleates the formation of the complete PIC (for review, see Hampsey 1998). Using the immobilized promoter system, we assayed the ability of an extract prepared from a strain carrying the TBP I143N temperature-sensitive mutation to form active PICs. The HIS4 immobilized template was preincubated with the I143N extract and Gal4–AH, in the presence or absence of recombinant TBP (rTBP), for 40 min, after which multi- and single-round transcription assays were performed as described for Figure ​Figure2b.2b. In the absence of rTBP, very low levels of RNA were produced in multi- and single-round assays (Fig. ​(Fig.3a,3a, lanes 4,6,8). Addition of rTBP to the mutant extract restored transcription to levels comparable with those of the wild-type extract (∼10-fold increase in single-round and washed reactions). In assays for PIC formation, the levels of all factors probed for were significantly reduced compared with the wild-type extract (Fig. ​(Fig.3b,3b, lane 2). Addition of rTBP restored the levels of all PIC components tested to wild-type levels (Fig. ​(Fig.3b,3b, lane 3). By densitometry, a 14-fold or greater stimulation was detected for all components except for TAF_II20 and TAF_II90, which showed a fourfold and twofold stimulation, respectively (Fig. ​(Fig.3b;3b; data not shown). Because TAF_II20 and TAF_II90 were not detected on the Promoterless template in the presence of Gal4–AH, it is possible that these TAF_IIs interact at a low level with the HIS4 promoter through a TBP-independent mechanism. Because these TAF_IIs are also components of the SAGA complex, it is possible that their recruitment to the promoter is facilitated by its association with SAGA (Grant et al. 1998).

Open in a separate window

Figure 3

An extract containing a TBP mutant is defective in stable PIC formation and activity. (a) Transcription activity of PICs formed in wild-type and TBP(I143N) extracts. Experimental design is outlined at top. The indicated extracts were incubated with the HIS4 immobilized template and Gal4–AH for 40 min. rTBP (200 ng) was included where indicated. For multiround measurements, PICs were incubated with NTPs for 40 min (lanes 1,4,5). Single round measurements were obtained by incubating PICs with NTPs for 2 min (lanes 2,6,7). Activity of washed complexes was measured by resuspending PICs in transcription buffer containing NTPs for 2 min (lanes 3,8,9). (Brackets) α-Amanitin sensitive start sites. (b) PIC assembly on the HIS4 template in wild-type and TBP(I143N) extracts. PICs were formed as described in a except that reactions were scaled up twofold. PICs were analyzed as described in Fig. ​Fig.22a.

The role of TFIIA in PIC formation and activity

The mechanism of TFIIA action in transcription has been controversial. TFIIA has been proposed to act as both an antirepressor and a coactivator (for review, see Orphanides et al. 1996; Hampsey 1998). TFIIA interacts with numerous factors in vitro, including TBP, a TAF_II, activators, and coactivators, and it can counteract the effect of a number of repressors. TFIIA can stabilize TBP binding to DNA (Weideman et al. 1997 and references therein), and it is required for an activator-dependent conformational change in TFIID–DNA complexes observed in vitro (Lieberman and Berk 1994; Chi and Carey 1996). To investigate the role of TFIIA in PIC formation and function, we prepared extracts from strains carrying conditional mutations in either of the two subunits of TFIIA, Toa1, or Toa2, for use in the immobilized template and transcription assays. Toa1-25 is a triple alanine mutation (K255, R257, K259) in the large subunit of TFIIA (Kang et al. 1995). The structure of the TFIIA–TBP–TATA complex and other biochemical studies suggest that the Toa1-25 mutant phenotype is due to a defective TFIIA–DNA interaction (Kang et al. 1995; Geiger et al. 1996; Tan et al. 1996). The other mutation used, Toa2-3, contains a double alanine mutation (D21 and D24) in the small subunit of TFIIA. These residues are located in the four helix bundle domain that faces away from TBP and DNA in the TFIIA–TBP–TATA complex (Geiger et al. 1996; Tan et al. 1996). In binding assays, TFIIA containing the Toa2-3 subunit formed stable complexes with TBP on DNA (Kang et al. 1995).

Extracts from strains containing the toa1-25 or toa2-3 alleles were assayed in multi- and single-round transcription experiments as described for Figure ​Figure2a.2a. Low, but detectable, levels of transcription were detected without recombinant TFIIA (rIIA), and rIIA stimulated transcription 3- to 4-fold in single round assays (Fig. ​(Fig.4a,4a, lanes 6–9 and 12–15), and 6- to12-fold in multi-round assays (Fig. ​(Fig.4a,4a, lanes 4–5 and 10–11). In PIC formation assays, both mutant extracts gave reduced levels of all factors probed for, and these levels were stimulated by rIIA (Fig. ​(Fig.4b,4b, lanes 2–5). The mutant TFIIAs were not detected in PICs formed in both extracts. For the Toa1-25 extract, this defect is not the result of instability of TFIIA, because both TFIIA subunits are readily detectable by Western blot (data not shown). Because our Toa2 antibodies do not react well with the Toa2-3 protein (Kang et al. 1995), we cannot quantitate the stability of this subunit in the extract. rIIA stimulated the levels of all factors tested approximately three- to fourfold [in lane 4 vs. 5, rIIA stimulates TAF_II90 approximately eightfold. However, in two other experiments TAF_II90 was stimulated approximately threefold (not shown)]. The correlation between the effect of TFIIA on both the activity of PICs and the levels of PIC components, suggests that TFIIA stimulates transcription by increasing the extent of PIC formation. Because we did not detect changes in the levels of NC2α (Gadbois et al. 1997) in PIC formation assays in the presence or absence of TFIIA (not shown), it seems unlikely that TFIIA stimulates transcription by removal of this regulator from the templates. Our results that TFIIA affects the stable recruitment of TFIID in nuclear extracts are consistent with a number of studies using purified factors that demonstrated a role for TFIIA in the extent of formation of stable activator-TFIID–TFIIA–TATA complexes (Lieberman and Berk 1994; Chi et al. 1995; Kobayashi et al. 1998). An alternative model in which the defect in TFIID recruitment is a secondary effect due to a role of TFIIA in holoenzyme recruitment, is not consistent with experiments presented in the next section.

Open in a separate window

Figure 4

Extracts containing mutant subunits of TFIIA are defective in stable PIC formation and activity. (a) Transcription activity of PICs formed in wild-type and TFIIA mutant extracts. Experimental design is outlined at the top. The indicated extracts were incubated with the HIS4 immobilized template and Gal4–AH for 40 min. rTFIIA (80 ng) was included where indicated. For multiround measurements, PICs were incubated with NTPs for 40 min (lanes 1,4,5,10,11). Single round measurements were obtained by incubating PICs with NTPs for 2 min (lanes 2,6,7,12,13). Activity of washed complexes was measured by resuspending PICs in transcription buffer containing NTPs for 2 min (lanes 3,8,9,14,15). Brackets indicate the α-amanitin sensitive start sites. (b) PIC assembly on the HIS4 template in wild-type and TFIIA mutant extracts. PICs were formed as described in a except that reactions were scaled up twofold. PICs were analyzed as described in Fig. ​Fig.22a.

The effect of TFIIA on the extent of PIC formation could be explained by an effect on the rate of complex assembly, and/or the stability of PICs. To probe the mechanism of TFIIA action, we measured the rate of active PIC formation in the presence and absence of rIIA. In the experiment shown in Figure ​Figure5a,5a, the Toa1-25 extract was incubated with the HIS4 template to which Gal4–AH was prebound, in the presence or absence of rIIA. At the indicated times, an aliquot of the reaction was removed to a tube containing NTPs to initiate transcription. To obtain single round measurements of RNA production, reactions were stopped after 2 min. The results show that during the ∼40 min window during which active PICs accumulate, TFIIA stimulates the rate of active PIC formation. Measurements of the slopes of the curves in Figure ​Figure5b5b show that TFIIA has at least a fivefold effect on the rate of active PIC formation. We next tested whether the stimulation of PIC formation by TFIIA could be explained by an increase in stability of PICs. To do this, we used the template challenge protocol outlined in Figure ​Figure5c.5c. PICs were allowed to form for 40 min in the presence or absence of rIIA, after which a 20-fold molar excess of challenger HIS4 template was added for up to 60 min. The number of active PICs that were stable to challenge was measured by adding NTPs to the reactions for 2 min, followed by stop buffer. Addition of the challenge template at the start of the reaction strongly inhibited transcription, showing that the challenge template could effectively compete for transcription factors in these reactions (Fig. ​(Fig.5c,5c, lanes 1 vs. 2 and 6 vs. 7). However, after 60 min of challenge, PICs formed in the absence and presence of rIIA were relatively stable (Fig. ​(Fig.5c,5c, lanes 4 vs. 5 and 9 vs. 10). By PhosphorImager analysis, ∼50% and 55% of the the active PICs remained stable to challenge in the absence and presence of rIIA, respectively. Together with previous results, these findings suggest that TFIIA stimulates transcription by promoting the stable association of TFIID with the promoter. As a result, the rate and extent of PIC formation is increased. However, once formed, PICs that lack TFIIA are stable and active to the same extent as complexes containing TFIIA.

Open in a separate window

Figure 5

Mechanism of TFIIA action. (a) Rate of formation of active PICs in a wild-type and a Toa1-25 mutant extract. Experimental design is outlined at top. The indicated nuclear extracts were incubated with the HIS4 template (pSH515) and Gal4–AH, in the presence or absence of rTFIIA (40 ng). At the indicated time, an aliquot of the reaction was removed and added to a tube containing NTPs. After 2 min the reaction was stopped and analyzed by primer extension. (Brackets) α-Amanitin- sensitive start sites. (b) Graph of the quantitated data from a. (c) Stability of PICs to challenge by a second template in the presence and absence of TFIIA. Experimental design is outlined at top. Toa1-25 mutant extract was incubated with the HIS4 template (pSH515) and Gal4–AH for 40 min. rTFIIA (40 ng) was included in reactions 6–10. PICs were challenged by addition of a 20-fold molar excess of a HIS4 promoter template (pSH388) for either 10 min (lanes 3,8) or 60 min (lanes 4,9). Transcripts produced from this template are not detected with the primer used in these experiments. Single-round transcription was measured by addition of NTPs for 2 min. (Lanes 1,6) Single-round transcription of PICs formed for 40 min with no challenge, in the absence and presence of rTFIIA. (Lanes 2,7) Identical to lanes 1 and 6 except that pSH388 was added at the start of the incubations. (Lanes 5,10) Single-round transcription of PICs that have incubated for the same time as reactions 4 and 9 without pSH388.

Extracts from conditional yeast strains define at least two intermediates in PIC assembly

PIC formation requires assembly of >50 polypeptides at the promoter. The identification of preassembled holoenzyme complexes containing Pol II and a subset of the transcription machinery suggested that at least some PIC assembly involves the recruitment of preassembled factors. However, a number of important questions about the assembly of the PIC remain unanswered. First, not all factors are in the holoenzyme complexes and it is unclear how the preassembled factors interact with the other factors to form the final PIC. Second, at least three forms of yeast holoenzymes have been isolated and it is not clear which form assembles at any particular promoter. Third, there is an excess of free general factors over holoenzyme complexes (Lee and Young 1998), suggesting that stepwise PIC assembly is also a viable mechanism.

We found that extracts containing mutations in TBP or TFIIA subunits showed defects in stable recruitment of all factors tested with the GAL4–AH activator. Under these conditions, holoenzyme was not recruited independently of TFIID and TFIIA. In contrast, extracts containing mutations in TFIIB, Pol II, Srb2, Srb4, and Srb5 showed defects in recruitment of all components except TFIID and TFIIA. These results indicate that in the yeast system, PIC assembly can occur in at least two steps: the holoenzyme-independent assembly of TFIID and TFIIA at the promoter, followed by holoenzyme recruitment. However, it is likely that alternative mechanisms exist depending on promoter architecture, the types of regulatory factors present, and the presence of chromatin (Struhl 1996). For example, holoenzyme recruitment could precede TFIID recruitment. The Gal4–VP16 activator can recruit high levels of holoenzyme components (Hengartner et al. 1995) and certain TAF_IIs in a manner that does not require promoter DNA (Fig. ​(Fig.2a).2a). In vivo, artificial recruitment of holoenzyme components can bypass the need for an activator (Keaveney and Struhl 1998 and references therein). One interpretation of these experiments is that activators such as Gal4–VP16 target holoenzyme components as well as TFIID for recruitment. Also, alternative mechanisms for PIC assembly likely exist at promoters that are transcribed independently of Srb factors (Lee and Lis 1998; McNeil et al. 1998).

As it has been estimated that TFIIB and the other general transcription factors are in 5- to 10-fold excess over the holoenzyme complex (Lee and Young 1998), it is surprising that little detectable binding of TFIIB to the TFIID–TFIIA-promoter complex occurs in the Pol II or Srb mutant extracts (Figs. ​(Figs.7b7b and ​and8b,c).8b,c). Western blot analysis showed that TFIIB is present at normal levels in these extracts. One possibility is that a repressor or some modification blocks binding of TFIIB to the TFIID–TFIIA-promoter complex when it is not part of a holoenzyme complex. One important unanswered question is why Srbs are required for transcription in crude yeast extracts (for review, see Koleske and Young 1995), but not required in highly purified systems. On the basis of our results, one possible reason for the Srb requirement in crude extracts is that the stepwise assembly pathway is blocked. Interestingly, it was found that mutation of two transcription repressors (NC2 or the NOT complex) could partially compensate for a temperature-sensitive mutation in Srb4 (Gadbois et al. 1997; Lee et al. 1998). Further work will be required to explore the relationship between these and other repressors and the requirement for Srb factors in transcription initiation. However, initial experiments have shown that a mutation in NC2α (Gadbois et al. 1997; Lee et al. 1998) will not compensate for a mutation in Srb4 in vitro (S. Hahn, unpubl.).

The role of TFIIA in PIC assembly and activity

In systems that are responsive to TFIIA, TFIIA stimulates both basal and activated transcription (DeJong et al. 1995; Kang et al. 1995). TFIIA tightly binds TBP on DNA and this binding competes with inhibitors of TBP–DNA binding, such as yTAF_II145/hTAF_II250 and Mot1 (Auble and Hahn 1993; Kokubo et al. 1998; Ozer et al. 1998). TFIIA binding to TBP also competes with other repressors that bind TBP and inhibit the assembly of active PICs such as, NC2, topoisomeraseI/Dr2, and HMG1 (for review, see Hampsey 1998). These results have led to the model that TFIIA functions to counteract inhibitors of transcription by blocking access to TBP. Several in vivo studies suggest that TFIIA may play a specific role in promoting high levels of activated transcription distinct from its mechanism of counteracting transcription inhibitors. In these studies, yeast and human TBP mutations were found that specifically inhibited activated transcription, and these mutations were defective in TFIIA binding (Stargell and Struhl 1995; Bryant et al. 1996). In vitro, TFIIA stimulates the rate and extent of purified TFIID binding to DNA, and activators can enhance this effect (Chi et al. 1995; Kobayashi et al. 1998). Additionally, it was found that the form of TFIID that binds in the presence of an activator and TFIIA is in a different conformation than that which weakly binds in the absence of TFIIA and/or activator (Lieberman and Berk 1994; Chi and Carey 1996). Furthermore, TFIIA has been found to interact directly with certain activators and coactivators in vitro, and it has been proposed that these protein–protein interactions are important for specific stimulation of activated levels of transcription (for review, see Hampsey 1998).

With extracts from strains carrying conditional mutations in TFIIA subunits, PIC formation was defective. The levels of all components examined were reduced, including TBP and TAF_IIs. It is unlikely that the defect in TFIID recruitment is a secondary effect due to a role of TFIIA in holoenzyme recruitment, because TFIID recruitment is unaffected in extracts from strains containing mutations in holoenzyme components. The defects in PIC formation caused by the TFIIA mutations correlated well with defects in the transcription activity of these PICs. TFIIA stimulated the rate, as well as the extent, of active PIC formation, but complexes that lacked TFIIA were just as stable as complexes that contained TFIIA. Our results, using a crude system, together with previous results using purified factors, demonstrate that TFIIA acts by stimulating the rate and extent of TFIID binding to the promoter.

Taken together with previous findings, our results lead to the following model for TFIIA action: TFIID exists in two forms, a productive form that is capable of stably interacting with the promoter, and a nonproductive form that cannot stably bind the promoter because of the inhibitory effect of TAF_IIs, other repressors, and dimerization. TFIIA shifts this equilibrium toward the productive form of TFIID resulting in more stable recruitment of TFIID. By this model, the productive form of TFIID that binds DNA would be analogous to the form observed in the presence of activator, TFIID, and TFIIA with purified factors (Lieberman and Berk 1994; Chi and Carey 1996). This model can account for TFIIA’s requirement for activated transcription because high levels of activator-TFIID–TFIIA complexes will be stably bound to the promoter in which they can support multiple rounds of transcription. It is unlikely that the only role of TFIIA is to counteract the effect of Mot1, because in extracts prepared from a TFIIA/Mot1 double-mutant strain, TFIIA still stimulated transcription (data not shown). Also, it is unlikely that TFIIA acts solely by removal of the repressor NC2 from the promoter, because the levels of this factor at the promoter were unaffected by the TFIIA mutations (data not shown). However, it is possible that TFIIA counteracts NC2 binding to TBP off of DNA, which could increase the pool of TFIID that can stably interact with the promoter.

A postrecruitment function for TFIIB

One extract containing the TFIIB mutation E62G has a strong transcription defect, but is unaffected for PIC assembly. These results suggest that TFIIB has a postrecruitment role that is independent of its role in PIC assembly. Residue E62 is in the amino-terminal conserved region of TFIIB, and it lies within a homology block (residues 61–68) that is conserved among all the TFIIB family members (Pardee et al. 1998). Another mutation of this same residue, E62K, results in a downstream shift in start site selection and a defect in Pol II interaction (Pinto et al. 1994; Bushnell et al. 1996). Other mutations in this homology block alter start site selection without affecting Pol II interaction (Pardee et al. 1998). Although we did not detect altered start site selection, our results with the E62G mutant are consistent with the latter results because, in both cases, a step after PIC recruitment was affected. Our data suggest that PICs assembled with the E62G mutant are defective at a step after PIC formation such as open-complex formation or promoter clearance.

Recently, a large deletion in the Pol III factor Brf (TFIIB related factor), was described that recruits Pol III normally, but results in a defect in open-complex formation (Kassavetis et al. 1998). This deletion removes the homology block described above in TFIIB as well as the zinc ribbon and the first repeat in the TFIIB-related region. It is possible that the defect in this Brf deletion and the E62G TFIIB mutant result in analogous defects in initiation for both Pol II and Pol III.

A special role for the TATA box

The role of the TATA box as an important promoter element and as a target for TBP binding is firmly established. Most, but not all, TATA mutations that decrease transcription have a corresponding effect on TBP binding (Hoopes et al. 1998). However, few studies on TATA box function have directly monitored the effects of TATA mutations on PIC formation. We found that removal of the TATA box severely reduced single- and multi-round transcription. Surprisingly, PIC assembly was decreased by only two- to fourfold, in contrast to the dramatic decrease in transcription activity. It is likely that other elements present at the promoter as well as protein–protein interactions, compensate for the TATA box mutation to permit recruitment of factors to form a stable PIC. Our results are consistent with a previous study in which a nuclease protection assay was used to monitor PIC assembly on a wild-type and mutant TATA-containing promoter (Jacob et al. 1994). The authors found a 12-fold defect in initiation compared to only a 4-fold defect in PIC assembly on the mutant TATA promoter, suggesting a role for the TATA box in a step after PIC assembly.

On the basis of our results, as well as previous studies, we propose that the conformation of the TBP–DNA complex is critical in determining the activity of the PIC. At the HIS4mTATA promoter, TBP cannot interact normally with the G–C sequence as the DNA binding surface of TBP is incompatible with the G–C DNA minor groove (Kim et al. 1993a,b). Although the transcription machinery is stably recruited to the mutant promoter, this lack of normal TBP binding gives rise to a complex in which Pol II is compromised in synthesis of 30 nucleotide transcripts. This may be due in part to a failure of TBP to distort the DNA as is seen on authentic TATA elements (Kim et al. 1993a,b). Thus, the TATA box may play a dual role in the initiation process, first as a target of TBP binding in which it nucleates the assembly of the PIC, and second, in promoting the activity of the PIC through its stereospecific interaction with TBP.

Preparation of nuclear extracts

Yeast nuclear extracts were prepared from 2–8 liter cultures of the indicated strains grown at the permissive temperature as described in Kang et al. (1995) and on the world wide web at www.fhcrc.org/science/basic/labs/hahn.

Recombinant proteins

All recombinant proteins were expressed in Escherichia. coli. rTBP was a gift from J. Geiger. rTFIIA was prepared as described in Ranish et al. (1992) after separately expressing the Toa1 and Toa2 subunits in E. coli. rTFIIB was purified by chromatography on BioRex 70 and Hydroxylapatite resins (Bio-Rad). His-tagged rSrb2 was purified by Nickel–agarose chromatography (Qiagen) under denaturing conditions, and renatured after purification. Gal4–VP16 was purified by polyethyleneimine precipitation followed by chromatography on S–Sepharose (Pharmacia). Gal4–AH was purified by polyethyleneimine precipitation followed by chromatography on DEAE–Sepharose, and S–Sepharose (Pharmacia).

Plasmid templates

For transcription assays on plasmid templates, pSH515 and pSH514 were used. pSH515 contains 144 bp of the HIS4 core promoter, (−141 to +3 with respect to the translation start site) including the TATA box and RNA start sites, cloned downstream of a single Gal4-binding site. In addition, the bases at four positions around the TATA box were changed to eliminate potential cryptic TATA sequences (sequence available upon request). pSH514 is identical to pSH515 except that the TATA box (TATATAATA) was replaced with the sequence TACCCGGGA. pSH388 contains 144 bp of the HIS4 core promoter, (−141 to +3 with respect to the translation start site) including the TATA box and RNA start sites, cloned downstream of three Gal4-binding sites. pSH388 lacks the sequence recognized by the primer used for primer extension in all of the experiments.

Biotinylated templates

Biotinylated templates were prepared by PCR with either pSH515 or pSH514 as templates, biotinylated, upstream primer p965 (5′-biotin-TAATGCAGCTGGCACGACAGG-3′) and various downstream primers. p965 is located ∼280 bp upstream from the Gal4 site. For synthesis of the HIS4 and HIS4mTATA templates, downstream primer pNot (5′-GGCCGCTCTAGCTGCATTAATG-3′) was used with both pSH515 and pSH514 to produce 594-bp products. For synthesis of the Promoterless template, p965 was used with primer BKS7 (5′-TACCGAGCTCGAATTCGGAGG-3′) and pSH515, which resulted in production of a 306-bp fragment that ends 14-bp downsteram of the Gal4 site. PCR products were extracted with phenol:chloroform (1:1), ethanol precipitated, and purified with S300 columns (Pharmacia). DNAs were quantitated by a fluorescent dye binding assay (PanVera Corp.)

Immobilized templates

Dynabeads M-280 Streptavidin (Dynal) were concentrated with a magnetic particle concentrator (MPC) (Dynal) and washed twice in Buffer T [10 mm Tris (pH 7.5), 1 mm EDTA, 1 m sodium chloride] (∼1 ml Buffer T/mg beads). The beads were resuspended in Buffer T at 10 mg/ml. NP-40 was added to 0.003% to prevent beads from sticking to the sides of tube. Dynabeads were incubated with ∼20 fmole biotinylated template/μg bead in Buffer T for 30 min at room temperature with constant agitation. The immobilized templates were concentrated with the MPC, and washed in Buffer T (∼1 ml Buffer T/mg beads). Immobilized templates were blocked in block buffer (1 ml/mg beads), for 15 min at room temperature. Block buffer consists of transcription buffer [10 mm HEPES (pH 7.6), 100 mm potassium glutamate, 10 mm magnesium acetate, 5 mm EGTA, 3.5% glycerol], containing 60 mg/ml casein (Sigma, C-5890), 5 mg/ml polyvinylpyrrolidone (USB, 20611), and 2.5 mm DTT. The beads were concentrated with the MPC, washed three times in transcription buffer, and resuspended in transcription buffer at 10 mg/ml. Immobilized templates were prepared fresh before each experiment.

In vitro transcription with plasmid templates

Transcription reactions with plasmid templates were performed as described in Ranish and Hahn (1991) and on the world wide web at www.fhcrc.org/science/basic/labs/hahn/. Reaction components were preincubated for 35–40 min at room temperature, after which 1 μl of 10 mm rNTPs (2.5 mm each) was added to the reactions. For single-round measurements, two methods were used. In the first method, sarkosyl was added to 0.2%, 45–60 sec after addition of rNTPs. Reactions were stopped 2–10 min later by the addition of 200 μl of stop solution (0.1 M sodium acetate, 10 mm EDTA, 0.5% SDS, 10 μg/ml tRNA). A time course showed that RNA synthesis was complete after 2 min. In the second method, stop solution was added 2 min after addition of rNTPs. Both methods gave comparable results. For multiround measurements, reactions were stopped 40–60 min after rNTP addition. Samples were phenol/chloroform extracted, and precipitated with ethanol. Products were analyzed by primer extension as described in Ranish and Hahn (1991), or by S1 nuclease protection as decribed in Cormack and Struhl (1992). S1 analysis was performed with a 5′-end-labeled oligonucleotide (5′-GTAAACTATTGTATTACTATTACACAGCGCAGGGTGTAG) with a short mismatch to the template strand at the 3′ end. Quantitation was performed with a PhosphorImager (Molecular Dynamics).

In vitro transcription with immobilized templates

Reactions were scaled up twofold from the standard transcription reaction to 50 μl. On ice, 120–180 μg of nuclear extract was mixed with 20 μl of 2.5× transcription mix, 0.05% NP-40, and 20–24 μl of transcription buffer to bring the final volume to 44 μl. After a 10 min incubation at room temperature, the mix was spun at 9K rpm for 2 min in an Eppendorf microfuge at 4°C. This step removed insoluble material that bound to the magnetic beads. The supernatant was transferred to a new tube on ice, and recombinant factors were added (where indicated). After addition of ∼250 ng HaeIII-digested E. coli DNA to reactions, 2.5 μl of immobilized template is added. For activated reactions, 2.5 μl of template is incubated with 60 ng Gal4–AH for ∼10 min at room temperature before addition to the reaction. After a 40 min incubation, immobilized templates were concentrated with the MPC, and washed three times with 150 μl of transcription buffer containing 0.05% NP-40, 0.5 mg/ml BSA, and 2.5 mm DTT. After resuspension in 50 μl of transcription buffer containing 12 mm phosphocreatine, 400 ng creatine phosphokinase, 2.5 mm DTT, 0.5 mg/ml BSA, and 20 units of ribonuclease inhibitor, transcription was initiated by addition of 100 μm of each rNTP. Reactions were stopped after 2–10 min by addition of 400 μl of stop solution. The supernatant was removed from the beads, phenol/chloroform extracted, and ethanol precipitated. Products were mapped by primer extension as described above, except that actinomycin C₁ was included (15 μg/ml) during the extension reaction. In Figure ​Figure8,8, primer extension products were resolved on a 6% Quickpoint sequencing gel according to the manufacturer’s protocol (NOVEX).

We thank members of the Hahn laboratory, past and present, for their thoughtful insights, and A. Krumm, R. Reeder, and T. Tsukiyama for comments on the manuscript. A. Berk suggested important controls. We thank E. Gadbois, R. Young, R. Knauss, L. Guarente, M. Carey, and J. Corden for strains. We also thank E. Gadbois, R. Young, J. Reese, M. Green, C.-Y. Ho, M. Swanson, Y. Nakatani, N. Thompson, D. Burgess, S. Buratowski, and C. Denis for antibodies. R. Kornberg and Y.-J. Kim generously provided E. coli strains expressing recombinant TFIIE and Med6 subunits used for antibody production. We thank K. Coachman and G. Schimmack for antibody production, J. Geiger for rTBP, S. Hurst for rTFIIB, and S. Miller for rSrb2.

This work was funded by a National Science Foundation predoctoral fellowship (J.A.R.), National Cancer Institute grant T32CA09657 (N.Y.), and National Institutes of Health grant GM42551 and a Leukemia Society Scholar Award to S.H. S.H. is an associate investigator of the Howard Hughes Medical Institute.