BRFU, a TFIIB-like Factor, Is Directly Recruited to the TATA-box of Polymerase III Small Nuclear RNA Gene Promoters through Its Interaction with TATA-binding Protein*

The human snRNA genes transcribed by RNA polymerase II (pol II) and III (pol III) have different core promoter elements. Both gene types contain similar proximal sequence elements (PSEs) but differ in the absence (pol II) or presence (pol III) of a TATA-box, which, together with the PSE, determines the assembly of a pol III-specific pre-initiation complex. BRFU is a factor exclusively required for transcription of the pol III-type snRNA genes. We report that recruitment of BRFU to the TATA-box of these promoters is TATA-binding protein (TBP)-dependent. BRFU in turn stabilizes TBP on TATA-containing template and extends the TBP footprint both upstream and downstream of the TATA element. The core domain of TBP is sufficient for BRFU·TBP·DNA complex formation and for interaction with BRFU off the template. We have mapped amino acid residues within TBP and domains of BRFU that mediate this interaction. BRFU has no specificity for sequences flanking the TATA-box and also forms a stable complex on the TATA-box of the pol II-specific adenovirus major late promoter (AdMLP). Furthermore, pol III-type transcription can initiate from an snRNA gene promoter containing an AdMLP TATA-box and flanking sequences. Therefore, the polymerase recruitment is not simply determined by the sequence of the TATA-box and immediate flanking sequences.

The human snRNA genes transcribed by RNA polymerase II (pol II) and III (pol III) have different core promoter elements. Both gene types contain similar proximal sequence elements (PSEs) but differ in the absence (pol II) or presence (pol III) of a TATA-box, which, together with the PSE, determines the assembly of a pol III-specific pre-initiation complex. BRFU is a factor exclusively required for transcription of the pol III-type snRNA genes. We report that recruitment of BRFU to the TATA-box of these promoters is TATA-binding protein (TBP)-dependent. BRFU in turn stabilizes TBP on TATA-containing template and extends the TBP footprint both upstream and downstream of the TATA element. The core domain of TBP is sufficient for BRFU⅐TBP⅐DNA complex formation and for interaction with BRFU off the template. We have mapped amino acid residues within TBP and domains of BRFU that mediate this interaction. BRFU has no specificity for sequences flanking the TATA-box and also forms a stable complex on the TATA-box of the pol II-specific adenovirus major late promoter (AdMLP). Furthermore, pol III-type transcription can initiate from an snRNA gene promoter containing an AdMLP TATA-box and flanking sequences. Therefore, the polymerase recruitment is not simply determined by the sequence of the TATA-box and immediate flanking sequences.
The core promoter regions of human snRNA 1 genes are sufficient to direct low levels of transcription in vitro and contain a binding site, called the proximal sequence element (PSE), for the multisubunit factor PBP/PTF/SNAP c (1)(2)(3). The PSE, usually located around Ϫ55, is interchangeable between snRNA gene promoters recognized by RNA polymerase II (e.g. U1 and U2) and RNA polymerase III (e.g. U6 and 7SK) (4,5) and purified or recombinant PTF functions as a basal transcription factor for both types of snRNA gene (6,7). The pol III-specific core promoters contain an additional TATA-box at Ϫ25, which in this context is responsible for the selective recruitment of pol III (5,8). Insertion of a TATA element into the pol II-transcribed U2 promoter converts it into a predominantly pol III promoter (8), whereas mutation of the 7SK TATA-box reduces pol III transcription and allows snRNA-type transcription by pol II to occur (5). TBP is required for transcription of both types of snRNA gene and is likely to be recruited to the TATAless pol II-specific promoters by interaction with PTF binding to the PSE (3). PTF also potentiates direct binding of TBP to the TATA-box of the pol III-specific promoters (9). Because loss of pol III transcription correlates with the loss of TBP binding to the mutated TATA-box (5, 10), the differential interaction of TBP with template DNA and the other proteins of the PIC is likely to play a key role in the ultimate recruitment of different polymerases.
For transcription of tRNA and 5 S rRNA genes, which have gene-internal pol III promoters, TBP is associated with TFIIIB90 (11) (also called hBRF (12)) and hBЉ (13) (also called TFIIIB150 (14)) within the TFIIIB-␤ complex (15). At these TATA-less promoters, the internal promoter recruits TFIIIC that results in the subsequent recruitment of TFIIIB-␤, which may then directly recruit pol III. TBP is a more loosely associated subunit of the less well characterized snRNA-specific TFIIIB form, designated hTFIIIB-␣ (15), which is required for transcription of the U6/7SK genes by pol III. Recently, a basal transcription factor known as BRFU (13), or TFIIIB50 (14), has been shown to be required for transcription of these snRNA genes but not an adenovirus 2 VA1 gene with an internal pol III promoter. Interestingly, BRFU/TFIIIB50 has sequence homology to both TFIIIB90/BRF and the pol II initiation factor TFIIB. A complex of TFIIIB50 and four tightly associated factors constitutes, together with TBP and TFIIIB150, the complete TFIIIB-␣ activity that transcribes pol III snRNA genes (14). However, Hernandez and colleagues (16) could obtain U6 transcription by combining a partially purified pol III fraction with recombinant PTF, TBP, hBЉ, and BRFU alone (16). In addition, another factor encoded by an alternatively spliced variant of hBRF (BRF2) may also be required for U6 transcription (17). Thus the exact "TFIIIB complex" requirement for pol III-transcribed snRNA genes remains to be determined.
The ϳ50-kDa human BRFU represents another member of the TFIIB-related protein family and has conserved zinc and core domains, and a divergent C-terminal domain (13), (18). Within the Zn 2ϩ -binding region, BRFU is 37.5 and 31.2% identical to human TFIIB and BRF, respectively, suggesting that this region of BRFU also adopts a zinc ribbon structure. The identity with the TFIIB core region is 19% (13), and as in TFIIB, the BRFU core domain consists of two direct repeats.
Here we show that BRFU interacts with TBP to form a complex on TATA-containing templates and have mapped amino acid residues within TBP and domains of BRFU that mediate this interaction. Strikingly, we found that BRFU, unlike TFIIB, appears to have no specificity for sequences outside the binding site for TBP. Together, the data presented here provide an insight into an important step in nucleation of a pol III-specific snRNA transcription initiation complex.
Expression of these recombinant proteins was induced by the addition of isopropyl ␤-D-thiogalactopyranoside to a final concentration of 1 mM in an exponentially growing NM544 bacteria culture at 30°C. Isolation of His-tagged proteins was carried out according to Bryant et al. (19) with some modifications. Cells were sonicated in buffer D (20 mM HEPES, pH 7.9, 100 mM KCl, 20% glycerol, 2 mM ␤-mercaptoethanol, and Complete mixture of inhibitors minus EDTA (Roche Molecular Biochemicals, 1873580)) containing 20 mM imidazole. Debris was removed by centrifugation, and sonicates were mixed with Ni 2ϩ -Sepharose (Amersham Pharmacia Biotech) and rotated at 4°C overnight. The beads were extensively washed in buffer D containing 500 mM KCl and 20 mM imidazole. Bound proteins were eluted with buffer D containing 1 M imidazole and dialyzed against buffer A (20 mM Tris-HCl at pH 8.0, 100 mM KCl, 0.5 mM EDTA, 1 mM DTT, and 0.5 mM phenylmethylsulfonyl fluoride). The amount of eluted protein was estimated by comparison with a bovine serum albumin (BSA) standard in Coomassie Bluestained SDS-polyacrylamide gels. When required, eluates were concentrated in Ultrafree-CL centrifugal concentrators (Millipore Corp.).
Purification of GST Fusion Proteins-The plasmid encoding glutathione S-transferase full-length BRFU fusion protein GST-BRFU was derived from construct His-BRFU by polymerase chain reaction amplification of the BRFU coding region and placing into pGEX-2T (Amersham Pharmacia Biotech.) using BamHI and EcoRI sites. The BRFU deletion mutants ⌬1-37, ⌬72-157, ⌬169 -266, and ⌬72-266 were fused to GST tag at the N terminus of the protein. Recombinant GST and GST-BRFU proteins were expressed in NM544 cells. Clarified bacterial lysates were incubated with glutathione-agarose (Sigma Chemical Co.) for 1 h at 4°C. The beads were then washed four times in NETN buffer (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, 0.5 mM DTT, and 0.5 mM phenylmethylsulfonyl fluoride). GST fusion proteins were eluted from beads using 50 mM reduced glutathione (Sigma) in washing buffer and dialyzed against buffer A. The amount of eluted proteins was estimated by SDS-PAGE as described above.
Electrophoretic Gel Mobility Assays-The probes for the EMSA studies were made as described previously (2). The 7SKwt probe was prepared using the O ϩ P ϩ construct (2). For the 7SK-TATA probe the template O ϩ P ϩ T Ϫ was used where the TATA-box is mutated to TTT-GCGTA (5). The U2wt probe is similar to the 7SK O ϩ P ϩ and contains sequences from Ϫ82 to ϩ20 of the U2 gene (GenBank TM accession number L37793) and an Oct-1 binding site 23 bp upstream of the PSE. In the U2ϩTATA probe the sequence between Ϫ26 and Ϫ18 was mutated to TTTATATAT as described by Lobo and Hernandez (8). The adenovirus major late promoter (AdMLP) probe was prepared utilizing the AflIII and Acc65I sites at positions Ϫ112 to ϩ34 relative to transcription start site of AdML gene.
Two Binding and Corresponding Gel Systems Were Used-Binding reactions for TBP⅐BRFU⅐DNA complex detection contained 10 fmol of probe, 0.2 g of poly(dG-dC), 10 mM HEPES, pH 7.9, 30 mM Tris-HCl, pH 8.4, 60 mM KCl, 7.5 mM MgCl 2 , 10% glycerol, 6 mM ␤-mercaptoethanol, 20 mM DTT, and 0.1 mg of BSA per ml. Reactions were initiated by the addition of proteins, and mixtures were incubated for 30 min at 30°C. The mixtures were electrophoresed on a 4% (37.5:1, acrylamidebisacrylamide) polyacrylamide gel, with 0.5ϫ Tris borate-EDTA (TBE) at 40 mA. In the text, these conditions are referred as a "TBE gel system." For detection of the TBP⅐DNA complex, various TBPs were incubated with the probe for 40 min at 30°C in buffer containing 0.2 g of poly(dG-dC), 10 mM HEPES, pH 7.9, 10 mM Tris-HCl, pH 8.0, 60 mM KCl, 5 mM MgCl 2 , 10% glycerol, 1 mM DTT, 0.01% Nonidet P-40, and 0.1 mg of BSA per ml. The binding reactions were resolved on 4% polyacrylamide gel in 1ϫ TGEM running buffer (50 mM Tris base, 380 mM glycine, 2 mM EDTA, and 5 mM MgCl 2 ) at 200 V. We refer to these conditions as a "TGEM gel system." DNase I Footprinting-The 5Ј-radiolabeled U2ϩTATA probe was employed. Reaction conditions were the same as described for EMSA using the TBE gel system. DNase I digestion was carried out as described by Murphy et al. (2), and the reaction products were analyzed on a 6% polyacrylamide-urea gel.
GST Pull-down Assays-The in vitro transcription-translation vector pT␤TBP for expression of full-length TBP has been described previously (20), and pET11d-cTBP, used to make labeled core TBP, was a kind gift from Alex Hoffmann. Luciferase T7 control DNA was from Promega. The [ 35 S]Met-labeled full-length TBP, core TBP, and Luciferase were produced using a TnT-coupled transcription-translation system (Promega). Equal amounts of GST-BRFU and GST proteins bound to glutathione beads were incubated with 35 S-labeled proteins at 4°C for 2 h. The beads were then washed five times in 1 ml volume each of 20 mM HEPES, pH 7.9, 100 mM KCl, 10% glycerol, 0.5 mM MgCl 2 , 0.5% Nonidet P-40, 0.4 mg of ethidium bromide per ml, boiled in SDS sample buffer, and resolved by SDS-PAGE. Gels were dried, and radiolabeled proteins were detected by autoradiography.
In Vitro Transcription-The U2 TATA/7SK transcription construct contains U2 gene sequences Ϫ556 to ϩ6 upstream from sequences ϩ1 to ϩ458 of the marked 7SK gene (2) between the EcoRI and PstI sites of pGEM4. The U2 sequence between Ϫ26 and Ϫ18 was mutated to TTTATATAT (8). The ML-S and ML-L templates are U2 TATA/7SK derivatives where the TATA-box and U2 flanking sequences were replaced by AdMLP TATA-box and adjacent sequences at Ϫ37 to Ϫ19 (ML-S) and Ϫ46 to Ϫ10 (ML-L) relative to the AdML gene transcription start site. ML-L(-TATA) is a ML-L derivative where TATA-box was inactivated by an A to G substitution in the second position, resulting in the sequence, TGTAAAAG.
The nuclear extract was prepared from HeLa cells as described previously (21). The transcription reactions were carried out as described by Murphy et al. (22), with 1 g/ml ␣-amanitin and 250 ng of template in 25 l. Following 1-h incubation at 30°C, the transcripts were analyzed by S1 analysis (23).

TBP Specifically Recruits BRFU to the TATA-box, and the TBP Core Domain Is Sufficient for BRFU⅐TBP⅐DNA Complex
Formation on pol III snRNA Promoters-In vitro studies on mRNA promoters indicate that the PIC assembles in an ordered stepwise fashion where TFIID binding to the TATA-box (via TBP) nucleates the assembly of a complex containing TFIIA and TFIIB, which in turn recruits other basal factors and pol II (reviewed by Orphanides et al. (24)). The presence of a TATA element in the promoters of the human snRNA genes is instead a critical determinant of transcription by pol III (see the introduction). Because TBP binds directly to the TATA-box of these genes (5, 10, 25), we have investigated the effect of TBP/template interaction on the recruitment of the pol III-and snRNA-specific factor BRFU to these promoters. We have used templates derived from the TATA-containing, pol III-transcribed 7SK gene and the TATA-less, pol II-transcribed U2 gene for this analysis (Fig. 1A). Mutation of the 7SK gene TATA-box or addition of a TATA-box to the U2 gene effectively switches the polymerase specificity of these templates in vivo (see Fig. 1A). We have monitored DNA⅐TBP⅐BRFU interactions on these templates in an EMSA, using purified untagged recombinant, full-length TBP (26), N-terminal truncated "core" TBP ( Fig. 1C) and BRFU proteins (Fig. 1D). TBP alone can be detected binding to the TATA-containing templates only when a TGEM gel system is used (shown in Fig. 4) and not in a TBE gel system (Fig. 1B, lane 6). However, TBP and BRFU form a distinct and stable complex on both the U2ϩTATA and 7SKwt probes in the TBE system (Fig. 1B, lanes 2 and 3). This complex is specific for the TATA-box, because no complex is observed when TATA-less probes: U2 wt (lane 1) and 7SK with a debil-itated TATA-box (lane 4), were used. BRFU alone is unable to form a stable complex in either gel conditions (Fig. 1B, lane 7, and data not shown) and incubation of TBP, BRFU, and DNA results in detection only of a binary TBP⅐DNA complex using the TGEM system (not shown).
The conserved C-terminal core domain of human TBP forms an interface with the TATA-box and core region of TFIIB (27). We were, therefore, interested whether cTBP can also recruit BRFU to TATA-box templates. Indeed, prominent bands are observed when cTBP and BRFU were co-incubated with the U2ϩTATA and 7SKwt probes (Fig. 1B, lanes 9 and 11), but not with TATA-less probes: U2wt and 7SK-TATA (lanes 8 and 10).
In summary, the above data suggest that TBP and BRFU bind snRNA pol III promoters cooperatively, core TBP is sufficient to create complex with TBP on DNA, and BRFU does not strongly bind to DNA on its own. Because the sequences flanking the TATA-box in the 7SK and U2 probes are different, the TBP⅐BRFU complex does not appear to have strict requirements for sequence motifs outside the TATA-box.
BRFU Stabilizes TBP on a TATA-containing Template and Extends the TBP Footprint-To determine whether the same effect is observed at equilibrium in solution, we performed DNase I footprinting on the U2 TATA probe using TBE gel system binding conditions for EMSA ( Fig. 2A). In accordance with the EMSA data, TBP on its own slightly protects the TATA region from digestion (compare lane 6 with the lanes 2 and 3). However, the addition of increasing amounts of BRFU results in a strong footprint over the TATA-box and extension of the protected region upstream (3 nucleotides) and downstream (5 nucleotides) of the TTTATATA motif (lanes 4 and 5), suggesting that either BRFU causes a conformational change in TBP to extend its interaction with DNA or that BRFU interacts directly with the DNA flanking the TATA-box.
BRFU Associates with TBP Independently of the Template, and the Core Domain of TBP Is Sufficient for Interaction with BRFU-TBP and BRFU can bind cooperatively to DNA, suggesting protein⅐protein interactions exist between these two factors on the template. To determine whether TBP and BRFU can also interact "off DNA," GST pull-down assays were performed with GST-BRFU and two forms of TBP (Fig. 1C). The firefly luciferase was used as a control for nonspecific binding. Full-length TBP, core TBP, and luciferase were expressed in rabbit reticulocyte lysates, and proteins were labeled with [ 35 S]methionine. These labeled proteins were then mixed with GST-BRFU and GST protein alone (Fig. 3B) that had been pre-bound to glutathione-agarose beads. To avoid nonspecific protein⅐DNA⅐protein interactions bridged by plasmid DNA, ethidium bromide was included in the reactions. Fig. 3A shows the results of these GST pull-down assays. Significant interactions are observed between GST-BRFU and both full-length (lane 2) and core TBP (lane 5). These interactions are specific, because little cross-reaction is detected between these proteins and GST samples (lanes 3 and 6, respectively). As an additional control, luciferase was tested for interaction with GST-BRFU and GST. There is no difference in minimal nonspecific signals originating from binding to GST-BRFU (lane 8) or GST alone (lane 9).
These data suggest that the core domain of TBP mediates interaction with BRFU both on and off template DNA.
Point Mutations in the Core Domain of TBP Inhibit and Modify TBP⅐BRFU⅐DNA Complex Formation-Having estab-lished that TBP interacts with BRFU via the conserved Cterminal domain, we were interested which residues in this region interface between TBP and BRFU. Because BRFU shares strong similarities with both TFIIB and BRF, we tested substitution E284R in the second repeat stirrup that inhibits TFIIB binding to TBP⅐TATA-box DNA (28) and substitutions R231E, R235E, and F250E that abolish interaction with yeast BRF (29). All four single-amino acid substitutions were in the context of altered specificity (AS) TBP (19) and do not affect interaction with the TATA-box. There is no difference between wtTBP and "wt"AS TBP in DNA binding affinity for TATA-box probes and in the efficiency of DNA⅐TBP⅐BRFU complex formation (data not shown). Mutant TBPs expressed in and purified from Escherichia coli were assayed for DNA binding alone and in the presence of recombinant GST-BRFU (Fig. 4). EMSA shown in the upper panel (ϪBRFU) was conducted in TGEM conditions and confirms that wild-type and all mutant AS TBPs are able to create complexes with DNA (lanes 1-5). Data in the lower panel (ϩBRFU) were obtained using TBE conditions and show that the R231E mutant still retains the capability, although reduced, to form a TBP⅐BRFU⅐DNA complex (compare lanes 1 and 2). In contrast, substitutions R235E and F250E completely inhibit GST-BRFU binding to TBP⅐TATA-box ( lanes  3 and 4, respectively). Interestingly, substitution E284R does not inhibit the TBP⅐BRFU⅐DNA formation but considerably changes complex conformation, which is apparent from its slow mobility (lane 5).
Direct Repeat 2 in the BRFU Core Is Necessary for the Assembly of a TBP⅐BRFU⅐DNA Complex-On the basis of TFIIB structure-function studies demonstrating that direct repeats 1 and 2 of the core are both required for TBP⅐TFIIB⅐DNA complex assembly (30), we designed a set of GST-BRFU mutants where important TFIIB-like domains are deleted (Fig. 5A). Recombinant proteins were then produced from these mutants (Fig. 5C) and tested in the gel retardation assay (Fig. 5B). Deletion of neither the Zn-ribbon (lane 2) nor repeat 1 (lane 3) resulted in an inhibition of TBP⅐BRFU⅐DNA complex formation. However, deletion of repeat 2 (lane 4) and the whole core (lane 5) completely eliminated formation of the complex.
These data suggest that, despite strong similarities, BRFU behaves differently to TFIIB in assembly of a TBP⅐BRFU⅐DNA complex.
BRFU Itself Is Not Sufficient to Select the Relevant TBP⅐TATA-box Promoter Template-Because TFIIB recognizes the BRE consensus sequence 5Ј-(G/C)(G/C)(G/A)CGCC-3Ј im-mediately upstream of the TATA element in protein-encoding genes (31) and there is no similar motif in TATA-box containing snRNA genes at the same position, we assumed there would be differential formation of TBP⅐TFIIB⅐DNA and TBP⅐BRFU⅐DNA complexes on these promoters. The adenovirus major late promoter was used as a representative of a typical pol II gene transcription system, and the U2 TATA promoter was used because it supports pol III-specific snRNA transcription (Fig.  6B). Probes prepared from these promoters were tested in the EMSA for recruitment of TBP⅐TFIIB and TBP⅐BRFU com- plexes (Fig. 6A). A TBP⅐TFIIB complex forms specifically on the AdMLP (lane 7) but not U2 TATA (lane 3). Intriguingly, however, BRFU creates a complex with TBP on the AdMLP (lane 6) as efficiently as on the U2 TATA promoter (lane 2). This is surprising, because the major late TATA-box is located next to a strong BRE that favors TFIIB binding. As indicated by arrows, there are two complexes of different size apparent in lane 7. In addition to the expected lower band, the upper band likely represents multimers or a different conformation of the TBP⅐TFIIB⅐DNA complex. Similarly to BRFU, TFIIB on its own does not bind to either AdMLP or U2 TATA probes under the conditions used in this experiment (data not shown).
Introduction of the AdMLP TATA-box and Flanking Sequences into the snRNA Promoter Retains pol III Transcription Specificity-The observation that AdMLP TATA-box and flanking sequences have no selective exclusion effect on recruitment of BRFU versus TFIIB prompted us to perform the following functional assay. We tested the effect of replacing the TATAbox region of a pol III-transcribed snRNA gene construct with AdML core promoter sequences on transcription in vitro. As a template we used a U2/7SK hybrid construct that gives a high level of pol III snRNA gene transcription in vivo only when a TTTATATAT is present between Ϫ26 and Ϫ18 (Fig. 6D, "wt"). 2 The TATA-box and U2 flanking sequences (Fig. 6B, U2 TATA) were replaced by the AdMLP TATA-box together with minimal 2 S. Murphy, unpublished observations. regions required for TFIIB binding (B, AdMLP), resulting in a U2/AdMLP/7SK chimeric construct (D, ML-S). To also evaluate the effect of a larger AdML core promoter region, we introduced a fragment that encompasses sequences 15 bp upstream and 14 bp downstream of the TATAAAAG element, into U2/7SK (D, ML-L). Transcription was then carried out in a HeLa cell nuclear extract containing 1 g/ml ␣-amanitin to inhibit transcription by pol II, and the products were analyzed by S1 assay. The pol III-dependent adenovirus VA1 gene was included as an internal control. As shown in Fig. 6C, all three templates are transcribed by pol III (lanes 1-3). Remarkably, the templates containing AdMLP sequences are transcribed even more efficiently than the U2 TATA/7SK template (compare lanes 2 and 3 with lane 1). When the TATA-box is mutated in ML-L (D, ML-L(-TATA)), the transcription is effectively abolished (lane 4) confirming that transcription is TATA-box-dependent. These results suggest that pol III snRNA-specific PICs, likely containing BRFU, can be formed on and direct transcription from an mRNA gene-specific TATA-box and flanking sequences.
In summary, these findings are in agreement with the results of the BRFU/TBP binding studies on the AdMLP template. DISCUSSION The exact mechanism of differential PIC assembly on pol II and pol III snRNA promoters is still awaiting elucidation. TBP is required for transcription of both types of snRNA genes but not as part of the TBP-containing complexes TFIID (32) or TFIIIB-␤ (3, 6, 15) that function in transcription of mRNA genes (by pol II) and tRNA/5 S RNA genes (by pol III), respectively. Because the same PTF is required for transcription of snRNA genes by both pol II and III, it seems likely that different modes of TBP recruitment by both PTF and promoter sequences set the stage for subsequent recruitment of polymerase-specific factors. Here we show that TBP bound to the TATA-box of pol III snRNA templates can recruit the pol IIIspecific snRNA-gene factor BRFU but not the pol II-specific TFIIB and that the intact TATA element is essential for BRFU⅐TBP⅐DNA complex formation. The stability of the BRFU⅐TBP⅐DNA complex resembles to some extent the yeast TFIIIB complex, which consists of strongly associated TBP and BRF and loosely associated BЉ (33,34). The non-conserved N-terminal domain of TBP was reported to be responsible for its cooperative binding with PTF to their respective binding sites on U6 promoter (9). In contrast, we find that the Cterminal core domain of TBP (cTBP) is sufficient to mediate interaction with BRFU both in solution and on DNA.
We could not detect direct interaction between BRFU and DNA, but DNase I analysis suggests that BRFU contacts sequences upstream and downstream of the TATA-box when TBP is bound. These sequences might play a role in the clamping of TBP to DNA by BRFU. We speculate there is a TBP-induced DNA bend in BRFU⅐TBP⅐DNA complex that places the upstream and downstream DNA segments in proper spatial register for simultaneous BRFU⅐DNA interactions with both DNA segments. It is therefore possible that, like TFIIB (27), binding of BRFU is synergetic with TBP requiring the distortion of the TATA-box (Fig. 2B). Biochemical studies have shown that cTBP recognizes the TATA-box of protein-encoding genes in both orientations (reviewed in Ref. 35) and TFIIB, as an essential factor in the assembly of a functional PIC, forms a stereospecific complex with TBP (36). It, therefore, follows that a specific TFIIB⅐BRE interaction (31) would contribute strongly to unique directionality in the assembly of the PIC and, hence, to the polarity of transcription. In the yeast Saccharomyces cerevisiae, TFIIIB is recruited to the U6 promoter through the interaction of its TBP subunit with a TATA-box, and the direction of this complex assembly is dictated by a TFIIIC-dependent mechanism (37). It remains to be determined whether the sequences flanking the TATA-box in human pol III-specific snRNA templates play a role in setting the orientation of the TBP⅐BRFU complex.
Although the BRFU core possesses a structure similar to the core of TFIIB, we can expect differences in composition of TFIIB⅐TBP⅐DNA and BRFU⅐TBP⅐DNA complexes. The TFIIB C-terminal domain, containing intact direct repeats and associated basic regions, is necessary for interaction with TBP⅐DNA complexes (30). We found that direct repeat 2 of the BRFU core, but not repeat 1, is required for formation of a TBP⅐BRFU⅐DNA complex. In hBRF, both repeats and more avidly the C-terminal half of the protein interact with TBP (11). The transcriptionally active BRF2 variant is also able to form a complex with TBP (17). In this regard it should be noted that the 23-kDa BRF2 does not contain all of the structural regions that are typical for TFIIB-related proteins (zinc-ribbon, repeat 1, and repeat 2) and includes only part of the second direct repeat. In TFIIB, the zinc-ribbon domain is required for direct recruitment of a TFIIF⅐pol II complex (38,39). BRF, like TFIIB, also directly contacts polymerase, in this case pol III (11,40). However, the Zn-ribbon in BRF plays a role in open complex formation in yeast (41) but is not required for pol III recruitment (41)(42)(43). Further experiments are therefore required to reveal the precise function of the Zn-ribbon in BRFU.
Earlier studies of TBP mutants have already revealed a great deal about how the protein functions in pol II and pol III transcription. Residues Glu-284, Glu-286, and Leu-287 at the tip of the second stirrup of the saddle-shaped molecule (44) are critical both for TFIIB binding and in vitro and in vivo transcription (19,28,45). In yeast, different residues Arg-231, Arg-235, and Phe-250 contribute to the TBP surface that interacts with BRF (29). In our assays, single-amino acid substitutions R235E and F250E in TBP prevent BRFU from entering a TBP⅐DNA complex. Mutation E284R did not affect the stability of the BRFU⅐TBP⅐DNA complex but caused significant changes in the complex conformation. Thus, these data are consistent with the BRFU sequence similarities to both BRF and TFIIB.
Clearly, the sequences flanking the TATA-box in the U2 TATA construct do not allow TBP⅐TFIIB⅐DNA complex formation and thus might exclude TFIIF-pol II recruitment, and in consequence, pol II-specific transcription. However, because BRFU forms a complex with TBP⅐AdMLP as well as TFIIB does and pol III-type transcription can initiate from an snRNA gene promoter containing an mRNA gene TATA-box and flanking sequences, additional mechanism(s) must direct BRFU to specifically assemble on pol III snRNA templates in the cell. Otherwise, BRFU would compete with TFIIB in PIC formation on the promoters of protein-encoding genes (Fig. 7). Thus, sequences further outside the core promoter of the AdMLP may ensure that only pol II-specific PICs can form on this template in vivo. For instance, factors binding to promoter and initiator sequences may interact specifically with pol II-specific basic factors and effectively exclude pol III-specific factors. However, at least one mRNA promoter, within the c-myc gene, can direct TATA-dependent transcription by pol III both in vitro and in vivo in some circumstances (46), suggesting that the balance can be tipped to favor pol III.
The availability of individual factors, TBP, PTF, BЉ, and pol III snRNA-specific BRFU offers a unique system to understand the structure and function of a basal transcription multisubunit complex specific for snRNA genes transcribed by pol III polymerase. It is possible that, in addition to TBP, BRFU or its associated factors directly contact transcription factors PTF and BЉ and, together with any recognition elements in the core of pol III snRNA promoters, provides a basis for selective recruitment of pol III.

FIG. 7. A model for selective recruitment of BRFU and TFIIB
to TATA-box containing promoters. pol III snRNA, TBP and BRFU cooperate to form a ternary complex. TATA-box flanking sequences do not allow TFIIB to enter TBP⅐DNA complex. pol II mRNA, when BRE element is present, the TFIIB⅐TBP⅐DNA complex is formed. An as yet unknown mechanism excludes BRFU from formation of partial PIC on promoters of protein-encoding genes in the cell.