Expression and purification of the RNA polymerase III transcription specificity factor IIIB70 from Saccharomyces cerevisiae and its cooperative binding with TATA-binding protein.

Transcription by RNA polymerase III (pol III) in yeast requires the assembly of an initiation complex comprising the TATA-binding protein (TBP), a 90-kDa polypeptide (TFIIIB90), and a 70-kDa polypeptide (TFIIIB70). TFIIIB70 interacts with TBP, a unique pol III subunit, C34, and the 131-kDa subunit of the pol III-specific complex, TFIIIC. TFIIIB70 was expressed in Escherichia coli and purified to homogeneity. The specific transcription activity of rTFIIIB70 is 22-58% that of the native yeast and in vitro synthesized factor. However, only a small fraction (0.07-0.32%) of the TFIIIB70 from these sources results in the synthesis of full-length RNA. The data suggest that TFIIIB70 function may be limited by an unfavorable recruitment equilibrium into the preinitiation complex. Quantitative DNase I “footprint” titrations of yeast TBP to the adenovirus major late promoter were conducted at a series of constant TFIIIB70 concentrations. A value of −0.7 ± 0.2 kcal/mol was determined for the cooperative free energy of formation of the TBP·TFIIIB70·DNA complex at concentrations of TFIIIB70 sufficient to partition all of the binding cooperativity to the TBP binding isotherm. A Kd of 44 ± 23 nM characterizes the TFIIIB70 concentration dependence of the TBP·TFIIIB70 cooperativity. The relationship δlog K/δlog (TFIIIB70) is consistent with the linkage of a single molecule of TFIIIB70 with the TBP-promoter binding reaction.

Gene transcription by RNA polymerase III (pol III) 1 in Saccharomyces cerevisiae requires four chromatographically separable and functionally distinct transcription factors (TFIIIA, TFIIIB, TFIIIC, and TFIIIE) that bind to the promoters of pol III genes and/or facilitate their transcription (1)(2)(3)(4). TFIIIA plays a unique gene-specific role in the transcription of the 5 S RNA genes, whereas the other factors play important functions in the transcription of all pol III genes in yeast. For TFIIIC, these functions include promoter recognition and the recruitment of TFIIIB, which is directed to a region upstream of the transcription start site (2,5). Promoter-bound TFIIIB can, by itself, recruit pol III for multiple rounds of transcription (6). TFIIIB, therefore, plays a role in pol III transcription that is analogous to the general transcription factors required for initiation of transcription by RNA pol II.
The limiting steps in the transcription of pol III genes in wild-type yeast cells have been defined by mutations or gene dosage effects that increase the synthesis of pol III gene products. Thus far, only missense mutations in one of the six subunits of TFIIIC (TFIIIC 131 ) (7,8) and increased levels of one of the three subunits of TFIIIB (TFIIIB 70 ) (9) have been found to illicit a stimulation of pol III gene transcription. For TFIIIB 70 , this and other data demonstrate that the factor is stoichiometrically limiting for transcription in vivo (9) and in whole-cell extracts (10). Because of the limiting nature of TFIIIB 70 , global control of pol III gene expression can potentially be achieved by regulating the amount of this factor (1). Indeed, this appears to be the case. The reduced transcription observed in extracts derived from cells that are approaching stationary phase or whose growth has been inhibited by cycloheximide can be accounted for, in part, by a reduction in the amount of TFIIIB 70 (10,11).
The individual polypeptides that comprise yeast TFIIIB have been identified and cloned (4,9,(12)(13)(14)(15)(16)(17)(18)(19)(20)(21). In addition to TFIIIB 70 noted above, the components of yeast TFIIIB include the TATA-binding protein (TBP) and TFIIIB 90 . These polypeptides bind to TFIIIC⅐DNA complexes in a stepwise manner in vitro beginning with TFIIIB 70 . The subsequent binding of TBP followed by TFIIIB 90 leads to progressive changes in: (i) the size of the upstream region protected from digestion by DNase I; (ii) the degree of protection conferred by these proteins at specific sites; and (iii) the efficiency of cross-linking of various TFIIIB and TFIIIC subunits by photoprobes positioned at specific locations in the DNA (2). Each of the components of yeast TFIIIB has been expressed in bacteria, and collectively, they suffice to support TATA box-mediated transcription in the presence of highly purified pol III. Additionally, when provided with highly purified TFIIIC, the recombinant TFIIIB components reconstitute the transcription of tRNA genes (4,21). TFIIIB 70 is a protein of 596 amino acids, the amino-terminal half of which is homologous to the entire sequence of the pol II general transcription factor, TFIIB (9,19,20). Analogous to TFIIB, TFIIIB 70 interacts directly with TBP, although this interaction appears to be mediated primarily by the unique carboxyl-terminal half of the protein (22). In addition, TFIIIB 70 interacts directly with the C34 subunit that is unique to pol III and with TFIIIC 131 (22)(23)(24). Thus, TFIIIB 70 can be thought of as a "polymerase specificity factor" by virtue of its ability to mediate interactions among the universal transcription factor, TBP, a pol III-specific assembly factor, and a unique subunit of RNA polymerase III.
The important functions of TFIIIB 70 in the assembly of pol III transcription complexes and in the regulation of pol III gene transcription, together with the possibility for comparative studies with TFIIB, make this protein attractive for biochemical investigations. We have, therefore, initiated a quantitative study of the interaction of TFIIIB 70 with other pol III transcription factors. To facilitate its biochemical characterization, TFIIIB 70 was expressed in, and purified to homogeneity from, Escherichia coli. Our initial biochemical studies have examined TBP and TFIIIB 70 binding to DNA containing the high affinity TATA sequence of the adenovirus major late promoter (AdMLP). Although the AdMLP is not a natural template for pol III transcription, its TATA sequence and numerous other TATA elements have been shown to direct complex assembly and transcription by pol III in a variety of experimental systems (for examples, see Refs. 13 and 25-27 and references therein). Additionally, the thermodynamic and kinetic properties of TBP-promoter interactions are the subject of extensive biochemical and biophysical analyses (28 -31) and provide a solid foundation for quantitative studies of the TBP⅐TFIIIB 70 interaction. In this study, we demonstrate that TFIIIB 70 and TBP bind cooperatively to the AdMLP in a manner analogous to that of TBP and TFIIB. 2

EXPERIMENTAL PROCEDURES
Transcription Factors-The BioRex70 (BR␣) fraction was prepared from a wild-type yeast strain (IW1B6) and further fractionated on DEAE-Sephadex A25 to obtain a combined TFIIIC/pol III fraction by step elution with 0.1-0.5 M NaCl (7,32). Heparin-agarose TFIIIB was purified from the 0.1 M NaCl flow-through fraction obtained after chromatography on DEAE-Sephadex (7). The Cibacron blue-agarose BЉ used for these studies and methods for preparing in vitro synthesized TFIIIB 70 have been described previously (10). Recombinant yeast TBP was prepared and stored as described (30,31).
Expression of TFIIIB 70 in E. coli-Plasmid pSH360 (a gift from Steve Hahn) is derived from pET-11d and contains the entire coding sequence of TFIIIB 70 fused to a carboxyl-terminal histidine tag. Site-directed mutagenesis was undertaken on pSH360 to change the sequence between nucleotides 453-474 (relative to the initiating AUG) from 5Ј-GGTGAGTGTGTATTCCATAGGA-3Ј to 5Ј-AGTCTCTGTGTACAG-CATCGGC-3Ј. The resulting plasmid, pIIIB 70 ⌬SD, retains the wildtype coding information while eliminating two potential prokaryotic ribosome-binding sites. BL21(DE3) cells containing the plasmids pIIIB 70 ⌬SD and pUBS520, which carries the argU gene (a gift from Prof. Ralf Mattes), were grown at 37°C in ZYG media with 20 g/ml ampicillin and 60 g/ml kanamycin as described previously (33). At a density of A 600 ϭ 0.5, recombinant TFIIIB 70 expression was induced by the addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 0.4 mM. Cells were grown for 2 h and harvested by centrifugation.
Purification of Recombinant TFIIIB 70 -Recombinant TFIIIB 70 was prepared under denaturing conditions and purified on Ni 2ϩ -agarose (Qiagen) following the manufacturer's recommendations (Qiaexpressionist purification scheme 7) with the modifications described below. Cell pellets from 4 liters of culture (ϳ8 g) were resuspended in 100 ml buffer A (6 M guanidine hydrochloride, 0.1 M NaH 2 PO 4 , 0.01 M Tris, and 10 mM ␤-mercaptoethanol, pH 8.0, containing protease inhibitors (1 g/ml leupeptin, 1 g/ml pepstatin A, 0.1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, and 1 mM benzamidine). This mixture of protease inhibitors was present in all purification and dialysis buffers. The cell suspension was sonicated for a total of 2 min at 15-s intervals and then centrifuged at 5000 ϫ g for 15 min at 4°C. The supernatant was added in batch to 2 ml (packed volume) of Ni 2ϩ -agarose resin and mixed slowly at room temperature for 1 h. The protein-bound resin was washed with 10 column volumes of buffer A without ␤-mercaptoethanol and then 10 column volumes of buffer B (8 M urea, 0.1 M NaH 2 PO 4 , and 0.01 Tris, pH 8.0). The TFIIIB 70 was eluted in 10 -20 ml of buffer E (8 M urea, 0.1 M NaH 2 PO 4 , and 0.01 M Tris-HCl, pH 4.5) and neutralized to pH 6.8. The nonionic detergent, Brij-58, was added to a final concentration of 0.01%, and the eluate was dialyzed into "denaturation buffer" (6 M guanidine hydrochloride, 50 mM sodium phosphate, pH 7.8, 5 mM magnesium acetate, 1 mM EDTA, 1 mM dithiothreitol, 150 mM potassium acetate, 5 M zinc sulfate, and 10% glycerol) and then "renaturation buffer" (50 mM sodium phosphate pH 7.8, 5 mM magnesium acetate, 1 mM EDTA, 1 mM dithiothreitol, 150 mM potassium acetate, 5 M zinc sulfate, and 10% glycerol) at 4°C.
The recombinant TFIIIB 70 was further purified by preparative SDS-PAGE in a Bio-Rad model 491 Prep Cell as described by the manufacturer (Bio-Rad). TFIIIB 70 prepared as described above was diluted into Laemmli buffer and heated to 50°C for 10 min. The protein solution was loaded onto a 4% polyacrylamide stacking gel above a 6% separating gel in the large Prep Cell column (37-mm internal diameter). Electrophoresis was conducted for ϳ 7 h. The fractions containing TFIIIB 70 (detected by analytical SDS-PAGE and silver staining) were pooled and concentrated using a Centriprep-30 concentrator (Amicon, Inc.). The concentrated sample was applied directly to an Extracti-Gel D detergent removing gel column (Pierce Chemical Company) for three passages. Brij-58 was added to a final concentration of 0.01%, and the sample was denatured and renatured as described above. TFIIIB 70 concentrations were determined from the A 280 using ⑀ ϭ 31,690 M Ϫ1 cm Ϫ1 . Recoveries of ϳ40% were routinely achieved using this procedure, providing sufficient quantities of pure and active (see below) protein for biochemical study.
Quantitation of TFIIIB 70 -The amount of TFIIIB 70 protein in purified yeast fractions and in rabbit reticulocyte lysates was determined by quantitative Western analysis. Standard techniques of Western transfer and enhanced chemiluminescence detection was used as described previously (10). The number of fmoles of TFIIIB 70 was determined per unit protein by comparing titrations of each fraction with a standard curve constructed using the recombinant protein. Autoradiograms of the Western signal were quantified by a Molecular Dynamics laser densitometer and the ImageQuant software. The signal for the standard curve was linear between 25 and 1000 fmol. Linear regression values for each sample were determined from a minimum of five data points within the linear portion of the standard curve.
Transcription Assays-Transcription assays were performed under single-round reaction conditions as described previously (10,34). Ternary complexes were assembled on the sup9-e gene in the presence of ATP, UTP (600 M of each) and 25 M [␣-32 P]GTP (16 Ci/mmol) for 30 min at 25°C. Heparin and CTP were then added for an additional 10 min at 15°C. The transcription products were analyzed on denaturing 8% polyacrylamide gels. After autoradiography, bands were excised from the gels and measured for radioactivity.
Quantitative DNase I Footprint Titration Experiments-The plasmid pML(C2AT) contains the AdMLP from Ϫ400 to ϩ10 relative to the cap site (35). A 656-bp DNA probe labeled at one end of the template strand was obtained by digestion of the plasmid with AflIII and incorporation of 32 P nucleotides with Klenow, followed by SapI digestion and purification using published protocols (36). The "TATA box" of the promoter is located 79 base pairs from the labeled 3Ј end of the probe. The concentration of 32 P-labeled DNA present in the assay mixtures is significantly lower than the K d values of the protein-DNA interactions being analyzed, allowing the assumption that [protein] total Ϸ [protein] free to be made in the analysis of the binding isotherms.
The DNase I footprint titration experiments were conducted following published protocols (Refs. 30 and 36 and references cited therein). All experiments were conducted at 30°C in an assay buffer containing 25 mM Bis-Tris, 5 mM MgCl 2 , 1 mM CaCl 2 , 2 mM dithiothreitol, 1 g/ml poly(dG⅐dC), 100 mM KCl, and 0.01% Brij-58 at pH 7.0. Titrations of the AdMLP promoter with TBP (in the presence of constant concentrations of TFIIIB 70 ) were conducted in 200-l volumes of assay buffer and incubated at 30°C for ϳ45 min. Each sample was exposed to DNase I for 2 min. The nuclease reaction was stopped by the addition of 40 l of 50 mM EDTA, followed by the precipitation solution. The reaction products were separated on denaturing 8% polyacrylamide gels, subsequently dried and visualized using phosphor storage plates and a Phos-phorImager (Molecular Dynamics).
Titration curves for the TBP-TATA binding reaction were obtained from the the digital images of the electrophoretograms following the protocols of Brenowitz et al. (37) implemented using the ImageQuant (Molecular Dynamics) software (30,36). The absorbance (A) of the bands corresponding to the TATA box were integrated for each concentration of TBP. The integrated absorbance values were corrected for background, standardized to the total amount of DNA loaded in each lane, and then paired with the corresponding TBP concentration. A plot of ⌬A versus [TBP] yields a transition curve describing the binding reaction (Fig. 4). Nonlinear least squares analysis of the transition curve using the Langmuir binding polynomial yields K eq , and the upper and lower limits to the transition yields the binding isotherm for the reaction, defined as the change in fractional saturation, Y TBP , as a function of [TBP].
The Gibbs free energy of binding, ⌬G°, is related to the equilibrium 2 V. Petri and M. Brenowitz, manuscript in preparation. association constant, K eq , by the well known expression, ⌬G°ϭ ϪRT ln K eq . In TBP titration experiments conducted in the presence of fixed concentrations of TFIIIB 70 , the total binding free energy (⌬G°t otal ) includes contributions from both the intrinsic affinity of TBP for the promoter and the cooperativity between the two proteins. This cooperativity is expressed as ⌬⌬G°ϭ ⌬G°t otal Ϫ ⌬G°T BP for each concentration of TFIIIB 70 analyzed. At high concentrations of TFIIIB 70 , Y TBP becomes independent of [TFIIIB 70 ]. Only at this limit is ⌬⌬G°ϭ ⌬G°T BPϪTFIIIB70 , the microscopic cooperative free energy between the two proteins (38,39). An exploration of this analysis, as applied to TBP interacting with general pol II transcription factors, will be presented elsewhere. 2

Expression and Purification of E. coli-expressed TFIIIB 70 -
The expression of TFIIIB 70 containing a carboxyl-terminal histidine tag in E. coli and its subsequent purification on Ni 2ϩchelating agarose gives rise to three major products; full-length TFIIIB 70 and two smaller proteins of ϳ50 and 30 kDa (20) (Fig.  1A). The 50-kDa protein is the predominant species in these preparations, and the yield of full-length TFIIIB 70 is low (about 0.1-0.2 mg protein/liter culture). Moreover, further purification of full-length TFIIIB 70 on a variety of resins is hindered by the similar chromatographic properties of the 50-kDa polypeptide. Sequence analysis of the 50-kDa protein revealed an amino terminus corresponding to valine 165, immediately adjacent to the third methionine of TFIIIB 70 . Initiation of translation at this site seemed plausible because of the presence of two upstream sites that could potentially function as ribosomebinding sites in E. coli. In an effort to improve the quality and quantity of TFIIIB 70 preparations, the plasmid pIIIB70⌬SD was constructed in which the two putative ribosome-binding sites were disrupted without altering the amino acid sequence of TFIIIB 70 . Production of the 50-kDa polypeptide was effectively abolished with this construct (Fig. 1A). However, the yield of full-length TFIIIB 70 remained unacceptably low.
An examination of the TFIIIB 70 coding sequence revealed a high number of codons that are rarely used by bacterial genes.
In particular, the arginine triplets AGA and AGG were found to account for 71% of all the arginine codons in TFIIIB 70 . In E. coli, these codons are translated by a rare arginine tRNA encoded by the argU gene (40). Previous studies have shown that E. coli cells overexpressing the argU gene more efficiently express eukaryotic genes containing many AGA/AGG codons (40,41). We, therefore, examined whether the yield of TFIIIB 70 could be improved in BL21(DE3) cells containing a compatible, argU overexpressing plasmid (pUBS520). Cell extracts of this strain, chromatographed on a Ni 2ϩ -affinity column as described under "Experimental Procedures" yielded ϳ18 mg of full-length TFIIIB 70 from 7.8 g of cells (or 4.5 mg protein/liter culture). This represents at least a 20-fold enhancement in the yield of full-length TFIIIB 70 compared to cells lacking the argU plasmid. Affinity column eluates of cell extracts prepared with and without argU overexpression are qualitatively indistinguishable by Coomassie Blue staining of SDS-polyacrylamide gels (data not shown).
Full-length TFIIIB 70 represents approximately one-third of the total protein obtained after the Ni 2ϩ -affinity column. Further purification of TFIIIB 70 by conventional chromatographic techniques to remove small quantities of truncated polypeptides was hampered by the broad elution characteristics of the protein and/or poor protein recovery. Similar properties had previously been observed in the purification of TFIIIB 70 from yeast. 3 However, preparative SDS-PAGE of the nickel column eluate, as described under "Experimental Procedures", yielded fractions containing a single 70-kDa band, as visualized by silver staining (Fig. 1B). These fractions were pooled and concentrated, the detergent was removed, and the protein was renatured from guanidine hydrochloride. SDS-PAGE of this material yielded a single band, as detected by Western blot analysis using TFIIIB 70 -specific antibodies (data not shown).
Transcriptional Activity of Native and Recombinant TFIIIB 70 -Pure recombinant TFIIIB 70 was assayed for transcription activity under single-round conditions in a reconstituted system. The reaction conditions were such that transcription was limited only by the amount of TFIIIB 70 . Accordingly, the synthesis of full-length sup9-e transcripts showed a linear dependence on the addition of this factor ( Fig. 2A). Transcription on a SUP4 template yielded similar results and demonstrated a quantitative conversion of the nascent transcript (17-mer) into full-length RNA (data not shown). Despite the rigorous purification of the recombinant protein, its transcription activity is comparable to that of native yeast and in vitro synthesized preparations (Fig. 2B). For this comparison, TFIIIB 70 in a heparin-agarose TFIIIB fraction and in vitro synthesized TFIIIB 70 were assayed in the reconstituted system as for the recombinant protein. Additionally, a BioRex fraction containing all of the activities necessary for transcription, including limiting amounts of TFIIIB 70 (10), was assayed. The recombinant protein showed 22% of the activity seen for TFIIIB 70 in the heparin-agarose TFIIIB fraction and 58% of the activity of the in vitro synthesized protein. These activities are comparable to those observed in the BioRex fraction prior to the separation of the factors. Interestingly, when the transcription activities of these TFIIIB 70 preparations are expressed as fmoles of full-length RNA product synthesized per fmole of TFIIIB 70 (Fig. 2B), it appears that the factor is very inefficiently utilized in transcription. Only 0.07-0.32% of the added TFIIIB 70 , or approximately 1 to 3 molecules per thousand are engaged in functional transcription complexes.
Cooperative Binding of TBP and TFIIIB 70 to a Promoter-To better understand the functions of TFIIIB 70 in transcription, 3  we have begun a quantitative biochemical study of its interactions with other pol III transcription components. As a first step in this work, we have examined the effect of TFIIIB 70 on the binding of TBP to the high affinity TATA sequence of the AdMLP. Titration of the DNA restriction fragment containing the AdMLP TATA box with TFIIIB 70 in the absence of TBP does not yield detectable DNase I protection at concentrations as high as 0.91 M anywhere on the restriction fragment (data not shown). From this observation, a lower limit to the DNA binding affinity of TFIIIB 70 on the order of 1.6 M (⌬G DNA/TFIIIB70°Ϸ Ϫ8 kcal/mol) can be estimated. This result is consistent with mobility-shift assays in which recombinant TFIIIB 70 does not "shift" a TATA-containing DNA probe at this concentration (data not shown).
The binding of TBP to the AdMLP TATA box results in clear and specific protection under the experimental conditions used in these studies (Fig. 3A). This binding reaction is described by the Langmuir binding polynomial (Fig. 4), as was observed previously for the E4 promoter (30). 4 The reproducibility of these experiments is illustrated by the six independent TBP titrations shown in the inset to Fig. 4. Global analysis of these binding isotherms yields ⌬G TBP°ϭ Ϫ11.3 Ϯ 0.1 kcal/mol. Nonspecific binding is not observed at TBP concentrations below saturation of the TATA box. The presence of 0.01% Brij-58 in the binding buffer (required to maintain solubility of recombinant TFIIIB 70 ) has no detectable effect on the affinity of TBP for the promoter in these experiments (data not shown). A TBP-dependent, DNase I-hypersensitive site is present just upstream of the AdMLP TATA (Fig. 3A).
TBP titrations conducted in the presence of constant concentrations of TFIIIB 70 reveal no extension of the footprint upstream or downstream of the TATA box (see below). However, protection of the TATA box commences at lower TBP concentrations compared to the TBP titration alone (Fig. 3B). This dependence of TBP binding upon TFIIIB 70 is clearly seen in the binding isotherms determined as described under "Experimental Procedures"; TFIIIB 70 shifts the TBP-binding isotherm to lower concentrations. This increase in the apparent affinity of TBP for the AdMLP reveals a positive cooperative interaction between the two proteins (Fig. 4). The specificity of this interaction is demonstrated by the absence of cooperativity for TFIIIB 70 preparations lacking significant transcriptional activity (data not shown). Similarly, the presence of bovine serum 4 M. Brenowitz, V. Petri, and E. Jamison, unpublished data. albumin at 50 g/ml does not affect the affinity of TBP for DNA. 5 The results of a series of TBP titrations conducted at fixed concentrations of TFIIIB 70 (ranging from 5 to 300 nM) are shown as the energy difference (⌬⌬G°) referenced to "TBP alone" titrations (Fig. 5). Moderate cooperativity (⌬G°T BP/ TFIIIB70 ϭ Ϫ0.7 Ϯ 0.2) between TBP and TFIIIB 70 is observed at the plateau of the transition; at these TFIIIB 70 concentrations, all of the binding cooperativity between the two proteins is partitioned into the TBP isotherm. Thus, in the presence of saturating TFIIIB 70 , the affinity of TBP for the AdMLP TATA box is increased 3-fold. The derivative ␦ln K eq /␦ln [TFIIIB 70 ] is consistent with the linkage of one molecule of TFIIIB 70 with the TBP binding reaction (42) although the data are not sufficient to unequivocally resolve this issue (Fig. 5, inset). Concentrations of TFIIIB 70 above those shown in Fig. 5 result in the beginning of a second transition in the observed values of ⌬⌬G°( data not shown). This additional cooperativity may be due to the binding of more than one TFIIIB 70 molecule at the AdMLP TATA box.
The fact that the DNase I footprints of TBP and the TBP⅐TFIIIB 70 complex are indistinguishable suggests that TFIIIB 70 does not interact with the DNA flanking the TATA box (Fig. 6, A and B). In the control titrations conducted with TBP alone, detectable DNase I cleavage, albeit at very low levels, is still present even at TBP concentrations sufficient to fully occupy the TATA box (Fig. 6A). 6 In these reactions, the concentration of the labeled DNA is several orders of magnitude lower than the TBP concentrations, such that all of the DNA is bound by protein (see "Experimental Procedures".) There is a reproducible further 10 -15% decrease observed in the densities of the electrophoretic bands representing bases within the TATA box under saturating conditions for both TBP and TFIIIB 70 (Fig. 6C). DISCUSSION An efficient method for expressing TFIIIB 70 in E. coli and purifying milligram quantities of the protein to homogeneity has been developed. A ϳ20-fold improvement in the yield of TFIIIB 70 was achieved by overexpressing the argU gene to compensate for an extreme codon usage bias. This strategy has now been used successfully in several cases including the expression of yeast TFIIIA (41). Additionally, the poor chromatographic properties of TFIIIB 70 (low recoveries and poor resolution), which have hampered the purification of both the recombinant and the native protein, were overcome using preparative SDS-PAGE. After detergent removal and renaturation of the protein, pure recombinant TFIIIB 70 was shown to be 22-58% as active as partially purified native preparations obtained using conventional chromatographic methods (Fig. 2). Under conditions where native or recombinant TFIIIB 70 was limiting for tRNA gene transcription (i.e. in a TFIIIC-dependent reaction), single-round initiation assays showed that only 0.07-0.32% of the TFIIIB 70 resulted in the synthesis of a fulllength RNA product. This finding suggests a biochemical basis for the previously reported limiting nature of TFIIIB 70 in vivo and in whole-cell extracts (9, 10). The ability of TFIIIB 70 to  (Fig. 4, inset) and is a measure of cooperativity. Calculations: binding isotherms obtained from footprint titration experiments are used to generate the Gibbs free energy of binding. The free energy is related to the equilibrium association constant, K eq , by the expression, ⌬G°ϭ ϪRT ln K eq . For TBP titrations in the presence of TFIIIB 70 the total binding free energy, ⌬G total°, includes the intrinsic affinity of TBP for the promoter, (⌬G TBP°, Fig. 4, inset) and the cooperativity between the two proteins. Cooperativity is expressed as ⌬⌬G°ϭ ⌬G total°Ϫ ⌬G TBP°f or each TFIIIB 70 concentration. At 50% cooperativity, a K d of 44 Ϯ 23 nM can be determined for the interaction of TFIIIB 70 with the TBP⅐DNA complex. The inset shows a plot of the equilibrium association constants, calculated as described above, as a function of TFIIIB 70 concentration. A dotted line with a slope of one representing the derivative ␦ lnK eq /␦ ln [TFIIIB 70 ] is consistent with the linkage of one molecule of TFIIIB 70 with the TBP binding reaction. Bars, S.D. function in transcription may be limited by an unfavorable equilibrium of association into the preinitiation complex. Based on the order of assembly of TFIIIB components onto TFIIIC⅐DNA complexes, i.e. TFIIIB 70 , TBP, and TFIIIB 90 (14), and the fact that TBP and TFIIIB 90 were present at saturating concentrations, the limiting equilibrium probably involves the binding of TFIIIB 70 to the TFIIIC⅐DNA complex. It should be noted that transcription is an indirect measure of this equilibrium and that inefficiencies at later stages in the transcription process may underestimate the amount of TFIIIB 70 assembled into TFIIIB⅐DNA complexes under our assay conditions. Alternatively, the low specific activity of TFIIIB 70 in vitro and its limiting function in vivo may indicate that the factor is largely inactive in binding to TFIIIC⅐DNA complexes. We do not favor this possibility, however, since gel-purified TFIIIB 70 that has been renatured from guanidine exhibits substantial activity in forming heparin-resistant TFIIIB⅐DNA complexes. 7 Since the thermodynamic studies presented herein on the TBP⅐TFIIIB 70 cooperative interaction are dependent upon the determination of the TBP-AdMLP binding isotherms, some discussion of this reaction is warranted. Studies of the fluores-cence anisotropy of the single tryptophan of yeast TBP suggest self-association of the protein, but at concentrations higher than those used in our studies (29). The TBP binding isotherms obtained under the experimental conditions used in this work are well described by the Langmuir "single-site" binding polynomial; self-association need not be invoked to describe the data (Fig. 4) (30). 4 Thus, the TBP-promoter binding reaction has been analyzed by the simplest model consistent with the available data. A second issue concerns the "specificity" of the TBP-promoter interaction (i.e. the ratio of binding constants for "specific" versus "nonspecific" sequences). The autoradiograms shown in Fig. 3 quite clearly demonstrate the high specificity of the TBP-AdMLP interaction under the experimental conditions used. This result contrasts with that of Coleman and Pugh (43), who have reported a low specificity for TBP binding to the AdMLP promoter. The basis of this difference is likely to reside in the thermodynamic variables, in particular the monovalent ion (our studies used 100 mM KCl, whereas those of Coleman and Pugh (43) used 75 mM potassium glutamate). Studies on the E. coli Lac repressor have shown that in buffer containing 75 mM KCl, the ratio of equilibrium binding constants for Lac repressor binding to a 20-base pair symmetric operator and to nonspecific sequence is ϳ10 7 ; substitution of glutamate for chloride as the counter ion reduces this ratio by several orders of magnitude (44). Preliminary studies suggest a similar effect of monovalent ion concentration and type on TBP-promoter binding. 5 Thus, the choice of experimental conditions is a key consideration in the design and interpretation of studies of TBP-promoter interactions and the interaction of TBP with polymerase-specific transcription factors.
The thermodynamic linkage of coupled binding reactions allows multiple pathways to be used to determine the cooperativity between ligands. Since binding of TFIIIB 70 to the AdMLP TATA box is not detectable in the DNase I footprinting assay (Figs. 3 and 6), cooperativity between TBP and TFIIIB 70 was quantitated by the effect of TFIIIB 70 on TBP-promoter binding isotherms. Under the experimental conditions of these studies, the TBP⅐TFIIIB 70 cooperativity (⌬G coop°ϭ Ϫ0.7 Ϯ 0.2) results in a moderate 3-fold increase in the affinity of TBP for the AdMLP TATA box. Given that the AdMLP is a high affinity binding site for TBP, it will be interesting to see if and how the magnitude of the cooperativity changes with other TATA-containing promoters. The linkage relationship described by Wyman (42) provides a method to evaluate the number of molecules linked to a binding reaction. The analysis of TFIIIB 70 shown in the insert to Fig. 5 is consistent with one TFIIIB 70 molecule per TBP⅐TATA complex. The 1:1:1 stoichiometry of TBP⅐TFIIIB 70 ⅐DNA is in agreement with the stoichiometry of the analogous TBP⅐TFIIB⅐DNA complex determined by x-ray crystallography (45). However, direct determinations of the complex stoichiometry are required to resolve this issue.
The TFIIIB 70 concentration dependence of ⌬⌬G°shown in Fig. 5 includes contributions to the ⌬G°of TFIIIB 70 binding to DNA as well as the ⌬G°for TFIIIB 70 binding to TBP. The observed 44 Ϯ 23 nM apparent K d for the TFIIIB 70 concentration dependence of the cooperativity suggests that a significant TBP⅐TFIIIB 70 interaction occurs in the absence of DNA since TFIIIB 70 binding to DNA was not detected at or below 0.91 M. An analysis of the TBP⅐TFIIIB 70 interaction in solution is under way.
Although cooperative binding between TBP and TFIIIB 70 is clearly established, the mechanism of this interaction remains to be determined. The footprinting data presented in Fig. 6C show that TFIIIB 70 inhibits cleavage of the AdMLP TATA sequence by DNase I in a TBP⅐TFIIIB 70 ⅐DNA complex. A plausible, but not exclusive, interpretation of this result is that 7 G. Kassavetis, personal communication.  A and B, the level of TATA box protection measured is equivalent. C, comparison of TATA box protection by TBP and TBP⅐TFIIIB 70 at saturation of the site. f, the protection achieved at saturation by TBP alone. Ⅺ, the protection achieved at saturation by the TBP⅐TFIIIB 70 complex. Calculations: at each protein concentration, the disappearance of absorbance within the site is referenced to a standard absorbance outside of the site. The difference between these ratios and the ratio calculated for the DNA-only control measures the protection of the DNA from DNase I cleavage due to protein binding. The data are normalized using the DNA-only control lanes to allow a comparison between the two experiments. TFIIIB 70 is positioned on the "underside" of the TBP-induced DNA bend, analogous to the structure of the TBP⅐TFIIB⅐DNA complex (45). This model is attractive given the comparable magnitudes of the TBP⅐TFIIIB 70 ⅐AdMLP and TBP⅐TFIIB⅐AdMLP cooperative interactions. 2 The model also suggests that the TBP⅐TFIIIB 70 binding cooperativity could result, in part, from TFIIIB 70 stabilizing the TBP-induced DNA bend (46,47). Alternatively, a TFIIIB 70 -mediated change in the structure of the TBP⅐DNA complex could account for the findings.
Recent studies of the kinetics of the TBP-promoter interaction have shown that the initial step of the binding reaction, the formation of a productive encounter complex, is rate-limiting (31). This model postulates that the absence of a detectable diffusion limited intermediate complex is due to its low concentration; TBP binds to only a subset of the DNA conformations transiently present in solution. If TFIIIB 70 transiently stabilizes a favorable DNA conformation or increases the population of DNA molecules to which TBP can productively bind, an increase in the association rate would be predicted. Alternatively, if TFIIIB 70 acts at a later step in the reaction pathway, its effect is expected to be predominantly manifest in the rate of dissociation. Kinetic studies are in progress to distinguish between these hypotheses.
The increased affinity of the TBP⅐TFIIIB 70 complex relative to TBP alone may define a new functional role for this protein in the assembly of pol III transcriptional initiation complexes. All yeast pol III genes contain either cryptic TATA boxes (A-T sequences) in the 5Ј-flanking region or a TATA element as in the yeast U6 snRNA gene. TBP will bind a variety of nonconsensus TATA elements, although with reduced affinity compared to the consensus (48,49). 8,9 TFIIIB 70 , as part of a TFIIIB 70 ⅐TFIIIC⅐DNA complex, may serve to recruit TBP to these pol III templates with significant affinity and stimulate the formation of initiation complexes.