The N-terminal Region of Yeast TFIIB Contains Two Adjacent Functional Domains Involved in Stable RNA Polymerase II Binding and Transcription Start Site Selection*

The general transcription factor IIB (TFIIB) is required for accurate and efficient transcription of protein-coding genes by RNA polymerase II (RNAPII). To define functional domains in the highly conserved N-terminal region of TFIIB, we have analyzed 14 site-directed substitution mutants of yeast TFIIB for their ability to support cell viability, transcription in vitro, accurate start site selection in vitro and in vivo, and to form stable complexes with purified RNAPII in vitro. Mutations impairing the formation of stable TFIIB·RNAPII complexes mapped to the zinc ribbon fold, whereas mutations conferring downstream shifts in transcription start site selection were identified at multiple positions within a highly conserved homology block adjacent and C-terminal to the zinc ribbon. These results demonstrate that the N-terminal region of yeast TFIIB contains two separable and adjacent functional domains involved in stable RNAPII binding and transcription start site selection, suggesting that downstream shifts in transcription start site selection do not result from impairment of stable TFIIB·RNAPII binding. We discuss models for yeast start site selection in which TFIIB may affect the ability of preinitiation complexes to interact with downstream DNA or to affect start site recognition by a scanning polymerase.

The general transcription factor IIB (TFIIB) is required for accurate and efficient transcription of protein-coding genes by RNA polymerase II (RNAPII). To define functional domains in the highly conserved Nterminal region of TFIIB, we have analyzed 14 site-directed substitution mutants of yeast TFIIB for their ability to support cell viability, transcription in vitro, accurate start site selection in vitro and in vivo, and to form stable complexes with purified RNAPII in vitro.

Mutations
impairing the formation of stable TFIIB⅐RNAPII complexes mapped to the zinc ribbon fold, whereas mutations conferring downstream shifts in transcription start site selection were identified at multiple positions within a highly conserved homology block adjacent and C-terminal to the zinc ribbon. These results demonstrate that the N-terminal region of yeast TFIIB contains two separable and adjacent functional domains involved in stable RNAPII binding and transcription start site selection, suggesting that downstream shifts in transcription start site selection do not result from impairment of stable TFIIB⅐RNAPII binding. We discuss models for yeast start site selection in which TFIIB may affect the ability of preinitiation complexes to interact with downstream DNA or to affect start site recognition by a scanning polymerase.
Eukaryotic RNA polymerase II (RNAPII) 1 requires the action of at least six accessory proteins to accurately initiate transcription. These accessory proteins, termed the general transcription factors (GTFs), have been the focus of much investigation and include TFIIA, TFIIB, TFIID, TFIIE, TFIIF, and TFIIH. The GTFs and RNAPII assemble in an ordered stepwise fashion on a class II promoter in vitro to form a functional preinitiation complex (PIC) (reviewed in Ref. 1). Assembly is initiated by the binding of TFIID to the TATA element via the TATA-binding protein (TBP) subunit, in some cases assisted by TFIIA. This complex is recognized by TFIIB, which binds and recruits RNAPII and TFIIF. PIC formation is completed by the association of TFIIE and then TFIIH, and the resulting complex can hydrolyze ATP and initiate mRNA syn-thesis. In contrast to this ordered-assembly model for PIC formation, it has been proposed that a preassembled holoenzyme, consisting of RNAPII, most of the GTFs, and additional factors, is recruited in one step to promoter-bound TFIID in vivo (reviewed in Ref. 1).
The general transcription factor TFIIB has an essential role in RNAPII transcription, and together with RNAPII and TBP, defines the minimal set of factors necessary for promoter-dependent transcription of a supercoiled DNA template in vitro (1). In both the ordered-assembly and holoenzyme-recruitment models of PIC formation, TFIIB recognizes promoter-bound TFIID and facilitates association of the remaining GTFs and RNAPII. Consistent with this role, TFIIB interacts with DNA adjacent to the TATA box (2) and binds to TBP (3)(4)(5), the TBP-associated factor TAF40 (6), RNAPII (4,7,8), and both subunits of TFIIF (4,9). TFIIB may also play a role in the regulation of transcription by gene-specific regulatory proteins, as many of these regulatory factors bind TFIIB directly (10 -17). In addition, TFIIB is involved in the selection of transcription start sites, as mutations in the Saccharomyces cerevisiae SUA7 gene, encoding TFIIB, can alter transcription start site selection in vivo and in vitro (7,18).
The conservation of TFIIB structure among eukaryotic organisms underscores its central role in transcription. The protein is comprised of a highly conserved N-terminal region and a C-terminal core domain (Fig. 1). Although no structural information has been obtained for full-length TFIIB, NMR structures have been obtained for the core domain of human TFIIB and for a portion of the N-terminal region of the archaeal TFIIB from Pyrococcus furiosus (19,20). In addition, the crystal structure has been reported for the core domain, as well as for a DNA⅐TBP⅐core TFIIB ternary complex (21). These studies have revealed several structural motifs, including a zinc ribbon fold in the N-terminal region, a pair of ␣-helical direct repeats in the core domain, and an amphipathic helix between the repeats.
The high degree of amino acid conservation of the N-terminal region of TFIIB suggests that important functions reside in this portion of the protein. Reflecting this, random PCR mutagenesis of the yeast TFIIB gene identified a number of cold-and temperature-sensitive mutants that mapped to this region (22). To determine the functions of the N-terminal region and to define functional domains, we have undertaken a site-directed mutagenesis study of yeast TFIIB. In previous work, we identified two substitution mutants, L50D and R64E, that exhibited impaired RNAPII binding and a downstream shift in transcription start site selection, respectively (7). To extend our analysis and to define the functional domains involved in stable RNAPII binding and transcription start site selection, we have in this work analyzed 14 additional site-directed substitution mutants for their ability to support cell viability, transcription in vitro, accurate start site selection in vitro and in vivo, and to bind purified RNAPII in vitro. Our data demonstrate that the N-terminal region of TFIIB contains two adjacent and separable functional domains involved in stable RNAPII binding and transcription start site selection.

EXPERIMENTAL PROCEDURES
Plasmids and TFIIB Site-directed Mutagenesis-Yeast TFIIB substitution mutants were generated using the megaprimer method of PCR mutagenesis and plasmid p314/yIIBN as template as described previously (7). For mutants that exhibited altered properties in vivo, Nterminally polyhistidine-tagged derivatives expressed from the normal SUA7 promoter were generated by replacing the 1100-base pair NdeI-PstI fragment of plasmid p314/H6wt (containing the SUA7 coding region) with the corresponding fragment from the p314/yIIBN mutant plasmid as described (7). For the production of recombinant polyhistidine-tagged TFIIB proteins, the 1100-base pair NdeI-PstI fragment from the relevant p314/yIIBN mutant plasmid was isolated and inserted between the NdeI and PstI sites of bacterial expression plasmid pQE/yIIB (7).
In Vivo Analysis of TFIIB N-terminal Mutants-To determine the ability of the TFIIB mutants to support yeast cell growth, a plasmidshuffle complementation assay was used (7,23). Test strain FP153 (MAT␣, ura3-52, trp1⌬63, sua7⌬1 and p316/yIIB[URA3]) contains a deletion of the chromosomal sua7 coding region (BclI-BamHI) with viability maintained by the presence of a wild-type SUA7 gene on a URA3-based vector. Strain FP153 was transformed with either wildtype or mutant TFIIB constructs in plasmid p314 (TRP1 selectable marker), selecting for Uraϩ, Trpϩ colonies on CAA medium lacking uracil and tryptophan at 30°C (7). Transformants were grown in liquid CAA medium lacking uracil and tryptophan at 30°C, dilutions were spotted on SC medium containing 5-fluoroorotic acid (5-FOA), and the plates were incubated at room temperature for 4 days. 5-FOA, toxic to cells containing a functional URA3 gene, selects for cells that have spontaneously lost the URA3-containing plasmid p316/yIIB with wildtype TFIIB. Thus, the appearance of 5-FOA-resistant colonies reflects the ability of a TFIIB mutant to support cell viability in the absence of wild-type TFIIB. To further analyze the growth properties of mutants that supported viability, 5-FOA-resistant colonies were grown to midexponential phase in liquid YPD (1% yeast extract, 2% Bactopeptone, 2% dextrose) medium at room temperature, and dilutions in sterile water were spotted on YPD plates. The plates were incubated at low (16°C), intermediate (30°C), or high (37°C) temperature (Table I). The mutant strains were also analyzed on a variety of other growth media, but the relative growth properties on these media were similar to those observed with YPD (data not shown).
Immunoblotting-To determine the steady-state protein levels of TFIIB mutants with altered growth properties, strains containing both wild-type and polyhistidine-tagged mutant SUA7 alleles were grown to mid-exponential phase in liquid CAA lacking tryptophan and uracil. Cells were harvested by centrifugation and whole-cell extracts were prepared as described previously (24). Extracts (50 g of protein) were resolved on a 10% SDS-polyacrylamide gel, transferred to Immobilon-P polyvinylidene difluoride membranes (Millipore), and probed with a mouse monoclonal antibody raised against the polyhistidine-tag (RGS-His; Qiagen, 1:2,000 dilution) followed by goat anti-mouse IgG coupled to horseradish peroxidase (Jackson Laboratories, 1:40,000 dilution). Immune complexes were visualized by the addition of SuperSignal Ultra chemiluminescent substrate (Pierce) and autoradiography.
Purification of Recombinant Proteins-Polyhistidine-tagged TFIIB proteins were expressed in Escherichia coli host strain JM109 under control of the phage T5 promoter in plasmid pQE32*. Culture growth, recombinant protein expression, and purification were carried out as described previously (7). Purified TFIIB proteins were judged to be Ͼ90% pure by SDS-PAGE.
Mapping of in Vivo mRNA Start Sites-Transcription start sites for the ADH1 and CYC1 genes were determined by primer extension analysis of total yeast RNA as described (25). The oligonucleotide primer for analysis of ADH1 mRNA contained the sequence 5Ј-dGTATTCCAACT-TACCGTGGGATTCG-3Ј, corresponding to positions ϩ63 to ϩ39 (where ϩ1 is the initiating ATG). The oligonucleotide primer for analysis of CYC1 mRNA contained the sequence 5Ј-dGTCTTGAAAAGTGTAG-CACC-3Ј, corresponding to positions ϩ53 to ϩ34 (where ϩ1 is the initiating ATG).
RNAPII Co-immunoprecipitation Assay-Purified yeast RNAPII (500 ng) and TFIIB (100 ng) were incubated in 250 l of P100 buffer (50 mM Tris acetate (pH 7.9), 10% glycerol, 100 mM potassium acetate, 14 mM magnesium acetate, 4 mM dithiothreitol, 1 mM EDTA, 50 g/ml bovine serum albumin, and protease inhibitors (0.5 g/ml pepstatin A, 0.5 g/ml leupeptin, 2 g/ml chymostatin, 1 g/ml bestatin, 2.5 g/ml antipain, 0.5 g/ml aprotinin, 1 mM PMSF, 2 mM benzamidine)) at 4°C overnight. Pansorbin (Calbiochem) was harvested by centrifugation and resuspended in blocking buffer (50 mM Tris acetate (pH 7.9), 100 mM potassium acetate, 25 mg/ml PVP-K-30 (U. S. Biochemical Corp.), 60 mg/ml casein, final solution brought to pH 7.9 with 1 M KOH). After incubation for 20 min at room temperature, the Pansorbin was harvested by centrifugation and resuspended to its original volume with P100 buffer, and 25 l was added to each reaction. The samples were incubated for 30 min at 4°C with gentle rocking and were centrifuged at 15,000 ϫ g for 2 min. Supernatants were transferred to fresh tubes containing 2.5 g of anti-RNAPII monoclonal antibody (8WG16, Research Diagnostics) and incubated for 90 min at 4°C with gentle rocking. Pansorbin (pre-blocked, 25 l in P100) was then added, and the reactions were incubated for 60 min at 4°C with gentle rocking. The samples were centrifuged at 15,000 ϫ g for 2 min, and the pellets were washed three times with 250 l of buffer P100 containing 0.1% Nonidet P-40 and once with 250 l of P100. The pellets were resuspended in 40 l of 1ϫ SDS sample buffer, and 10 l of each sample was resolved on a 10% SDS-polyacrylamide gel and transferred to Immobilon-P polyvinylidene difluoride membranes. Immunoblotting was performed as described above, using an affinity-purified anti-yeast TFIIB antibody (1:5,000), followed by goat anti-rabbit IgG conjugated to horseradish peroxidase (Jackson Laboratories, 1:40,000).

Identification of TFIIB N-terminal Substitution Mutants with Altered Properties in Vivo-
The N-terminal region of TFIIB contains several highly conserved sequence motifs (Fig.  1). To address the role of these motifs in the function of yeast TFIIB, we constructed and analyzed 14 site-directed substitution mutants. Substitutions were introduced at conserved residues that reside within the second rubredoxin knuckle of the zinc ribbon, the C-terminal ␤-strand of the zinc ribbon, and an adjacent homology block that is highly conserved among seven eukaryotic species. The mutants were initially tested for their ability to support yeast cell viability using a plasmid-shuffle complementation assay (see "Experimental Procedures"). Plasmid-shuffling of the mutants revealed that substitution mutants C45S/C48S and L52P did not support cell growth (Table  I). The remaining 12 mutants, capable of supporting viability, were tested for their growth rates at several temperatures. Seven mutants were indistinguishable from the wild-type, whereas five mutants (W63P, W63R, R64A, F66D, and H71E) exhibited varying degrees of cold sensitivity, with the W63P and H71E mutants also displaying sensitivity to elevated temperature. The mutants were also analyzed on a variety of other growth media, but the relative growth properties on these media were similar to those observed with YPD (data not shown).
To confirm that the observed growth phenotypes were due to inherent functional impairments and not due to instability of the mutant proteins, we examined the steady-state levels of the mutant proteins by immunoblotting. Polyhistidine-tagged versions of the mutant proteins expressed from the wild-type SUA7 promoter were introduced into the plasmid-shuffle strain. The resulting polyhistidine-tagged mutant strains were retested and confirmed to confer the same growth properties as the non-tagged versions (data not shown). Immunoblotting of whole-cell extracts was performed with a monoclonal antibody directed against the polyhistidine tag. As shown in Fig. 2, all of the mutants exhibited steady-state protein levels comparable to wild-type TFIIB. Taken together, these results suggest that the C45S/C48S double substitution in the second rubredoxin knuckle of the zinc ribbon, the L52P substitution in the Cterminal ␤-strand of the zinc ribbon, and the W63P, W63R, R64A, F66D, and H71E substitutions in the adjacent homology block impair one or more essential functions of yeast TFIIB.
In Vitro Transcriptional Analysis of TFIIB Mutants-To determine the transcriptional activity of the TFIIB mutants conferring altered in vivo properties, we expressed and purified recombinant polyhistidine-tagged versions of the mutant TFIIB proteins and tested their abilities to support basal and activated transcription using a TFIIB-depleted whole-cell extract. The C45S/C48S double substitution mutant did not support any detectable level of transcription above background, whereas the L52P and W63P mutants were severely compromised for transcriptional activity (Fig. 3). The H71E mutant was indistinguishable from wild-type TFIIB, but significantly, the W63P, W63R, R64A, and F66D mutant proteins all conferred preferential usage of the more downstream transcription start site. The observed defects in transcriptional activity of the mutant proteins are unlikely to be due to gross misfolding, as all of the recombinant mutant proteins were competent to form stable TBP⅐TFIIB⅐DNA ternary complexes in a gel mobilityshift assay (data not shown). In addition, all of the mutant proteins that supported some basal transcription also supported a strong response to the transcriptional activator GAL4-VP16 (Fig. 3). These results demonstrate that the second rubredoxin knuckle and C-terminal ␤-strand of the zinc ribbon are critical motifs for basal transcription activity, whereas the adjacent homology block participates in transcription start site selection.
Determination of Transcription Start Sites in Vivo-The results of the in vitro transcription assays revealed that several of the mutants displayed an alteration in transcription start site selection on the CYC1 promoter. To confirm and extend these results, we examined in vivo start site utilization at the ADH1 and CYC1 promoters in the mutant strains. For comparison, we also analyzed RNA from strains containing the E62K and R64E substitution mutants, shown previously to alter start site utilization (7,18). Consistent with the in vitro results, mutants W63P, R64A, and F66D displayed a downstream shift in start site selection from both the CYC1 and ADH1 promoters, whereas the shifts observed for mutant W63R were less pronounced (Fig. 4).
Analysis of RNAPII-TFIIB Interaction-In previous work,  we identified a substitution mutant in the C-terminal ␤-strand of the zinc ribbon (L50D) that was unable to bind purified yeast RNAPII in vitro (7). Although this result suggests that the zinc ribbon is involved in the interaction between TFIIB and RNA-PII, additional mutants in the zinc ribbon would further establish its participation in RNAPII binding. Therefore, we determined the ability of the N-terminal mutants to form stable complexes with purified RNAPII using a co-immunoprecipitation assay. The C45S/C48S and L52P mutants did not form stable complexes with RNAPII, whereas the remaining mutants did complex with RNAPII, with the W63P protein exhibiting only slight impairment (Fig. 5). To confirm participation of the N-terminal region of yeast TFIIB in stable RNAPII binding, we tested the ability of recombinant core domain (residues 119 -346) to form stable complexes with RNAPII. Fulllength wild-type or L50D TFIIB protein was included in the binding reactions to act as internal positive and negative controls for stable RNAPII binding, respectively. As shown in Fig.  6, the core domain of TFIIB did not detectably bind RNAPII in these assays. Taken together, these results demonstrate that the second rubredoxin knuckle and the C-terminal ␤-strand of the zinc ribbon in the N-terminal region of TFIIB are involved in the direct interaction between TFIIB and RNAPII.

DISCUSSION
In this work, we have analyzed the in vivo and in vitro properties of 14 site-directed substitution mutants in the conserved N-terminal region of yeast TFIIB. Several structural motifs were targeted for mutagenesis, including the second rubredoxin knuckle and C-terminal ␤-strand of the zinc ribbon fold and an adjacent homology block that is conserved among seven eukaryotic species (Fig. 1). In vivo analyses revealed that two of the mutants, C45S/C48S and L52P, were incapable of supporting yeast cell growth, while an additional five mutants exhibited cold sensitivity and, in two cases, temperature sensitivity as well (Table I). In vitro transcriptional analysis demonstrated that both the C45S/C48S and L52P mutant proteins were severely impaired in their ability to support basal transcription (Fig. 3). These transcriptional defects correlated with an inability of these two mutant proteins to form stable complexes with purified RNAPII in vitro (Fig. 5). Similarly, the core domain, which lacks the N-terminal region and the zinc ribbon, also failed to bind RNAPII (Fig. 6). These results, combined with our previous determination that the L50D mutant is impaired for RNAPII binding, strongly suggest that the zinc ribbon fold, and in particular the C-terminal ␤-strand of the ribbon, are critical structural motifs for stable interaction of TFIIB with RNAPII.
Four of the five mutants that displayed conditional growth properties contain a substitution in the homology block adjacent to the zinc ribbon (Fig. 1). The one conditional mutant with a substitution outside of this homology block, H71E, exhibited a temperature-sensitive phenotype but was comparable to the wild-type with respect to transcription activity in vitro, RNA-PII binding, and transcription start site selection. The precise biochemical defect of the H71E mutant protein remains to be determined. The four conditional mutants with a substitution in the homology block (W63R, W63P, R64A, and F66D) exhibited cold-sensitive phenotypes and a downstream shift in transcription start site selection both in vivo and in vitro (Table I, Figs. 3 and 4). These results demonstrate that multiple residues within the highly conserved homology block adjacent to the zinc ribbon participate in selection of transcription start FIG. 2. Steady-state levels of wild-type and mutant TFIIB proteins in vivo. TFIIB protein levels were determined in strains containing both wild-type TFIIB and a wild-type or mutant polyhistidinetagged version as indicated. Whole-cell extracts were prepared and analyzed (50 g of total protein) by immunoblotting with a monoclonal antibody that recognizes the polyhistidine tag.  4. Analysis of in vivo start site selection from the CYC1 and ADH1 promoters. Total RNA (30 g) from strains containing the indicated SUA7 allele were analyzed by primer extension to map the start sites for both the CYC1 (A) and ADH1 (B) promoters. The numbers to the right of each panel denote the major start sites and indicate the distance from the initiating ATG (ϩ1). sites. In previous work, Hampsey and co-workers reported that E62K or R78C mutations conferred downstream shifts in start site selection (18). Residue Glu-62 is also contained within this homology block, whereas Arg-78 is C-terminal to this block, but also is an invariant residue among seven species (Fig. 1). In ongoing studies, we have utilized a random PCR-generated mutant TFIIB library and a genetic selection scheme to directly select for TFIIB mutants conferring downstream shifts in start site selection. To date, all of the mutants identified in this selection contain substitutions of residues in the homology block adjacent to the zinc ribbon or of residues Arg-78 or Val-79. 2 Our results demonstrate that two separable and adjacent functional domains in the N-terminal region of yeast TFIIB govern stable RNAPII binding and transcription start site selection (Fig. 7). Mutations that impair stable RNAPII binding, as determined by a co-immunoprecipitation assay, extend from the second rubredoxin knuckle of the zinc ribbon through the C-terminal ␤-strand of the ribbon to residue Glu-62. We suggest that mutation of the first rubredoxin knuckle of the zinc ribbon would also impair RNAPII binding, because it is highly likely that the integrity of the ribbon fold is required for proper RNAPII interaction. Mutations altering start site selection encompass residues Glu-62 to Phe-66, and also include the more C-terminal pair of conserved residues Arg-78 and Val-79. Mutants W63R, W63P, R64A, and F66D, all conferring downstream shifts in start site selection, were proficient for stable RNAPII binding. Moreover, the L52P mutant was defective for stable RNAPII binding, but the low level of transcription supported by this mutant protein initiated at normal start sites in vitro (Fig. 3). Thus, our results indicate that downstream shifts in start site selection are not caused by impairment in the stable association of TFIIB and RNAPII. It was reported previously that the E62K and R78C mutants, both conferring downstream shifts in start site selection, were impaired for RNAPII binding in a surface plasmon resonance assay (8). We suggest that residue Glu-62 is involved in both start site selection and RNAPII binding, i.e. the adjacent domains for stable RNAPII binding and start site selection overlap at residue Glu-62.
We have determined that substitutions of residue Arg-78 other than R78C, such as R78L, confer downstream shifts in start site selection without impairing stable RNAPII binding by our assay (7). 2 The failure of the R78C mutant to bind RNAPII perhaps results from the introduction of a cysteine residue adjacent to the zinc ribbon. The presence of this additional cysteine could interfere with normal coordination of zinc by the adjacent four cysteines in the zinc ribbon, thereby perturbing the ribbon structure and impairing RNAPII binding.
The mechanism by which transcription start sites are selected by RNAPII remains to be determined. In higher eukaryotes, transcription initiation usually occurs at a discrete start site located 25-30 base pairs downstream of the TATA element. In contrast, transcription initiation in S. cerevisiae frequently occurs at multiple start sites within a window of 30 -120 base pairs downstream of the TATA element (26,27). The window for transcription initiation in S. cerevisiae could result from preinitiation complexes that loop out a limited amount of intervening downstream DNA to interact with potential initiation sites. If so, TFIIB mutants could confer down- Purified recombinant TFIIB proteins were tested for their ability to form stable TFIIB⅐RNAPII complexes by co-immunoprecipitation. Binding reactions contained 100 ng of the indicated recombinant TFIIB protein and 500 ng of purified yeast RNAPII where indicated. Proteins were precipitated with a monoclonal antibody specific for RNAPII (8WG16, see "Experimental Procedures"). The input TFIIB protein (25% of total) and the precipitates (25% of total) were analyzed by immunoblotting using an affinity-purified anti-yeast TFIIB polyclonal antibody. The RNAPII-independent species of greater molecular weight than TFIIB corresponds to the heavy chain of the 8WG16 mouse IgG, which weakly cross-reacts with the goat anti-rabbit IgG secondary antibody. Shown in the left and right vertical panels are the results from two separate experiments.
FIG. 6. The core domain of yeast TFIIB does not detectably bind RNAPII. Binding reactions contained 35 ng of recombinant core TFIIB (residues 119 -346), 50 ng of wild-type or L50D proteins as positive or negative internal controls, respectively, and 500 ng of purified RNAPII where indicated. The input TFIIB proteins (50% of total) and the precipitates (50% of total) were analyzed by precipitation and immunoblotting as described in the legend to Fig. 5. The RNAPIIindependent species of greater and lower molecular weight than TFIIB correspond to the heavy and light chains of the 8WG16 mouse IgG, respectively, which weakly cross-react with the goat anti-rabbit IgG secondary antibody. The minor species in lanes 1 and 3 that co-migrates with core domain is a degradation product of the full-length proteins. FIG. 7. Summary of mutations that impair stable RNAPII binding or confer downstream shifts in start site selection. Yeast TFIIB residues 11-90 are marked with asterisks designating the positions of substitution mutants that impair stable RNAPII binding (asterisks above the amino acid sequence) or confer downstream shifts in transcription start site selection (asterisks below the amino acid sequence). stream shifts in start site selection by altering preinitiation complex structure so as to impair the ability to loop out shorter segments of DNA, thereby impairing the ability of the complexes to utilize start sites closer to the TATA element. However, it has been reported that promoter melting for the S. cerevisiae GAL1 and GAL10 genes in vivo is comparable to other eukaryotes, occurring approximately 20 base pairs downstream of the TATA element (28). The extent of promoter melting appears to be independent of the distance between the TATA box and the transcription start sites, suggesting that the positioning and subsequent promoter melting of S. cerevisiae preinitiation complexes is similar to mammalian preinitiation complexes. Thus, transcription initiation in the downstream window from S. cerevisiae TATA elements may be the result of a "scanning" polymerase (28). If start site selection in S. cerevisiae involves a scanning polymerase, upstream or downstream shifts in start site utilization could result from RNAPII with enhanced or diminished initiation site recognition, respectively. Upstream shifts in start site selection have been observed both in vitro and in vivo with yeast strains containing a deletion of the small RPB9 subunit of RNAPII or mutation of the C-terminal zinc-binding motif in RPB9 (29 -31). Mutation or deletion of the RPB9 subunit might alter the structure of polymerase on the DNA such that initiation sites are more efficiently recognized early in the scanning process. Conversely, downstream shifts, conferred by mutations in TFIIB or the RPB1 subunit of RNAPII (32), may be a consequence of a polymerase with diminished start site recognition. TFIIB might function in start site selection as a component of the scanning polymerase complex and assist in initiation site recognition and/or utilization (18). If so, mutations that impair the stable association of TFIIB and RNAPII might produce unstable scanning complexes, resulting in alterations in start site selection.
Although it remains possible that TFIIB is part of a scanning polymerase complex, our results demonstrate that most mutations that confer downstream shifts in start site selection do not affect stable association of TFIIB and RNAPII. Moreover, the L50D and L52P mutants, defective for stable RNAPII binding yet able to support low levels of transcription, support initiation at normal start sites. Therefore, if a scanning polymerase is involved in start site selection, our results are more consistent with TFIIB conferring properties for initiation site recognition upon RNAPII at the time of preinitiation complex assembly. In this model, the zinc ribbon of TFIIB has the major function in the stable interaction with RNAPII, whereas the adjacent homology block may interact with RNAPII in a more subtle manner to affect start site selection. Such an interaction between the homology block of TFIIB and RNAPII could affect the conformation of the polymerase on the DNA, thereby affecting subsequent start site recognition upon the onset of scanning. Continued biochemical analyses of the TFIIB mutants described here should provide additional insight into the mechanism of transcription start site selection and the role played by TFIIB.