Amino Acid Substitutions in Yeast TFIIF Confer Upstream Shifts in Transcription Initiation and Altered Interaction with RNA Polymerase II

Abstract

Transcription factor IIF (TFIIF) is required for transcription of protein-encoding genes by eukaryotic RNA polymerase II. In contrast to numerous studies establishing a role for higher eukaryotic TFIIF in multiple steps of the transcription cycle, relatively little has been reported regarding the functions of TFIIF in the yeast Saccharomyces cerevisiae. In this study, site-directed mutagenesis, plasmid shuffle complementation assays, and primer extension analyses were employed to probe the functional domains of the S. cerevisiae TFIIF subunits Tfg1 and Tfg2. Analyses of 35 Tfg1 alanine substitution mutants and 19 Tfg2 substitution mutants identified 5 mutants exhibiting altered properties in vivo. Primer extension analyses revealed that the conditional growth properties exhibited by the tfg1-E346A, tfg1-W350A, and tfg2-L59K mutants were associated with pronounced upstream shifts in transcription initiation in vivo. Analyses of double mutant strains demonstrated functional interactions between the Tfg1 mutations and mutations in Tfg2, TFIIB, and RNA polymerase II. Importantly, biochemical results demonstrated an altered interaction between mutant TFIIF protein and RNA polymerase II. These results provide direct evidence for the involvement of S. cerevisiae TFIIF in the mechanism of transcription start site utilization and support the view that a TFIIF-RNA polymerase II interaction is a determinant in this process.

Transcription of protein-encoding genes by eukaryotic RNA polymerase II (RNAPII) is a multistep process involving the concerted action of RNAPII and a host of auxiliary proteins (29). One of these auxiliary proteins, the general transcription factor IIF (TFIIF), has been shown to play a functional role during multiple steps of the RNAPII transcription cycle. Biochemical studies have shown that mammalian TFIIF, a heteromeric complex composed of the RAP74 and RAP30 subunits, facilitates the efficient entry of RNAPII into the preinitiation complex (PIC) and is required for the subsequent association of general transcription factors TFIIE and TFIIH (5, 7, 8, 12, 27, 31). Upon completion of PIC assembly, mammalian TFIIF is reported to induce the wrapping of promoter DNA around RNAPII in the PIC (34) and to facilitate the efficient escape of RNAPII from the promoter subsequent to the formation of the first phosphodiester bond of the nascent transcript (41). Lastly, a number of studies have shown that TFIIF can stimulate the activity of the RNAPII C-terminal domain (CTD) phosphatase Fcp1 (6) and can enhance the efficiency of transcript elongation by interacting with the elongating RNAPII to suppress pausing along the DNA template (1, 2, 21, 22, 26, 32, 38). Thus, the involvement of TFIIF in many mechanistic steps of the transcription cycle underscores the central importance and complexity of function for this factor.

Consistent with the multiple mechanistic roles played by TFIIF, biochemical analyses of higher eukaryotic TFIIF have identified multiple domains in the RAP74 and RAP30 subunits that mediate interactions with multiple transcription factors and DNA. The human RAP74 polypeptide (517 amino acids; calculated mass, ∼58 kDa; apparent mass by sodium dodecyl sulfate-polyacrylamide gel electrophoresis [SDS-PAGE], ∼74 kDa) comprises a globular N-terminal domain that binds RAP30 (40), a highly charged central linker region, and a globular CTD involved in the interaction with RNAPII (11), TFIIB (40), and Fcp1 (6, 24). The human RAP30 polypeptide (249 amino acids; calculated mass, ∼28 kDa; apparent mass by SDS-PAGE, ∼30 kDa) is composed of an N-terminal domain that binds RAP74 and TFIIB (11), a central domain that interacts with RNAPII (28, 35), and a cryptic C-terminal DNA-binding domain with similarity to that found in members of the σ⁷⁰ family of bacterial sigma factors (16, 17, 39).

Although numerous investigations have been reported regarding the functions of higher eukaryotic TFIIF, relatively few studies have been conducted regarding the analysis of the TFIIF homolog from the yeast Saccharomyces cerevisiae. The purification of native S. cerevisiae TFIIF, also known as Factor g, led to the identification of three copurifying polypeptides that were designated Tfg1, Tfg2, and Tfg3 (19). The Tfg1 polypeptide (735 amino acids; calculated mass, ∼82 kDa; apparent mass by SDS-PAGE, ∼105 kDa) is homologous to the human TFIIF RAP74 subunit yet is significantly larger than RAP74 (735 versus 517 amino acids) (Fig. 1A). Similarly, the Tfg2 polypeptide (400 amino acids; calculated mass, ∼47 kDa; apparent mass by SDS-PAGE, ∼54 kDa) is homologous to, but significantly larger than, the human TFIIF RAP30 subunit (400 versus 249 amino acids) (Fig. 1B). Tfg1 and Tfg2 are both required for yeast cell viability, whereas the Tfg3 subunit, also known as yeast TAF14, has no known counterpart in mammalian TFIIF, and strains containing deletions of Tfg3 are viable (19).

FIG. 1.

Comparison of S. cerevisiae, Schizosaccharomyces pombe, and human TFIIF homologs. Presented are proposed sequence alignments comparing the primary structures for (A) the TFIIF large subunit Tfg1 (human RAP74) from S. cerevisiae (S.c), S. pombe (S.p), and Homo sapiens (H.s) and (B) the TFIIF small subunit Tfg2 (human RAP30). Sequence alignments were generated using alignment program T-coffee version 1.41, and the shaded areas denote identical or conserved residues. Arrowheads indicate the residues in S. cerevisiae Tfg1 or Tfg2 that were subjected to site-directed mutagenesis, and those that gave rise to mutants exhibiting altered properties in vivo are highlighted by asterisks. (C) Depicted are the overall structures of the S. cerevisiae Tfg1 and Tfg2 proteins and their potential functional domains as suggested from the proposed sequence alignments and the reported functional domain boundaries of human RAP74 and RAP30.

In the work described in this paper, we utilized site-directed mutagenesis, plasmid shuffle complementation assays, and primer extension analysis to initiate an investigation of the functions of S. cerevisiae Tfg1 and Tfg2. We report the analyses of 54 mutant S. cerevisiae strains containing single amino acid substitutions in Tfg1 or Tfg2 and the identification of five amino acid substitutions that conferred altered growth properties. Three of these mutations, tfg1-E346A, tfg1-W350A, and tfg2-L59K, were found to confer pronounced upstream shifts in transcription initiation in vivo. Analyses of double mutant strains demonstrated functional interactions between the Tfg1 mutations and mutations in Tfg2, TFIIB, and RNA polymerase II that also confer alterations in start site utilization. Lastly, we report the results from in vitro experiments demonstrating an altered interaction between mutant TFIIF protein and RNA polymerase II. The results from this study firmly establish a role for S. cerevisiae TFIIF in the mechanism of transcription start site utilization, and we discuss them in the context of possible models where a TFIIF-RNAPII interaction plays an important role in the utilization of a transcription start site.

MATERIALS AND METHODS

Plasmids and construction of S. cerevisiae TFIIF mutants. (i) Tfg1 plasmids.

For p306/g1 and p316/g1, a 4.2-kb DNA fragment containing S. cerevisiae TFG1 was amplified from genomic DNA by PCR with the introduction of an XbaI site approximately 900 bp upstream of the coding region. The PCR product was digested with XbaI and XhoI, and the resulting 4-kb fragment was cloned into the yeast-Escherichia coli shuttle vectors pRS306 (URA3) and pRS316 (URA3) digested with the same enzymes. p306/g1Δ is a derivative of p306/g1 that has a deletion of the approximately 900-bp BamHI-BglII TFG1 fragment. p314/g1 was constructed by transferring the 4-kb TFG1 gene from p306/g1 into the yeast-E. coli shuttle vector pRS314 (TRP1) using XhoI and EagI. p314/g1-MH is a derivative of p314/g1 that contains both a Myc epitope tag at the N terminus and a hexahistidine tag at the C terminus of the Tfg1 coding sequence. Accompanying changes that were introduced by PCR include NheI-(Myc)-BglII at the beginning of the coding region and AgeI-(6His)-Stop-SalI at the end of the coding region.

(ii) Tfg2 plasmids. For p306/g2 and p316/g2, a 2.8-kb DNA fragment containing S. cerevisiae TFG2 was amplified from genomic DNA by PCR, with the introduction of an XbaI site and an XhoI site approximately 800 bp upstream and downstream of the coding region, respectively. The PCR product was digested with XbaI and XhoI, and the 2.8-kb fragment was cloned into vectors pRS306 and pRS316 digested with the same enzymes. For p306/g2Δ, plasmid p314/g2-H (see below) was digested with AflII and MluI to generate a 700-bp deletion, and the ends were then made blunt with avian myeloblastosis virus reverse transcriptase and religated. The resulting plasmid (p314/g2Δ) was digested with XbaI and XhoI, and the 2.1-kb fragment was cloned into pRS306 digested with the same enzymes. p314/g2 was generated by transferring the 2.8-kb TFG2 gene from p316/g2 into vector pRS314 using XhoI and EagI. p314/g2-H is a derivative of p314/g2 that contains a hexahistidine tag at the N terminus of the Tfg2 coding sequence. Accompanying changes that were introduced by PCR include NcoI-(6His)-AgeI at the beginning of the coding region and Stop-MluI-BamHI at the end of the coding region.

(iii) Tfg1 and Tfg2 site-directed mutants. Plasmids containing site-directed amino acid substitutions in Tfg1 or Tfg2 were constructed using the QuikChange Mutagenesis Kit (Stratagene) and plasmid p314/g1-MH or p314/g2-H as a PCR template, respectively.

(iv) Tfg1-Tfg2 coexpression plasmids. For the coexpression of nontagged versions of Tfg1 and Tfg2 (URA3 vector; p316/g1-g2), the 2.8-kb TFG2 gene from p306/g2 was PCR amplified with the introduction of a SacII site immediately downstream of the XhoI site, and the product was cloned into p316/g1 by using NotI and SacII. For the coexpression of epitope-tagged versions of Tfg1 and Tfg2 (TRP1 vector; p314/g1-g2), the 2.8-kb NotI-SacII fragment from p314/g2-H was cloned into p314/g1-MH digested with the same enzymes. The same approach was used in the construction of plasmids that coexpress combinations of Tfg1 and Tfg2 mutations [p314/g1-g2(59); p314/g1-g2(303); p314/g1(346)-g2; p314/g1(350)-g2; p314/346-59; p314/346-303; p314/350-59; p314/350-303]. All DNA fragments generated by PCR amplification in this study were verified by DNA sequencing (Health Research Inc., Roswell Park Cancer Institute). The complete lineages and DNA sequences of the plasmids used in this report are available upon request.

Yeast strains and media.

All S. cerevisiae strains used in this study are derivatives of S288C. The following strains were used: FP153 (MATα ura3-52 trp1Δ63 sua7Δ1 [+p316/yIIB {URA3}]), FP274 (MATα ura3-52 trp1Δ63 tfg1Δ1 [+p316/g1 {URA3}]), FP290 (MATa ura3-52 trp1Δ63 tfg2Δ1 [+p316/g2 {URA3}]), FP291 (MATα ura3-52 trp1Δ63 tfg1Δ1 tfg2Δ1 [+p316/g1-g2 {URA3}]), FP308 (MATa ura3-52 trp1Δ63 his3Δ200 leu2Δ1 rpb9::HIS3), FP310 (MATα ura3-52 trp1Δ63 his3Δ200 tfg1-E346A), FP321 (MATα ura3-52 trp1Δ63 tfg1-E346A sua7Δ1 [+p316/yIIB {URA3}]), FP325 (MATa ura3-52 trp1Δ63 his3Δ200 tfg1-E346A rpb9::HIS3), and FY251 (MATa ura3-52 trp1Δ63 his3Δ200 leu2Δ1).

Rich media (yeast extract-peptone-dextrose [YPD] plates and liquid) and 5-fluoroorotic acid (5-FOA; synthetic complete medium containing 1.2 mg of 5-fluoroorotic acid/ml) plates were prepared as described previously (27). YPG medium is equivalent to YPD with the exception of 3% glycerol substituted for dextrose. Casamino Acids (CAA) medium contained 0.6% Casamino Acids, 0.68% yeast nitrogen base without amino acids, 25 μg of adenine per ml, 25 μg of uracil per ml, 80 μg of tryptophan per ml, 2% dextrose, and for solid medium, 2% agar.

Phenotypic analyses.

To determine the effect of Tfg1 and Tfg2 mutations on yeast cell growth, the plasmid shuffle complementation assay was used (3). Plasmids containing Tfg1 mutations were analyzed in the TFG1 plasmid shuffle strain FP274, and plasmids containing Tfg2 mutations were analyzed in the TFG2 plasmid shuffle strain FP290. Plasmids containing various Tfg1-Tfg2 combinations were analyzed in the TFG1-TFG2 shuffle strain FP291. Yeast strains were transformed with the relevant test plasmid (TRP1-containing vector) and plated on CAA medium lacking uracil and tryptophan. The plates were incubated for 3 days at 30°C to select for cells harboring both the mutant test plasmid (TRP1-containing) and the endogenous plasmid (URA3) containing the wild-type S. cerevisiae TFG1 and/or TFG2 genes. The Ura⁺ Trp⁺ transformants were purified once on CAA medium lacking uracil and tryptophan and plated on 5-FOA medium, and the plates were incubated at room temperature for 4 days. 5-FOA, which is toxic to cells containing a functional URA3 gene, selects for cells that have spontaneously lost the URA3 plasmid containing wild-type TFG1 and/or TFG2. The appearance of 5-FOA-resistant colonies reflected the ability of a TFG1-TFG2 variant to support cell viability in the absence of its wild-type counterpart. 5-FOA-resistant colonies were purified once by streaking on CAA medium lacking tryptophan and incubating the plates at room temperature for 3 to 4 days. The growth properties of 5-FOA-resistant strains (confirmed to be Trp⁺ Ura⁻) were analyzed by growing the strains in liquid YPD at room temperature to an optical density at 600 nm of 1.0, spotting 5 μl of 10-fold serial dilutions on YPD plates, and incubating the plates for 3 days at 16, 23, 30, or 37°C. To determine the growth properties of Tfg1-TFIIB (sua7) double mutant strains, strains FP153 and FP321 were transformed with the indicated TFIIB plasmid (TRP1), and the Ura⁺ Trp⁺ transformants were plated on 5-FOA medium and analyzed as described above.

Immunoblotting.

To determine the steady-state levels of Tfg1 and Tfg2 mutant proteins, the relevant shuffle strain harboring a plasmid with the wild-type TFG1 or TFG2 gene (untagged in URA3 vector) in addition to the plasmid containing the test variant (Myc tagged and/or polyhistidine tagged in TRP1 vector) were grown to mid-exponential phase in liquid CAA lacking both tryptophan and uracil. Cells were harvested by centrifugation, and whole-cell extracts were prepared as described previously (28). Extracts (50 μg of protein) were resolved on an SDS-10% polyacrylamide gel, and the proteins were transferred to Immobilon-P polyvinylidene difluoride membranes (Millipore). The membranes were probed with a monoclonal antibody raised against the polyhistidine tag (RGS-His, 1:2,000 dilution; QIAGEN) or with monoclonal antibody raised against the Myc tag (1:2,000 dilution; Covance), followed by goat anti-mouse immunoglobulin G coupled to horseradish peroxidase (1:40,000 dilution; Jackson Laboratories). Immune complexes were visualized by the addition of SuperSignal Ultra chemiluminescent substrate (Pierce) and autoradiography.

Primer extension analysis.

Total RNA was isolated from yeast cultures (50 ml) grown in either YPD (for analysis of ADH1 transcripts) or YPG (for analysis of CYC1 transcripts). Strains were grown at 30°C to an optical density at 600 nm of 2.0, the cells were harvested by centrifugation and washed with diethyl pyrocarbonate-treated glass-distilled water, and the cell pellet was frozen at −80°C (all subsequent buffers for RNA isolation and primer extension analysis were prepared using diethyl pyrocarbonate-treated glass-distilled water). Cell pellets were thawed on ice and resuspended with 400 μl of buffer (10 mM Tris-HCl [pH 7.5], 10 mM EDTA, 0.5% SDS), and 400 μl of water-saturated (acidic) phenol was added. The samples were incubated at 68°C for 30 min with occasional mixing, incubated on ice for 5 min, and centrifuged at 16,000 × g for 5 min at room temperature. Approximately 400 μl of the aqueous phase was harvested and extracted successively, once with acidic phenol, twice with buffered phenol-chloroform, and once with chloroform. Total RNA was precipitated with ethanol and resuspended in 50 μl of 10 mM Tris-HCl (pH 8.0)-1 mM EDTA. For the mapping of in vivo mRNA 5′ ends, 30 μg of total RNA was analyzed by primer extension as described previously (30) by using an ADH1-specific primer (5′-dGTATTCCAACTTACCGTGGGATTCG-3′, corresponding to positions +63 to +39, where +1 is defined as the A in the translation-initiating ATG) or a CYC1-specific primer (5′-dGTCTTGAAAAGTGTAGCACC-3′, corresponding to positions +53 to +34, where +1 is defined as the A in the translation-initiating ATG).

In vitro transcription and electrophoretic mobility shift assays.

Reconstituted transcription reactions with purified RNAPII and yeast general transcription factors were performed as described previously (42). Electrophoretic gel mobility shift assays for RNAPII-TFIIF binding in the absence of additional factors were performed as described previously (42) and contained an internally labeled E1B TATA probe and combinations of 100 ng of wild-type or ΔRpb9 RNAPII and 100 ng of wild-type or Tfg1-E346A recombinant TFIIF.

RESULTS

Isolation and characterization of S. cerevisiae Tfg1 and Tfg2 mutants exhibiting altered growth properties.

In contrast to the numerous studies that have established a role for mammalian TFIIF in multiple steps of the RNAPII transcription cycle, relatively few studies have been conducted regarding the functions of TFIIF in S. cerevisiae. Since the S. cerevisiae RNAPII machinery exhibits unique aspects of transcription initiation, and since the S. cerevisiae TFIIF subunits Tfg1 and Tfg2 are both significantly larger than their mammalian counterparts (Fig. 1), a genetic analysis of S. cerevisiae TFIIF was initiated in order to investigate the unique and conserved functions of S. cerevisiae TFIIF. Presented in Fig. 1A and B are proposed alignments of the amino acid sequences for the Tfg1 and Tfg2 homologs from S. cerevisiae, Schizosaccharomyces pombe, and humans. The alignments identify regions of the two proteins exhibiting the greatest homology and divergence between the species and highlight the higher degree of overall conservation between the S. cerevisiae and Schizosaccharomyces pombe homologs (Tfg1, 20% identity and 27.5% similarity; Tfg2, 26.5% identity and 27.3% similarity).

In order to probe the functional domains of S. cerevisiae TFIIF and to isolate mutants with altered transcriptional properties in vivo, site-directed mutagenesis was employed to introduce single amino acid substitutions at conserved or similar residues in the Tfg1 and Tfg2 subunits (Fig. 1A and B). In selecting the residues to undergo mutagenesis in these initial studies, we chose to avoid the extended highly charged regions of Tfg1 and Tfg2 using the rationale that single amino acid substitutions in these regions would be less likely to confer detectable alterations. For Tfg1, 35 (mostly conserved) residues were chosen for mutagenesis and included 20 residues in the N-terminal domain, 7 residues near the end of the C-terminal domain, and 8 residues in the proposed central region (Fig. 1A and C). The selection of six of these eight central region residues was based upon their location in a region that is homologous to the human RAP74 α1 helix, which previous studies have shown is sensitive to amino acid substitutions that reduce the activities of RAP74 for both initiation and elongation stimulation in an in vitro system (14, 25, 33).

For Tfg2, residues chosen for mutagenesis included 7 in the C-terminal DNA binding region, 1 in the central region, and 11 in the N-terminal domain (Fig. 1B and C). Given the fact that the isolation of yeast strains exhibiting altered growth phenotypes has proven extremely successful in the identification of numerous transcription factors with altered properties in vivo, plasmids containing the Tfg1 or Tfg2 mutations were initially tested for their ability to support cell viability using the plasmid shuffle assay and strain FP274 (for Tfg1) or strain FP290 (for Tfg2). The results demonstrated that all 35 of the plasmids containing Tfg1 alanine substitution mutations could support at least minimal cell growth in the absence of wild-type TFG1 on 5-FOA medium (Fig. 2A and data not shown). Similarly, 18 of the 19 plasmids containing Tfg2 mutations were competent for supporting cell viability, with the lone exception being a plasmid containing an aspartic acid substitution for Val-71 (Fig. 3A and data not shown). The viable mutant strains were then tested for their relative growth rates on rich (YPD) medium at 16, 23 (ambient), 30, and 37°C. Of the 35 tfg1 mutants examined, 33 mutants exhibited growth properties that were indistinguishable from the corresponding wild-type TFG1 strain, whereas 2 mutants, which contain alanine substitutions for Glu-346 (E346A) or Trp-350 (W350A), exhibited dramatically reduced growth rates at low or elevated temperature, respectively (Fig. 2B and data not shown). Immunoblotting confirmed that the observed growth phenotypes were not due to instability of the mutant proteins, as the steady-state levels of both the E346A and W350A mutant proteins were comparable to that of wild-type Tfg1 protein (Fig. 2C).

FIG. 2.

Growth properties of S. cerevisiae tfg1 mutants. (A) Plasmid shuffle complementation assay. Shown is the 5-FOA plate after 3 days of incubation at 23°C for transformants of the TFG1 shuffle strain FP274 containing a plasmid expressing epitope-tagged versions of wild-type S. cerevisiae Tfg1 or the indicated Tfg1 variant. (B) Relative growth rates at different temperatures. Strains were grown in liquid YPD, and 10-fold serial dilutions were spotted on YPD plates and incubated for 2 to 3 days at the indicated temperatures. Mutants with the following mutations had growth properties that were indistinguishable from those of the corresponding wild-type TFG1 strain: R322A, R326A, R327A, K328A, L332A, P349A, G357A, G363A, G368A, S372A, Y373A, D381A, G382A, F384A, P388A, Y393A, F395A, T396A, T403A, L404A, T405A, E410A, M413A, L424A, M425A, H427A, I717A, K720A, L721A, R723A, K724A, D728A, and M730A. (C) Immunoblotting. Whole-cell extracts were prepared from the indicated strains containing the indicated tfg1 variant and analyzed (50 μg of total protein) by immunoblotting using a monoclonal antibody recognizing the Myc epitope tag.

FIG. 3.

Growth properties of S. cerevisiae tfg2 mutants. (A) Plasmid shuffle complementation assay. Shown is the 5-FOA plate after 3 days of incubation at 23°C for transformants of the TFG2 shuffle strain FP290 containing empty vector or a plasmid expressing epitope-tagged versions of wild-type S. cerevisiae Tfg2 or the indicated Tfg2 variant. (B) Relative growth rates at different temperatures. Strains were grown in liquid YPD, and 10-fold serial dilutions were spotted on YPD plates and incubated for 2 to 3 days at the indicated temperatures. Mutants with the following mutations had growth properties that were indistinguishable from those of the corresponding wild-type TFG2 strain: E36A, D58A, D60A, W69R, L70E, P74A, G93D, R96A, I97R, P226A, K290A, R293A, D301A, Q324A, P325R, and E354A. (C) Whole-cell extracts were prepared and analyzed (50 μg of total protein) by immunoblotting with a monoclonal antibody recognizing the hexahistidine tag. The arrow indicates the position of Tfg2p.

Of the 18 tfg2 mutants analyzed for relative growth at different temperatures, 16 exhibited growth properties that were indistinguishable from those of the corresponding wild-type TFG2 strain (Fig. 3B and data not shown). However, the mutants containing a lysine substitution for Leu-59 (L59K) or an arginine substitution for Leu-303 (L303R) exhibited significantly reduced growth at elevated or low temperature, respectively (Fig. 3B). Immunoblotting confirmed that the steady-state levels of the L59K and L303R mutant proteins, as well as the V71D mutant protein that was unable to support viability on 5-FOA, were comparable to that of wild-type Tfg2 protein (Fig. 3C). Taken together, these results indicate that the E346A and W350A substitutions in Tfg1 and the L59K, V71D, and L303R substitutions in Tfg2 impair one or more important functions of S. cerevisiae TFIIF in vivo.

To test for potential functional interactions between the identified Tfg1 and Tfg2 mutations, the growth properties of strains containing various combinations of the tfg1 and tfg2 alleles were determined. Plasmids containing wild-type or mutant TFG1 in combination with wild-type or mutant TFG2 (TRP1 vector) were constructed and introduced into the TFG1-TFG2 plasmid shuffle strain FP291 (see Materials and Methods). As anticipated, strains containing a wild-type subunit in combination with a mutant subunit on the TRP1 plasmid (i.e., TFG1 and tfg2-L59K; TFG1 and tfg2-L303R; tfg1-E346A and TFG2; and tfg1-W350A and TFG2) were viable on 5-FOA medium (Fig. 4A), and their relative growth rates were essentially the same as those of their corresponding mutant strains that were previously analyzed (in FP274 or FP290) (Fig. 4C). One exception, however, was the FP291 derivative that contained TFG1 and tfg2-L303R, which exhibited reduced growth rates at both 37 and 16°C (Fig. 4C), in contrast to the relatively proficient growth at elevated temperature that was exhibited by the FP290 derivative containing tfg2-L303R (Fig. 3B).

FIG. 4.

Growth properties of S. cerevisiae tfg1 tfg2 double mutants. (A and B) Complementation assays for combined Tfg1 and Tfg2 function. Shown are the 5-FOA plates after 3 days of incubation for transformants of the TFG1-TFG2 shuffle strain FP291 containing a plasmid coexpressing a wild-type subunit in combination with a mutant of the other subunit (A) or a mutant of both the tfg1 and tfg2 subunits (B). (C) Relative growth rates at different temperatures. Strains containing the indicated combination of TFG1 and TFG2 alleles were grown in liquid YPD, and 10-fold serial dilutions were spotted on YPD plates and incubated for 2 to 3 days at the indicated temperatures. (D) Immunoblotting. Whole-cell extracts were prepared, and 50 μg of total protein was analyzed using monoclonal antibodies recognizing the Myc epitope tag (for Tfg1) or the hexahistidine tag (for Tfg2). The arrows indicate the position of the Tfg1 and Tfg2 proteins.

Analysis of the tfg1 tfg2 double mutant strains revealed that strains containing the combination of the tfg1-E346A mutation with either tfg2-L59K or tfg2-L303R were viable on 5-FOA medium, whereas the combination of the tfg1-W350A allele with either tfg2-L59K or tfg2-L303R was lethal (Fig. 4B). When tested for growth at different temperatures, the tfg1-E346A tfg2-L59K double mutant strain grew poorly at both cold and elevated temperatures, consistent with the additive defects conferred by the tfg1-E346A mutation (cold sensitive) and the tfg2-L59K mutation (temperature sensitive) (Fig. 4C). Interestingly, however, the tfg1-E346A tfg2-L303R double mutant strain exhibited relatively good growth at both cold and elevated temperatures, in contrast to the temperature sensitivity exhibited by the tfg2-L303R strain and the cold sensitivity exhibited by both the tfg1-E346A and tfg2-L303R strains (Fig. 4C). This suppression of the conditional phenotypes that are associated with the single mutants suggests an interaction between the tfg1-E346A and tfg2-L303R mutant subunits that partially restores an important TFIIF function.

Amino acid substitutions in S. cerevisiae Tfg1 or Tfg2 can confer upstream shifts in transcription initiation in vivo.

To analyze the effects of the Tfg1 and Tfg2 mutations on transcription initiation in vivo, strains were grown in either liquid YPG (for the analysis of CYC1 transcripts) or YPD (for the analysis of ADH1 transcripts), and total RNA was isolated and analyzed by primer extension using an ADH1-specific (Fig. 5A) or CYC1-specific (Fig. 5B) primer. Compared to the wild-type strain, a striking increase in the levels of transcripts mapping further upstream than the major start sites was observed at both the ADH1 and CYC1 promoters in the tfg1-E346A, tfg1-W350A, and tfg2-L59K mutant strains (Fig. 5A and B). The tfg1-E346A strain exhibited the most dramatic increase in the levels of these upstream transcripts, followed by a lesser yet substantial increase of these transcripts in the tfg2-L59K strain. These results strongly suggest that S. cerevisiae TFIIF plays a role in the mechanism of transcription start site utilization and indicate that mutations in either the Tfg1 or Tfg2 subunit can alter this process.

FIG. 5.

Amino acid substitutions in Tfg1 or Tfg2 can confer upstream shifts in transcription initiation in vivo. Total RNA (30 μg) from strains containing the indicated alleles of S. cerevisiae TFG1 or TFG2 were analyzed by primer extension utilizing an ADH1-specific (A) or CYC1-specific (B) oligonucleotide primer. The numbers to the left of each panel indicate the position of transcription start sites, where +1 is defined as the A in the translation-initiating ATG. The arrows indicate the positions of the enhanced upstream transcripts.

The tfg1-E346A mutation functionally interacts with mutations in RNAPII and TFIIB that confer alterations in start site utilization.

To further investigate the potential role of TFIIF in the mechanism of start site utilization, we tested whether a TFIIF mutation conferring an upstream shift in start site utilization functionally interacts with mutations in RNAPII and TFIIB that have previously been shown to confer shifts in transcription initiation. This was addressed by analyzing the growth properties and in vivo transcription initiation patterns of double mutant strains containing the tfg1-E346A mutation in combination with an RNAPII Δrpb9 mutation that confers upstream shifts in transcription initiation (20, 42) or in combination with various TFIIB mutations that confer various degrees of downstream shifts (10, 30). The results demonstrated that the tfg1-E346A Δrpb9 double mutant strain exhibited a more severe growth impairment at several temperatures than either the tfg1-E346A or Δrpb9 mutant strains (Fig. 6A). Primer extension analysis revealed that the more severe growth defect of the tfg1-E346A Δrpb9 double mutant strain was associated with an exacerbation of upstream shifts in transcription initiation in vivo (Fig. 6B). In contrast, double mutant strains containing the tfg1-E346A mutation in combination with one of several TFIIB mutations that confer downstream shifts were healthier than the corresponding single mutant strains (Fig. 7A), and the double mutant strains exhibited diminished upstream shifts and, to a lesser extent, diminished downstream shifts in vivo (Fig. 7B). Taken together, these results demonstrate functional interactions between the tfg1-E346A mutation and mutations in both TFIIB and Rpb9, further supporting a role for S. cerevisiae TFIIF in the mechanism of start site utilization.

FIG. 6.

Combining the tfg1-E346A and Δrpb9 mutations results in a severe growth defect and exacerbated upstream shifts in transcription initiation in vivo. (A) Relative growth at different temperatures. Strains containing the indicated combination of TFG1 and RPB9 alleles were grown in liquid YPD, and 10-fold serial dilutions were spotted on YPD plates and incubated for 2 to 3 days at the indicated temperatures. (B) Primer extension analysis. Total RNA (30 μg) from strains containing the indicated alleles of TFG1 and RPB9 were analyzed utilizing an ADH1-specific primer. The numbers to the left of each panel indicate the position of transcription start sites, where +1 is defined as the A in the translation-initiating ATG. The arrows indicate the positions of the enhanced upstream transcripts.

FIG. 7.

Functional interactions between tfg1-E346A and mutations in the TFIIB finger domain result in improved growth and a partial suppression of shifts in transcription initiation in vivo. (A) Relative growth at different temperatures. Strains containing the indicated combination of TFG1 and SUA7 (TFIIB) alleles were grown in liquid YPD, and 10-fold serial dilutions were spotted on YPD plates and incubated for 2 to 3 days at the indicated temperatures. (B) Primer extension analysis. Total RNA from the indicated strains was analyzed as described in the legend to Fig. 6. The arrows indicate the positions of the enhanced upstream transcripts conferred by the tfg1-E346A mutation, and the asterisks indicate the positions of the enhanced downstream transcripts conferred by TFIIB finger mutations.

Upstream shifts in transcription initiation conferred by the tfg1-E346A and RNAPII Δrpb9 mutations are associated with impaired TFIIF-RNAPII interaction.

Recent studies from our laboratory demonstrated that upstream shifts in transcription initiation conferred by the Δrpb9 mutation are associated with an impaired interaction between TFIIF and the ΔRpb9 mutant polymerase in vitro (42). To begin an investigation into the biochemical basis by which TFIIF mutations cause upstream shifts, we analyzed the activities of ΔRpb9 RNAPII and mutant TFIIF containing a Tfg1-E346A subunit in both reconstituted transcription reactions and electrophoretic mobility shift assays. Transcription assays were performed using a plasmid template containing the S. cerevisiae ADH1 promoter, purified recombinant factors (TBP, TFIIB, TFIIE, and wild-type or Tfg1-E346A TFIIF) and factors purified from yeast extracts (wild-type RNAPII, the ΔRpb9 RNAPII, and TFIIH) (42). Primer extension analysis of the transcription reaction products revealed that, compared to the reaction containing wild-type factors, the reactions containing Tfg1-E346A or the ΔRpb9 RNAPII yielded increased levels of more-upstream transcripts (Fig. 8A, lanes 2 to 4), while a reaction containing both Tfg1-E346A TFIIF and ΔRpb9 RNAPII yielded dramatically reduced levels of overall transcription (Fig. 8A, lane 5). To address the hypothesis that a TFIIF-RNAPII interaction plays a role in the mechanism of upstream shifts, TFIIF-RNAPII binding was assayed using electrophoretic mobility shift assays that contained combinations of purified wild-type or ΔRpb9 RNAPII and wild-type or Tfg1-E346A TFIIF in the absence of any additional general transcription factors. The results shown in Fig. 8B demonstrated that, compared to the reaction containing wild-type TFIIF and RNAPII (Fig. 8B, lane 3), the reactions containing wild-type RNAPII and Tfg1-E346A TFIIF (Fig. 8B, lane 4) or ΔRpb9 RNAPII and wild-type TFIIF (Fig. 8B, lane 6) yielded dramatically reduced levels of stable DNA-RNAPII-TFIIF (PolF) complexes. Moreover, essentially no stable PolF complexes were detected in the reaction containing the combination of Tfg1-E346A TFIIF and ΔRpb9 RNAPII (Fig. 8B, lane 7). These results indicate that mutations in TFIIF or RNAPII that confer upstream shifts are associated with an impaired TFIIF-RNAPII interaction, and they strongly suggest that this interaction plays an important role in the productive utilization of a transcription start site.

FIG. 8.

Upstream shifts in transcription initiation conferred by the tfg1-E346A and Δrpb9 mutations are associated with impaired TFIIF-RNAPII interaction. (A) In vitro transcription. Reconstituted transcription reactions with purified RNAPII and yeast general transcription factors were performed as described previously (42), and transcript 5′ ends were mapped by primer extension using an ADH1-specific primer. (B) Electrophoretic mobility shift assay for RNAPII-TFIIF binding. Reactions were performed as described previously (42) and contained an internally labeled E1B TATA probe and the indicated combination of 100 ng of wild-type or ΔRpb9 RNAPII and 100 ng of wild-type or Tfg1-E346A recombinant TFIIF. The mobilities of the RNAPII-DNA (Pol) and RNAPII-TFIIF-DNA (PolF) complexes are indicated.

DISCUSSION

In this study, we initiated a structure-function analysis of S. cerevisiae TFIIF by utilizing site-directed mutagenesis to introduce single amino acid substitutions at mostly conserved residues in the Tfg1 and Tfg2 subunits (Fig. 1). Using plasmid shuffle complementation assays, 19 tfg2 and 35 tfg1 mutant strains were constructed and initially analyzed for their relative growth properties. One of the tfg2 mutants was found to be inviable (tfg2-V71D), and an additional two mutants exhibited conditional growth phenotypes (tfg2-L59K, temperature sensitive; tfg2-L303R, cold sensitive) (Fig. 3B). All 35 tfg1 mutants were viable, but 2 of them exhibited conditional growth phenotypes (tfg1-E346A, cold sensitive; tfg1-W350A, temperature sensitive) (Fig. 2B). We did not observe any discernible difference in the growth properties of a wild-type TFG1 strain and tfg1 mutant strains containing substitutions in the region homologous to the human RAP74 α1 helix (Thr-403, Leu-404, Thr-405, Glu-410, Met-413, and Leu-424). As noted earlier, mutant RAP74 proteins containing substitutions in the α1 helix were reported to exhibit approximately 20 to 30% of the initiation and elongation stimulation activities that are observed with wild-type RAP74 in an in vitro system (14, 25, 33). The essentially wild-type growth properties displayed by the mutant strains containing substitutions in this region of Tfg1 may have several explanations. First, the α1 helix may play a more important role in the functions of human RAP74 than it does in S. cerevisiae Tfg1. In this regard, it is noteworthy that RAP74 contains a highly charged region of approximately 40 amino acids immediately C-terminal to the α1 helix that is not present in either S. cerevisiae or Schizosaccharomyces pombe Tfg1 (Fig. 1A). Alternatively, the amino acid substitutions in this region of Tfg1 may reduce TFIIF activity, but the degree of impairment is not sufficient to significantly alter the rate of cell growth. Additional biochemical and mutational studies will be needed to elucidate the potential role of the α1 helix in S. cerevisiae Tfg1 function.

Using primer extension, we directly analyzed the effects of the Tfg1 and Tfg2 substitution mutations on transcription initiation in vivo by mapping the transcription start sites utilized at the ADH1 and CYC1 promoters in the mutant strains. Strikingly, these analyses revealed that the tfg1-E346A, tfg1-W350A, and tfg2-L59K mutations all conferred significant upstream shifts in start site utilization at both the ADH1 and CYC1 promoters (Fig. 5). The shifts conferred by the tfg1-E346A substitution were especially dramatic and were significantly more pronounced than the upstream shifts that are observed in strains containing a deletion of the RNAPII Rpb9 subunit (Fig. 6B) (20, 42). These results provided the first direct evidence for a role for TFIIF in the mechanism by which a potential start site is productively utilized in S. cerevisiae. Although such a role for TFIIF was suggested by the isolation of a Tfg1 mutant that suppressed the downstream shifts conferred by a TFIIB mutation, the nature of the mutation in Tfg1 and its effects on start site utilization in a wild-type TFIIB strain were not reported (37). Thus, the results from the studies reported here provide the first identification of substitutions in Tfg1 and Tfg2 that alter the mechanism of start site utilization and the first direct demonstration that substitutions in either Tfg1 or Tfg2 can confer upstream shifts in an otherwise wild-type strain.

To further investigate the effects of the TFIIF mutations on transcription in vivo, we tested for functional interactions between the TFIIF mutations and previously identified mutations in TFIIB or RNAPII that confer alterations in start site utilization by constructing double mutant strains and determining their relative growth properties and in vivo transcription initiation patterns. Analyses of tfg1 tfg2 double mutant strains revealed that the combined alterations conferred by the tfg1-W350A mutation with either tfg2-L59K or tfg2-L303R were lethal to the cell, whereas interestingly, the tfg1-E346A tfg2-L303R double mutant strain grew significantly better at the restrictive temperatures than either of the corresponding single mutant strains (Fig. 4B and C). The phenotypic suppression exhibited by the tfg1-E346A tfg2-L303R double mutant strain is apparently not due to a suppression of the alteration in start site utilization conferred by the tfg1-E346A mutation, since the degree of upstream shifts at the ADH1 and CYC1 promoters in the double mutant strain were not significantly different than those observed in the tfg1-E346A single mutant strain (data not shown). Further studies will be needed to elucidate the underlying biochemical basis for the functional interactions between the Tfg1 and Tfg2 mutant subunits resulting in the observed phenotypic suppression or lethality.

A role for TFIIF in start site utilization was further substantiated by the analyses of double mutant strains containing tfg1-E346A in combination with the Δrpb9 mutation or in combination with TFIIB (sua7) mutations that confer various degrees of downstream shifts. Compared to the tfg1-E346A or Δrpb9 single mutants, the tfg1-E346A Δrpb9 double mutant exhibited more severe growth impairment and more pronounced upstream shifts in transcription initiation in vivo (Fig. 6). Although it is possible that the observed additive effects of the tfg1-E346A and Δrpb9 mutations reflect the action of these mutations on two separate pathways leading to start site utilization, their effect on a common pathway is suggested by the results demonstrating that both mutations impair RNAPII-TFIIF interaction (Fig. 8) (see below).

In contrast to the tfg1-E346A Δrpb9 double mutant, the tfg1-E346A sua7 double mutants grew better at the restrictive temperatures than the corresponding single mutant strains and displayed diminished upstream shifts in vivo (Fig. 7). In utilizing four sua7 alleles that confer various degrees of downstream shifts, it was also evident that the extent of suppression of the upstream shifts conferred by tfg1-E346A was directly related to the degree of downstream shifts conferred by a given sua7 mutation. Of the sua7 mutations examined, the sua7-E62V substitution conferred the most-pronounced downstream shifts in a wild-type TFG1 strain and the greatest degree of suppression of upstream shifts in a tfg1-E346A strain (Fig. 7B, lanes 3 and 4). The sua7-F66D mutation, which conferred the least-dramatic downstream shift in a wild-type TFG1 strain, was likewise the least effective sua7 allele in suppressing the upstream shifts in a tfg1-E346A strain (Fig. 7B, lanes 7 and 8). Moreover, in contrast to the other sua7 alleles, the downstream shifts conferred by sua7-F66D were also partially suppressed by the tfg1-E346A mutation (Fig. 7B, lanes 7 and 8). Taken together, the results from these studies provide further support for a functional role for TFIIF in the mechanism of start site utilization and demonstrate the additive or offsetting effects of the mutations in TFIIF, RNAPII, and TFIIB.

Although a high-resolution crystal structure for the S. cerevisiae Tfg1-Tfg2 complex is yet to be reported, such a structural determination has been made for the human RAP30/RAP74 dimerization domains that contain RAP30 residues 2 to 119 and RAP74 residues 2 to 172 (15). Despite the noted divergence in the primary structures between S. cerevisiae Tfg1 and RAP74 and between Tfg2 and RAP30, insight into the potential biochemical alterations of our S. cerevisiae TFIIF mutants was sought by examining the locations of the (conserved) residues in the RAP30/RAP74 structure that correspond to those mutated in Tfg1 and Tfg2. With the exception of Tfg2 Leu-303, which resides in the proposed C-terminal DNA-interaction domain (Fig. 1C), the residues mutated in the identified Tfg1 and Tfg2 mutants are proposed to correspond to conserved residues in RAP30 and RAP74 that reside within the determined RAP30/RAP74 structure. Tfg2 Val-71, which when mutated in the tfg2-V71D mutant is unable to support cell viability, is homologous to RAP30 Val-21 that is positioned in the middle of the β2 strand (15). The position of Val-21 in the β2 strand suggests that it is most likely involved in the important hydrophobic interactions that stabilize the stacking of multiple β strands in this region. Accordingly, the lethality associated with the tfg2-V71D mutant is most likely due to a significant alteration in the proper folding of the Tfg2 subunit. Importantly, however, the RAP30 and RAP74 residues corresponding to those mutated in the Tfg1 and Tfg2 mutants conferring upstream shifts in start site utilization are all proposed to reside on the same side of the RAP30/RAP74 dimer in a solvent-exposed region for potential interaction with other factors. These include RAP30 Leu-9 (Tfg2 Leu-59), located at the beginning of the α1 helix, RAP74 Glu-94 (Tfg1 Glu-346), located in the loop between β strands β5 and β6, and RAP74 Trp-98 (Tfg1 Trp-350), which resides in the beginning of β6 (15). Thus, we anticipated that the Tfg1-E346A, Tfg1-W350A, and Tfg2-L59K substitutions were not impairing the association of the Tfg1 and Tfg2 subunits but rather were conferring upstream shifts in start site utilization by virtue of an impaired interaction with another factor. To initiate a biochemical analysis of mutant TFIIF proteins, recombinant TFIIF complexes containing both wild-type subunits (Tfg1 and Tfg2), a mutant Tfg1 subunit (Tfg1-E346A and Tfg2), or a mutant Tfg2 subunit (Tfg1 and Tfg2-L59K) were produced and purified as described previously (42). We did not observe any significant difference in the 1:1 stoichiometric recovery of the Tfg1 and Tfg2 subunits from the purified wild-type and mutant TFIIF complexes, supporting the view that the Tfg1 and Tfg2 mutations were not impairing the association of the Tfg1 and Tfg2 subunits (data not shown). Since we recently determined that the upstream shifts in transcription initiation conferred by RNAPII lacking the Rpb9 subunit were associated with an impaired interaction between the ΔRpb9 mutant polymerase and TFIIF, we utilized mobility shift assays to test the hypothesis that the Tfg1 and Tfg2 substitutions conferring upstream shifts were similarly altering the interaction between RNAPII and TFIIF. The results confirmed that there was an impaired interaction between mutant TFIIF and wild-type RNAPII, as evidenced by a dramatic reduction in the amount of stable PolF complexes (Fig. 8B, lanes 3 and 4). Moreover, the combination of mutant TFIIF (Tfg1-E346A and Tfg2) and ΔRpb9 RNAPII, which in vivo resulted in severe growth impairment and an exacerbation of upstream shifts (Fig. 6), resulted in a dramatic reduction in the level of overall transcription in vitro (Fig. 8A, lane 5) and essentially undetectable levels of stable PolF complexes (Fig. 8B, lane 7). The observed correlation between the degree of impairment in the formation of stable PolF complexes and the extent of upstream shifts strongly suggests that an RNAPII-TFIIF interaction plays an important role in the utilization of a potential start site.

If an RNAPII-TFIIF interaction indeed plays a role in S. cerevisiae start site utilization, what might be the mechanism by which an alteration of this interaction leads to an increase in the number of transcripts with 5′ ends mapping further upstream than normal? In considering this question, it is important to note the unique feature of RNAPII transcription in S. cerevisiae. In contrast to the case in higher eukaryotes, where transcription initiation usually occurs at a discrete site located 25 to 30 bp downstream of the TATA element, the positions of mRNA 5′ ends in S. cerevisiae frequently map to multiple sites ranging from 45 to 120 bp or greater downstream of the TATA element (18, 36). One possibility that may account for this difference is that the architecture of S. cerevisiae PICs is fundamentally different from that of higher eukaryotic PICs. Although the architecture of yeast PICs remains to be determined, published reports from numerous laboratories strongly support a model for the architecture of mammalian PICs that involves the wrapping of approximately 90 bp of promoter DNA around the PIC (4, 9, 13, 23, 34). The wrapping of promoter DNA around a mammalian PIC is reportedly induced by TFIIF and involves DNA sequences ranging from about 30 bp upstream to 55 bp downstream of the TATA element (34). Since mammalian TFIIF functions as a heterodimer of the RAP30 and RAP74 subunits, and since both the Tfg1 and Tfg2 subunits are significantly larger than their mammalian counterparts, it is possible that S. cerevisiae PICs have the potential to wrap significantly more promoter DNA than mammalian PICs in order to contact transcription start sites further downstream of the TATA element. In this case, an altered RNAPII-TFIIF interaction, resulting either from TFIIF or RNAPII (ΔRpb9) mutations, might reduce the efficiency and/or extent of wrapping of promoter DNA around RNAPII that is induced by TFIIF. Such a reduction could lead to an increase in the number of PICs with the RNAPII active center in contact with more-upstream DNA, thereby resulting in an increase in the number of transcripts with 5′ ends mapping further upstream than normal. Another possibility is that the architecture of S. cerevisiae PICs is similar to that of higher eukaryotes but that, unlike in higher eukaryotes, RNAPII can be brought into contact with downstream sequences through an energy-dependent translocation subsequent to PIC assembly. Since TFIIF can interact with RNAPII to suppress pausing along the DNA template (1, 2, 21, 22, 26, 32, 38), an impaired RNAPII-TFIIF interaction might increase the amount of pausing of a translocating RNAPII. In this case, RNAPII may spend a longer time in contact with potential start sites encountered early in the translocation process, thereby increasing the probability that they will be productively utilized. Continued biochemical studies should prove invaluable in elucidating the concerted roles of TFIIF, TFIIB, and RNAPII in the mechanism of S. cerevisiae start site utilization.