|Home | About | Journals | Submit | Contact Us | Français|
The transcription factors TFIIB, Brf1, and Brf2 share related N-terminal zinc ribbon and core domains. TFIIB bridges RNA polymerase II (Pol II) with the promoter-bound preinitiation complex, whereas Brf1 and Brf2 are involved, as part of activities also containing TBP and Bdp1 and referred to here as Brf1-TFIIIB and Brf2-TFIIIB, in the recruitment of Pol III. Brf1-TFIIIB recruits Pol III to type 1 and 2 promoters and Brf2-TFIIIB to type 3 promoters such as the human U6 promoter. Brf1 and Brf2 both have a C-terminal extension absent in TFIIB, but their C-terminal extensions are unrelated. In yeast Brf1, the C-terminal extension interacts with the TBP/TATA box complex and contributes to the recruitment of Bdp1. Here we have tested truncated Brf2, as well as Brf2/TFIIB chimeric proteins for U6 transcription and for assembly of U6 preinitiation complexes. Our results characterize functions of various human Brf2 domains and reveal that the C-terminal domain is required for efficient association of the protein with U6 promoter-bound TBP and SNAPc, a type 3 promoter-specific transcription factor, and for efficient recruitment of Bdp1. This in turn suggests that the C-terminal extensions in Brf1 and Brf2 are crucial to specific recruitment of Pol III over Pol II.
In eukaryotes, the job of transcribing nuclear genes is divided among three different RNA polymerases: RNA polymerase I (Pol I), RNA Pol II, and RNA Pol III. A common step in transcription initiation by all three eukaryotic RNA polymerases is the recruitment of the polymerase to the proper promoters by a specific factor or complex. This transcription factor serves as a bridge between promoter-bound factors that nucleate preinitiation complex assembly and the RNA polymerase enzyme itself. For Pol II and Pol III transcription, the factors that accomplish this task are TFIIB and TFIIIB, respectively.
Although TFIIB is a single polypeptide, the TFIIIB activity is minimally composed of three polypeptides (see references 11 and 33 for reviews). In human cells, two forms of the TFIIIB complex have been characterized thus far: Brf1-TFIIIB, which functions on type 1 and 2 Pol III promoters, and Brf2-TFIIIB, which functions on type 3 Pol III promoters such as the human U6 small nuclear RNA (snRNA) promoter. Brf1-TFIIIB and Brf2-TFIIIB both contain the TATA-box binding protein TBP and the SANT domain protein Bdp1. They differ by the presence of different members of a family of TFIIB-related factors: Brf1 in Brf1-TFIIIB and Brf2 in Brf2-TFIIIB.
TFIIB, Brf1, and Brf2 each contain a Zn ribbon domain at their N terminus, followed by a core domain consisting of two imperfect repeats. In this region, human Brf2 is 20% identical with human TFIIB and 18% identical with human Brf1. In addition, Brf1 and Brf2 contain a C-terminal extension. In Brf1, this C-terminal segment contains two regions called homology domains II and III, which are conserved among yeast and human proteins. Moreover, a third region called homology domain I is conserved among several different yeast species (25, 28). Such domains have thus far not been characterized in Brf2.
Although all TFIIB family members share the function of bridging promoter-bound factors and the Pol, TFIIB and yeast Brf1 accomplish this function differently, even on TATA-box-containing Pol II and Pol III promoters. Indeed, each TFIIB family member can associate with a TBP/TATA box complex, but TFIIB and Brf1 do so primarily through different regions. TFIIB assembles with the TBP/TATA box complex through its core domain, which interacts directly with a region within the C-terminal stirrup of the TBP core domain (3, 12, 17, 30, 31, 42). The core domain of yeast Brf1 also interacts with the TBP/TATA box complex, but the highest-affinity anchorage point is between homology domain II within the C terminus of Brf1 and surfaces on the N-terminal lobe of the TBP core domain, on the face of the TBP/TATA box complex opposite the TFIIB binding site (9, 19, 20, 22, 36).
TFIIB and Brf1 also recruit the RNA Pol differently. With TFIIB, the Zn ribbon domain is crucial for the recruitment of RNA Pol II: it interacts with the dock domain of Pol II and allows the positioning of the TFIIB core domain on the polymerase (3, 5, 7, 8, 12, 17, 31, 42; see reference 13 for a review). In contrast, with Brf1 the zinc ribbon is not necessary for PolIII recruitment but instead plays an essential role at a later stage in transcription initiation, that of promoter opening (14, 20, 23). It is likely that interactions between the C34 subunit of Pol III and the Brf1 core and homology regions I and II (1, 25, 40, 41), as well as between the C17 subunit of Pol III and the Brf1 core (10), are strong enough for recruitment of the polymerase even in the absence of any protein-protein contact contributed by the zinc ribbon.
Like Brf1, Brf2 is a Pol III transcription factor, but it is specifically required for transcription from type 3 promoters such as the human U6 promoter. These promoters are located upstream of the transcribed region and their core region consists of a proximal sequence element (PSE) and a TATA box, which are recognized by a multisubunit complex we call SNAPc and by TBP, respectively. We have now examined the functions of various regions of Brf2 for U6 transcription and for assembly of a preinitiation complex containing SNAPc, TBP, Brf2, and Bdp1, and we have tested the ability of Brf2/TFIIB chimeric proteins to perform these functions. The results suggest that a common feature among TFIIB family members involved in Pol III transcription is the use of their C-terminal extension to associate efficiently with a TBP/TATA box complex and to recruit Bdp1.
Full-length Brf2 and TFIIB, as well as the Brf2 truncations and the four TFIIB/Brf2 chimeras, were all expressed in Escherichia coli with the T7 system from (37). The proteins contained an N-terminal FLAG tag and a C-terminal His tag, except for full-length Brf2, which had only the C-terminal His tag. The proteins were purified via their His tags by Ni-nitrilotriacetic acid (NTA) agarose (QIAGEN) chromatography and analyzed by sodium dodecyl sulfate (SDS)-12.5% polyacrylamide gel electrophoresis. The gels were stained with the GelCode Blue Stain reagent (Pierce) for 1 h according to the manufacturer's instructions and were scanned with an Odyssey Infrared Imager (LI-COR) to measure the intensities of the appropriate bands. The proteins were also analyzed by immunoblotting with the Odyssey Infrared Imaging System (LI-COR).
TBP was expressed in E. coli as a glutathione S-transferase (GST)-fusion protein and bound to glutathione agarose beads. The beads were washed, and TBP was released from the beads by thrombin cleavage at a site located just after the GST tag. Bdp1 was expressed in E. coli and purified first through a C-terminal His tag by Ni-NTA agarose chromatography and then through an N-terminal FLAG tag by anti-FLAG Antibody M2 agarose (Sigma) chromatography. Both TBP and Bdp1 were dialyzed against buffer D100 (50 mM HEPES [pH 7.9], 0.2 mM EDTA, 20% glycerol, 0.1% Tween 20, 100 mM KCl, 3 mM dithiothreitol [DTT], 0.5 mM phenylmethylsulfonyl fluoride).
SNAPc was expressed in Sf9 insect cells as described previously (15) and purified via a His tag fused to the C terminus of SNAP190 by Ni-NTA agarose (QIAGEN) chromatography. The RNA Pol III complex used in the reconstituted U6 transcription system was purified from a HeLa cell line expressing a FLAG- and His-tagged RPC4 (RPC53) subunit of Pol III, as described previously (18).
For the GST pull-down assay, GST, GST-tagged PCMT, and GST-tagged TBP were all expressed in E. coli with the T7 system from (37) and bound to glutathione-agarose beads. The beads were then washed four times with HEMGN150 buffer (25 mM HEPES [pH 7.9], 150 mM KCl, 12.5 mM MgCl2, 0.1 mM EDTA, 10% glycerol, 0.1% NP-40, 1 mM phenylmethylsulfonyl fluoride, 2 mM DTT, and protease inhibitors), resuspended in the same buffer, and stored at 4°C.
The reconstituted U6 transcription assay and RNA analysis were performed as described by (18).
The binding reactions contained the proteins indicated in the figure legends and 9 μg of fetal bovine serum, 25 ng of poly(dG-dC)-poly(dG-dC), and 75 ng of pUC118 in a buffer containing 20 mM HEPES (pH 7.9), 100 mM KCl, 10 mM MgCl2, 0.2 mM EDTA, 10% glycerol, 1 mM DTT, 0.07% Tween 20, and 0.15 mM ATP. The total reaction volume was 20 μl. The mixtures were incubated at 4°C for 20 min before addition of the radiolabeled probe and further incubation at 30°C for 30 min. The samples were then fractionated on a 4% polyacrylamide gel (39:1 acrylamide-bisacrylamide) in TGE buffer (50 mM Tris base, 380 mM glycine, 2 mM EDTA).
Either full-length Brf2, full-length human TFIIB, the B/2/− chimera, or the B/2/2 chimera were mixed with 10 μl of glutathione-agarose beads bound by either GST, GST-PCMT, or GST-TBP in HEMGN150 buffer containing 400 μg of ethidium bromide/ml in a total reaction volume of 60 μl. The reactions were incubated for 2 h at 4°C, rotating end over end. The beads were then washed two times with HEMGN150, two times with HEMGN300 (containing 300 mM KCl instead of 150 mM KCl), and two times once again with HEMGN150, always in the presence of 400 μg of ethidium bromide/ml. The beads were then resuspended in 5× Laemmli buffer (325 mM Tris-HCl [pH 8], 10% SDS, 0.025% bromophenol blue, 50% glycerol, 500 mM β-mercaptoethanol) and boiled. The proteins eluted from the beads were then fractionated on a 10% SDS-polyacrylamide gel, transferred to a nitrocellulose membrane, and visualized by immunoblot with an anti-His tag monoclonal antibody (QIAGEN) at a 1:3,000 dilution. The immunoblots were developed with the Odyssey Infrared Imaging System (LI-COR).
The top of Fig. Fig.1A1A shows a comparison of the structures of human TFIIB, Brf1, and Brf2. These three proteins each contain a zinc-binding domain (shown in blue) and a core domain (shown in purple). Brf1 and Brf2 additionally contain unique C-terminal domains. To determine the role of different parts of Brf2, we synthesized the truncated Brf2 proteins illustrated in the lower part of Fig. Fig.1A.1A. Brf2(66-419) lacks the region N-terminal to the core domain, which includes the Zn ribbon motif. The second truncation, Brf2(2-289), lacks the C-terminal domain of the protein. The full-length and truncated proteins all contain a histidine tag at their C terminus (indicated in red), which was used for purification after expression in E. coli. The resulting proteins are shown in Fig. Fig.1B1B and, as revealed by immunoblot with an anti-Brf2 antibody after normalization of the amounts, in Fig.Fig.1C.1C. These proteins were then tested for U6 transcription in a minimal in vitro transcription system, which consists of recombinant TBP, Bdp1, Brf2, and SNAPc combined with highly purified Pol III from a HeLa cell line expressing tagged RPC4 (RPC53) (18).
As seen in Fig. Fig.1D,1D, TBP, Bdp1, and SNAPc combined with purified Pol III in the absence of Brf2 did not give rise to any detectable U6 transcript in the reconstituted in vitro transcription assay (lane 1). As expected, addition of recombinant full-length Brf2 resulted in efficient U6 transcription (lanes 2 to 4). However, when either Brf2(66-419) or Brf2(2-289) were used in place of full-length Brf2, U6 transcripts were not detectable (lanes 5 to 7 and lanes 8 to 10, respectively), suggesting that both of the regions N and C terminal to the Brf2 core domain are important for transcriptional activity.
To enable us to study the functions of the N- and C-terminal regions of Brf2 in the formation of the transcription initiation complex, we set out to establish an electrophoretic mobility shift assay (EMSA) in which we could visualize the assembly of the various U6 transcription factors on the U6 promoter. We first focused on the Brf2-TFIIIB components. As shown in Fig. Fig.2A,2A, unlike Brf2 or Bdp1, TBP on its own was able to form a protein-DNA complex on a probe containing the wild-type U6 PSE and TATA box (lanes 2 to 4). As observed previously, the addition of both TBP and Brf2 resulted in the formation of a prominent complex (henceforth referred to as the “two-factor complex,” labeled 2 in lane5; see also Fig. Fig.2E,2E, which summarizes the nomenclature used for some of the complexes), reflecting the cooperative binding of these two factors (6, 27, 43). With TBP and Bdp1, a weak complex migrating much more slowly than the TBP complex was obtained (lane 6, complex labeled TBP/Bdp1). Since Bdp1 on its own does not bind detectably to the probe (lane 4), this suggests protein-protein interactions between Bdp1 and TBP on the DNA. With TBP, Brf2, and Bdp1, a robust complex migrating even more slowly than the TBP/Bdp1 complex was obtained (henceforth referred to as the “three-factor complex,” labeled 3 in lane 7; see Fig. Fig.2E).2E). This complex was supershifted by anti-Bdp1 and anti-TBP antibodies (which as expected did not bind to the probe on their own, see lanes 12 and 13), confirming that it contains Bdp1 and TBP (compare lanes 8 to 11 to lane 7). On the other hand, we did not observe a supershift with any of our anti-Brf2 antibodies (data not shown). Nevertheless, since the three-factor complex was not observed in the absence of Brf2 and migrated more slowly than the TBP/Bdp1 complex (compare lanes 6 and 7), it is highly likely that it indeed contains Brf2 but that the Brf2 epitopes recognized by the antibodies we tested are masked. None of the complexes were observed with a probe containing a mutated TATA box (lanes 15 to 17), indicating that they all depend on TBP-TATA box interactions.
We next combined the Brf2-TFIIIB components with SNAPc. As shown in Fig. Fig.2B,2B, as we titrated increasing amounts of SNAPc into a binding reaction containing the three Brf2-TFIIIB components, we observed a gradual increase in the formation of a large complex (henceforth referred to as the “four-factor complex,” labeled 4 in lane 4; see Fig. Fig.2E)2E) migrating more slowly than the three-factor complex containing TBP, Brf2, and Bdp1 (compare complex 4 in lanes 4 to 6 with complex 3 in lane 1). This complex was not observed in the absence of the Brf2-TFIIIB components (lanes 7 to 11) and could be supershifted with anti-TBP, anti-Bdp1, and anti-SNAP190 antibodies (Fig. (Fig.2C,2C, compare lanes 3 to 8 with lane 2), confirming that it contains TBP, Bdp1, and SNAPc. Like complex 3, complex 4 was not supershifted by the anti-Brf2 antibodies we tested (data not shown), most likely because within the complex, the relevant Brf2 epitopes are not accessible to the antibodies.
In addition to complex 4, binding reactions containing SNAPc and Brf2-TFIIIB gave rise to four prominent smaller complexes (see, for example, Fig. Fig.2B,2B, lane 5). As shown in Fig.Fig.2D,2D, comparison with binding reactions containing various subsets of the four components on either a wild-type probe or a probe with a mutated TATA box indicated that, of these four smaller complexes, the one migrating the fastest is the two-factor complex containing TBP and Brf2 (compare lanes 3 and 6, complex 2, see also lanes 8 to 10), whereas the one above it corresponds to just SNAPc bound to the DNA (compare lanes 1 and 6; see also lanes 4, 7, 9, and 10). The third complex from the top in lane 6 comigrates with one formed by TBP, Brf2, and SNAPc (compare lanes 4 and 6, complex labeled with a white circle; see also lanes 9 and 10), whereas the complex just below the four-factor complex in lane 6 corresponds to the three-factor complex (compare lanes 5 and 6, see also lanes 8 to 10). Thus, by EMSA we can visualize a complex containing the basal transcription factors SNAPc and Brf2-TFIIIB, which are known to be sufficient to reconstitute U6 transcription when combined with a highly purified Pol III complex (18), as well as visualize various subcomplexes.
We used the EMSA system described above to determine whether the N- and C-terminal regions of Brf2 might be involved in initiation complex assembly. In the interest of space, all EMSA experiments presented hereafter show only the region corresponding to the upper half of the gel containing the various complexes. Lanes 1, 2, and 3 in Fig. Fig.3A3A show the two-, three-, and four-factor complexes (containing, respectively, TBP and Brf2; TBP, Brf2 and Bdp1; and TBP, Brf2, Bdp1 and SNAPc) obtained with full-length Brf2 (complexes labeled 2, 3, and 4). With the Brf2(66-419) truncated protein, the two-, three-, and four-factor complexes were all observed (lanes 10 to 12), indicating that the region N-terminal to the core domain is not required for binding to the TBP-TATA box complex (lane 10), recruitment of Bdp1 (lane 11), or assembly of the full complex (lane 12). In contrast, with a Brf2(66-289) truncated protein containing just the Brf2 core domain and with the Brf2(2-289) truncated protein, none of these complexes were obtained (lanes 4 to 9). In particular, we did not observe a two-component complex containing TBP and a truncated Brf2 (lanes 4 and 7). The only complexes observed were those containing either just SNAPc (compare lanes 6 and 9 with lane 16) or SNAPc and TBP (compare lanes 6 and 9 with lane 13), a finding consistent with previous observations that SNAPc and TBP bind cooperatively to the U6 promoter (27, 29). Thus, the Brf2 region C-terminal of the core domain is required for detectable binding of Brf2 together with TBP to the U6 TATA box in the EMSA. Moreover, it is required for the formation of all subsequent complexes containing both Brf2 and TBP.
Brf2, which does not bind to DNA on its own, can be recruited to the U6 promoter not only through protein-protein interactions with TBP binding to the TATA box but also through protein-protein interactions with SNAPc binding to the PSE (16). However, with the truncated Brf2 proteins missing the C-terminal region, we did not observe novel complexes that might contain Brf2 but lack TBP. We wondered, therefore, whether the C-terminal region of Brf2 might be required for cooperative binding with SNAPc. As shown in Fig. Fig.3B,3B, when only SNAPc and Brf2 were added to the U6 promoter probe, we observed a weak complex migrating more slowly than the SNAPc-only complex and slightly faster than the TBP/SNAPc complex (lane 4, complex labeled with a white arrowhead, compare with lanes 1 and 3). This complex most likely contains SNAPc and Brf2 and presumably forms due to interactions between Brf2 and the PSE-bound SNAPc. Notably, a similar, even darker complex was observed when full-length Brf2 was replaced by Brf2(66-419) but not when it was replaced by Brf2(2-289) (compare lane 4 with lanes 5 and 6). Thus, the C-terminal region of Brf2 is not only required for cooperative binding with TBP bound to the TATA box but also for cooperative binding with SNAPc bound to the PSE.
These results indicate that the C-terminal domain of Brf2 is necessary for the protein to form a complex with TBP, as well as with SNAPc, on a U6 promoter. Because of the failure to form these initial complexes, higher-order complexes in turn cannot form. Thus, the inability of Brf2(2-289) to direct U6 transcription (Fig. (Fig.1D1D above) can be ascribed to a failure of transcription initiation complex assembly. On the other hand, the inability of Brf2(66-419) to direct U6 transcription probably results from a defect at a step after transcription factor assembly, such as Pol III recruitment or promoter opening.
As discussed above, Brf2 and TFIIB have several features in common but they are used by different RNA polymerases—TFIIB by Pol II and Brf2 by Pol III. To learn more about the functions of the various Brf2 regions and the determinants of RNA polymerase specificity, we constructed a set of TFIIB/Brf2 chimeric proteins, which are shown in Fig. Fig.4A.4A. We divided Brf2 into core region and regions N and C terminal to the core and TFIIB into core region and region N terminal to the core and constructed four chimeras by swapping these regions between the two proteins. The chimeras were named based on the origin of the regions from which they were comprised. Thus, for example, 2/B/− consists of the N-terminal region of Brf2, the core domain of TFIIB, and no C-terminal domain, whereas B/2/2 consists of the N-terminal region of TFIIB and the core and C-terminal domains of Brf2. The four chimeras, as well as full-length Brf2 and TFIIB, were expressed in E. coli and partially purified through a C-terminal His tag. The resulting proteins are shown in Fig. Fig.4B4B as visualized by staining with the GelCode Blue Stain reagent (Pierce) and in Fig. Fig.4C4C as visualized by immunoblot with an anti-His tag antibody after normalization of the amounts.
We then tested the four chimeras for U6 transcription in the minimal in vitro transcription system, and the results are shown in Fig. Fig.4D.4D. Combining full-length Brf2, but not full-length TFIIB, with TBP, Bdp1, SNAPc, and purified Pol III reconstituted U6 transcription (compare lanes 2 to 4 to lanes 5 to 7 and to lane 1), consistent with previous observations showing that TFIIB is not active for Pol III-directed U6 transcription (26). Both the 2/B/− and B/2/− chimeras were inactive in this assay (data not shown) as was the B/2/2 chimera (lanes 8 to 10). The 2/B/2 chimera showed barely detectable activity (lanes 11 to 13). This prompted us to widen our titration of the 2/B/2 chimera, but in all cases the U6 signal was extremely weak (lanes 15 to 19), although clearly detectable in the original film. Thus, the chimeras are generally inactive for U6 transcription, except for the 2/B/2 protein, which displays extremely weak activity.
To understand the defects of TFIIB in terms of U6 transcription activity, as well as those of the chimeric proteins, we tested their capacity to assemble a U6 preinitiation complex by using EMSA. Figure Figure5A,5A, lanes 1 to 6, show the locations of the SNAPc complex alone (lane 1), the two-, three-, and four-factor complexes (lanes 3, 5, and 6) and the complexes containing TBP and SNAPc (lane 4, complex labeled TBP/SNAPc) and TBP, Brf2, and SNAPc (lane 4, complex labeled with a white dot on the left). We first focused on TFIIB. As expected, TFIIB alone was unable to bind to the probe but bound cooperatively with TBP to the U6 TATA box (lanes 7 and 8). However, upon addition of Bdp1, no additional complex was observed (lane 9), indicating that the TBP/TFIIB complex cannot recruit Bdp1. Interestingly, when TBP, TFIIB, Bdp1, and SNAPc were present in the binding reaction, we obtained a complex containing TBP, TFIIB, and SNAPc (lane 10, complex labeled with a white dot on the left). Thus, on the Pol III U6 promoter, TFIIB can assemble efficiently with TBP and SNAPc but is unable to recruit Bdp1.
We next examined the B/2/− and 2/B/− chimeras. In Fig. Fig.5B,5B, the locations of the two-, three- and four-factor complexes in the EMSA are shown (lanes 1 to 3), as are those of the complexes containing just SNAPc (lane 13); TBP and SNAPc (lane 10, complex labeled TBP/SNAPc on the left of the panel); and TBP, Brf2, and SNAPc (lane 11, complex labeled with a white dot on the left). The B/2/− chimeric protein contains the core of Brf2 but lacks its C-terminal region. Consistent with the deletion results obtained above, which indicate that the Brf2 C-terminal region is necessary for efficient cooperative binding with TBP, as well as with SNAPc (Fig. (Fig.3),3), the B/2/− chimera was unable to form any of the Brf2-dependent complexes (lanes 4 to 6). The only complexes visible were those containing either just SNAPc or both SNAPc and TBP (see lane 6). With the 2/B/− chimeric protein, which contains the core domain of TFIIB, we observed a two-factor complex containing TBP and 2/B/− (lane 7), a finding consistent with the known ability of the TFIIB core to bind to a TBP-TATA box complex (3, 12, 17, 42). This complex was, however, unable to recruit Bdp1 (lane 8), although it could assemble with SNAPc to form a complex containing TBP, 2/B/−, and SNAPc (lane 9, complex labeled with a white dot). This is consistent with the inability of full-length TFIIB to recruit Bdp1 (Fig. (Fig.5A).5A). Moreover, the results obtained above with Brf2(66-419) (Fig. (Fig.3A)3A) indicate that the N-terminal region of Brf2 is not required for Bdp1 recruitment but do not address the possibility that this region is capable of recruiting Bdp1 in a manner redundant with the core or C-terminal regions of Brf2. The results with the 2/B/− chimera show that the N terminus of Brf2 cannot recruit Bdp1 when targeted to a TBP-TATA box complex.
The activities of the B/2/− and 2/B/− chimeras in transcription complex assembly indicated an essential role for the Brf2 C-terminal domain in binding to the TBP-TATA box complex and perhaps in recruiting Bdp1. We therefore tested in the EMSA the B/2/2 and 2/B/2 chimeras, which differ from B/2/− and 2/B/− only by addition of the Brf2 C-terminal domain. As shown in Fig. Fig.5A,5A, both proteins were capable of forming a complex with TBP on the U6 TATA box (lanes 12 and 16, complex 2), although the complex formed with B/2/2 was weaker than that obtained with either Brf2 (lane 3) or 2/B/2. This is surprising since Brf2(66-419), which lacks the N-terminal region of Brf2 altogether, binds as efficiently as full-length Brf2 to a TBP/TATA box complex (Fig. (Fig.3).3). Moreover, as shown below, B/2/2 binds as efficiently as Brf2 to TBP in a GST pull-down assay. These differences may indicate that the presence of the N-terminal region of TFIIB in place of that of Brf2 somehow reduces the B/2/2 chimera's ability to assemble with TBP on the U6 TATA box but not off the DNA. We have not, however, pursued this observation further. For the 2/B/2 chimera, its binding to the TBP-TATA box complex may be due to the TFIIB core region, the Brf2 C-terminal region, or both since these two regions both contain TBP-binding activity. However, as with B/2/2, the two-factor complex obtained with 2/B/2 migrated as a tight band close to the position observed with full-length Brf2 (compare lanes 12 and 16 with lane 3) and very different from the diffuse and slower-migrating complex observed with TFIIB and TBP (lane 8). Although we do not know what causes the TFIIB-TBP-TATA box complex to be diffuse, this suggests that 2/B/2 binds to the TBP-TATA box complex in a manner more similar to Brf2 than TFIIB.
The promoter-bound B/2/2 and 2/B/2 proteins could recruit Bdp1 (Fig. (Fig.5A,5A, lanes 13 and 17, complex 3) and SNAPc (lanes 14 and 18, complex 4), although the recruitment of SNAPc by 2/B/2 is not efficient (lane 18). The B/2/2 protein's ability to promote assembly of the four-factor complex is consistent with the results above (Fig. (Fig.3)3) showing that the core and C-terminal regions of Brf2 together are sufficient for interaction with TBP, Bdp1, and SNAPc on the DNA. Its inactivity in U6 transcription indicates that the essential function(s) of the N-terminal region of Brf2 cannot be efficiently performed by the N-terminal region of TFIIB. In the case of the 2/B/2 protein, its ability to recruit Bdp1 contrasts with the inability of 2/B/− to perform this same function and indicates that the Brf2 C-terminal region is required for Bdp1 binding. Moreover, the formation of a four-factor complex with 2/B/2 indicates that when the C-terminal region of Brf2 is present the core region of Brf2 can be replaced with that of TFIIB. Indeed, the 2/B/2 chimera is capable of directing a very low level of U6 transcription (Fig.(Fig.4D).4D). However, the efficiency of both four-factor complex assembly and transcription is very low, indicating that either the TFIIB core region is not placed optimally for full function in the chimeric protein or that there is some function of the Brf2 core region that cannot be fully replaced by the TFIIB core region. One possibility is that the interactions between the 2/B/2 chimera and both Bdp1 and SNAPc together may not be stable enough to support efficient reconstituted U6 in vitro transcription. Another possibility is that the core domain of Brf2 is involved in Pol III recruitment, which was not assayed in our EMSA experiments.
To test which Brf2 region is required to bind to TBP off the promoter DNA, we performed the GST pull-down experiment shown in Fig. Fig.6.6. GST-TBP, as well as the negative controls GST and GST-protein carboxyl methyltransferase (GST-PCMT), were immobilized on glutathione-agarose beads, and the binding of various histidine-tagged proteins to these beads was visualized by immunoblotting with anti-HT antibodies. Lanes 1 to 4 show the proteins tested, namely, TFIIB, Brf2, B/2/−, and B/2/2. Both TFIIB and Brf2 bound specifically to GST-TBP in this assay, with Brf2 binding even more strongly than TFIIB (lanes 5 to 8). Strikingly, however, the B/2/− chimera, which contains the Brf2 core but not the Brf2 C-terminal region, bound very weakly to GST-TBP (lane 11), although the signal was clearly above the background seen with both GST alone and GST-PCMT (lanes 9 and 10). With the B/2/2 chimera, which contains the Brf2 C-terminal domain, very efficient and specific binding to GST-TBP was observed (lanes 12 to 14). These results provide further evidence for the crucial role of the C-terminal region of Brf2 in TBP binding and show that the region performs this function even off the promoter DNA.
TFIIB, Brf1, and Brf2 are all members of a family of transcription factors that share a number of common features, yet they are used by different classes of promoters: TFIIB by Pol II-recruiting promoters and Brf1 and Brf2 by Pol III-recruiting promoters. We have analyzed here the functions of various regions of Brf2 for U6 transcription complex assembly and in vitro transcription, and we have tested the ability of Brf2/TFIIB chimeric proteins to perform these functions. The results, which are summarized in Fig. Fig.7,7, reveal an essential role for the Brf2 C-terminal region and provide mechanistic insights on how RNA polymerase specificity is determined at the human snRNA promoters.
We find that the N-terminal region of Brf2 can be deleted or replaced by the TFIIB N-terminal region without major effects on the ability of Brf2 to bind to a TBP/TATA box complex, recruit Bdp1, or interact with SNAPc (Fig. (Fig.7).7). The N-terminal region is, however, required for U6 transcription, and for this function it cannot be replaced by the TFIIB N-terminal domain. These results are consistent with what is known about the roles of the N-terminal domains of human and yeast TFIIB and yeast Brf1. Indeed, in these proteins, the N-terminal domain is dispensable for association with a TBP-TATA box complex and, in the case of yeast Brf1, for recruitment of Bdp1, but plays essential roles at later stages (9, 20).
In TFIIB, the N-terminal domain associates tightly with PolII and is required for Pol II recruitment (3, 4, 12, 17, 31, 42). The crystal structure of a Pol II/TFIIB complex (5) and site-specific photo-cross-linking studies of both a Pol II/TFIIB complex and a complete preinitiation complex (7, 8; see reference 13 for a review) indicate that within the N-terminal region of TFIIB, the zinc ribbon is located within the center of the pocket formed by the clamp, wall, and dock domains of Pol II, most likely making contacts with one surface of the dock, whereas the B finger, which is located immediately downstream of the zinc ribbon on the linear TFIIB sequence, extends within the space from the RNA exit channel to the active site of Pol II. The zinc ribbon and the B finger appear to form the highest-affinity anchorage point between TFIIB and Pol II, which then allows, with the help of other components of the preinitiation complex, the proper positioning of the TFIIB core region on the polymerase. This in turn likely orients the promoter DNA over the central cleft of the polymerase (7, 8). Thus, in the Pol II preinitiation complex, the N-terminal region of TFIIB does not contribute essential contacts with TBP or the promoter DNA but is intimately associated with the catalytic core of the enzyme.
In yeast Brf1, the N-terminal domain is dispensable for recruitment of Pol III but is required at a later stage, during promoter opening (14, 20). In the case of human Brf2, the function performed by its N-terminal domain is not known, but it clearly cannot be performed by the TFIIB N-terminal region. It is possible that the N-terminal regions of both Brf1 and Brf2 contribute in fact to specific recruitment of Pol III over Pol II. The TFIIB, Brf1, and Brf2 zinc ribbon domains (but not the B finger regions) are highly conserved (14, 34, 38), and so are the Pol II and Pol III dock regions (see references 7), a finding consistent with the Brf1 and Brf2 zinc ribbons associating with Pol III in much the same way as the TFIIB zinc ribbon associates with Pol II. Nevertheless, the specific amino acids involved in the interactions may be different; indeed, it is striking that, for example, the yeast TFIIB N-terminal region fails to recruit human Pol II to the adenovirus 2 major late promoter, even though both the yeast and human TFIIB N-terminal regions can recruit yeast Pol II to this promoter (39). Similarly, subtle differences within the zinc ribbons of TFIIB, Brf1, and Brf2, as well as the lack of an obvious B finger in Brf1 and Brf2, may contribute to specific recruitment of the correct RNA polymerase. However, with Brf1, and perhaps Brf2, contacts between the core and C-terminal domain and Pol III (1, 10, 25, 41) are clearly strong enough to allow recruitment of the polymerase even in the absence of the N-terminal region.
The N-terminal region of Brf2 may also be required, like that of Brf1, for promoter opening, a function specific to PolIII transcription initiation complexes which, unlike Pol II transcription initiation complexes, do not require hydrolyzable ATP to form the transcription bubble (21, 23, 24).
In TFIIB, which does not contain a C-terminal extension, the core region is sufficient for efficient binding to the TBP/TATA box complex. This is different in yeast Brf1, where the association of a truncated protein missing the C-terminal domain is so weak that it is only detected after photochemical cross-linking (9, 20, 22, 35). Brf2 appears to be very similar to Brf1. Deletion of either the core (6) or the C-terminal domain (see Fig. Fig.33 and and7)7) prevents association with the TBP/TATA box complex as detected in an EMSA in the absence of cross-linking, indicating that the core is required but not sufficient for high-affinity association with the TBP/TATA box complex.
How does the Brf2 core associate with TBP? In the case of yeast Brf1, the weak association of a truncated protein corresponding to the N-terminal half of the protein (including the core domain) with TBP is prevented by radical mutations in the TBP surface that interacts with TFIIB, suggesting that the TFIIB and yeast Brf1 core domains contact human TBP in a very similar manner (35). However, such mutations affect the association of human full-length TFIIB family members differently. Radical mutations of TBP residues E284, E286, or L287 do not prevent association with full-length human Brf1 nor Brf1-dependent transcription, but they reduce association with TFIIB and change the mobility of the TBP/TFIIB/DNA complex in an EMSA, suggesting that they somehow affect the conformation of the complex (43). The different effects of TBP mutations observed with truncated yeast Brf1 and full-length human Brf1 may result from the ability of the Brf1 C-terminal region to compensate for loss of contacts with TBP within the Brf1 N-terminal region (35), or they may reflect different TBP contacts with yeast and human Brf1 in truncated and full-length proteins. Whichever the case, in the case of Brf2, radical mutations of TBP residues E284 and E286 have an effect very similar to that observed with TFIIB (43): reduction of the intensity of the complex and change in conformation, as suggested by a mobility change in the EMSA. Mutation of TBP residue L287, on the other hand, has little effect, as it does with Brf1 (43). Thus, the association of full-length Brf2 with TBP appears to share some features with TFIIB and others withBrf1.
When we replaced the core region of Brf2 with the core region of TFIIB, which has TBP/TATA box complex binding activity, the resulting chimera (2/B/2) was only able to sustain a barely detectable level of transcription, even though it was capable of recruiting Bdp1 to the promoter. The very low levels of transcription may be due to the very inefficient recruitment of SNAPc by this chimeric protein (Fig. (Fig.5A).5A). In addition, however, it may also result from inefficient recruitment of Pol III. Indeed, the yeast Brf1 core domain associates with both the C34 (1, 25) and the C17 (10) subunits of Pol III. The core domain of Brf2 may be similarly involved in interactions with Pol III-specific subunits.
In Brf1, the C-terminal region—more specifically, the conserved region II within the C-terminal region—provides the main anchor for binding to a TBP/TATA box complex (1, 9, 22, 25). The structure of a yeast Brf1 region II/TBP/TATA box complex has been solved and shows region II in Brf1 binding to the convex and lateral surfaces of TBP (19). We find that removing the C-terminal region of Brf2 inhibits U6 transcription (Fig. (Fig.1D)1D) and prevents Brf2 from associating efficiently with a TBP/TATA box complex (Fig. (Fig.3A)3A) and with TBP in solution (Fig. (Fig.6).6). An alignment of the C-terminal domains of human Brf2 (HsBrf2) and putative Brf2 homologues from mouse (Mm), rat (Rn), zebra fish (Dr), and fugu (Tr) is shown in Fig. Fig.8A8A and reveals a striking concentration of conserved residues close to the end of the protein. Such a high degree of sequence conservation between such varied species suggests an important function and prompted us to compare this segment to Brf1 conserved region II. One possible alignment is shown in Fig. Fig.8B.8B. The yeast and human Brf1 sequences are shown on top, together with the location of α-helices as seen in the yeast Brf1 region II/TBP/TATA box complex (19). Clearly, the Brf2 conserved sequence can be aligned with the Brf1 conserved region II, with identical and similar amino acids clustering in the region corresponding to Brf1 α-helices H22, H23, and the beginning of H24, which are all involved in interactions with the convex and lateral surfaces of TBP (19). This strongly suggests that the conserved Brf2 region near the end of the protein corresponds to the Brf1 conserved region II and provides the high-affinity contacts withTBP.
Apart from interactions with TBP, Brf2 is likely also involved in interactions with SNAPc because it can be recruited to the U6 promoter by a subcomplex of SNAPc referred to as mini-SNAPc and consisting of SNAP50, SNAP43, and the N-terminal third of SNAP190 (16). Although we have not previously observed recruitment of Brf2 by mini-SNAPc, perhaps because the mini-SNAPc/Brf2 complex was obscured by a comigrating complex containing what appeared to be an alternative conformation of mini-SNAPc (27), our present data indicate that the complete SNAPc can recruit Brf2 to the U6 promoter and that this is dependent on the C-terminal domain of Brf2 (Fig. (Fig.3B).3B). Thus, the C-terminal domain of Brf2 allows recruitment of the protein to the U6 promoter, probably through direct contacts with SNAPc, even in the absence of TBP.
In the absence of the entire TFIIB-related N-terminal half of the protein, the Brf1 C-terminal domain retains some ability to bind to a TATA box/TBP complex and to recruit Bdp1 (20). Our results suggest that the C-terminal domain of Brf2 is similarly involved in Bdp1 recruitment. Indeed, we find that the chimeric protein 2/B/− binds strongly to the TBP/TATA box complex but fails to incorporate Bdp1 (Fig. (Fig.5B).5B). However, when the C-terminal region of Brf2 is added to the 2/B/− chimera to form 2/B/2, Bdp1 can then be recruited (Fig. (Fig.5A).5A). Thus, the C-terminal domain of Brf2 appears to play a crucial role in the assembly of the preinitiation complex by providing stabilizing interactions with TBP, SNAPc, and Bdp1.
RNA polymerase specificity is determined at two levels. The first level is the arrangement of the promoter elements themselves. In the human snRNA promoters, it is the presence or absence of a TATA box downstream of the PSE that largely determines Pol III or Pol II transcription, respectively. This in turn somehow results in the specific recruitment of either TFIIB in the case of Pol II snRNA promoters or Brf2 in the case of Pol III snRNA promoters. Ironically, we do not yet understand how the presence of a TATA box specifies Brf2 recruitment over TFIIB recruitment. Indeed, we show here that it is possible to assemble efficiently a complex containing TBP, SNAPc, and TFIIB (or the 2/B/− chimeric protein) on the U6 promoter. This is somewhat reminiscent of the situation in the yeast U6 snRNA promoter, which consists of both a TATA box upstream of the transcription start site and a downstream binding site for the factor TFIIIC. The yeast U6 upstream promoter has no specificity for recruitment of Brf1 over TFIIB. Rather, it is the binding of TFIIIC to the downstream site that specifies Brf1-TFIIIB recruitment and Pol III transcription (32). It seems likely that, in human cells, as-yet-unidentified factors associated with either TFIIB, Brf1, TBP, or SNAPc prevent assembly of TFIIB and/or favor assembly of Brf2 into the U6 initiation complex.
The second level of RNA polymerase specificity determination is protein-protein interactions among factors recruited by the promoter elements. Thus, even though the U6 promoter is capable of assembling complexes containing either TBP, SNAPc, and TFIIB or TBP, SNAPc, and Brf2, only the latter complex (or complexes formed with TFIIB/Brf2 chimeras containing the Brf2 C-terminal region) is capable of recruiting Bdp1. Moreover, the C-terminal domain of Brf2 may also be involved in the recruitment of Pol III. With yeast Brf1, the C-terminal domain of the protein, and in particular homology domains II and III, have been shown to associate with the yeast Pol III subunit C34, which has no counterpart in Pol II (1, 25). Thus, the C-terminal extensions present in the TFIIB family members involved in RNA Pol III transcription, i.e., Brf1 and Brf2, are key to favoring recruitment of Pol III over that of PolII because they allow recruitment of Bdp1 and, probably, of Pol III itself through Pol III-specific subunits.
We thank Robert S. Haltiwanger, Patrick Hearing, and Winship Herr for helpful comments during the course of this project. We also thank Yuling Sun for generating the constructs expressing the TFIIB-Brf2 chimeras and J. Duffy for artwork and photography.
This study was funded in part by NIH grant GM38810. N.H. and B.M. are supported by the Howard Hughes Medical Institute.