A Cryptic DNA Binding Domain at the COOH Terminus of TFIIIB70 Affects Formation, Stability, and Function of Preinitiation Complexes*

TFIIIC-dependent assembly of yeast TFIIIB on class III genes unmasks a high avidity of TFIIIB for DNA. TFIIIB contains TATA-binding protein (TBP), TFIIIB90/B", and TFIIIB70/Brf1, which is homologous to TFIIB. Using limited proteolysis, we have found that the COOH terminus of TFIIIB70 (residues 510–596) forms a protease-resistant domain that binds DNA tightly as seen by Southwestern, DNase I footprinting, and gel shift assays. Consistent with a role for this DNA binding activity, preinitiation complexes were formed less efficiently with truncated TFIIIB70 lacking the COOH-terminal domain and displayed an increased sensitivity to heparin. B′ (TFIIIB70 + TBP)·TFIIIC·DNA complexes were also particularly unstable. In addition, TFIIIB·TFIIIC·DNA complexes containing truncated TFIIIB70 were impaired in promoting transcription initiation.

TFIIIC-dependent assembly of yeast TFIIIB on class III genes unmasks a high avidity of TFIIIB for DNA. TFIIIB contains TATA-binding protein (TBP), TFIIIB90/ B؆, and TFIIIB70/Brf1, which is homologous to TFIIB. Using limited proteolysis, we have found that the COOH terminus of TFIIIB70 (residues 510 -596) forms a protease-resistant domain that binds DNA tightly as seen by Southwestern, DNase I footprinting, and gel shift assays. Consistent with a role for this DNA binding activity, preinitiation complexes were formed less efficiently with truncated TFIIIB70 lacking the COOH-terminal domain and displayed an increased sensitivity to heparin. B (TFIIIB70 ؉ TBP)⅐TFIIIC⅐DNA complexes were also particularly unstable. In addition, TFIIIB⅐TFIIIC⅐DNA complexes containing truncated TFIIIB70 were impaired in promoting transcription initiation.
The auxiliary transcription factors required for class III gene activation have been much investigated, especially in yeast. Essentially all the components of the yeast RNA polymerase III (pol III) 1 transcription system have been identified and characterized (1,2). The cascade of interactions leading to transcription complex formation on some prototypical class III genes is known in general outline (2)(3)(4)(5). Two factors, TFIIIB and TFIIIC, are required to direct transcription of yeast tRNA genes. The binding of TFIIIC to the intragenic promoter elements is the primary step of transcription complex assembly. The TFIIIC⅐DNA complex projects its 131 subunit past the transcription start site to promote the binding of TFIIIB at a fixed distance upstream of the start site (6 -9). TFIIIB can not bind to a tRNA gene by itself. It needs to be assembled on tDNA, but the resulting TFIIIB⅐DNA complex is exceptionally stable and resists disruption by heparin or high salt concentrations that dissociate TFIIIC (6,7). Hence, it could be shown that TFIIIB⅐DNA complexes can direct accurate transcription and reinitiation by RNA polymerase III, in the absence of TFIIIC (7). Therefore, TFIIIB is the central initiation factor responsible for proper pol III recruitment.
Yeast TFIIIB comprises three components, the TATA-binding protein (TBP), TFIIIB70/Brf1 homologous to TFIIB, and TFIIIB90/BЉ. TBP was first shown to participate in pol III transcription in the case of the yeast SNR6 gene that contains a TATA box at bp Ϫ30 (10). Specific binding of TBP to the TATA box can direct the correct assembly of TFIIIB in the absence of TFIIIC in vitro (10,11). TBP also interacts with DNA on the TATA-less tRNA genes (9), but the assembly of TFIIIB is then dependent on TFIIIC, via its interaction with TFIIIB70 (12). TFIIIB70 was shown to interact with the 131 subunit of TFIIIC (13,14) as well as with TBP (14,15). The weak TFIIIB70⅐TFIIIC⅐tDNA complex is stabilized after recruitment of TBP and probably structurally modified, as seen by changes in the accessibility of both TFIIIB70 and 131 to DNA cross-linking (12). The BЈ (TFIIIB70 ϩ TBP)⅐C⅐DNA complex is further stabilized and again modified upon binding of BЉ that confers the heparin and salt resistance property typical of B⅐C⅐DNA complexes (12,16). The recruitment of BЉ is probably mediated both by BЈ (17) and the 131 subunit of TFIIIC (18). This succession of interaction steps, artificially decomposed in vitro, might in fact represent the in vivo situation, since both genetic (19,20) and biochemical (15,21) evidence suggests that TFIIIB, when not bound to DNA, is not a stable molecular entity.
The striking stability of the TFIIIB⅐DNA complex contrasts with the lack of affinity of TFIIIB or BЈ alone for TATA-less promoters (6,21,22). It has been proposed that TFIIIB is convertible between two states: one binding very tightly to DNA without sequence specificity, the other hiding the DNA binding domain(s) to avoid random dispersion of TFIIIB on irrelevant sites (5). Cryptic DNA-binding sites have been found to exist in general transcription factors like 70 , K , and 32 (23,24) and RAP30 (25). Neither intact Escherichia coli 70 nor RAP30 exhibits detectable DNA binding activity, but this property is unmasked upon removal of the NH 2 -terminal part of these polypeptides (23,25). A masked DNA-binding potential has also been revealed in the protease-resistant core of TFIIB (26), in the zinc ribbon of TFIIS (27,28), and in protein p53 (29). To understand how TFIIIB can be locked in a highly stable protein⅐DNA complex, we sought for a masked DNA binding domain in TFIIIB70 using limited proteolysis. We found that the TFIIIB70 COOH terminus folds into a protease-resistant domain that binds DNA tightly. Deletion of this domain affects the rate of formation, stability, and function of preinitiation complexes.

EXPERIMENTAL PROCEDURES
Plasmids and Yeast Strains-Plasmid pET3b(rTFIID) expressing TBP was a gift from J.-M. Egly (Institut de Génétique et de Biologie Moléculaire et cellulaire, Strasbourg, France). Plasmid pSH360 expressing TFIIIB70 was kindly provided by S. Hahn (Fred Hutchinson Cancer Research Center, Seattle, WA). Plasmid pCC-10, obtained by cloning of the SalI/XhoI DNA fragment from YpCC7 (see below) into pET28b (Novagen), encodes a TFIIIB70 derivative from residues 1-509 (r⌬C86) tagged with six histidine residues and the T7-TAG epitope (Novagen). PCC-Leu3 was constructed by inserting into pGEM-T (Promega) the tRNA 3 Leu gene from residues Ϫ150 to ϩ228 on an NcoI-EcoRI DNA fragment obtained by polymerase chain reaction.
Oligonucleotide-mediated mutagenesis was performed as described by the manufacturer, using a Muta-Gene kit (Bio-Rad), on uracil-enriched single-stranded YpCC1 DNA (15). The oligonucleotide 5Ј-GGAG-GCAGATATCtaactcgagGCCACAGGTAACAC-3Ј containing sequences complementary to PCF4/TDS4/BRF1 DNA (uppercase) and a stop codon followed by a XhoI restriction site (lowercase) was used to introduce a stop codon at position ϩ1528 of the PCF4 open reading frame. After sequencing inserts of several transformants, the SacI/SmaI mutagenized DNA fragment was cloned into the same sites of a multicopy plasmid creating YpCC7. Centromeric or multicopy plasmids harboring the deleted version of the PCF4 gene were used to transform the SHy76 haploid strain (31) containing the PCF4 gene on a plasmid with the chromosomal copy disrupted. The modified copies of PCF4 were substituted for wild type PCF4 by plasmid shuffling on 5-fluoroorotic aid plates. The resulting strains isolated at 30°C have also been tested for growth at 37 and 16°C.
Protein Purification and Limited Proteolysis-Highly purified RNA polymerase III and affinity-purified TFIIIC were prepared as described by Huet et al. (52) and Gabrielsen et al. (53), respectively. TFIIIB activity was reconstituted from the three components BЉ, TBP, and TFIIIB70. Fraction BЉ was extracted from chromatin pellets and partially purified according to the protocols of Kassavetis et al. (12). Recombinant TBP expressed from pET3b(rTFIID) was purified by the procedure of Burton et al. (54). Recombinant histidine-tagged TFIIIB70 or ⌬C86 was expressed in E. coli cells from pSH360 or pCC-10, purified by chromatography on Ni 2ϩ -NTA-agarose (Qiagen) under native conditions (55). For Southwestern experiments (Figs. 1 and 2) QIAGEN, rTFIIIB70 was further purified by chromatography on a heparin column. The protein fraction (0.67 mg of protein) was adsorbed on a 1-ml heparin column (UGH; Pharmacia Biotech Inc.) equilibrated in buffer A (25 mM Tris-HCl pH8, 0.12 M KCl, 0.2 mM EDTA, 10 mM ␤-mercaptoethanol, 15% glycerol, 1 M leupeptin, 1 M pepstatin). After the column was washed with 8 volumes of buffer A, the proteins were step-eluted with buffer A containing 0.7 M KCl into 250-l fractions. Alternatively, rTFIIIB70 was purified on Ni 2ϩ -NTA-agarose under denaturing conditions followed by Mono S chromatography according to Colbert and Hahn (31). The recombinant 90-kDa component of TFIIIB (TFIIIB90/ BЉ) was purified as described by Rü th et al. (18).
To purify p10, the 10-kDa protease-resistant domain of rTFIIIB70, fast liquid protein chromatography on a Superdex 75 column (Smart System, Pharmacia) was performed on 5 g (50 l) of rTFIIIB70 fraction purified under denaturing conditions. Polypeptides were eluted with a buffer containing 50 mM Tris-HCl, pH 8, 500 mM KCl, and 10% glycerol. About 50 fractions (43 l) were collected and analyzed by SDS-PAGE and silver staining. Fraction 17 (27.5 g/ml) was found to contain most of the 10-kDa polypeptide.
Limited proteolysis of heparin-purified rTFIIIB70 was performed according to the procedure of Cleveland et al. (56) with the following modifications. After SDS-PAGE (13%), the full-length rTFIIIB70 polypeptide was located by Coomassie Blue staining. The gel was washed with water, and the protein band was excised and loaded on a SDS-polyacrylamide gel (13%) together with proteinase K or protease V8 as described except that 2 mM EDTA and 5% glycerol were added in the upper gel (57). The polypeptides generated during the co-migration of rTFIIIB70 with the proteases were transferred onto nitrocellulose and revealed with antibodies directed to TFIIIB70 or probed with labeled tDNA.
Amino-terminal Sequence Analysis-High pressure liquid chromatography was performed using a chromatograph 130-A (Applied Biosystems). Chemicals for buffer preparation were purchased from Pierce (trifluoroacetic acid), J. T. Baker Inc. (acetonitrile), and Merck (isopropyl alcohol). A Millipore system supplied high pressure liquid chromatography quality water. Chromatographic separations were performed at 35°C. Recombinant TFIIIB70 fraction purified under denaturing conditions (10 g) was diluted with water up to 450 l and then applied to an Aquapore RP300 column (2.1 ϫ 30 mm; Brownlee Labs). The column was developed in solvent A with a linear gradient from 25 to 50% of solvent B for 30 min at a flow rate of 200 l/min (solvent A was 0.1% trifluoroacetic acid in water, and solvent B was 0.075% trifluoroacetic acid in 40% acetonitrile, 40% isopropyl alcohol, and 20% water). Peaks of absorbance at 214 nm were collected and analyzed for protein content by SDS-PAGE. For amino-terminal sequence analysis, the fraction containing the purified 10-kDa fragment of rTFIIIB70 was partially dried in an evaporator (Speed-Vac, Savant) and spotted onto a Polybrene-pretreated filter disc. Sequence analysis was performed on an Applied Biosystems model 477-A liquid phase sequenator interfaced with a model 120-A on-line phenylthiohydantoin-amino acid analyzer.
Interaction of rTFIIIB70 with TBP or DNA-Far Western analysis was performed using in vitro synthesized 35 S-labeled TBP as described previously (15). For Southwestern analysis, rTFIIIB70 fractions were subjected to SDS-PAGE and blotted onto nitrocellulose. The filters were then incubated for 1 h at 4°C in buffer B (20 mM Hepes-KOH, pH 7.9, 0.1 mM EDTA, 5% glycerol, 1 mM dithiothreitol, 5 mM MgCl 2 ) containing 100 mM KCl, in the presence of 100,000 cpm/ml of a 32 P-labeled DNA fragment harboring the SUP4 tRNA Tyr gene (270-bp BamHI-BamHI fragment from plasmid pTZ1; Ref. 6). The filters were washed for 10 min at 4°C, three times with buffer B containing 100 mM KCl, and twice with the same buffer without glycerol. Labeled polypeptides were revealed by autoradiography.
In Vitro Transcription Assays-Plasmid DNAs used for in vitro transcription were the following: pRS316-SUP4, containing the yeast SUP4 tRNA gene (a gift from S. Shaaban); pUC-Glu (58) Gel Retardation Assays and Footprinting Analysis-For p10⅐DNA complex analysis by DNase I footprinting, 5 fmol (12,000 cpm) of a 5Ј-end 32 P-labeled DNA fragment carrying the yeast SUP4 gene (212-bp BamHI-SmaI fragment from the pTZ1 plasmid; Ref. 6) were incubated for 30 min at 22°C in 20 l of binding buffer containing 20 mM Tris-HCl, pH 8, 0.5 mM EDTA, pH 8, 100 mM KCl, 10% glycerol with varying concentrations of p10 polypeptide purified by gel filtration. The DNase I treatment was initiated by the addition of 5 mM MgCl 2 , 0.5 mM CaCl 2 , and 6 ng of DNase I (BDH 39101 RNase-free). After 30 s at 22°C, reactions were stopped with 10 mM EDTA, and the cleavage products were purified by two phenol extractions and ethanol precipitations, separated on 8% polyacrylamide gels containing 8 M urea, and revealed by autoradiography.
Preinitiation complexes were assembled on a 284-bp 32 P-labeled XbaI-EcoRI DNA fragment from the pTZ1 plasmid carrying the yeast SUP4 gene. A mixture (10 l) of rTFIIIB90 (30 ng), rTBP (30 ng), TFIIIC (50 ng), and rTFIIIB70 or r⌬C86 (40 -400 ng, as indicated) preincubated for 10 min at 22°C was then incubated for 20 min at 22°C in binding buffer (20-l final volume) with 2 fmol (ϳ3000 cpm) of 32 P-labeled template and 150 ng of pBR322. With the tRNA 3 Leu gene as a template, TFIIIBЈ⅐TFIIIC⅐DNA complexes were formed in the absence of rTFIIIB90, using a 358-bp 32 P-labeled NcoI-EcoRI DNA fragment from the pCC-Leu3 plasmid. When indicated, variable concentrations of heparin (Sigma, H-2149) were added to the reaction mixtures and incubated for 5 min at 22°C. Complexes were analyzed by nondenaturing gel electrophoresis at 4°C in 4% polyacrylamide gels (53).
For two-dimensional footprinting assays, BЈ⅐C⅐DNA complexes were formed as described above with the 5Ј-end labeled tRNA 3 Leu gene as a template. Complexes were then treated with DNase I (6 ng) for 1.5 min at 22°C, and reactions were stopped with 20 mM EDTA. Complexes were separated from free DNA by nondenaturing gel electrophoresis in 4% polyacrylamide gels and located by autoradiography. Retarded and nonretarded DNAs were excised from the native gel, passively eluted in 10 mM Tris-HCl, pH 8, 1 mM EDTA, precipitated with ethanol, analyzed on a 10% polyacrylamide gel containing 8 M urea, and revealed by autoradiography. Fig. 1, preparations of recombinant, histidine-tagged TFIIIB70 purified under denaturing (lane 1) or native (lane 2) conditions contained polypeptides ranging from 10 to 70 kDa. Most of these polypeptides were recognized by antibodies directed to TFIIIB70 (lanes 3 and 4) or to the carboxyl-terminal histidine stretch (lanes 7 and 8). We took advantage of this partial proteolysis of TFIIIB70 to investigate the TBP and DNA binding properties of TFIIIB70 domains. In previous work, we found that TBP interacts with the carboxyl-terminal extension (CTE) of TFIIIB70 (13). Consistent with this result, we found that TBP interacted with carboxyl-terminal fragments of TFIIIB70 larger than 30 kDa (lanes 5 and 6). The same rTFIIIB70 polypeptides were blotted and probed in parallel with a 32 P-labeled DNA fragment bearing the SUP4 tRNA gene. Wild type TFIIIB70 and most of the large polypeptide fragments did not bind tDNA. However, both preparations of rTFIIIB70 contained a 10-kDa fragment that bound the tDNA probe. Note that the 30-kDa subdomain of rTFIIIB70 that bound TBP and corresponded to the carboxylterminal part of the CTE (lane 10) also bound to tDNA. The recombinant fraction enriched in the 10-kDa fragment (lane 1) also interacted with the tDNA probe in gel shift assays to give one retarded band of complex (data not shown). The same retarded band was obtained with the purified 10-kDa fragment (see below). These results confirmed that the entire TFIIIB70 protein is not able to bind DNA by itself and suggested that TFIIIB70 contains a cryptic DNA binding domain.

Recombinant TFIIIB70 Contains a Protease-resistant DNA Binding Domain-As shown in
To ascertain that the 10-kDa DNA binding polypeptide corresponded to a domain of rTFIIIB70 and was not a contaminant E. coli protein, the full-length TFIIIB70 protein was gelpurified and subjected to limited proteolysis (Fig. 2). Proteolytic fragments derived from rTFIIIB70 were revealed with antibodies or probed with labeled tDNA. As shown in Fig.  2, proteinase K or protease V8 generated fragments of TFIIIB70 of various molecular weights that for the most part did not bind DNA. However, two or three polypeptides of low molecular weight were revealed by the tDNA probe (Fig. 2). Cleavage by both proteases gave rise to a DNA binding domain of about 10 kDa that probably corresponded to the recombinant protease-resistant 10-kDa fragment. The 10-kDa polypeptide (p10) from the recombinant TFIIIB70 preparation was purified by gel filtration and used in subsequent work.
TFIIIB-DNA interaction is exceptionally stable to high electrolyte concentrations (6,7). We therefore explored the effects of salt concentration on p10⅐DNA complex formation or stability, in Southwestern experiments. The purified p10 polypeptide was subjected to SDS-PAGE and transferred onto a membrane. Filters were incubated in binding buffer containing labeled tDNA and different concentrations of KCl. The p10 polypeptide bound DNA optimally in the presence of 0.1 or 0.2 M KCl. Complex formation was only partially reduced in the presence of 0.4 M KCl but totally inhibited at 1 M salt. Once formed, the complexes were stable for up to 1 h in 0.4 M KCl but were destroyed by 0.6 M salt, heparin (0.1 mg/ml), or proteinase K (data not shown). Thus, the 10-kDa domain of rTFIIIB70 was capable of forming tightly bound, high salt-resistant complexes.
The p10⅐tDNA complex was further analyzed by DNase I footprinting. The DNase I digestion patterns of the transcribed strand of the SUP4 gene in the presence of increasing amounts of bound p10 polypeptide is shown Fig. 3. p10 did not bind to DNA in a sequence-specific way, since no specific footprint could be precisely observed. Interestingly, protein-bound DNA showed a series of enhanced DNase I cleavage sites, regularly spaced every 15-20 bp (indicated with black dots, for example at bp Ϫ25, Ϫ9, and ϩ14). Many of these cleavage sites were not present in the absence of p10. This was the case at bp Ϫ25, just 3Ј of one of the AT-rich regions present in the upstream sequence of the SUP4 gene (indicated in Fig. 3 as TATA) or at bp ϩ37/ϩ38. This suggests that p10 molecules bound in a regular pattern along the DNA and possibly induced a structural change in the DNA favoring DNase I attack.
To map the protease cleavage site, the p10 polypeptide was purified by high pressure liquid chromatography and subjected to NH 2 -terminal sequence analysis. Based on sixteen consecutive residues that could be clearly determined, p10 was a well defined COOH-terminal fragment starting at A510 and encompassing the last 87 amino acids of TFIIIB70. The precise proteolytic cleavage of TFIIIB70 COOH-terminal domain suggests a very stable structural domain. The carboxyl-terminal end of rTFIIIB70 from residue 510 to 596 was expressed in E. coli, purified under native conditions on a nickel-agarose column, and tested in Southwestern or gel retardation assays for tDNA binding. All results described with p10 polypeptide purified by gel filtration were reproducibly obtained with recombinant p10 (data not shown). p10 encompasses one of three regions strongly conserved in the CTE of TFIIIB70 from different yeast species (Fig. 4), starting at the end of region II, as defined by Khoo et al. (14). As seen in Fig. 4, sequence similarities could also be found with a human equivalent of TFIIIB70 (30) and with a putative protein of Caenorhabditis elegans. The CTE-(510 -596) corresponds to a region of low sequence similarity with HMG2 in human TFIIIB90 (30). Since sequence conservation was a sign of functional significance, we explored the ability of ⌬C86 (truncated TFIIIB70 deprived of the CTE-(510 -596) sequence) to assemble into stable preinitiation complexes and to direct transcription initiation.
Transcriptional Activity and DNA Binding Properties of the Deleted Form of TFIIIB70 -Previous work has shown that a yeast strain expressing a derivative of TFIIIB70 deleted of 50 residues at its carboxyl-terminal end was thermosensitive and cryosensitive, whereas a larger deletion of 100 residues was lethal (31). To study the phenotype of strains harboring TFIIIB70 truncated from residue 510 to 596 (thereafter named odd-numbered lanes) or native (2 g; even-numbered lanes) conditions was subjected to SDS-PAGE (13% polyacrylamide) and stained with Coomassie Blue (lanes 1 and 2) or transferred to nitrocellulose (lanes 3-10). The filters were incubated with different protein or DNA probes as indicated. 35 S-Labeled TBP and 32 P-labeled tDNA binding were revealed by autoradiography. Immune complexes were revealed by antibodies tagged with alkaline phosphatase. ⌬C86), we constructed this derivative by site-directed mutagenesis as described under "Experimental Procedures." Centromeric or multicopy plasmids harboring the ⌬C86 mutant copy of the TFIIIB70 gene were tested for their ability to functionally replace, at different temperatures, a chromosomally disrupted copy of the gene. We found that the ⌬C86 deletion was thermosensitive at 37°C and cryosensitive at 16°C when the mutant gene was on a multicopy plasmid but lethal at 30°C when expressed from a centromeric plasmid. Therefore, the carboxyl-terminal end of TFIIIB70 was essential for cell viability when ⌬C86 was expressed from a low copy number plasmid from its own promoter.
The ⌬C86-deleted form of TFIIIB70 was tested in vitro for its ability to replace full-length TFIIIB70 for specific transcription of various pol III genes. Fig. 5 shows a comparison of the transcriptional efficiencies of rTFIIIB70 and r⌬C86. Several pol III genes were transcribed in a reconstituted transcription system, in the presence of rTFIIIB70 or r⌬C86 at about the same concentration. The deleted form of TFIIIB70 showed reduced levels of transcription over wild type with all of the templates tested. The difference in transcription efficiency varied from 5-fold for the tRNA 3 Lue gene to 8-fold for the SUP4 or SNR6 genes, and no transcription at all could be detected for the tRNA 3 Ser or tRNA 2 Arg genes. The faint transcript seen with the tRNA 3 Glu gene was not quantified. This variable response of the different templates to truncated TFIIIB70 suggested a critical involvement of their upstream TFIIIB-binding sequences. The SUP4 tRNA gene was chosen to further investigate the reasons for the low transcription efficiency of r⌬C86, since transcription of this gene has been extensively studied in vitro (3). First, varying the concentration of r⌬C86 over a large range (20 -400 ng) did not restore the level of transcription observed with 50 ng of rTFIIIB70 (a maximum of 20% of wild type transcription level was reached); in competition experiments, the addition of an excess of r⌬C86 (140 ng) over rT-FIIIB70 (50 ng) did not inhibit transcription; however, in preemption experiments, where r⌬C86 was first preincubated with all of the factors, the addition of rTFIIIB70 could not restore normal transcription efficiency, although in the absence of BЉ, rTFIIIB70 was dominant over r⌬C86 to form productive BЈ⅐C⅐DNA complexes (data not shown). These results suggested that r⌬C86 was poorly assembled in preinitiation complexes and that complexes containing the deleted form of TFIIIB70 were somehow deficient in transcription initiation. The following experiments were designed to confirm these conclusions.
The ability of r⌬C86 to direct transcription initiation was examined after preassembly of TFIIIB. TFIIIC⅐DNA complexes, formed on a labeled SUP4 tDNA gene, were preincubated for 20 min with rTBP, rBЉ, and increasing amounts of rTFIIIB70 or r⌬C86, before the addition of pol III and nucleotides. Ternary complex formation, monitored by the synthesis of a 17-mer nascent RNA in the absence of GTP, was initiated less efficiently with the deleted r⌬C86 protein than by the wild type (Fig. 6). Increased amounts of pol III or preincubation of pol III with preinitiation complexes before the addition of the three nucleotides did not restore wild type synthesis of the 17-mer product (data not shown). These results suggest that the defect in transcription initiation caused by r⌬C86 was not due to a lower affinity of pol III for the preinitiation complexes that could be compensated by increasing pol III concentration but that it may correspond to differences in the number, stability, and/or conformation of preinitiation complexes.
To test this hypothesis, TFIIIB⅐TFIIIC⅐tDNA complexes were separated from free SUP4 tDNA by native electrophoresis and visualized by autoradiography. As shown in Fig. 7A, the B⅐C⅐DNA complex was fully formed in the presence of 80 ng of rTFIIIB70 (lane 4), whereas 420 ng of r⌬C86 were necessary to obtain a similar result (lane 12). Detectable complex formation with r⌬C86 required 240 ng of protein (lane 11). That concentration of r⌬C86 stabilized all C⅐DNA complexes, as seen by the disappearance of free DNA, but only ϳ50% of the complexes were fully shifted, while an intermediate complex, migrating slightly slower than C⅐DNA complexes, was formed (lane 11). Since we had verified by silver-stained SDS-PAGE and by Full-length rTFIIIB70 (the band labeled 70 in the Western blot of rTFIIIB70) was isolated by SDS-PAGE, excised from the gel, and subjected to co-electrophoresis with proteinase K or V8 protease as indicated. Proteolyzed polypeptides derived from rTFIIIB70 (arrow and bracket) were transferred to nitrocellulose membranes and probed with anti-TFIIIB70 antibodies or 32 P-labeled tDNA as described under "Experimental Procedures." Short arrows show the proteolyzed by-products of rTFIIIB70 that bind DNA.
Western blot analysis that the protein content of both recombinant fractions was similar (data not shown), r⌬C86 was clearly less efficient than wild type TFIIIB70 in assembling B⅐C⅐DNA complexes on the SUP4 tDNA gene. This result accounted in part for the low levels of transcription initiation observed previously and was also consistent with the in vivo phenotype of mutant yeast strains that could survive only when ⌬C86 was overexpressed. Nevertheless, we noted previously that 400 ng of r⌬C86, corresponding to 100% of B⅐C⅐DNA complexes formed on a SUP4 template (lane 12), did not restore wild type levels of transcription. Furthermore, even if the tRNA 3 Ser tDNA gene was not detectably transcribed in transcription mixtures containing r⌬C86 (see Fig. 5), a B⅐C⅐DNA complex could be formed with r⌬C86 on this template (data not shown). These observations indicated that B⅐C⅐DNA complexes assembled in the presence of r⌬C86 were not functionally identical to the ones formed with rTFIIIB70.
The properties of TFIIIB⅐TFIIIC⅐tDNA complexes were therefore further investigated. Complexes assembled on a la-beled SUP4 tDNA gene in the presence of TFIIIC, rTBP, rBЉ, and rTFIIIB70 (400 ng) or r⌬C86 (525 ng) were incubated for 5 min with increasing amounts of heparin. It was shown previously that incubation of a B⅐C⅐DNA complex with heparin stripped off TFIIIC from DNA and resulted in the formation of stable TFIIIB⅐tDNA complexes (6,7). As shown in Fig. 7B, heparin treatment of B⅐C⅐DNA complexes assembled in the presence of r⌬C86 resulted in the formation of lower amounts of TFIIIB⅐tDNA complexes than with the wild type protein and in release of free DNA (compare lanes 6 and 7 to lanes 11 and  12). Quantification showed that the ratio of B⅐C⅐DNA complexes (lane 3 or 8; no heparin) to B⅐DNA complexes (lane 7 or 12; 20 g/ml heparin) was 1.2 and 4.4 for the wild type and truncated protein, respectively. Four independent experiments gave similar results: B⅐C⅐DNA complexes containing r⌬C86 gave rise, when incubated with heparin (20 g/ml), to ϳ3.5-fold fewer TFIIIB⅐tDNA complexes than wild type TFIIIB70. These results indicated that B⅐C⅐DNA complexes were less stable with r⌬C86 than with TFIIIB70 or were more sensitive to disruption by polyelectrolytes. B⅐C⅐DNA complexes may also be structurally different because, when subjected to mild proteolysis, the complexes formed with r⌬C86 generated a protein⅐DNA electrophoretic band pattern noticeably different from that obtained with rTFIIIB70 (data not shown).
We also explored the stability of the intermediate complex BЈ⅐C⅐DNA, formed in the absence of BЉ ( Fig. 7C; for this experiment, we used Leu3 tDNA, since BЈ⅐C⅐DNA complexes are well resolved from C⅐DNA complexes). This type of complex is much less resistant to heparin than the B⅐DNA complex (12). Interestingly, heparin treatment of BЈ⅐C⅐DNA complexes containing r⌬C86 generated a new complex, CЈ, migrating almost like TFIIIC⅐DNA complexes and similar to the intermediate complex observed above (Fig. 7A, lane 11). The CЈ complex was resistant to concentrations of heparin that totally disrupted TFIIIC⅐DNA complexes (1 g/ml). The CЈ complex was also observed when B⅐C⅐DNA complexes were treated with 2 or 4 g/ml heparin (see Fig. 7B, lanes 9 and 10). Its faster migration rate may reflect the loss of polypeptides from BЈ⅐C⅐DNA complexes or, more probably, a large conformational change like the loss of DNA bending characteristic of B⅐C⅐DNA complexes (32).
We finally investigated whether DNA protection was altered in BЈ⅐C⅐DNA complexes formed with r⌬C86 on the tRNA 3 Leu gene. Preformed BЈ⅐C⅐DNA complexes were treated with DNase I and separated from free DNA by native polyacrylamide electrophoresis (Fig. 8A) before analysis of the DNase I cleavage pattern on a sequencing gel. The DNase I digestion patterns of the transcribed strand of the tRNA 3 Leu gene in free DNA or in BЈ⅐C⅐DNA complexes is shown in Fig. 8B. The footprint of TFIIIC over the A and B blocks was clearly observed. BЈ extended DNase protection from bp ϩ1 to bp Ϫ35 much as described for SUP4 tDNA (12,21). Two differences were seen with r⌬C86. First, although preformed complexes were separated from free DNA before DNA analysis, the footprint over the ϩ1/Ϫ38 region was incomplete, which suggested an instability of BЈ⅐C⅐DNA complex during DNase treatment. In addition, the size of the footprint was slightly reduced on both sides. In particular, there was an enhanced cleavage site at Ϫ35 that was absent in free DNA and in complexes containing TFIIIB70. This footprint analysis, therefore, confirmed the alteration of BЈ-DNA interaction. DISCUSSION TFIIIB70 or Brf1 is a pivotal component of TFIIIB, since it interacts with multiple components of the transcription complex including TFIIIC, TBP, and the C34 subunit of RNA polymerase III (13-15, 33). The amino-proximal half of FIG. 3. Interaction of p10 polypeptide with SUP4 tDNA. Purified p10 polypeptide was incubated with a 32 P-labeled SUP4 DNA probe, and the complex was analyzed by DNase I footprinting as described under "Experimental Procedures." The cleavage products were resolved, next to the G ϩ A sequencing ladder, by short run (right) as well as by long run electrophoresis (left) to better visualize the ϩ1/ TATA region. Lane 1, G ϩ A sequencing ladder; lane 2, DNase I treatment of the DNA probe; lanes 3, 4, and 5, p10⅐DNA complexes formed with 42, 85, and 127 ng of p10, respectively. Enhanced cleavage sites are indicated by black dots. The locations of the TATA-rich sequence (TATA) and transcription start site (ϩ1) are indicated next to the long run panel.
TFIIIB70 is structurally related to TFIIB, with a putative Zn 2ϩ -binding sequence, followed by two imperfect repeats (19,31,34). The proteolytically stable core TFIIB, which consists of the two repeats, interacts with TBP and with DNA upstream and downstream of the TATA box (26,(35)(36)(37)(38)(39). DNA bending by TBP is critical for fitting TFIIB in the complex. Based on its homology with TFIIB, the amino-terminal half of TFIIIB70 is probably inserted similarly in the TFIIIB⅐DNA complex. However, a major structural difference is the existence in TFIIIB70 of a CTE that almost doubles the size of the polypeptide compared with TFIIB. Unexpectedly, it is the CTE that most strongly interacts with TBP (13,14). Here we show that the COOH-terminal region of TFIIIB70 (CTE-(510 -596)) forms a protease-resistant domain that has the property to bind DNA tightly, in contrast to full-length TFIIIB70. This domain is important to form and stabilize TFIIIB⅐DNA complexes and to direct efficient initiation of transcription.
The existence of a DNA binding domain located at the COOH terminus of TFIIIB70 was revealed by limited proteolysis, a method often used to delineate functional regions in proteins (26,29,35). The precise proteolytic cleavage occurring naturally in E. coli cells is strongly suggestive of a tightly folded structural domain. The interaction of the p10 polypeptide with DNA displayed several interesting properties: it was remarkably resistant to electrolytes (0.4 M salt), apparently nonspecific (since binding was detected by footprinting all over the SUP4 tRNA gene), and possibly caused a distortion of the DNA backbone as suggested by the regularly spaced DNase I enhanced cleavage sites. Alternatively, these enhanced cleavage sites may also simply correspond to hypersensitive sites between closely spaced p10 molecules.
The hypothesis that CTE-(510 -596) contributes by its DNAbinding property to TFIIIB⅐DNA complex formation and stability was supported by several observations. First, the deletion of residues 510 -596 impaired the formation of B⅐C⅐DNA complexes. Five times as much r⌬C86 was needed for full complex formation on SUP4 tDNA as compared with intact TFIIIB70. Second, B⅐C⅐DNA complexes containing the truncated form of TFIIIB70 were less resistant to heparin than wild type complexes. Third, BЈ⅐C⅐DNA complexes formed with r⌬C86 were particularly unstable. This instability was observed in twodimensional footprinting experiments where isolated complexes displayed only partial protection of the upstream region; it was also apparent in transcription-competition experiments where intact rTFIIIB70 was dominant over r⌬C86 for BЈ⅐C⅐DNA complex formation (data not shown). Interestingly, heparin treatment of BЈ⅐C⅐DNA complexes generated a new form of complex (called CЈ, since it migrates just slightly slower than C⅐DNA complexes) that was not obtained with wild type TFIIIB70. CЈ complex was also observed as an intermediate during B⅐C⅐DNA complex formation with limiting levels of r⌬C86. There is the possibility that heparin treatment induced the loss of r⌬C86 from the BЈ⅐C⅐DNA complex that may explain the migration shift. Alternatively, the large down-shift of CЈ⅐DNA complexes compared with BЈ⅐C⅐DNA could reflect an important conformational change of the complex. TFIIIB was shown to bend DNA (32), and BЈ contributes to this bending (40). A TBP-induced bend in the TATA-less upstream region and then incubated for 5 min at 22°C with varying concentrations of heparin and analyzed by electrophoresis. Lane 1, DNA probe; lane 2, TFIIIC⅐DNA complex; lanes 3 and 8, control B⅐C⅐DNA complexes. B⅐C⅐DNA complexes treated with heparin at 2 g/ml (lanes 4 and 9); 4 g/ml (lanes 5 and 10); 10 g/ml (lanes 6 and 11); 20 g/ml (lanes 7 and 12) are shown. C, heparin resistance of BЈ⅐TFIIIC⅐Leu3 tDNA complexes. Factor⅐DNA complexes were formed with TFIIIC alone (lane 2), TFIIIC ϩ rTBP ϩ rTFIIIB70 (400 ng, lane 3), or r⌬C86 (523 ng, lane 8) as described under "Experimental Procedures." BЈ⅐C⅐DNA complexes were treated with heparin at 1 g/ml (lanes 4 and 9), 2 g/ml (lanes 5 and 10), 4 g/ml (lanes 6 and 11), or 10 g/ml (lanes 7 and 12). The migration of the different factor⅐DNA complexes is indicated. could require TFIIIB70 to be stabilized. Complete or partial dissociation of r⌬C86 by heparin may then cause a loss of DNA bending and increase the migration rate of the complex. Whatever the cause, the formation of the CЈ complex further underscores the role of the COOH terminus of TFIIIB70 in complex stability. One could imagine that the COOH-terminal DNA binding domain of TFIIIB70 is unmasked during the recruitment of TBP and holds the BЈ⅐DNA complex in place (note that p10 by itself does not bind TBP; see Fig. 1).
Transcription studies suggest that the role of the CTE-(510 -596) domain may not be restricted to a firmer anchoring of TFIIIB70 on the DNA. The addition of an excess of r⌬C86, leading to full B⅐C⅐DNA preinitiation complex formation, in fact, did not restore wild type transcription efficiency. In addition, different templates were transcribed with largely varying efficiencies in the presence of r⌬C86. Hence, the tRNA 3 Ser gene could form a B⅐C⅐DNA complex of apparently normal electrophoretic migration with r⌬C86 but was very poorly transcribed. Transcription initiation, monitored by synthesis of a short nascent RNA on SUP4 DNA remained impaired in the presence of an excess of RNA polymerase III (data not shown). Thus, B⅐C⅐DNA complexes containing r⌬C86 were deficient at some stage of transcription initiation. If p10 polypeptide indeed distorts the DNA backbone, there is the possibility that the COOH terminus of TFIIIB70 facilitates DNA melting by RNA polymerase. Attempts to reconstitute TFIIIB70 transcriptional activity by supplementing r⌬C86 with excess p10 were only partially successful (2-3-fold stimulation; data not shown).
Considering the striking structural and functional homology of TFIIB and TFIIIB70, the adjunction of the CTE to the NH 2 -terminal TFIIB-like domain is intriguing. The COOHterminal extension of TFIIIB70 could possibly play two types of function. One function would be to orient TBP⅐DNA complexes irreversibly toward pol III transcription. In this respect, the CTE can be considered equivalent to a class III TBP-associated factor. In addition, the CTE could have a basic function played by a pol II factor distinct from TFIIB. The best candidate for a functional homologue would be RAP30, the small subunit of TFIIF(RAP30/74). TFIIF interacts with RNA polymerase II and cooperates with TFIIB to recruit the enzyme into the preinitiation complex (41). RAP30 is critical in this process, since it was shown to bind both TFIIB and RNA polymerase II (42)(43)(44) and to be sufficient for specific recruitment of the enzyme to a TBP⅐TFIIB⅐promoter complex (45)(46)(47). The CTE of TFIIIB70 provides a similar link between TBP, the TFIIB-like region of TFIIIB70, and the C34 subunit of pol III (13)(14)(15)33). In addition, like the CTE of TFIIIB70, RAP30 has a cryptic DNA binding domain of similar size (ϳ80 amino acids) located at the COOH terminus (25,48). These domains, CTE-(510 -596) and RAP-(162-249), bind DNA nonspecifically, which is not unexpected, since, in both cases, sequence specificity is provided by other components, and their deletion impairs transcription initiation (Ref. 48 and this work). RAP30 is homologous to transcription factor 70 (49,50). It has been proposed that the DNA binding domain of RAP30 is related to an evolutionary conserved DNA binding domain of members of the 70 family of factors (51). A structural and mutational comparison of the DNA binding domains of RAP30 and TFIIIB70 would shed light on the evolution of the eucaryotic transcription machineries.