Cloning and functional characterization of the gene encoding the TFIIIB90 subunit of RNA polymerase III transcription factor TFIIIB.

The yeast RNA polymerase III (pol III) general transcription factor TFIIIB is composed of three subunits; the TATA-binding protein (TBP)1, the TFIIB-related factor (BRF1), and a third factor termed TFIIIB90 or B". Here we report the purification of yeast TFIIIB90, cloning of the gene encoding TFIIIB90, and reconstitution of TFIIIB from recombinant polypeptides. The TFIIIB90 open reading frame encodes a 68-kDa polypeptide and has no obvious similarity to any other known protein sequences. The gene encoding TFIIIB90 is essential for viability of yeast. Using recombinant TFIIIB subunits, we found that TFIIIB90 interacts weakly with TBP in the absence of BRF1, and that this interaction is enhanced at least 25-fold by BRF1. In addition, TFIIIB90 showed pol III specificity as it could not interact with the pol II-specific TFIIB-TBP-DNA complex. To localize the regions of the TBP-DNA complex that interact with BRF1 and TFIIIB90, we tested whether the pol II factors TFIIA and TFIIB interfered with the binding of BRF1 and TFIIIB90 to TBP-DNA. Our results suggest that the binding sites for BRF1 and TFIIIB90 on TBP-DNA both overlap the binding sites for TFIIA and TFIIB.

The yeast RNA polymerase III (pol III) general transcription factor TFIIIB is composed of three subunits; the TATA-binding protein (TBP) 1 , the TFIIB-related factor (BRF1), and a third factor termed TFIIIB90 or B؆. Here we report the purification of yeast TFIIIB90, cloning of the gene encoding TFIIIB90, and reconstitution of TFIIIB from recombinant polypeptides. The TFIIIB90 open reading frame encodes a 68-kDa polypeptide and has no obvious similarity to any other known protein sequences. The gene encoding TFIIIB90 is essential for viability of yeast. Using recombinant TFIIIB subunits, we found that TFIIIB90 interacts weakly with TBP in the absence of BRF1, and that this interaction is enhanced at least 25-fold by BRF1. In addition, TFIIIB90 showed pol III specificity as it could not interact with the pol II-specific TFIIB-TBP-DNA complex. To localize the regions of the TBP-DNA complex that interact with BRF1 and TFIIIB90, we tested whether the pol II factors TFIIA and TFIIB interfered with the binding of BRF1 and TFIIIB90 to TBP-DNA. Our results suggest that the binding sites for BRF1 and TFIIIB90 on TBP-DNA both overlap the binding sites for TFIIA and TFIIB.
Binding of eukaryotic RNA polymerase III (pol III) 1 to a promoter requires the ordered assembly on DNA of the multisubunit factors TFIIIC and TFIIIB, as well as TFIIIA in the case of 5 S rRNA genes (reviewed in Refs. 1 and 2). TFIIIC functions in recruitment and positioning of TFIIIB to DNA upstream of the transcription start site. After recruitment of yeast TFIIIB in vitro, TFIIIC can be artificially removed from its intragenic binding site leaving TFIIIB capable of directing multiple rounds of transcription (3). In fact, the yeast U6 gene, which contains a TATA box, can be transcribed by TFIIIB in vitro in the absence of TFIIIC (4 -6). In general, however, TFIIIB binding exhibits little sequence specificity, and proper positioning of TFIIIB is mediated by TFIIIC.
Yeast TFIIIB appears to consist of three factors: the TATAbinding protein (TBP), the TFIIB-related factor BRF1 (also known as TDS4 or PCF4; Refs. [7][8][9], and a protein termed TFIIIB90 or (BЉ), which is chromatographically separable from TBP and BRF1 and migrates on SDS gels with an apparent molecular mass of 90 kDa (10). TFIIIB90, which had been renatured from an SDS gel slice, has been shown, in conjunction with recombinant TBP and BRF1, to reconstitute the activities characteristic of TFIIIB (11). These activities include the ability to bind to TATA⅐TBP⅐BRF1 complexes assembled on the U6 TATA box, to bind to tRNA promoters in a TFIIIC-dependent fashion, and to reconstitute pol III transcription along with recombinant TBP, BRF1, and purified TFIIIC and pol III.
Here we report purification of the yeast TFIIIB90 polypeptide and isolation of the gene encoding TFIIIB90. Recombinant TFIIIB90 in conjunction with TBP and BRF1 reconstitutes the expected activities of TFIIIB in vitro. In addition, recombinant TFIIIB90 allowed us to test for specific interaction between TFIIIB90 and TBP, BRF1, and the pol II-specific factor TFIIB.
Recently, Kassavetis and colleagues also reported cloning of the gene encoding TFIIIB90 and found that the recombinant protein possessed the anticipated properties of TFIIIB90 in vitro (12).

MATERIALS AND METHODS
Purification of Recombinant BRF1-30 liters of Escherichia coli containing BRF1-His 6 under control of the T7 promoter (8) were grown to an A 600 of 0.7 and induced with 0.5 mM isopropyl-1-thio-␤-galactopyranoside for 2 h. Cells were lysed by incubation in 700 ml of Buffer A (6 M guanidine HCl, 0.1 M NaH 2 PO 4 , pH 8) at room temperature for 1 h. The lysate was spun at 4000 ϫ g for 10 min, and the supernatant was batch-bound to 10 ml of Ni-NTA-agarose (Qiagen) for 1 h at room temperature. After washing the resin in binding buffer, BRF1 was eluted in Buffer B (6 M guanidine HCl, 0.1 M NaH 2 PO 4 , pH 4.5). Fractions containing full-length BRF1 were pooled and an equal volume of buffer M added (4 M guanidine, 4 M urea). This fraction was split into two equal fractions, which were separately loaded to a 1.6-ml Poros R2-10 reverse phase column equilibrated in 0.1% trifluoroacetic acid, 9% acetonitrile and eluted with a 40-column volume gradient from 9% to 90% acetonitrile. BRF1 eluted at about 40% acetonitrile. Fractions containing full-length BRF1 were pooled and dried at room temperature under vacuum. Protein was resuspended in 6 M guanidine HCl, 0.1% Brij 58, 10% glycerol, 2 mM DTT, and renatured by dialysis against 40 mM HEPES, pH 7.8, 5 mM MgCl 2 , 1 mM EDTA, 10% glycerol, 0.1 M KCl, and 0.1% Brij 58. Approximate yield is 2 mg of ϳ90% pure BRF1.
Yeast Whole Cell Extract-30 liters of yeast strain BJ5626 (13) was grown at 30°C in YEP (1% yeast extract, 2% Bactopeptone) containing 3% dextrose to an A 600 of about 7.0. Cells (ϳ400 g) were harvested and resuspended in 500 ml of 50 mM Tris, pH 7.5, 30 mM DTT and incubated at 30°C with gentle shaking for 15 min. Cells were spun at 1600 ϫ g for 10 min and resuspended in 200 ml of YPD/S (1% yeast extract, 2% Bactopeptone, 2% dextrose, 1 M sorbitol). 150 ml of 2 M sorbitol and 150 ml of recombinant lyticase (ϳ1 mg/ml) (14) were added and cells incubated at 30°with gentle shaking until about 70% of cells were converted to spheroplasts (typically 1.5-2 h). 1 liter of YPD/S was added and cells * This work was supported by Grant GM53451 from the National Institutes of Health, a Leukemia Society Scholar award (to S. H.), and a Human Frontier Science Program fellowship (to S. R.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U37533.
Purification of TFIIIB90 -A whole cell extract was made as above from 1. The fractionation on HS-20 was repeated multiple times until all the DEAE flow-through fraction was fractionated on HS-20. The peak of TFIIIB90 activity (100 mg) was pooled, concentrated in a Centriprep-10 (Amicon), and diluted to a conductivity equivalent to D ϩ 0.1 M KCl. This fraction was split into three equal volumes, and these fractions were bound to a 1.6-ml Poros HE-20 column equilibrated in D ϩ 0.1 M KCl. The column was eluted with a 40-column volume gradient between 0.1 and 1.0 M KCl. TFIIIB90 activity eluted at 0.45 M KCl. Fractions containing TFIIIB90 activity were pooled and an equal volume of Buffer M added. The sample was split into two fractions, and each fraction was loaded to a Poros R2-10 reverse phase column equilibrated in 0.1% trifluoroacetic acid, 9% acetonitrile and eluted with a 40-column volume gradient between 9% and 90% acetonitrile. TFIIIB90 activity eluted at about 36% acetonitrile. Fractions were dried under vacuum and separated by SDS-PAGE. Final yield of TFIIIB90 was ϳ4 g.
Trypsin Digestion, HPLC Separation, and Microsequencing-Purified proteins were separated by SDS-PAGE and electrotransferred to polyvinylidene difluoride membrane. Proteins were visualized by staining with Ponceau S. The band corresponding to TFIIIB90 (90 kDa) was excised and submitted to in situ digestion with trypsin (15). The resulting peptide mixture was separated by narrow-bore high performance liquid chromatography using a Vydac C18, 2.1 mm by 150-mm reversephase column on a Hewlett-Packard 1090 HPLC/1040 diode array detector. Optimum fractions from the chromatogram were chosen based on differential UV absorbance at 210, 277, and 292 nm, peak symmetry, and resolution. Peaks were further screened for length and homogeneity by matrix-assisted laser desorption time-of-flight mass spectrometry on a Finnigan Lasermat 2000 (Hemel, United Kingdom), and selected fractions were submitted to automated Edman degradation on an Applied Biosystems model 477A sequencer (Foster City, CA). Details of strategies for the selection of peptide fractions and their Edman microsequencing have been described previously (16). Alternatively, tryptic peptide sequences were determined by electrospray ionization/tandem mass spectrometry on a Finnigan TSQ700 (San Jose, CA) triple quadrupole mass spectrometer as described (17).
Isolation, Mapping, and Sequencing of the Gene Encoding TFIIIB90 -The peptide sequences (see "Results") were used to design degenerate primers for use in PCR amplification. Oligonucleotides corresponding to peptides 2 and 3 amplified a DNA fragment of 600 bp. The resulting 600-bp fragment was used to probe colonies containing a yeast genomic library constructed in vector YEP24 (18). Eight clones were obtained, all of which contained an identical 8-kb insert. After restriction mapping analysis, the 8-kb insert was subcloned as three fragments, a 4-kb ApaI fragment, a central 3-kb ApaI-SalI fragment, and a 2-kb SalI fragment, into corresponding sites in pBluescriptSKII or pBluescriptK-SII (Promega), generating plasmids pKS/Apa4, pSK/ApaSal3, and pKS/ Sal2, respectively. Restriction mapping analysis identified the 600-bp TFIIIB90 fragment in pSK/ApaSal3. Automated DNA sequencing was performed, extending outward in each direction from the 600-bp region. The TFIIIB90 gene straddled the 3-kb ApaI-SalI fragment and the 4-kb ApaI fragment. 2522 base pairs of the TFIIIB90 gene were sequenced on both strands, which included 580 bp of sequence upstream of the translation start codon, the open reading frame, and 158 bp of sequence downstream of the stop codon.
Construction of Expression Plasmids-As the TFIIIB90 gene was divided into pKS/Apa4 and pSK/ApaSal3, a Bluescript derivative containing a continuous version of the gene was made prior to construction of expression plasmids. pKS/Apa4 was mutagenized to generate a XhoI site 486 bp upstream of the methionine start codon, and a BamHI site was inserted 108 bp downstream of the stop codon in pSK/ApaSal3. Oligonucleotides for mutagenesis were: 3bXho1, 5Ј-CGAACACTGGCT-CACTCGAGATTCTTTCGGAATCGGG; 3bBamHI, 5Ј-TGAAGCGAT-GTTCGGAAGGATCCATTGGCAACGAAAACAG. XhoI-ApaI and ApaI-BamHI fragments from the resulting constructs were inserted into the XhoI and BamHI sites of pBluescriptKSII by three-way ligation, to generate pKS/XB-TFIIIB90. In pKS/XB-TFIIIB90, the sequence around the start codon was mutagenized to create an NcoI site with or without a 6-histidine tag at the start of the open reading frame. Oligonucleotides for this mutagenesis were 3bNco6his and 3bNco, respectively. Where the 6-His tag was omitted from the N terminus, it was inserted at the end of the open reading frame immediately upstream of the stop codon, using oligonucleotide 3bC6his. The sequences of the oligonucleotides were: 3bNco6his, 5Ј-CCACTTTTATTAACAATAC(GTG) 6 ACCCA-TGGATACCTGGTAATCAGTGGC; 3bNco: 5Ј-CCACTTTTATTAACA-ATACTACCCATGGATACCTGGTAATCAGTGGC; 3bC6his: 5Ј-GTAT-GCATATAAATGTCTCTTA(GTG) 6 ATCAATCTCAGGCTCTTC.
NcoI-BamHI fragments from the mutagenized derivatives of pKS/ XB-TFIIIB90 were subcloned into corresponding sites of the pET11d (Novagen) expression vector, downstream of the T7 promoter to generate expression plasmids pET11d/NcoN6His and pET11d/NcoC6His. (19) harboring the pET11d/NcoN6His and pET11d/NcoC6His expression constructs were grown to an A 600 of 0.5 and induced for 150 min with 0.4 mM isopropyl-1-thio-␤-galactopyranoside. Induction of expression of the TFIIIB90 gene was detectable by SDS-PAGE of a crude bacterial lysate, and the recombinant protein was found to be soluble. Cells were harvested, and pellets resuspended in 150 ml of lysis buffer (20) and lysed by sonication. The lysate was cleared by centrifugation, and the supernatant (150 ml) was incubated with 20 ml of a 50% slurry of Ni 2ϩ -NTAagarose (Qiagen) for 2 h at 4°C. The slurry was applied to a column and washed and eluted as described (20). Fractions containing peak TFIIIB90 concentrations as determined by SDS-PAGE were pooled and dialyzed into a buffer consisting of 20 mM HEPES, pH 7.8, 1 mM EDTA, 10% glycerol, 1 mM DTT, 1 mM PMSF, and 100 mM KCl. This material was loaded to a 1.6-ml Poros HS-20 column equilibrated in the dialysis buffer. The column was washed with 10 column volumes of the same buffer and eluted with a 30-column volume gradient from 0.1 M to 1.0 M KCl. TFIIIB90 activity eluted at about 0.6 M KCl. The protein was about 40% pure (Fig. 3A) and the final yield was ϳ1 mg.

Expression of the TFIIIB90 Gene in E. coli and Purification of the Recombinant Protein-20 liters of BL21(DE3)pLysS cells
Mobility Shift Assays-The mobility shift experiments in Figs. 3B and 4A were performed as described previously (21), except that for the experiments in this paper BRF1 and TFIIIB90 were prepared as described above, and the gel and running buffers contained 1.5 mM MgOAc. In Figs. 3C and 4B, TFIIIB⅐DNA and TFIIB⅐TBP⅐DNA complexes were detected using the identical binding buffer and overall method as in Fig. 3B, except that magnesium and EDTA were omitted from the gel and running buffers, and the gel and running buffers contained TG (25 mM Tris, 190 mM glycine, pH 8.3) instead of TBE.
In Vitro Transcription-pol III transcription was performed by the method described for Pol I transcription (22) except that the templates were unlinearized pGE2wt (23) carrying the tRNA 3 Leu gene and pCH6 (24) containing the U6 gene, each at 30 g/ml, and ␣-amanitin was used at 10 g/ml. Additionally, transcription was performed with the combination of recombinant proteins and fractions described in the figure legends.
Disruption of TFIIIB90 Gene-Disruption of the gene encoding TFIIIB90 was accomplished by replacing the 1688 base pairs between the Mlu1 and NdeI sites in pKS/XB-TFIIIB90 with a XhoI fragment containing the TRP1 gene, from pSH398. 2 This was done by ligating a SalI linker to blunt-ended MluI and NdeI sites, followed by ligation to the XhoI fragment. The resulting plasmid was linearized with NsiI and BglI and transformed to the diploid yeast strain YPH501 (25). The substitution, which resulted in removal of amino acids 1-511, was confirmed by PCR and Southern blotting analysis. Disruptants were transformed with the URA ϩ YEP24 derivative isolated from the library screen for the TFIIIB90 gene. Transformants were sporulated, dissected tetrads were found to segregate 2:2 for the TRP ϩ phenotype, and all TRP ϩ haploids were also shown to be URA ϩ . Mating tester strains were employed to confirm the haploidy of the cells. The URA3 plasmid was replaced by plasmid shuffle with a LEU2 plasmid carrying the TFIIIB90 gene. Cells in which the LEU2 vector contained the TFIIIB90 gene survived on 5-fluoroorotic acid, while those carrying the vector alone did not. The LEU2 plasmid used to shuffle out the URA3 plasmid was constructed by cloning the XhoI-BamHI fragment from pKS/XB-TFIIIB90 into corresponding sites in pRS315 (25) to generate p315/XB-TFIIIB90.

RESULTS
Purification of Yeast TFIIIB90 -TFIIIB90 activity was purified from yeast whole cell extracts as detailed under "Materials and Methods" (Fig. 1A). Activity was monitored by TFIIIB complex formation in the presence of recombinant TBP and BRF1 at the yeast U6 promoter using a mobility shift assay (Fig. 1C) and by complementation of the 0.5 M heparin fraction for transcription of the SUP4 tRNA gene (data not shown and Fig. 5). A key step in the purification was fractionation by reverse phase chromatography under denaturing conditions. Renaturation of these reverse phase fractions and assay of activity by mobility shift assay showed that a ϳ90-kDa polypeptide exactly co-fractionated with TFIIIB90 activity. In addition, elution and renaturation of the 90-kDa polypeptide from SDS gels also reconstituted TFIIIB complex formation (not shown). Chromatography of TFIIIB90 transcription activity exactly paralleled mobility shift activity from both the Poros HS-20 and Poros HE-20 columns. Unexpectedly, complementation of transcription activity at the SUP4 tRNA promoter was not reconstituted with denatured and renatured TFIIIB90 from the reverse phase column, even when activity was normalized to undenatured TFIIIB90 using the mobility shift assay. This is in contrast to an earlier report where TFIIIB90 transcription activity was recovered by renaturation from SDS gels (11). The reason for this difference is unknown but might be due to the difference in purity of components used to reconstitute transcription.
Cloning and Expression of the Gene Encoding Yeast TFIIIB90 -After reverse phase chromatography, fractions containing TFIIIB90 activity were fractionated by SDS-PAGE and electrotransferred to polyvinylidene difluoride membrane. After staining and excision of the 90-kDa polypeptide, TFIIIB90 was subjected to tryptic digestion, HPLC separation, mass spectrometry, and amino acid sequence analysis. The sequence of three peptides were determined: 1) ADVIEDNVTLKPAP-LQTHR; 2) ALSMWGTDFNLISQLYPYR; 3) EADENENYVI-SK (Fig. 2). Degenerate primers designed from these sequences were used to PCR amplify a probe that was used to screen a yeast genomic library (see "Materials and Methods"). A clone containing an open reading frame consistent with the peptide sequences was obtained and sequenced (Fig. 2). The open reading frame predicted a protein of 68 kDa, compared with the approximately 90-kDa size of TFIIIB90 on SDS gels. A search of the DNA sequence for the splicing signal of consensus TAC-TAAC did not reveal potential introns. The protein has a calculated isoelectric point of 6.93, and its middle region contains a stretch of charged amino acids. The tBLASTn computer alignment program (26) was used to search the DNA sequence data bases for similarities with TFIIIB90. The gene encoding TFIIIB90 did not share sequence identity or bear significant homologies with any known sequences. Sequencing also revealed that the TFIIIB90 gene lies in close proximity to the gene IDH1 (encoding isocitrate dehydrogenase) which was previously mapped to chromosome XIV (27). IDH1 was present in the antisense orientation to the gene encoding TFIIIB90, with the stop codons of the two genes located approximately 1 kb apart.
Recombinant TFIIIB90 protein was produced in E. coli under control of the T7 promoter. For ease of purification, the open reading frame was tagged with a 6-histidine moiety at either the N or C terminus prior to its subcloning into the E. coli PET11d expression vector. The expressed N-terminally tagged protein was purified under native conditions by a combination of nickel affinity chromatography followed by further separation on a cation exchange resin. This N-terminally tagged and purified version of the protein was used in the experiments in this paper except where stated otherwise. For the sake of comparison, we also tested the C-terminally tagged derivative, purified over the nickel affinity column (see Fig. 5). Despite its predicted molecular mass of 68 kDa, recombinant TFIIIB90 migrated aberrantly as 90 kDa using SDS-PAGE (Fig. 3A). Mobility Shift Assays of TBP, BRF1, TFIIB, and TFIIIB90 -To confirm the identity of the cloned gene as encoding a TFIIIB subunit, mobility shift assays were performed in which recombinant TBP, BRF1, and TFIIIB90 were bound to a probe comprising the TATA-containing yeast U6 promoter. The mobility of the ternary complex was compared with that obtained using the highly purified R2-10 native yeast TFIIIB90 fraction (Fig. 1). Successive addition of TBP, BRF1, and native TFIIIB90 to the probe produced complexes of decreasing mobility concurrent with incorporation of each protein into the complex (Fig. 3B, lanes 1-4). For reasons unknown to us, we consistently observed a doublet of bands produced by binding of TBP and BRF1 to this probe. Binding of native TFIIIB90 was dependent on inclusion of TBP and BRF1 in the assay (data not shown). In conjunction with TBP and BRF1, recombinant TFIIIB90 (rTFIIIB90) produced a ternary complex of equivalent mobility to that formed with native TFIIIB90 (compare lanes 3 and 4 with lanes 5 and 6). These results show that TBP, BRF1, and recombinant TFIIIB90 reconstitute the gel shift activity of TFIIIB.
Binding of rTFIIIB90 at concentrations sufficient for it to induce the observed mobility shift required inclusion of both TBP and BRF1 (lanes 7 and 8 and data not shown). However, at higher concentrations of rTFIIIB90, a TFIIIB90-TBP-DNA complex was observed in the absence of BRF1 (Fig. 3, B, lane 10, and C, lane 7). This complex was specific for addition of TBP (lanes 7 and 8). Since less than 12 ng of TFIIIB90 was sufficient to induce a mobility shift in the presence of BRF1 (lane 5 and data not shown) and 300 ng was sufficient in the absence of BRF1, this result shows that BRF1 increases the affinity or stability of TFIIIB90 for the complex by at least 25-fold. The interaction between TBP and TFIIIB90 is consistent with our finding that native TFIIIB90 bound specifically to an Affi-Gel TBP affinity column; in an analogous experiment TFIIIB90 did not bind to a BRF1 affinity column. 3 Since the pol II factor TFIIB is related to BRF1 (8), we explored the specificity of TFIIIB90 by testing whether it could bind to a TATA⅐TBP⅐TFIIB complex. As we have shown previously (21), TFIIB bound a TBP⅐TATA complex assembled on the U6 promoter (Fig. 3C, lane 2). However, as much as 60 ng of recombinant TFIIIB90 did not stably bind the TBP⅐TFIIB complex, whereas 5-fold less TFIIIB90 was sufficient for the formation of the TFIIIB⅐DNA complex under the same conditions (lanes 3 and 4 compared with lanes 5 and 6). In other experiments as little as 2.5 ng of TFIIIB90 was sufficient to produce a TFIIIB⅐DNA mobility shift (not shown).
Mutually Exclusive Binding of BRF1⅐TFIIIB90 and pol II Factors to TBP-We attempted to identify regions on TBP that interact with the subunits of TFIIIB. We took advantage of the fact that the crystal structures of TBP and TFIIB are solved (28 -30), and areas of TBP and DNA likely to interact with yeast TFIIA have been suggested by mutagenesis and footprinting experiments (31)(32)(33)(34)(35). We tested whether TFIIA or TFIIB can bind to a TFIIIB-DNA complex in a mobility shift experiment. If binding occurs this would suggest that TFIIA and/or TFIIB bind to different surfaces of TBP and DNA than do BRF1 and TFIIIB90. Alternatively, if these proteins bind in a mutually exclusive manner, this would suggest that the TBP and/or DNA contacts made by BRF1 and/or TFIIIB90 lie in 3  regions overlapping those that interact with TFIIA or TFIIB.
In a mobility shift experiment, the mobility of the TFIIA⅐ TBP⅐DNA complex was distinguishable from that of the TFIIIB⅐DNA complex (Fig. 4A, lanes 3 and 5). TFIIIB was allowed to form a stable complex with the U6 probe for 30 min, then TFIIA was added and the reaction allowed to proceed for an additional 15 min. Addition of TFIIA to the TFIIIB⅐DNA complex did not produce a TFIIA⅐TFIIIB⅐DNA complex; moreover, the amount of the TFIIA⅐TBP⅐DNA complex was reduced (lane 4). The same result was obtained when TBP and TFIIA were prebound to the template and BRF1 and TFIIIB90 were added later (data not shown). A similar experiment was performed using amino acids 110 -345 of yeast TFIIB (⌬TFIIB), which corresponds to the version of human TFIIB used to determine its crystal structure. ⌬TFIIB did not bind to a preformed TFIIIB⅐DNA complex (Fig. 4B, lane 3). Furthermore, the amount of the ⌬TFIIB⅐TBP⅐DNA complex was severely reduced by TFIIIB, whether TFIIIB (lane 3) or ⌬TFIIB (data not shown) was prebound to the template. These results suggest that BRF1 and/or TFIIIB90 bind to regions on TBP and/or DNA overlapping those bound by ⌬TFIIB and TFIIA.
Since TFIIIB⅐DNA complexes did not bind TFIIA and ⌬TFIIB, we examined the individual effects of TFIIIB90 and BRF1. The mobilities of TBP⅐BRF1 or TBP⅐TFIIIB90 complexes were similar to that of the TBP⅐TFIIA complex (Fig. 4A, lanes  2, 5, and 7). Therefore, we could not conclude whether these proteins competed with each other for binding to TBP⅐DNA complexes. However, TFIIA did not detectably bind to and retard a preformed TBP⅐BRF1⅐DNA complex (lane 6) or a TBP⅐TFIIIB90⅐DNA complex (lane 8). The same result was obtained when TFIIA was added prior to BRF1 or TFIIIB90 (data not shown). In addition, ⌬TFIIB did not bind to a TBP⅐TFIIIB90 complex and binding of both ⌬TFIIB and TFIIIB90 was mutually inhibitory (Fig. 4B, lanes 6 -8). The simplest explanation of these results is that both TFIIIB90 and BRF1 contact surfaces of TBP and/or DNA that overlap regions which interact with TFIIA and TFIIB.
Reconstitution of TFIIIB Activity in pol III Transcription-As further proof of its identity as a TFIIIB subunit, rTFIIIB90 was tested for transcriptional activity in conjunction with recombinant TBP and BRF1, and a yeast fraction consisting of partially purified TFIIIC and pol III (heparin 0.5 M, Fig.  1). Although a low level of TFIIIB90 activity detectably con- The silver-stained gel shows 5 l of a preparation, which was obtained by purification of the TFIIIB90 protein, tagged at its N terminus with a 6-histidine moiety over Ni-NTA-agarose and Poros HS-20 resin. The mobilities of protein molecular mass standards are shown on the left. The arrow indicates the band corresponding to full-length TFIIIB90 polypeptide. B, recombinant TFIIIB90 reconstitutes the gel retardation activity of TFIIIB in the presence of TBP and BRF1. The probe consisted of yeast U6 promoter sequences from Ϫ82 to ϩ33 relative to the transcription start site. 1 ng of the conserved portion of TBP and 3 ng of BRF1 were incubated with the probe and the indicated amounts of the Poros R2-10 fraction containing yeast TFIIIB90 (Fig. 1A), or the recombinant TFIIIB90 preparation shown in panel A of this figure. The gel and running buffers contained 1.5 mM MgOAc (see "Materials and Methods"). The positions of protein-DNA complexes are marked by arrows. C, TFIIIB90 does not bind to a DNA-TBP-TFIIB complex. A gel retardation experiment is shown, in which the U6 probe was incubated with the indicated proteins as in B. The preparation of yeast TFIIB was described previously in Ref. 42. Magnesium and EDTA were omitted from the gel and running buffers, and the gel and running buffers contained TG (25 mM Tris, 190 mM glycine, pH 8.3) instead of TBE (see "Materials and Methods").

FIG. 4. TFIIA and TFIIB do not bind to TFIIIB or its individual subunits complexed with DNA.
A, mobility shift experiment to test whether TFIIA binds to TFIIIB or individual TFIIIB subunits complexed with the U6 promoter. TFIIIB or its components were bound to U6 DNA for 30 min, following which TFIIA was bound for another 15 min. In lane 5, where TFIIA was bound only to TBP and DNA, TFIIA was added for the same 15-min period as in the other lanes. 2 ng of the conserved portion of TBP and 4 ng of BRF1 were used. 24 ng of recombinant TFIIIB90 was used in lanes 3 and 4, and 300 ng was used in lanes 7 and 8. In lanes 4 -6 and 8 reactions contained 0.15 ng of TFIIA. The gel and running buffer contained 1.5 mM MgOAc (see "Materials and Methods"). The positions of protein-DNA complexes are marked by arrows. B, binding of ⌬TFIIB to TFIIIB or TFIIIB90 complexed with DNA. The experiment was the same as for A, except that 15 ng of ⌬TFIIB was added after the 30-min incubation of TFIIIB or TFIIIB90 with TBP and DNA. ⌬TFIIB was allowed to bind for 15 min in all reactions to which it was added. 24 ng of recombinant TFIIIB90 was used in lanes 2 and 3, and 300 ng was used in lanes 7 and 8. The gel and running buffers contained TG and no magnesium or EDTA as in Fig. 3C. taminated this fraction, this system is nevertheless limiting for TFIIIB90 activity. The activity of the recombinant protein was compared to that of the HE20 fraction containing yeast TFIIIB90. As noted above, although further purification of the protein by reverse phase chromatography permitted reconstitution of gel shift activity, its ability to function in transcription was lost, so the most highly purified yeast TFIIIB90 was not used in this assay. The activities of recombinant and native TFIIIB90 used in this experiment were previously compared by gel retardation assays. Equivalent concentrations of the native TFIIIB90 fraction used in lane 2 and the preparation of recombinant TFIIIB90 used in lanes 8 -11 contained similar gel retardation activities (data not shown). Transcription of the tRNA 3 Leu gene stimulated by this HE20 preparation of yeast TFIIIB90 is shown in lanes 1 and 2 of Fig. 5. Recombinant TFIIIB90 carrying a C-terminal 6-histidine tag stimulated tRNA transcription in a concentration-dependent manner, to levels comparable with those obtained by native TFIIIB90 (compare lanes 1 and 2 with lanes 3-6). Likewise, an N-terminally tagged derivative of the protein reconstituted transcription activity (compare lanes 3 and 8). We noted that addition of higher amounts of recombinant TFIIIB90 resulted in inhibition of transcription (lanes 7 and 9). Recombinant TFIIIB90 also reconstituted transcription of the yeast U6 gene (lanes 10 and  11). These experiments demonstrate that recombinant TFIIIB90 substitutes for the native factor in reconstitution of yeast pol III transcription.
Disruption of the Gene Encoding TFIIIB90 in Vivo-We tested whether the gene encoding TFIIIB90 is required for yeast viability by replacing amino acids 1-511 with the TRP1 locus. A haploid strain containing the disrupted locus and bearing a URA3 vector carrying the TFIIIB90 gene was derived as outlined under "Materials and Methods." The URA3 vector was replaced by plasmid shuffle with either an empty LEU2 vector or one carrying the TFIIIB90 gene. Only cells containing the latter survived on 5-fluoroorotic acid, which selects against the URA plasmid (Fig. 6). These data show that the TFIIIB90 gene encodes one or more functions essential for the growth of yeast. DISCUSSION We have described cloning of the gene encoding the TFIIIB90 subunit of TFIIIB. As expected of an essential TFIIIB subunit, the gene encoding TFIIIB90 is necessary for yeast viability. Recombinant TFIIIB90 is indistinguishable from the native factor in its size upon migration through SDS gels, its ability together with TBP and BRF1 to reconstitute gel shift activity of TFIIIB, and its requirement in transcription by pol III.
Since U6 mobility shift activity is reconstituted using recombinant TBP, BRF1 and TFIIIB90, this activity of TFIIIB is supplied solely by these three proteins. However, it is formally possible that our relatively impure TFIIIC⅐pol III fraction contributes as yet unidentified factors, which are stimulatory or essential for transcription. Our final stage of purification of natural TFIIIB90, which required its denaturation, resulted in loss of its transcriptional but not gel shift activity. One explanation for this result is that the activity of TFIIIB90 in transcription was destroyed by denaturation while its capacity for protein-DNA binding remained intact. However, transcription activity of TFIIIB90 has been shown previously to be renaturable from an SDS gel slice (11). We are currently testing whether our denaturing chromatography conditions destroyed transcriptional activity of yeast TFIIIB90, by using these same denaturing conditions to purify the recombinant protein. Alternatively, denaturation may have caused separation from or inactivation of a factor required for transcription but not gel retardation. For instance such a factor might counteract a putative transcriptional inhibitor in our fractions. One potential candidate might be a factor called TFIIIE, which was recently identified as distinct from TFIIIB90 but required for pol III transcription under certain conditions (36,37).
A number of studies have aimed to delineate the proteinprotein and protein-DNA contacts within the TFIIIB⅐DNA complex. In order of assembly experiments BRF1 was recruited by TFIIIC to the upstream promoter of a tRNA gene; TFIIIB90 could not bind the complex until TBP was bound (11). It was not clear from these studies whether TFIIIB90 made direct contact with TBP or whether alteration in the conformation of BRF1 by TBP (11) mediated binding of TFIIIB90 to the complex. Our mobility shift assays showed that TFIIIB90 has weak but detectable affinity for TBP-DNA in the absence of BRF1; BRF1 increased the affinity or stability of TFIIIB90 in the complex by at least 25-fold. This suggests that TBP and TFIIIB90 contact one another in the TFIIIB complex. Such an interaction is consistent with our finding that native TFIIIB90 bound specifically to an Affi-Gel TBP affinity column (S. Miller, unpublished). In our transcription experiments we observed a delicate balance between the amount of TFIIIB90 required to stimulate versus inhibit transcription. Perhaps at high concentrations, TFIIIB90 bound to TBP⅐DNA complexes in the absence of BRF1, resulting in squelching of transcription through formation of non-productive complexes.
Our binding studies suggest that BRF1 and TFIIIB90 interact with TBP-DNA surfaces overlapping those which contact both TFIIA and TFIIB. It is not yet known whether any parts of TFIIA and TFIIB contact adjacent regions of TBP. The crystal structure of TFIIB bound to TBP and DNA shows that TFIIB contacts the C-terminal stirrup of TBP, helix H1Ј and the C terminus of TBP as well as DNA both upstream and downstream of the TATA box (30,38). Thus BRF1 and TFIIIB90 may bind in a manner overlapping or close to these regions of TBP and/or DNA. Mutations in the linker region of TBP connecting the two repeats of TBP, disrupted the interaction between TBP and TFIIA (31)(32)(33)(34). Interestingly BRF1 was isolated as a suppressor of one of these mutant TBPs (referred to as TDS4; Ref. 7). This overlap between the binding sites of TFIIA and BRF1 on TBP is consistent with our lack of binding of TFIIA to TBP⅐BRF1⅐DNA complexes. Another possibility, which we cannot distinguish by our experiments, is that the surface of TBP contacted by these proteins is non-overlapping but the conformation of TBP or the DNA is altered by BRF1 and TFIIIB90, such that TFIIA and TFIIB are no longer able to bind stably to a TBP⅐DNA complex. We think the latter possibility is unlikely, as TBP has not been observed to undergo any major conformational changes upon protein or DNA interaction. The structure of TBP off DNA, on a TATA box, or in complexes with TFIIB or TFIIA is very similar (28 -30). 4 The specificity of TFIIIB90 binding suggests its contribution toward RNA polymerase specificity in transcription. It has been shown previously that in the absence of the TFIIIC site, the TATA-containing promoter of the yeast U6 gene could be transcribed by pol II (21) or pol III (4 -6) in vitro, depending partly on the transcription system used. This was consistent with the finding that BRF1 and TFIIB bound equally well to U6 TATA⅐TBP complexes in vitro (21). Here we found that in the next step of factor assembly, recombinant TFIIIB90 did not bind detectably to DNA⅐TBP⅐TFIIB complexes. Thus TFIIIB90 can distinguish between complexes containing the related TFIIB and BRF1 factors. This result suggests that TFIIIB90 does not bind aberrantly to TBP⅐TFIIB complexes in vivo, and presumably pol II factors do not bind to BRF1⅐TBP complexes either. In our previous experiments, we found that TFIIIC determined the in vivo pol III specificity of U6 transcription (21). In addition, the pol III-specific C34 subunit was shown to interact with BRF1 but not TFIIB (39,40). Thus, an obvious model for pol III specificity involves a chain of specific protein interaction events whereby TFIIIC recruits BRF1, after which binding of TBP occurs, leading to specific interaction of the complex with TFIIIB90 and binding of the C34 subunit of pol III.
What is the role of TFIIIB90 in pol III transcription? By analogy with the pol II system, it might be expected that a complex of TBP and BRF1, like TBP and TFIIB, would recruit polymerase. The binding of successive factors may effect subsequent steps in transcription. However, a TFIIIC⅐TBP⅐BRF1 complex did not recruit pol III; TFIIIB90 was required to mediate binding of polymerase (41). This seems to rule out a model where TFIIIB90 would function solely at some later stage of initiation. Since, as mentioned above, BRF1 also interacts with polymerase subunit C34, the combination of TFIIIB90 and BRF1 appears to serve in polymerase recruitment. The func-tion of TFIIIB90 may be mediated by altering the conformation of the preinitiation complex, as TFIIIB90 increased photocrosslinking of BRF1 and reduced cross-linking of TBP to a region of the SUP4 tRNA promoter (11). TFIIIB90 and/or BRF1 may then direct a subsequent step in initiation such as open complex formation or promoter clearance of the preinitiation complex. The availability of recombinant TFIIIB will assist analysis of the mechanisms by which its individual components function in transcription.