Inhibition of TATA Binding Protein Dimerization by RNA Polymerase III Transcription Initiation Factor Brf1*

The Brf1 subunit of TFIIIB plays an important role in recruiting the TATA-binding protein (TBP) to the upstream region of genes transcribed by RNA polymerase III. When TBP is not bound to promoters, it sequesters its DNA binding domain through dimerization. Promoter assembly factors therefore might be required to dissociate TBP into productively binding monomers. Here we show that Saccharomyces cerevisiae Brf1 induces TBP dimers to dissociate. The high affinity TBP binding domain of Brf1 is not sufficient to promote TBP dimer dissociation but in addition requires the TFIIB homology domain of Brf1. A model is proposed to ex-plain how two distinct functional domains of Brf1 work in concert to dissociate TBP into monomers. Eukaryotic genes are transcribed by one of three RNA poly-merases, pol 1 I, II, or III. In higher eukaryotes, there are two main pathways of assembling a pol III transcription complex, which are governed in large part by distinct promoter elements located either upstream or downstream of the transcriptional start site (1–3). In the budding yeast Saccharomyces cerevisiae , one major assembly pathway exists, and it involves the binding of TFIIIC to promoter elements internal to tRNA genes. TFIIIC then recruits TFIIIB to a location immediately upstream of the transcriptional start site (4, 5). TFIIIB then directs pol III to the transcriptional start site. TFIIIB is composed of three sub-units:

The Brf1 subunit of TFIIIB plays an important role in recruiting the TATA-binding protein (TBP) to the upstream region of genes transcribed by RNA polymerase III. When TBP is not bound to promoters, it sequesters its DNA binding domain through dimerization. Promoter assembly factors therefore might be required to dissociate TBP into productively binding monomers. Here we show that Saccharomyces cerevisiae Brf1 induces TBP dimers to dissociate. The high affinity TBP binding domain of Brf1 is not sufficient to promote TBP dimer dissociation but in addition requires the TFIIB homology domain of Brf1. A model is proposed to explain how two distinct functional domains of Brf1 work in concert to dissociate TBP into monomers.
Eukaryotic genes are transcribed by one of three RNA polymerases, pol 1 I, II, or III. In higher eukaryotes, there are two main pathways of assembling a pol III transcription complex, which are governed in large part by distinct promoter elements located either upstream or downstream of the transcriptional start site (1)(2)(3). In the budding yeast Saccharomyces cerevisiae, one major assembly pathway exists, and it involves the binding of TFIIIC to promoter elements internal to tRNA genes. TFIIIC then recruits TFIIIB to a location immediately upstream of the transcriptional start site (4,5). TFIIIB then directs pol III to the transcriptional start site. TFIIIB is composed of three subunits: the TATA binding protein (TBP), Brf1, and Bdp1 (6 -10).
TBP is important for the expression of essentially all eukaryotic genes. It is a saddle shaped protein having aminoand carboxyl-terminal stirrups and a concave/convex surface (11,12). TBP binds to a TATA box promoter element when it is present. However, most pol III genes with an internal promoter lack a TATA box. TBP is delivered to the upstream region of these promoters by Brf1, and Brf1 is recruited by TFIIIC (13)(14)(15).
The amino-terminal half of Brf1 is homologous to the pol II general transcription factor TFIIB (16 -18). TFIIB binds to TBP along its carboxyl-terminal stirrup, pinning it to DNA (19). In contrast to TFIIB, the TFIIB homology domain of Brf1 has not been shown to independently bind TBP. Where or whether it interacts with TBP is largely unknown. Some evidence places it near the carboxyl-terminal stirrup of TBP (20,21). The amino-terminal domain of Brf1 interacts with the 131 subunit of TFIIIC as well as the C34 subunit of pol III (14,22). This domain also plays an important role in open complex formation (23,24), the stage at which the template and transcribed strands are separated in preparation for transcription initiation. The carboxyl-terminal half of Brf1 is unrelated to TFIIB but is conserved among Brf proteins. A portion of the carboxyl-terminal half (437-506) of Brf1 binds with high affinity to the convex surface of TBP, snaking along the "top" of TBP and down its amino-terminal stirrup (25).
TBP self-associates into dimers and possibly higher order structures through its concave DNA binding surface and carboxyl-terminal stirrup (26 -39). Dimers must dissociate into monomers prior to DNA binding. In vivo, TBP dimerization might play an important role in preventing unregulated expression of quiescent pol II genes (30,37,39). Mutations in TBP that disrupt TBP dimerization cause increased expression of lowly expressed genes, suggesting that dimer dissociation is at least partly rate-limiting in the expression of these genes. Dimer dissociation does not appear to be rate-limiting at highly expressed genes, presumably because of direct or indirect alleviation of TBP auto-inhibition by transcription factors. Indeed, the pol II transcription factor TFIIA stimulates TBP/TATA interactions in part through dissociation of TBP dimers (29). Because TBP homodimerization is an intrinsic property of TBP, any polymerase system that utilizes TBP is expected to be endowed with an activity that dissociates TBP dimers.
Here we explore the possibility that Brf1, the primary TBPinteracting protein within the pol III transcription system, inhibits TBP dimerization when it engages TBP. We explore the mechanism by which Brf1 affects TBP dimerization by examining which domains of Brf1 are important for this reaction and whether Brf1 can bind TBP dimers and induce their dissociation. Our results suggest that Brf1 binds to TBP dimers and induces their dissociation. The presence of this activity in Brf1 suggests that TBP dimers are a physiological target of the pol III transcription machinery. 1-262, 263-596, and 263-431. For the 1-596 and 1-262 constructs, the 5Ј primers contained an XhoI site and the 3Ј primers contained a ClaI site at the ends of the amplified fragments. Restriction sites, which follow, are underlined. 1-596 and 1-262 utilized the following upstream  primer (5Ј-AGCATAATCTCGAGATGCCAGTGTGTAAGA-3Ј) and  downstream primers (5Ј-ATGATGATCGATTTACCTAAACAAACCGT-CAA-3Ј and 5Ј-AGCCTTATCGATTTAGAATTCGTTCAACCGTT-3Ј).  263-596 and 263-431 contained an NdeI site in the 5Ј primer (5Ј-CG-TGAGC TCATATGAAAAATACA AAGGCTGCTAA-3Ј). The 3Ј primer contained a ClaI site at the ends of the amplified fragment and contained the following sequences: 263-596, 5Ј-CTGAGTGCATCGATTTA-CCTAAACAAACCG-3Ј and 263-431, 5Ј-TGGGTCATCGATTTAAGAG-TAACCATCAATGGCATCT-3Ј. The PCR-amplified deletions were inserted in frame with the polyhistidine coding sequence of pEt-16b. The integrity of the entire open reading frame was confirmed by DNA sequencing.
Cells were thawed and mixed with 0.8 mg of lysozyme per ml for 10 min at 4°C and with 0.2% IGEPAL CA-630 for 5 min. Extracts were sonicated to reduce viscosity and then centrifuged in an SS34 rotor (RC5C centrifuge) at 15,000 rpm for 30 min at 4°C. Supernatants were removed and pellets were transferred to a type A Dounce homogenizer and resuspended in H.35 buffer (20% glycerol, 2 mM magnesium chloride, 20 mM HEPES, pH 8.0, 350 mM potassium chloride, 1% sodium hydroxide, 0.1 mM phenylmethylsulfonyl fluoride, 0.1% IGEPAL CA-630) and centrifuged in an SS34 rotor (RC5C centrifuge) at 15,000 rpm for 15 min at 4°C. The supernatant was again removed and the pellet was transferred to a type A Dounce homogenizer, washed with H.35 buffer, and centrifuged in an SS34 rotor (RC5C centrifuge) at 15,000 rpm for 15 min at 4°C. The supernatant was removed and inclusion bodies containing Brf1 were extracted from the pellet with G6 buffer (6 M guanidine hydrochloride, 100 mM Tris-HCl pH 7.5, and 2 mM ␤-mercaptoethanol) and centrifuged in an SS34 rotor (RC5C centrifuge) at 15,000 rpm for 15 min at 4°C. The supernatant was bound in a batch solution to TALON resin (Clontech) overnight at 4°C.
After incubation, the resin was washed in batch with G6 buffer. The resin was transferred to a column, and the protein was eluted in fractions with elution buffer (6 M guanidine hydrochloride, 100 mM Tris-HCl, pH 7.5, 10 mM ␤-mercaptoethanol, 500 mM imidazole, and 250 l of protease inhibitor mixture (Sigma)). Renaturation of the protein in solution was achieved through dialysis against 20 mM Tris acetate (80% cation), 200 mM potassium chloride, 2 mM magnesium chloride, 20% glycerol, 0.1 mM phenylmethylsulfonyl fluoride, and 0.1% IGEPAL CA-630 to lower the guanidine concentration to less than 0.1 M. Brf1 aliquots were frozen in liquid nitrogen and stored at Ϫ80°C.
Chemical Cross-linking Assay-Reactions were performed in buffer A (20 mM Tris acetate, pH 7.5, 4 mM magnesium chloride, 4 mM spermidine, 0.1 mM EDTA, 75 mM potassium glutamate, 50 pg/l heparin, 0.025% IGEPAL CA-630, and 5% glycerol) in a volume of 10 l at 30°C. In Fig. 5 the fraction of cross-linked TBP was determined from densitometric scans of autoradiographs. Local background was sub-tracted, and the data were normalized to total TBP recovered. Data from quadruplicate assays were plotted as a function of time and globally fit to the equation S x ϭ ⌬Se Ϫkt ϩ S bkd , using Kaleidagraph software. S x is the normalized data. ⌬S is the computer-fitted change in total signal during the reaction timecourse, and S bkd is the computer-fitted background at equilibrium. The data were converted to a reaction coordinate by the algebraic equation: reaction coordinate ϭ (S x Ϫ S bkd )/⌬S.
GST Pull-down Assay-Reactions using GST-180C were performed in buffer A. Reactions using GST-181C contained 20 mM Tris acetate, pH 7.5, 3.25 mM magnesium chloride, 10.7% glycerol, 122 mM potassium chloride, 48 mM potassium glutamate, 0.05% IGEPAL CA-630, 0.038 mM phenylmethylsulfonyl fluoride, 0.5 g heparin per ml, 97.7 g bovine serum albumin per ml, and 4 mM spermidine. Brf1 derivatives, polyhistidine-tagged TBP, and GST-180C, GST-181C, or GST bound to glutathione-agarose resin were used at concentrations indicated in the legends to Fig. 1, C and D, and Fig. 3. All reactions were performed in a volume of 450 l. Reactions were incubated at 4°C for 1 h with mixing. Resins were centrifuged briefly and supernatant was discarded. Resins were washed once with 500 l of buffer A, and the resin was collected by centrifugation. Bound proteins were eluted with 2ϫ PSB and nonspecific carrier protein and subjected to SDS-PAGE. TBP was probed by immunoblotting with TBP antibodies. Reactions were performed at least three times, and representative data are shown.
Brf1-TBP Interaction Assay-Brf1 derivatives and polyhistidinetagged TBP were incubated together for 1 h on ice in 10 mM Tris acetate (80% cation, pH 8.0), 1 mM magnesium chloride, 10% glycerol, 100 mM potassium chloride, 0.05% Nonidet P-40, 0.05 mM phenylmethylsulfonyl fluoride, 2.5 mM sodium metabisulfite, 2.5 mM benzamidine. The reaction mixture was then added to a 20-l bed volume of nickel-Sepharose equilibrated to the same buffer, incubated at room temperature for 1 min, washed with 20 mM imidazole in cross-linking buffer, and eluted with 500 mM imidazole in 2ϫ PSB.
Plasmids expressing "null," V161R, and V161E derivatives of TBP under control of the GAL10 promoter are pCALF-T(M1stop)(GAL10), pCALF-T(V161R)(GAL10), and pCALF(V161E)(GAL10), respectively, and are described elsewhere (30). Each was transformed into SJZ4. pCALF-T(M1stop)(GAL10) was also transformed into SJZ2 and SJZ6. 100-ml cultures were grown at 30°C in complete synthetic medium-Leu-Trp-Ura plus 3% raffinose to an OD 600 of ϳ0.6, and then were either induced with 2% galactose or mock treated for 2 h. Following induction, cells were harvested and total RNA was extracted with hot acidic phenol. Approximately 5 g of total RNA was treated with DNase I and then subjected to first strand cDNA synthesis for 2 h at 42°C in 20-l reactions with a primer (P1) common to ⌿-WT SNR6 and SNR6 or its derivative, snr6⌬BB. Following reverse transcription, 1, 0.1, and 0.01% of each cDNA sample was added to two separate reactions for PCR amplification using the common P1 primer and primer P2 (specific to ⌿-WT SNR6) or primer P3 (specific to SNR6 and snr6⌬BB). Primer sequences include the following: P1, 5Ј-AAAACGAAATAAATCTCTTT-G-3Ј; P2, 5Ј-ATCTTCGGATCATTTGG-3Ј; P3, 5Ј-GAAGTAACCCTTCG-TG-3Ј. PCR was performed for 25 cycles at 52°C annealing temperature. ⌿-WT PCR reactions and test reactions were combined and electrophoresed onto 15% polyacrylamide Tris-acetate-EDTA gels. Bands were detected with ethidium bromide staining, and quantification was done with NIH Image software. Local background intensity was subtracted from each band intensity and normalized to ⌿-WT.

Brf1
Inhibits TBP Dimerization-Previously we employed two distinct assays to measure TBP dimerization, chemical Brf1 Inhibits TBP Dimerization cross-linking and a GST pull-down assay (28,31,37). As shown in Fig. 1A (lane 1), purified TBP cross-links into dimers in the presence of the cysteine-specific homobifunctional cross-linker BMH. However, increasing concentrations of purified recombinant Brf1 inhibited the cross-linking of yeast TBP dimers (lanes 2-4). Inhibition was observed at roughly equimolar concentrations of purified Brf1 and TBP. Nonspecific proteins failed to inhibit the cross-linking (not shown and Ref. 30), indicating that Brf1 was not simply titrating out the crosslinker. In addition, we did not detect any cross-linking of TBP to Brf1, possibly reflecting that cysteines on Brf1 were at suboptimal distances from cysteines located on TBP.
To address whether the inhibition was caused by Brf1 rather than some contaminant in the preparation, Brf1 was depleted with either Brf1 antibodies or mock depleted with nonspecific antibodies (Fig. 1B). As shown in lanes 5-7 of Fig. 1A, Brf1 depletion abolished the inhibitory effect, whereas the control depletion (lanes 8 -10) gave the same level of inhibition as untreated Brf1. As a further control, TBP was titrated back into the inhibited reaction, restoring TBP dimerization (lanes [11][12][13], suggesting that Brf1 inhibition of TBP dimerization involved stoichiometric interactions. As a second measure of TBP dimerization, we used a GST pull-down assay in which the conserved DNA binding/dimerization core of TBP was fused to GST. Because we have found that human TBP tends to form more stable dimers in vitro than yeast TBP (37), the assay was performed with both the human (Fig. 1C) and yeast (Fig. 1D) core. In both cases, little or no full-length TBP was retained on the resin in the absence of the core TBP dimerization domain (Fig. 1, C and D) lane 1 versus  2). In both experiments, titration of purified Brf1 inhibited the pull down of TBP (Fig. 1, C and D, lanes 3 and 4), confirming that Brf1 inhibits TBP dimerization.
Both the TFIIB Homology Domain and the Carboxyl-terminal Domain of Brf1 Are Required to Inhibit TBP Dimerization-To address the specificity of dimer inhibition by Brf1 and determine what structural domains on Brf1 are important, we constructed several previously described deletion mutants of Brf1 ( Fig. 2A) (22). Region 1-262 corresponds to the TFIIBhomology domain, and region 263-596 corresponds to the TBPbinding carboxyl-terminal domain. A smaller region (263-431) with diminished affinity for TBP has also been described (22). These polyhistidine-tagged deletion derivatives of Brf1 were expressed and purified from recombinant E. coli (Fig. 2B). To verify their binding affinity for TBP, the Brf1 derivatives were immobilized on nickel-agarose resin and incubated with TBP. As shown in Fig. 2C, 263-596 2. Brf1 deletion constructs. A, schematic of the deletion derivatives used in this study, which are based upon similar constructs described previously (22). His denotes a polyhistidine tag, Zn F denotes zinc finger, the arrows reflect a structural repeat, and I, II, and III denote conserved domains in the carboxyl terminus. B, silver-stained polyacrylamide gel of purified recombinant Brf1 deletion derivatives. C, 110 nM TBP was incubated for 1 h at 0°C with nickel-Sepharose either in the absence (lane 2) or presence of 37.5 nM full-length Brf1 (lane 3) or Brf1 deletion mutants (lanes 4 -6), which were retained on the resin via their polyhistidine tag. The resins were then washed and TBP was analyzed by immunoblotting. 10% of the input TBP is shown in lane 1.
followed by full-length Brf1 and the other derivatives. This hierarchy of affinities is essentially identical to that described earlier (22).
Next we addressed whether the various deletion derivatives of Brf1 could inhibit TBP dimerization. As shown in Fig. 3 using the human or yeast TBP core pull-down assay, we found that although full-length Brf1 inhibited TBP dimerization, none of the truncation mutants were inhibitory. Therefore, both the TFIIB-homology domain and the carboxyl-terminal TBP binding domain of Brf1 were required to inhibit TBP dimerization. We also mixed derivatives 1-262 and 263-596 and found that they did not inhibit TBP dimerization (data not shown). Thus, despite Brf1 (263-596) having a high affinity for TBP, it must be physically connected to the amino-terminal TFIIB-homology domain to inhibit TBP dimerization. The fact that individual or combined fragments of Brf1 failed to inhibit TBP dimerization indicated that the inhibitory effect of the full-length protein is unlikely to be caused by nonspecific binding of Brf1 to TBP.
Brf1 Binds TBP Dimers-To explore the mechanism by which Brf1 might inhibit TBP dimerization, we first examined whether Brf1 could stably bind to TBP dimers or whether Brf1 binding and TBP dimerization were competitive events. Stabilization of TBP dimers was achieved by incubating full-length (and untagged) TBP with BMH, which covalently cross-links TBP into dimers. The quenched and dialyzed cross-linking reaction was then incubated with polyhistidine-tagged Brf1 bound to nickel-Sepharose. As shown in Fig. 4A, cross-linked (as well as uncross-linked) TBP bound to the Brf1 resin but not to resin alone, indicating that Brf1 binds TBP dimers. To verify that any Brf1 bound to TBP in this state (i.e. immobilized on a solid support) was still capable of inhibiting TBP dimerization, uncross-linked TBP was incubated with immobilized Brf1. The resin was washed to remove free TBP, and the proteins were subjected to BMH cross-linking. As shown in Fig. 4B (lane 1), TBP cross-linking was inhibited by Brf1 compared with a control reaction lacking Brf1 and omitting the wash step (lane 3). Taken together, these results suggest that Brf1 binds to TBP dimers and subsequently inhibits dimerization.
Brf1 Induces TBP Dimer Dissociation-The second step in examining Brf1 inhibition of TBP dimerization was to determine whether Brf1 induces dimer dissociation or simply sequesters monomers as they dissociate. The intrinsic dissociation kinetics of TBP dimers was measured by incubating TBP with a 2-fold excess of a DNA oligonucleotide containing a TATA box as reported previously (28). At various times after addition of the TATA DNA, the dimerization state of TBP was assayed by BMH chemical cross-linking. Because the intrinsic rate of dimer dissociation is slow in comparison to the incubation time in a chemical cross-linker, the dimer dissociation rate can be measured accurately. As shown in Fig. 5A, TBP dimers dissociated with a t1 ⁄2 of ϳ7 min. No net dissociation was observed in the presence of a mutant TATA oligonucleotide, indicating that the reaction was specific.
The time course was then repeated in the presence of a 2-fold excess of Brf1. As shown in Fig. 5B, TBP dissociated with a t1 ⁄2 of ϳ1 min, significantly faster than the intrinsic rate observed with the TATA oligonucleotide. This result suggests that Brf1 induces TBP dimer dissociation rather than simply trapping monomers.
A Dimer Defective Mutant of TBP Is More Active on a Basal Pol III Promoter-Previously we have shown that a truncated pol II promoter lacking its upstream activating sequence and transcribed at a low basal level is stimulated by TBP dimer/ DNA mutants (30). In contrast, highly active promoters are inhibited by these mutants. The data suggest that TBP homodimerization contributes to transcriptional repression by blocking access of TBP to promoters. This inhibitory interaction is expected to be eliminated under conditions of activated transcription. A new rate-limiting step such as TBP/TATA stability might dictate transcriptional output, and this would be susceptible to mutations along the DNA binding surface of TBP, as was observed.
To test whether TBP dimer dissociation is potentially relevant to pol III transcription in vivo, we examined the effect of a dimer/DNA defective mutant TBP(V161R) on transcription of the pol III-transcribed SNR6 gene. Like most pol III genes, SNR6 is highly transcribed in growing yeast cells, and so it is not expected to be stimulated by TBP(V161R). In an effort to mimic the basal pol II system that was stimulated by TBP(V161R), we used a variant of SNR6 that lacked one of its  2 and 3) 100 nM Brf1. The resins were then washed, treated with BMH, and then analyzed for TBP by immunoblotting. In lane 3, the wash step was omitted, and 25% of the sample was loaded. The migration of TBP dimers is denoted by XL and uncross-linked TBP by un-XL.
critical control elements, the B box (41,45). snr6(⌬BB) is transcribed at ϳ10% of the level of SNR6 and thus is significantly impaired in transcription. This allele does not support cell viability and thus must be accompanied by a wild type SNR6. The wild type SNR6 gene is modified (referred to as ⌿-WT) making it distinguishable in reverse transcriptase-PCR reactions from snr6(⌬BB) (41,45). As shown in Fig. 6, when transcript levels are normalized to ⌿-WT, the presence of TBP(V161R) caused a modest but reproducible increase in expression of snr6(⌬BB). This level of increase was not reproduced in the presence of a different TBP mutant V161E, which shows defects in DNA binding but is less defective in dimerization than V161R (30). The use of V161E helps control for effects caused by defects in DNA binding. This observation is consistent with the notion that although TBP dimer dissociation might not be rate-limiting for activated pol III transcription, potentially because of the activity of Brf1, it can be at least partially rate-limiting at pol III promoters when they are not activated.

DISCUSSION
Through interactions with promoter-bound TFIIIC, Brf1 plays an important role in loading TBP onto the promoters of TATAless tRNA and other pol III-transcribed small RNA-encoding genes. Inasmuch as TBP has a relatively high affinity for nonspecific DNA (46) and in this capacity can assemble pol II transcription complexes, the cell is likely to employ a number of mechanisms to prevent this nonproductive and potentially detrimental assembly. Our previous studies (26,30,37,39,40) suggest that TBP dimerization represents a physiologically important auto-inhibitory mechanism by which unregulated binding of TBP to DNA is prevented. If so, then there are likely to be mechanisms that dissociate TBP dimers as a prelude to productive binding. In the pol II system, we have found that TFIIA  1 and 2) or TATA (lanes 3-13) oligonucleotide at 22°C for the indicated time. Oligonucleotides were 28 base pairs and contained either a strong TATA box TATAAAAG or mutant version TAAGAAAG (43). Similar reactions were preformed in B except the DNA was replaced by either buffer (lanes 1 and 2) or 110 nM Brf1 (lanes 3-13). BMH (1 mm) was then added for 15 s at 30°C and then quenched with 2ϫ PSB. Cross-linked TBP dimers and uncross-linked TBP were separated by SDS-polyacrylamide gel electrophoresis and TBP analyzed by immunoblotting. Representative data are shown with the uncross-linked TBP signal shown at a low autoradiographic exposure, which being in the linear response range was used for quantification.
FIG. 6. Effect of TBP mutants on pol III transcription. A, reverse transcriptase-PCR reactions were performed on derivatives of S. cerevisiae strains containing a ⌿-WT SNR6 and wild type TBP (41). Lanes 1-6, strains also contain a null TBP and an empty SNR6 vector (lanes 1-3) or wild type SNR6 (lanes 4 -6). Lanes 7-12, strains contained the indicated TBP mutant (null, V161R, V161E) and snr6(⌬BB). Note that the TBP and SNR6 derivatives are present in a wild type TBP and SNR6 background. Null TBP is a ϩ1 frameshift (40). B, quantitation of 12 replicates of data presented in A in which the median ratio of snr6(⌬BB)/ ⌿-WT, normalized to null TBP, is graphed. Standard errors are presented.
dissociates TBP dimers, thereby enhancing the binding of TBP to DNA (29). The regulation of TBP dimerization in the pol II system prompted us to examine whether the same was true in the pol III transcription machinery. Brf1 is a prime candidate to alleviate TBP dimerization because it directly binds TBP and is responsible for loading TBP onto pol III promoters.
Our findings indicate that Brf1 binds to TBP dimers and accelerates the kinetics of dimer dissociation by a factor of at least seven. Strikingly, the high affinity carboxyl-terminal TBP binding domain of Brf1 is not sufficient to promote dimer dissociation but in addition requires that the TFIIB homology domain be attached. The carboxyl-terminal domain of Brf1 binds to the convex surface of TBP as defined by mutagenesis and x-ray crystallography (25,43,44). The crystallographic structures of the Brf1 (437-506)-TBP-TATA and TBP dimer complexes indicate that TBP-Brf1 (437-506) interactions do not overlap with the TBP dimer interface. Brf1 (437-506) meanders along the upper convex and amino-terminal stirrup of TBP, whereas the TBP dimerization interface occupies its concave surface and carboxyl-terminal stirrup. Our finding that Brf1 (263-596) binds TBP but does not inhibit dimerization is consistent with these interactions.
The carboxyl-terminal stirrup of TBP binds TFIIB within a TBP/DNA complex. Because TFIIB is homologous to the aminoterminal half of Brf1, it is possible that this portion of Brf1 binds TBP in a similar manner. Potential interactions of the TFIIB homology domain along the TBP carboxyl-terminal stirrup have been reported (30,37). Other studies, which demonstrate that the primary interactions of Brf1 are along the amino-terminal stirrup, did not address the location of the TFIIB homology domain (43,44). Fig. 7 illustrates one speculative mechanism by which Brf1 might induce TBP dimers to dissociate. Brf1 is proposed to make primary contact along the amino-terminal stirrup and convex surface of TBP dimers, away from the dimer interface. The amino-terminal TFIIB homology domain of Brf1 induces TBP dimers to dissociate through interactions that might overlap with the dimerization interface of TBP. One possible interaction site might be the TBP carboxyl-terminal stirrup if TFIIB is a precedent. Interactions of the TFIIB-homology domain anywhere near or along the concave surface of TBP, including the amino-terminal stirrup, could in principle dissociate dimers (illustrated as such in Fig. 7 for simplicity). This aspect of the model remains quite speculative because it is not known exactly where the TFIIB homology domain binds on TBP if it binds TBP at all. Any interactions of TFIIB along the concave surface would likely need to be removed upon DNA binding along this region. Potential release of the TFIIB-homology domain when TBP binds DNA could allow this domain to make contact with downstream DNA regions and facilitate open complex formation (23,24). FIG. 7. Model for how Brf1 might dissociate TBP dimers. In this model, the Brf1 carboxyl-terminal domain binds TBP dimers along the exposed convex surface of TBP dimers. The TFIIB homology region is then proposed to dissociate dimers into monomers, possibly through interactions with one or the other stirrup of TBP. Brf1 is denoted as B.