TAF7: A possible transcription initiation check-point regulator

Abstract

Transcription consists of a series of highly regulated steps: assembly of a preinitiation complex (PIC) at the promoter nucleated by TFIID, followed by initiation, elongation, and termination. The present study has focused on the role of the TFIID component, TAF7, in regulating transcription initiation. In TFIID, TAF7 binds to TAF1 and inhibits its intrinsic acetyl transferase activity. We now report that although TAF7 remains bound to TAF1 and associated with TFIID during the formation of the PIC, TAF7 dissociates from the PIC upon transcription initiation. Entry of polymerase II into the assembling PIC is associated with TAF1 and TAF7 phosphorylation, coincident with TAF7 release. We propose that the TFIID composition is dynamic and that TAF7 functions as a check-point regulator suppressing premature transcription initiation until PIC assembly is complete.

In eukaryotic cells, expression of most protein-encoding genes depends on the RNA polymerase II (Pol II)-dependent transcription machinery. The RNA Pol II machinery assembled on the promoter is composed of distinct complexes that are responsible for effecting the sequential steps of transcription initiation and elongation (1–5). The first step in transcription is the recognition of the core promoter by the TFIID complex and the assembly of a preinitiation complex (PIC) through an ordered recruitment of the general transcription factors (GTFs) TFIIB and TFIIA, followed by the RNA Pol II holoenzyme, the mediator and the remaining GTFs, TFIIF, TFIIE, and TFIIH. Once the PIC has been assembled, transcription initiation ensues: local melting of the promoter DNA, formation of the first phosphodiester bond, followed by the synthesis of a short nascent RNA at which point the polymerase pauses. The initiation of transcription is accompanied by the phosphorylation of serine 5 in the C-terminal domain (CTD) heptad repeat of RNA Pol II (CTD) by the kinase subunit of TFIIH, CDK7 (6–9).

Despite the extensive understanding of the general requirements of transcription, many details remain unresolved and there is considerable variation among different promoters and cell types. Promoter recognition is mediated by members of either the TFIID complex family (TFIID, TFIID-like) or the SAGA complex family (e.g., TFTC, PCAF, SAGA) (10–12). In yeast, 90% of gene expression is TFIID-dependent transcription; the remaining 10% is largely dependent on the SAGA complex (13). The TFIID complexes are composed of either the TATA-binding protein (TBP) or a TBP related protein (TRF1, TRF2) and several TBP-associated factors (TAFs) (14–17). The SAGA family complexes do not contain TBP or TBP-related proteins; rather, they contain a GCN5-related histone acetyl transferase (AT) subunit, several adapter and Spt proteins and a subset of TAFs. The composition of the TAFs present in these different complexes varies depending on the structure of the promoter and the cell cycle and tissue-specific gene expression requirement. For example, in yeast, the SAGA complex associated with stress-induced gene expression contains five of the TAFs described in the TFIID complex (13, 10–12). Conversely, a tissue-restricted TAF, TAF4b is found in the TFIID complexes purified from a mammalian mature B cell line (18).

Four major roles have been ascribed to the individual TAFs. First, they may act as specific coactivators interacting with transactivators (19–22). Second, some are critical for the stability of either the TFIID or SAGA complexes (e.g., TAF10 which contains a histone folding motif; ref. 23). Third, some TAFs are involved in promoter recognition through direct contact with different core promoter elements (24–27). Finally, TAFs may be selectively required by subsets of genes (28, 29). For example, the largest TAF, TAF1, has been reported to be absolutely indispensable for expression of 18% of the mammalian genes (30). TAF1 is known to have kinase activity and acetyltransferase activity (31–33), but the roles of these activities and the function of TAF1 are still not understood. Although relatively little is known about the functions of any of the TAFs, TFIID structure and composition are thought to be relatively invariant on the assembled PIC.

We have previously demonstrated that TAF1 is necessary for the transcription of an MHC class I gene, and that its intrinsic acetyltransferase activity is essential for both in vivo and in vitro transcription of naked DNA (34). Thus, in ts13 cells with a temperature-sensitive mutation in TAF1, MHC class I expression is abrogated at the nonpermissive temperature. Similarly, MHC class I transcription is inhibited, both in vitro and in vivo, by the transactivator, HIV Tat, which binds to the TAF1 AT domain and inhibits its enzymatic activity (35). Of particular interest, we recently reported that TAF7, a TFIID component, also binds to TAF1 and inhibits its AT activity, resulting in repression of MHC class I transcription (36). Thus, TAF7 is an intrinsic regulator of MHC class I transcription.

These findings lead to the hypothesis that repression of TAF1 AT activity by TAF7 must be relieved upon completion of PIC assembly to allow the transition to transcription initiation. The present studies were designed to determine the fate of TAF7 during transcription. We report that TAF7, which is initially associated with TFIID in the PIC, is released upon initiation of transcription. Release of TAF7 from the complex is coincident with the phosphorylation of TAF1 in the complex; TAF1 phosphorylation reduces the binding of TAF7. We propose a model in which TAF7 is a transcription check-point regulator, that prevents transcription initiation until the PIC is fully assembled.

Results

TAF7 Inhibits Transcription in Vivo and in Vitro. TAF7 is a component of the general transcription factor, TFIID. We have shown that TAF7 binds to TAF1, inhibiting its acetyl transferase activity which is required for basal MHC class I transcription (36). This finding leads to the surprising prediction that TAF7 is a transcriptional inhibitor. To test this possibility, CHO cells were cotransfected with an MHC class I promoter construct and a TAF7 expression vector (Fig. 1). As predicted, TAF7 overexpression resulted in a significant inhibition of class I promoter activity. This inhibition was a direct effect of TAF7 on transcription, because addition of TAF7 to an in vitro transcription assay using HeLa nuclear extract similarly inhibited transcription (Fig. 8, which is published as supporting information on the PNAS web site). Furthermore, TAF7 inhibition is not unique to the MHC class I promoter: it similarly inhibited a number of other promoters (data not shown). This is consistent with the widespread (≈24%) dependence of yeast genes on TAF7, which contrasts with TAF1, which is required for only ≈10% of genes (37). These findings raise the question of how TAF7 inhibition is relieved in order for transcription to proceed. We considered two possibilities: (i) TAF7 dissociates from TFIID during PIC assembly, or (ii) TAF7 remains associated with TFIID during PIC assembly, but is released upon initiation of transcription.

Fig. 1.

TAF7 inhibits transcription from an MHC class I promoter. An MHC class I promoter construct (1) consisting of 516 bp of extended promoter sequences ligated to the CAT reporter (5 μg) was cotransfected with either the TAF7 expression vector (2 or 5 μg) or the empty vector (5 μg) into CHO cells. After 48 h, CAT activity was measured and corrected relative to protein concentration.

TAF7 Is Associated with the DNA-Bound PIC. To determine whether TAF7 remains associated with TFIID on the PIC, we assembled PICs on an immobilized MHC class I promoter template. The MHC class I DNA fragment used to nucleate the preinitiation complex was a 731-bp segment, encompassing the MHC class I extended promoter (core and promoter-proximal regulatory elements) and part of the first exon (–516/+44) linked to the 67 first base pairs of the cat reporter gene (Fig. 9, which is published as supporting information on the PNAS web site). The PICs were assembled with HeLa nuclear extract in the absence of nucleotides to prevent transcription initiation. The assembled PIC was purified and associated proteins analyzed by SDS/PAGE and Western blotting. As shown in Fig. 2A, the PIC contained major TFIID components, TAF1 and TAF4, and RNA Pol II, as well as other GTFs (data not shown). Of note, TAF7 was also a component of the isolated PIC, indicating that TAF7 is still associated with TFIID. Although some background association of Pol II and TAF7 was observed to the promoterless control, this binding is probably due to some nonspecific association of Pol II with DNA and consistently was significantly below the extent of binding to the class I promoter (see below).

Fig. 2.

TAF7 is associated with a transcriptionally active PIC assembled on a class I promoter. (A) PICs were assembled as described (Supporting Text and Fig. 9) on either the MHC class I promoter or a control promoter-less fragment PSV0. Associated proteins were revealed by Western blotting with antibodies to the indicated proteins. The extents of nonspecific binding of Pol II and TAF7 to the PSV0 control were comparable, as assessed by densitometry (data not shown). (B) Plasmid DNA containing the MHC class I promoter region (lane 1) or naked beads (lane 2) or two different concentrations of PIC assembled on the immobilized promoter DNA template (420 and 600 ng, lanes 3 and 4) were used as templates in in vitro transcription reactions. The synthesized RNA products were revealed by primer extension using an oligonucleotide primer corresponding to the cat gene region of the template. The two major MHC class I start sites are indicated.

The assembled, purified PICs were active in transcription: addition of the four NTPs after PIC assembly and isolation resulted in correct transcription initiation and elongation (Fig. 2B, lanes 3 and 4). Two transcripts of 67 and 55 bp were synthesized, corresponding to the dominant initiation sites at +1 and +12 bp, observed both in vivo (38) and on naked DNA templates (Fig. 2B lane 1 and ref. 35). Thus, TAF7 remains associated with TFIID during active PIC formation.

Addition of TAF7 to an in vitro transcription reaction decreases MHC class I transcription (Fig. 9 and ref. 36). To determine whether the TAF7 acts directly on the PIC, we determined the effect of adding exogenous TAF7 on the activity of the purified, assembled PIC. Addition of excess TAF7 to the PIC assembly reaction before transcription initiation results in decreased transcription (Fig. 3, compare lanes 1 and 2). However, addition of TAF7 30 sec after initiation of transcription has no effect, suggesting that TAF7 targets the PIC during initiation (Fig. 3, compare lanes 1 and 3). To directly determine whether TAF7 inhibits at initiation or at a subsequent stage in transcription, reinitiation of PIC transcription was inhibited by the addition of sarkosyl at various times after the addition of NTPs. Addition of low concentrations of sarkosyl prevents reinitiation but allows completion of elongating transcripts (Fig. 3, lanes 4 and 6). TAF7 decreased transcription when added before initiation (Fig. 3, lane 4 and 5). In contrast, in the presence of sarkosyl, TAF7 had no effect when added 30 sec after transcription initiation, (Fig. 3, lanes 6 and 7). Thus, excess TAF7 blocks transcription initiation from the assembled PIC, consistent with its inhibition of the TAF1 AT activity.

Fig. 3.

TAF7 inhibits transcription initiation on an assembled PIC. TAF7 and/or sarkosyl at a concentration of either 0.05% or 0.5%, were added either during PIC assembly (0 sec) or 30 sec after the start of transcription, as indicated. In all cases, the reactions were incubated for 30 min to allow completion of transcription. The positions of the two major MHC class I start sites (+1 and +12) are indicated with arrowheads (lane M); the relative level of transcription at each site was determined by densitometry, normalized to the level of transcription observed in the absence of either TAF7 or sarkosyl, and indicated under the appropriate lane.

Endogenous TAF7 Is Released from the PIC During Transcription. Because endogenous TAF7 remains associated with the assembled PIC but is a negative regulator of the TAF1 AT activity necessary for transcription, its inhibition must be relieved to allow transcription to initiate normally. Therefore, we considered the possibility that TAF7 dissociated from the TFIID complex during transcription initiation. To address this question, PICs were assembled on the immobilized class I promoter template (Fig. 9). After PIC assembly and isolation, transcription was initiated by addition of nucleotides. After incubation to allow completion of transcription, proteins that remained associated with the immobilized template were removed by magnetic bead separation. Proteins released into the supernatant during transcription (as well as those remaining associated with the template and recovered on beads), were assessed by SDS/PAGE and Western blotting. Before transcription initiation (in the absence of rNTPs), the assembled and purified PIC contained TAF7, TAF4, and Pol II (Fig. 4). After transcription, an average of 40% of the PIC-associated Pol II was recovered in the supernatant (based on eight independent experiments). A slow-mobility form of Pol II, corresponding to a phosphorylated form as shown by immunoblotting with a phospho-Ser-2-specific antibody (data not shown), was consistently detected in posttranscriptional supernatants and represented an average of 42.5 ± 2.3% (over six experiments) of the total Pol II in the supernatant. The presence of Pol II in the supernatant presumably reflects the dissociation of Pol II as a result of reaching the end of the template and completion of transcription. The Pol II that remained bound to the template is presumed to be associated with inactive PICs, because it was largely unphosphorylated (Fig. 4).

Fig. 4.

TAF7, but not TAF4, is released from the MHC class I PIC after transcription initiation. In vitro transcription reactions of PICs assembled on the MHC class I promoter were done in absence or presence of rNTP. Proteins remaining associated with the promoter (beads) or released from the promoter (sup) during transcription were analyzed by Western blot using antibodies to the indicated proteins. Two forms of RNA Pol II polymerase were detected: the hyperphosphorylated upper band and the hypophosphorylated lower band.

Importantly, a significant fraction of TAF7 was released into the supernatant concomitant with transcription initiation and elongation (Fig. 4 and Fig. 10A, which is published as supporting information on the PNAS web site). In eight independent experiments, an average of 18% of the TAF7 was released into the supernatant after transcription. In contrast to TAF7, only a very small amount (8%) of the TFIID component TAF4 was recovered in the supernatant (Figs. 4 and 10A). When normalized to the amount of Pol II recovered in the supernatant after transcription, the amount of TAF7 released into the supernatant was significantly greater (P < 0.0025) than that of TAF4. To control for the possibility that the simple addition of rNTPs to the PIC nonspecifically released TAF7, purified PICs were incubated with 20 mM rATP. No premature dissociation of either TAF7 or Pol II from the PIC was observed in the presence of rATP alone (data not shown). Furthermore, TAF7 release requires transcription, because it was not released in the presence of the inhibitor α-amanitin (Fig. 10B). Therefore, TAF7 is specifically released from the TFIID complex during transcription.

TAF1 Phosphorylation Results in the Dissociation of the TAF1/TAF7 Complex. We next examined the mechanism by which TAF7 is released from the PIC during transcription. TAF1 is known to have both a kinase activity and an AT activity. Because TAF1 does not acetylate TAF7 (36), we asked whether TAF7 was phosphorylated by TAF1 and, if so, whether it affected the interaction. We performed in vitro kinase assays using purified recombinant TAF7 and TAF1 proteins, either alone or in combination. TAF1 alone autophosphorylated through its intrinsic kinase activity (Fig. 5A, lane 2). (Coomassie blue staining of purified recombinant TAF1 revealed only a single band; data not shown). In contrast, TAF7 did not autophosphorylate (Fig. 5A, lanes 3–5). In the presence of increasing amounts of TAF1, TAF7 was increasingly phosphorylated (Fig. 5A, lanes 6–8). Thus, the interaction of TAF7 with TAF1 resulted in its phosphorylation.

Fig. 5.

TAF1 phosphorylates itself and TAF7. (A) Kinase reactions were performed with recombinant Flag-tagged TAF1 alone (200 ng; lane 2), recombinant Flag-tagged TAF7 with increasing amounts of control extract (lanes 3–5), and Flag-tagged TAF7 in presence of increasing amounts of Flag-tagged TAF1 (lanes 6–8, 40, 80, and 200 ng, respectively). The control was extract purified from uninfected SF9 cells. (B) Kinase reactions were performed as in A, but using recombinant Flag-TAF7 peptide fragments, as diagrammed, derived from the N terminus, central region and C terminus of TAF7. The additional bands observed in lane 8 result from minor degradations observed in this fragment.

The TAF7 phosphorylation site(s) were mapped by using a series of TAF7 deletion mutants (Fig. 5B). The N terminus of TAF7 (amino acids 1–109) contains the domain responsible for inhibition of TAF1 AT activity. The central region of the molecule is a serine-rich region that contains the TAF1 interacting domain (39). As shown in Fig. 5B, TAF7 peptides derived from the central TAF1-interacting domain and the C-terminal domain are efficiently phosphorylated by TAF1, whereas the N terminal domain is not a substrate for phosphorylation. These results identify TAF7 as a substrate of the TAF1 kinase activity.

The finding that the TAF1-interacting domain of TAF7 is phosphorylated by TAF1 suggested that phosphorylation might affect the interaction between TAF7 and TAF1. To test this possibility, recombinant GST-TAF7 was incubated with Flag-tagged TAF1 in the presence of increasing amounts of [γ-³²P]ATP. After the kinase reaction, Flag-tagged TAF1 was recovered on anti-Flag M2 beads, and the amount of TAF7 that copurified with TAF1 was determined by SDS/PAGE and Western blotting. The extent of phosphorylation of both TAF1 and TAF7 (as assessed by extent of ³²P incorporation) increased as a function of ATP concentration (data not shown). The amount of TAF7 protein bound to TAF1 (as determined by Western blotting) was calculated as a function of ATP concentration in the kinase reaction. As shown in Fig. 6A, the binding of TAF1 to TAF7 decreased with increasing phosphorylation of the TAF1 and TAF7 proteins. Thus, the interaction of TAF7 and TAF1 is negatively regulated by phosphorylation.

Fig. 6.

TAF1 autophosphorylation results in the dissociation of TAF7. (A) Kinase reactions were performed by combining recombinant Flag-tagged TAF1 and recombinant GST-tagged TAF7 in presence of increasing amounts of ATP. Flag-tagged TAF1 was immobilized on anti-Flag M2 agarose beads, and the amount of both TAF1 and associated TAF7 recovered were determined by Western blot analysis using the anti-Flag M2 antibody and a GST antibody, respectively. The graph represents the amount of TAF7 associated with TAF1 (corrected to the amount of TAF1 present on the beads) as a function of the ATP added to the kinase reaction. Phosphorylation of the TAF1 and TAF7 was monitored by including [γ-³²P]ATP in the assay. (B) Recombinant Flag-tagged TAF1 was phosphorylated in presence of increasing amount of ATP, immobilized on anti-Flag M2 agarose beads; after extensive washing, GST-tagged TAF7 was added. The amount of TAF1 and associated TAF7 recovered was determined by Western blot analysis using the anti-Flag M2 antibody and a GST antibody, respectively. The graph represents the amount of TAF7 associated with TAF1 (corrected to the amount of TAF1 present on the beads) as a function of the ATP added to the kinase reaction.

Under the conditions of this experiment, it was not possible to distinguish whether the transphosphorylation of TAF7 or autophosphorylation of TAF1, or both, was responsible for the loss of binding. To address this question, the following experiment was performed (see Fig. 6B Inset): Flag-tagged TAF1 alone was incubated in the presence of increasing concentrations of [γ-³²P]ATP to permit autophosphorylation. The phosphorylated Flag-tagged TAF 1 was bound to anti-Flag M2 agarose beads and washed to remove unincorporated ATP. Unphosphorylated GST-TAF7 was then incubated with the phosphorylated TAF1 immobilized on the beads. The ability of TAF7 to bind to TAF1 was determined by recovering the beads and analyzing the amount of TAF7 bound to TAF1 by SDS/PAGE and Western blotting. Surprisingly, with increasing ATP concentration and phosphorylation of TAF1, the subsequent binding of TAF7 to TAF1 decreased, reaching a plateau at ≈50% of the initial binding to unphosphorylated TAF1 (Fig. 6B). The extent of the decrease in binding approximated that observed when both TAF1 and TAF7 were phosphorylated, suggesting that TAF1 phosphorylation is the primary determinant in the extent of interaction. These results indicate that although TAF1 phosphorylates TAF7, TAF1 autophosphorylation is sufficient to reduce the binding of TAF7 to TAF1. The functional consequences of TAF7 phosphorylation remain to be determined; the possibility that TAF7 phosphorylation further reduces the interaction with TAF1 cannot be excluded.

Although recombinant TAF1 undergoes autophosphorylation and transphosphorylates TAF7, the phosphorylation states of TAF1 and TAF7 in the context of TFIID has not been examined previously. Therefore, we determined whether (and when) TAF1 and/or TAF7 are phosphorylated in TFIID by sequential addition of purified GTFs in the presence of [γ-³²P]ATP to assemble PICs on the class I promoter fragment. Phosphorylation of TAF1 and TAF7 were monitored at each step in the assembly (Fig. 7 Left). Neither TAF1 nor TAF7 were phosphorylated in the presence of only TFIID and the GTFs. Phosphorylation of both TAF1 and TAF7 depended on the addition of Pol II to the PIC assembly reaction. Thus, TAF1 and TAF7 in TFIID are phosphorylated, but only after the entry of Pol II to the assembled PIC. It should be noted that these experiments do not distinguish TAF1-dependent phosphorylation from phosphorylation by another kinase. Further, and importantly, TAF7 was only released from the complex upon addition of Pol II (Fig. 7 Right). From these experiments, we conclude that completion of PIC assembly signals TAF1 and TAF7 phosphorylation and release of TAF7 allowing transcription to initiate.

Fig. 7.

TAF1 phosphorylation in TFIID occurs upon entry of Pol II into the PIC and is coincident with TAF7 release. (A) Transcription complexes were assembled from various combinations of purified transcription factors on the immobilized MHC class I promoter in presence of [γ-³²P]ATP, as indicated above each lane. Phosphorylated proteins were analyzed by gel electrophoresis and autoradiography. The position of RAP74, which is known to be phosphorylated by TAF1, is shown as a control and indicated based on mobility. The positions of TAF1 and of TAF7 are indicated based on both mobility and Western blot analysis of the same gel; it is the lower band of the doublet in lanes 3, 6, and 7. The purified transcription factors were TFIID, D; TFIIB, B; TFIIE, E; TFIIF, F; TFIIH, H; and RNA polymerase II. (B) Purified transcription complexes were assembled as in A. Phosphorylation of proteins that either remained associated with the PIC (B) or were released into the supernatant (S) was identified by separation of the assembled PIC from the supernatant and analysis on PAGE, followed by autoradiography. The position of TAF7 is indicated based on both mobility and Western blot analysis of the same gel.

Discussion

Transcription depends on the temporal and spatial integration of preinitiation complex assembly, initiation, elongation, and termination. Proper regulation of the transition from PIC assembly to initiation is necessary to prevent aberrant or abortive transcription. Here, we have demonstrated that TAF7, which inhibits the AT activity of TAF1 required for TFIID-dependent transcription, is associated with TFIID during PIC formation. Upon addition of Pol II to the assembling PIC, TAF1 and TAF7 undergo phosphorylation, releasing TAF7 from the PIC. We propose that the association of TAF7 with TFIID during PIC assembly prevents transcription initiation before the successful completion of the recruitment of the other GTFs and Pol II. The release of TAF7 from TFIID relieves the inhibition of TAF1 AT activity and allows transcription to initiate. Thus, we propose that TAF7 serves as a checkpoint regulator in the transition from PIC assembly to initiation.

The release of TAF7 from TFIID in an assembled PIC is coincident with TAF1 phosphorylation. In vitro, autophosphorylation of recombinant TAF1 reduces its binding to TAF7. This finding, which reports a role for TAF1 autophosphorylation, suggests that autophosphorylation may regulate the release of TAF7 from TFIID during the transition from PIC assembly to initiation. What regulates TAF1 phosphorylation in the assembled PIC and whether it is by the TAF1 kinase or another kinase remain to be determined. The mechanism by which phosphorylation of TAF1 alters the avidity of TAF7 binding also remains to be determined. It is possible that a conformational change in autophosphorylated TAF1 obscures or destroys the TAF7-binding site. Although phosphorylation-mediated changes in molecular interactions are well known, in most cases, phosphorylation results in intermolecular associations, not dissociations. Even less common are phosphorylation-mediated dissociations that lead to activation. Perhaps the best characterized example is that of the ATM molecule involved in repair of double-strand breaks in DNA. In the absence of DNA damage, ATM exists as an inactive dimer; phosphorylation of the ATM dimer triggers its dissociation and activation (40). The interaction between TAF7 and TAF1 provides another example of phosphorylation-mediated dissociation, in this case resulting in activation of TAF1 AT activity (36).

The present findings also raise the question of the role of TAF1 in PIC assembly and transcription initiation. As the largest component of the TFIID complex, TAF1 was originally thought to function primarily as a scaffold for the other TFIID constituents. It is now clear that TAF1 plays an active role in transcription. The AT activity of TAF1 is necessary for transcription, although its exact role is still not known (35). The TAF1 AT activity is inhibited by the binding of TAF7 (36). The present studies suggest that the release of TAF7 after completion of PIC assembly relieves the inhibition of TAF1 AT activity, allowing transcription to proceed. Indeed, excess TAF7, both in vitro and in vivo, blocks initiation, presumably by maintaining the association with TAF1.

The extent of TAF7 release from the PIC parallels the extent of productive in vitro transcription, suggesting that the two processes are closely linked. Transcription from an immobilized MHC class I promoter DNA fragment resulted in the release of 20% of the TAF7 initially associated with the PIC. This amount corresponds approximately to the fraction of hyperphosphorylated Pol II that mediated productive transcription. Thus, productive transcription is accompanied by the release of TAF7 from TFIID and the hyperphosphorylation of Pol II. In similar experiments in yeast, using an immobilized HIS4 promoter and yeast extracts to reconstitute the PIC, Yudkovsky et al. found that TFIID (and particularly TBP, TAF5, TAF7) and TFIIA remained associated with the promoter (41) during transcription elongation. They did not observe the release of TAF7. One possible explanation for the discrepancy between their results and ours may relate to differences in the TFIID dependency of the promoters examined. The release of TAF7 from TFIID may only be observed on promoters such as the MHC class I promoter whose transcription requires TAF1 and its AT activity (35). The TAF1-dependence of the HIS4 promoter was not examined.

In conclusion, we propose that TAF7 acts as a checkpoint regulator in the transition from PIC assembly to transcription initiation. We propose that, during PIC assembly, TAF7 remains bound to TFIID, thereby inhibiting the AT activity of TAF1 and preventing premature transcription initiation. Once PIC assembly is completed, TAF1 undergoes phosphorylation, resulting in the release of TAF7 and activation of its AT activity, which allows transcription to initiate. Further studies are needed to test this model.

Materials and Methods

For further details, see Supporting Text, which is published as supporting information on the PNAS web site.

Assembly and Purification of PICs. Biotinylated PCR product (600 ng), dIdC (600 ng), and HeLa nuclear extract (150 μg, Promega) were incubated for 40 min at 23°C in in vitro transcription buffer (20 mM Hepes, pH 7.9/50 mM KCl/6.24 mM MgCl₂/0.5 mM EDTA/2mMDTT/10 μM ZnSO₄/10 mM creatine phosphate/100 μg/ml creatine kinase/8.5% glycerol). The assembled PIC was captured on streptavidin-coated magnetic beads that had been preequilibrated in binding buffer (20 mM Hepes, pH 7.9/80 mM KCl/10 mM MgCl₂/2mMDTT/10 μM ZnSO₄/100 μg of BSA/0.5% Nonidet P-40, and 10% glycerol) (BB) by incubation for an additional 30 min. The captured DNA/PIC complex was washed five times in BB.

In Vitro Transcription (IVT) with Purified PIC. IVT reactions were performed by resuspending purified PICs in 25 μl of IVT buffer (8 mM Hepes, pH 7.9/40 mM KCl/6 mM MgCl₂/0.2 mM DTT/80 μM EDTA/8% glycerol) in presence of 1.6 mM of each rNTP and incubating for 30 min at 20°C. Transcripts were revealed by primer extension (2). Proteins either released into the supernatant or remaining on the beads after IVT were separated by SDS/PAGE gel and analyzed by Western blotting. When added, 2 μg of α-amanitin was presented during PIC assembly and IVT.

TAF1 Kinase Assays. Flag-tagged TAF7 (250 ng) was incubated for 30 min at 30°C with increasing amounts of Flag-tagged TAF1 (50, 100, 250 ng) or control extract immobilized on anti-Flag M2 agarose in 20 μl of kinase buffer (25 mM Hepes, pH 7.9/100 mM KCl/12.5 mM MgCl₂/0.1 mM EDTA/0.1% Nonidet P-40) in the presence of 10 μCi of [γ-³²P]ATP (6,000 Ci/mM); the proteins were resolved by SDS/PAGE gel, and labeling was quantitated by PhosphorImager. All kinase assays of TAF7 or different TAF7 mutants were performed with the same molar amount of proteins.

TAF1 Phosphorylation and Binding to TAF7. Flag-TAF1 (250 ng) and GST-TAF7 (200 ng) were mixed together in presence of increasing amounts of [γ-³²P]ATP in the same conditions as described above. One-fifth of the reaction was loaded on an SDS/PAGE to test the efficiency of the kinase reaction. The remainder was subject to anti-Flag M2 agarose in 1× kinase buffer supplemented with 17% glycerol and a mixture of phosphatase inhibitors. Agarose-bound complexes were washed twice, and proteins were separated on SDS/PAGE and revealed by Western blotting using either anti-Flag M2 antibody to detect TAF1 or an GST antibody to detect TAF7. The amount of TAF7 bound to TAF1 was normalized according to the amount of TAF1 on the agarose beads.

TAF7 Binding to Autophosphorylated TAF1. Flag-TAF1 (250 ng) was incubated in presence of increasing amount of [γ-³²P]ATP as described above. One fifth of the reaction was loaded on an SDS/PAGE to test the efficiency of the kinase reaction. The remainder was captured on anti-Flag M2 agarose in kinase buffer supplemented with 17% glycerol and a mixture of phosphatase inhibitors. Agarose-bound [³²P]TAF1 was washed four times and then mixed with 200 ng of GST-TAF7 the same buffer. Agarose-bound complexes were separated on SDS/PAGE. TAF1 and TAF7 were revealed by Western blotting using anti-Flag M2 or GST antibodies, respectively. The amount of TAF7 bound to TAF1 was normalized according to the amount of TAF1 on the agarose beads.

PIC Assembly and in Vitro Kinase Assay. Biotinylated PCR product (30 ng) was immobilized on streptavidin magnetic beads and then incubated in presence of 10 μCi of [γ-³²P]ATP (6,000 Ci/mM) with combinations 100 ng of TFIID, 20 ng of TFIIB, 10 ng of TFIIE, 10 ng of TFIIF, 50 ng of TFIIH, and 50 ng of Pol II as indicated. Purified GTFs were obtained from Protein One, Bethesda, MD. The reactions were incubated for 1 h in 20 μl of kinase buffer (50 mM Tris, pH 7.9/5mMDTT/10 μM ZnSO₄/5 mM MnCl/4 mM MgCl₂/1 mM ATP). The proteins were resolved by SDS/PAGE gel, and labeling was revealed by PhosphorImager.