INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The structures and functions of eukaryotic RNA polymerases I, II, and III are well conserved during evolution (reviewed in references 6, 41, and 53). Structurally, these three enzymes contain two large subunits which are homologous to the two largest subunits,  and ', of Escherichia coli RNA polymerase (RNAP) and which contain the active site of the enzyme (reviewed in references 6, 41, and 53). In addition to containing the two largest subunits, eukaryotic RNAPs also contain approximately 10 smaller subunits, some of which are present in both RNAPI and RNAPII or in all three enzymes (reviewed in references 41 and 53).

Mutations affecting the two largest subunits of Saccharomyces cerevisiae RNAPII were isolated by Arndt et al. (8) in a genetic selection for sit mutations (sit stands for suppressor of initiation of transcription defect) that can increase transcription of the HIS4 gene in a strain in which the BAS1, BAS2, and GCN4 genes were deleted. The last three of these genes encode transcription factors which activate transcription of HIS4 through specific DNA elements located in the promoter region of the gene (8). A yeast strain which lacks BAS1, BAS2, and GCN4 is auxotrophic for histidine because HIS4 is transcribed poorly (7, 8). Transcription of HIS4 can be increased by mutations in SIT1 (also known as RPO21, RPB1, and SUA8) and SIT2 (also known as RPO22 and RPB2), which encode the largest and second-largest subunits of RNAPII, respectively. Transcription of HIS4 in sit1 and sit2 mutant strains, as in a wild-type strain, requires the presence of a functional TATA element as well as a binding site for RAP1 in the promoter. The requirement for these two promoter elements most likely reflects a need for RAP1 to antagonize the repressing effect of chromatin and allow the general transcription machinery to gain access to the HIS4 promoter, at least in part through an interaction between TBP and the TATA box (22). RAP1 binds to the HIS4 promoter but does not efficiently stimulate its transcription in the absence of BAS1, BAS2, and GCN4 (8, 22).

One can envisage two general mechanisms by which the sit1 and sit2 mutations affect RNAPII in order to increase synthesis of HIS4 mRNAs. First, these mutations could increase the number of initiation complexes that assemble at or are recruited to the HIS4 TATA box. Results of experiments performed in vivo in which high levels of transcription were obtained by artificially tethering the RNAPII holoenzyme to promoter DNA suggested that recruitment of the holoenzyme to the promoter is a rate-limiting step in transcription (9).

Second, the sit1 and sit2 mutations could affect other steps that can be rate limiting following transcription initiation, such as an early block to elongation or premature termination (14, 28, 36, 39, 52). In this case, the sit1 mutations could increase the number of RNAPII molecules that reach the end of the transcription unit, without affecting the overall rate of initiation, by increasing the processivity of RNAPII during elongation. The increased processivity of the enzyme could affect either early elongation, by increasing the frequency at which the initiation complexes leave the HIS4 promoter and enter the elongation mode (promoter clearance), or later stages of elongation, by decreasing the rate of premature termination. As a first step toward understanding how the sit1 mutations affect transcriptional activation, we have identified regions of the largest subunit of RNAPII that are affected by these mutations.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

S. cerevisiae strains and yeast manipulations. All sit1 strains used in this study were isolated as spontaneous His⁺ revertants of strain L3110 (MATa gcn4-2 bas1-2 bas2-2 ura3-52) as previously described (7, 8). The yeast strain YF2047 (MAT gcn4-2 bas1-2 bas2-2 rpo21::LEU2 leu2 trp1::hisG ura3-52 [pJS121; RPO21 on URA3 CEN/ARS plasmid]) was constructed in several steps. First, the TRP1 locus of L3110 was changed to trp1::hisG as previously described (1) to create yeast strain YF2037. YF2037 was then mated with yeast strain CY843 (MAT gcn4-2 bas1-2 bas2-2 ura3-52 leu2). The resulting diploid was sporulated, and tetrads were dissected. A Leu Trp haploid progeny, YF2044 (MAT gcn4-2 bas1-2 bas2-2 ura3-52 trp1::hisG leu2), was then transformed to uracil prototrophy with plasmid pJS121 (RPO21 URA3 CEN/ARS) (4). The RPO21 locus of this merodiploid strain was then disrupted (rpo21::LEU2) in one step to create strain YF2047. The yeast strain YDW383, which carries the sua8-1 mutation as well as an isogenic SUA8 wild-type strain, T16, have been described previously (13). YF2277 (MAT rpo21::ADE2 [pJS121]), a strain congenic with W303-1B (MAT ade2-1 trp1-1 leu2-3,112 his3-11,15 ura3-52 can1-100 ssd1-d) contains a disruption of the chromosomal RPO21 allele caused by a replacement by ADE2 of the RPO21 sequence from positions 718 to +660 (the A of the translation initiation codon is +1). YF2278 (MAT rpo21::ADE2 ppr2::hisG [pJS121]) is isogenic to YF2277 except for the disruption of PPR2 by hisG. This was introduced by using plasmid pJD3, which contains ppr2::hisG-URA3-hisG (a gift of C. M. Kane, University of California at Berkeley). YF2277 and YF2278 were generated by standard genetic techniques. Details of their construction are available upon request. Growth media used were as described previously (45). Yeast transformation was performed as described previously (26).

Plasmids. Plasmid pJAY61, which was used for the isolation of sit1 alleles, was constructed in two steps. First, an integrating plasmid carrying URA3 as a selectable marker and the wild-type SIT1 gene on a HindIII-EcoRI genomic DNA fragment (the HindIII and EcoRI sites are located 1.6 kb upstream of the SIT1 translation initiation codon and 1.7 kb downstream of the translation termination codon, respectively) was constructed. Second, a deletion was introduced in the SIT1 open reading frame by removing the DNA sequences contained between an AvrII site (located 162 nucleotides downstream of the translation initiation codon) and a SnaB1 site (located 16 nucleotides downstream of the translation termination site) and replacing them with a BamHI linker (5'-GGGATCCC-3') after filling in the protruding ends of the AvrII site with the Klenow fragment of DNA polymerase I. Plasmid pJA483 was created by subcloning a 5.7-kb EcoRI-HindIII fragment encompassing RPO21 into the EcoRI and HindIII sites of pFL39 (TRP1 CEN6 ARS) (15). Plasmids pM50 and pM107, which carry CYC1-lacZ and cyc1-5000-lacZ, respectively, cloned in a single-copy (URA3) plasmid, have been described previously (38).

DNA manipulation and sequencing. All DNA manipulations were performed essentially as described previously (30). DNA sequencing of plasmid DNA was performed by the chain termination method (40) using primers synthesized on the basis of the SIT1 sequence.

PCR amplification of the DNA sequence encoding conserved region F. The DNA sequence between nucleotides 2155 and 2907 (in the numbering system of Allison et al. [3]), which encodes the conserved region F of RPO21, was amplified by PCR from genomic DNA of several sit1 alleles by using the following two oligonucleotides: 5'-CGGGATCCGGTGTAGTAGAGAAAAAAAC-3' and 5'-CGGGATCCTGAATAACGTTACCCAATG-3'. The resulting amplified fragments containing BamHI sites at both ends (generated by the two primers) were then cut with BamHI and cloned into the BamHI site of plasmid YEp24 (17). For each sit1 allele amplified and cloned, two independently isolated plasmids were sequenced. The amplified region F DNAs were also cut with PpuMI and XbaI and subcloned between the PpuMI and XbaI sites of pJA483 (RPO21 TRP1 CEN6 ARS).

PCR amplification of the DNA sequence encoding regions B, C, D, and E. The DNA sequence between the SpeI and PpuMI sites of RPO21, which encodes conserved regions B, C, D, and E, was amplified by PCR using the following two oligonucleotides: 5'-AGAGCGAAAATTGGTGGTC-3' and 5'-GCCTCTGCAATTGTCTCTG-3'. The amplified fragment was cut with SpeI and PpuMI and was subcloned into pJA483 (RPO21 TRP1 CEN6 ARS) cut with SpeI and PpuMI. For each sit1 allele, the DNA sequence between SpeI and PpuMI was determined in two independently isolated plasmids.

-Galactosidase assays. -Galactosidase activity was measured as described by Miller (34). Cells were grown to an optical density (at 600 nm) of 0.6 in SC-Ura medium (45) with glucose as a carbon source. For each measurement, -galactosidase activity was determined on three independent cultures.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Molecular cloning of sit1 alleles. As a first step toward determining the mechanism by which mutations which affect RNAPII can increase transcription of HIS4 in the absence of the BAS1, BAS2, and GCN4 proteins, we set out to identify which regions of the largest subunits are altered by the sit1 mutations. A gap repair strategy (37) was used to clone five independent sit1 alleles (sit1-4, -5, -7, -8, and -9) as well as the wild-type SIT1 allele from the parental yeast strain, L3110 (7). In order to localize the sit1 mutations, two independently cloned copies of each sit1 allele were digested with several restriction endonucleases, each of which recognizes a different 4-bp site (AluI, HaeIII, TaqI, and Sau3A), in the hope that certain sit1 mutations would either create or destroy a site so as to give rise to a restriction fragment length polymorphism. A Sau3A restriction fragment was detected in the sit1-7 allele that was not present in any of the other mutant alleles or in the wild-type gene (data not shown). This Sau3A polymorphic fragment was cloned and sequenced. The mutation creating the new Sau3A site results in the replacement of glycine 730 with an aspartic acid residue in the conserved region F of the largest subunit of RNAPII (Table 1), one of eight regions of this polypeptide that have been conserved during evolution (27).

Six sit1 alleles carry mutations affecting conserved region F of RPO21. Additional sit1 mutants were then examined for the presence of mutations affecting region F (nucleotides 2200 to 2900 in the numbering system of Allison et al. [3]). When the other cloned sit1 alleles (sit1-4, -5, -8, and -9) were examined, region F mutations were found in sit1-4, sit1-5, and sit1-9 (Table 1 and Fig. 1). In the case of sit1 mutants for which the sit1 allele was not cloned, the DNA sequence encoding region F (between nucleotides 2200 and 2900) was amplified from genomic DNA by PCR and cloned into a plasmid and two independent plasmid isolates were sequenced. Of five sit1 mutants analyzed in this fashion (sit1-290, -246, -252, -261, and -278), two (sit1-290 and -246) were found to carry mutations that affect region F (Table 1 and Fig. 1). One of them (sit1-290) had two mutations, one in region F and one just upstream of region F (Table 1). Since four of the sit1 alleles tested (sit1-8, -252, -261, and -278) did not carry a mutation affecting region F, at least one other region of SIT1 can be altered so as to confer the Sit phenotype (see below).

View larger version (57K): [in this window] [in a new window]   FIG. 1.   Locations of sit1 amino acid substitutions. The amino acid sequences of regions F, D, and G of S. cerevisiae RNAPII (S. cerevisiae) are compared to those of other eukaryotes: Schizosaccharomyces pombe RNAPII (S. pombe), Arabidopsis thaliana RNAPII (A. thaliana), Caenorhabditis elegans RNAPII (C. elegans), D. melanogaster RNAPII (D. melano.), Mus musculus RNAPII (M. musculus), Sulfolobus acidocaldarius (S. acidocal.) subunit A or C as indicated, E. coli RNAP, and S. cerevisiae RNAPIII (S. c. RNAPIII). Residues identical to those of S. cerevisiae RNAPII are indicated by dots; gaps are indicated by hyphens. For each sit1 mutation, the substituted amino acid is indicated above the wild-type amino acid and the designation of the allele is given in parentheses. A second substitution in the sit1-290 polypeptide that is located upstream of region F is not shown (Table 1). Amino acid substitutions that confer resistance to the transcriptional inhibitor -amanitin (Ama) in C. elegans (19), D. melanogaster (19), and M. musculus (10, 11) are also indicated. Single amino acid substitutions that affect elongation and termination by E. coli RNAP (49) are written below the sequence of the largest subunit of E. coli RNAP (E. coli mut.). Substitutions that confer streptolydigin resistance in E. coli (E. coli St1) (44) and Bacillus subtilis (B. sub. St1) (51) are indicated. The double substitutions in the S. cerevisiae RNAPIII C160-270 mutant (RNAPIII C160-270) (47) and C160-112 mutant (RNAPIII C160-112) (23) are indicated. The invariant Mg2+ binding motif NADFDGD in region D is underlined. Also indicated are the locations of the conditional lethal rpb1-17 substitution (43), the sua8-1 and sua8-2 substitutions (13), and the mutations (rpb1-501 and -502) that confer an Spt phenotype (25).

/MAT sit1

View larger version (68K): [in this window] [in a new window]   FIG. 2.   Region F and D substitutions are necessary to confer the Sit phenotype. (A) Region F substitutions. Reconstructed sit1 alleles carrying a region F mutation or the wild-type SIT1 alleles were was assayed for the ability to confer the Sit phenotype by introduction into the diploid strain YF2201 (MATa/MAT sit1::LEU2/SIT1 gcn4-2/gcn4-2 bas1-2/bas1-2 bas2-2/bas2-2 ura3-52/ura3-52 leu2/leu2 trp1::hisG/trp1::hisG). The resulting transformants were then sporulated, 10 or more tetrads were dissected, and spores were allowed to germinate on YPD medium (45). Viable Trp+ (sit1 allele on plasmid) Leu+ (sit1::LEU2) haploid progeny were then tested by streaking onto solid SD medium (45) containing (+ His) or lacking ( His) histidine. Cells were allowed to grow for the indicated number of days at 30°C. sit1-9 required a longer incubation to show visible single colonies. (B) Region D substitutions. Reconstructed sit1 alleles carrying region D mutations or the wild-type SIT1 alleles were assayed for the ability to confer the Sit phenotype as described above. Cells were allowed to grow on solid SD medium (45) containing (+ His) or lacking ( His) histidine for the indicated number of days at 30°C. sit1-8G did not grow on medium lacking histidine, even when incubation was extended to 24 days. A region F substitution (sit1-4) is shown for comparison.

MAT gcn4-2 bas1-2 bas2-2 sit1

Mapping the sit1 mutation in the sit1-8 allele. The above analysis indicated that in some sit1 alleles a region of the largest subunit other than region F is altered so as to bring about the Sit phenotype. Hybrid genes were constructed that contained various portions of sit1-8 and wild-type SIT1 (Fig. 3) in order to localize the mutation in the sit1-8 allele, which does not contain a region F mutation (see above). These chimeric genes were tested by the two assays described above for the ability to support growth on medium lacking histidine. As shown in Fig. 3, the smallest region of sit1-8 that was capable of conferring a His⁺ phenotype in an assay that relied on the semidominance of the Sit phenotype was located between nucleotides 971 and 2224 (hybrid g). The nucleotide sequence of this region was determined, and a mutation was found that replaces Asn 479 by Tyr in conserved region D of the largest subunit (Table 1 and Fig. 1). This mutation is referred to as sit1-8D. By the same assay, it was also noticed that yeast cells carrying a chimeric gene containing only the sit1-8D mutation (hybrid g) did not grow as well on medium lacking histidine as did cells carrying the entire sit1-8 allele (hybrid a). This enhancement of growth was associated with a second region of sit1-8, which was mapped to a region between nucleotides 3433 and 4621 (compare hybrid b with hybrid c and hybrid g with hybrid i). This region, however, was not sufficient by itself to bring about the Sit phenotype (hybrid h). The region located between nucleotides 3433 and 4621 was sequenced, and a mutation was found that results in the replacement of alanine 1076 by valine in conserved region G of the largest subunit (Table 1 and Fig. 1). The presence of both the sit1-8D and sit1-8G mutations was confirmed in other, independently isolated sit1-8 alleles (data not shown).

View larger version (22K): [in this window] [in a new window]   FIG. 3.   Localization of mutations in the sit1-8 allele. Schematic representation of the SIT1 locus (EcoRI-HindIII fragment) showing the positions (in the numbering system of Allison et al. [3]) of various endonuclease sites that were used in the construction of the chimeric genes. The encoded SIT1 protein is diagrammed below the restriction map. The gray boxes represent regions (A to H) of the polypeptide that are most conserved evolutionarily (27), and the diagonally striped box represents the carboxy-terminal domain. The structures of the chimeric genes are indicated by open and filled boxes representing wild-type (RPO21) and mutant (sit1-8) sequences, respectively. Each hybrid gene is designated by a letter (a to i). The ability of each chimeric gene to confer a semidominant Sit phenotype was assayed by introducing it into yeast strain YF2047 (MAT gcn4-2 bas1-2, bas2-2 ura3-52 leu2 trp1::hisG sit1::LEU2 [pJS121; SIT1 on URA3 CEN/ARS plasmid] [4]) and testing the resulting transformants for the ability to grow on solid SD medium (45) lacking histidine: ++, growth rate similar to that of a cell carrying the entire sit1-8 allele; +, growth rate slower than that of a cell carrying the entire sit1-8 allele; , absence of growth.

Identification of other region D mutations. Other sit1 mutant strains which did not carry a region F mutation (sit1-252, -261, and -278) were then tested for the presence of a region D mutation. For each of these mutants, a portion of SIT1 encoding conserved regions B, C, D, and E (nucleotides 971 to 2224 [SpeI-PpuMI fragment]) was amplified and subcloned into an otherwise wild-type SIT1 gene. When tested for the Sit phenotype by the two assays described above, all three reconstructed sit1 alleles were capable of supporting growth on medium lacking histidine. DNA sequencing revealed that two alleles, sit1-252 and sit1-261, carry the same mutation which changes methionine 487 to leucine (Table 1 and Fig. 1). The mutation in the sit1-278 allele replaces asparagine 445 by threonine (Table 1 and Fig. 1). The Sit phenotype imposed by region D substitutions is shown in Fig. 2B.

Effect of sit1 mutations on start site selection. Amino acid substitutions located in and near conserved region D of the largest subunit were identified previously that suppress the reduction in expression of CYC1 brought about by the insertion of a nonfunctional AUG codon into the 5' untranslated sequence of the gene (Sua phenotype [Sua stands for suppression of upstream ATG]) (38). Mutations in SUA8 affect the position of the transcription start site at many yeast promoters in such a way that downstream start sites are favored relative to upstream ones (13). It was reported previously that some sit1 mutations have a similar effect in that they increase the number of transcripts that start downstream of the HIS4 translation initiation codon (8). This observation and the fact that some of the mutations characterized in this study change amino acids in region D which are also changed by sua8 mutations (sua8-1 and sit1-278 change asparagine 445 to Ser and Thr, respectively) prompted us to determine whether sit1 mutations can confer an alteration of start site selection at CYC1. Yeast strains carrying either the sit1-8 (regions D and G), the sit1-278 (region D), the sit1-4 (region F), the sit1-5 (region F), or the wild-type SIT1 gene were transformed separately with a plasmid carrying a CYC1-lacZ reporter gene that either contains (cyc1-5000 allele; plasmid pM107 [38]) or lacks (wild type CYC1; plasmid pM50 [38]) a nonfunctional ATG codon in the CYC1 5' untranslated region. As controls, the same two plasmids were also introduced into a strain carrying the sua8-1 mutation (YDW383 [13]) and an isogenic SUA8 wild-type strain (T16 [13]). For each strain, the levels of -galactosidase activity expressed from the CYC1 (wild-type) allele or from the cyc1-5000 allele were measured (Table 2). The ratio of the level of expression of the cyc1-5000 allele to that of the wild-type CYC1 gene increases in strains with sua8 mutations. A low ratio, such as that measured in strains carrying a wild-type SIT1 gene (SUA8 or SIT1 in Table 2), indicates that most of the transcripts that are initiated at the cyc1-5000 promoter start upstream of the nonfunctional ATG. In contrast, a high ratio, such as that measured in a sua8-1 mutant strain, indicates that a substantial proportion of cyc1-5000 transcripts start downstream of the nonfunctional ATG and hence that the mutation affects the position of the transcriptional start site.

Synthetic lethality of sit1-8D and ppr2. The clustering of sit1 mutations in regions of RNAPII which form part of its catalytic center suggest that these mutations may affect the elongation properties of the enzyme (see Discussion). We reasoned that if the sit1 mutations affect transcriptional elongation in vivo, this defect may be exacerbated by a deletion of the nonessential gene PPR2, which encodes transcription elongation factor IIS (TFIIS). This was tested by shuffling plasmids carrying the sit1 alleles (sit1-4, -5, -7, -9, -290, -252, -278, -8D, -8G, and -8DG) into a strain carrying a disruption of both SIT1 and PPR2 (YF2278) or, as a control, into a strain carrying only a disruption of SIT1 (YF2277). In these experiments, the ability of the various sit1 alleles to support growth in the absence or presence of TFIIS was tested by determining the ability of cells to grow on medium containing 5'-fluoroorotic acid, which selects for loss of the wild-type SIT1 maintenance plasmid. Two observations were made. First, in the genetic background used in these experiments, the sit1-8D mutation is not lethal (Fig. 4). Second, this mutation is, however, lethal when combined with a deletion of PPR2 (Fig. 4). To our knowledge, sit1-8D is the first mutation in a gene encoding a subunit of RNAPII, or for that matter in any gene, reported to be synthetically lethal with a PPR2 deletion. This synthetic lethality, between a sit1 mutation and a known elongation factor, lends further support to the notion that sit1 mutations affect transcriptional elongation in vivo. We also noted that this synthetic lethal phenotype could be suppressed by the sit1-8G substitution (Fig. 4), indicating that sit1-8G and PPR2 may perform a similar function. In this context, the sit1-8G substitution lies in a portion of the largest subunit which is close to, and may even be part of, a binding site for TFIIS (5, 50). None of the other sit1 mutations tested (sit1-4, -5, -7, -9, -290, -252, -278, -8G, and -8DG) were synthetically lethal in combination with a PPR2 deletion (Fig. 4, data not shown), perhaps because their effect on elongation is not as pronounced as that imposed by sit1-8D. The notion that the sit1-8D mutation imposes a more pronounced effect on transcription than the other sit1 alleles is supported by the observation that it is the only sit1 mutation which can be lethal.

View larger version (147K): [in this window] [in a new window]   FIG. 4.   Synthetic lethality of sit1-8D and ppr2. TRP1 CEN/ARS plasmids bearing SIT1 (RPO21) alleles (indicated by white letters on a black background) were introduced into YF2277 (MAT rpo21::ADE2 [pJS121, RPO21 on URA3, CEN/ARS]), shown on the left side of the plate or YF2278 (MAT rpo21::ADE2 ppr2::hisG [pJS121, RPO21 on URA3, CEN/ARS]), shown on the right side of the plate, and grown on SD medium (45) lacking tryptophan and containing 5'-fluoroorotic acid for 3 days at 30°C in order to select for loss of the URA3 maintenance plasmid (pJS121).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The results presented above, which indicate that both conserved regions D and F of the largest subunit are altered by the sit1 mutations, suggest that these mutations exert their effect by altering the overall structure of the RNAPII active center. Previous functional studies have suggested that regions D and F, together with other conserved regions of the largest and second-largest subunits of multimeric RNA polymerases, form the active site of the enzyme as broadly defined (23) to include the catalytic center together with the DNA and RNA binding sites (reviewed in references 6, 41, and 53). Region D contains a segment, NADFDGD, that is invariant among multimeric RNA polymerases (33) and is involved in binding Mg²⁺ ions, which are essential for RNA catalysis (the three aspartate residues that chelate Mg²⁺ are underlined) (54). Within the active site, region D is located very close to the transcription initiation site (54) and the 3' end of the RNA transcript (16, 31). Cross-linking studies with E. coli RNAP indicated that region G is also in close proximity to the 3' end of the transcript; while region G was the predominant cross-linking site within arrested complexes, region D was favored when cross-linking was performed on elongating complexes (31). Our finding that a region D mutation (sit1-8D) can be suppressed by a region G mutation (sit1-8G) is consistent with the notion that both regions participate in similar functions. Indeed, mutational studies have indicated that both regions are involved in the selection of the start site (13, 25; this study) and play a role in transcription elongation and termination (23, 49, 54). In S. cerevisiae, a mutant RNAPIII (C160-112) harboring a conditional lethal double substitution in region D (T to I at position 506 [T506I] and N509Y [Fig. 1]) was characterized biochemically (23). At a natural promoter, the mutant enzyme was as proficient as the wild-type enzyme in carrying the initiation reaction but was more prone to catalyze a slippage reaction during synthesis of the first few nucleotides. The mutant enzyme also had a markedly reduced elongation rate, which was attributable to increased pausing at intrinsic pausing sites. It was hypothesized that the double amino acid substitution could either affect the correct positioning of the 3' end of the RNA transcript in the catalytic center of the enzyme or have a more direct effect on the formation of phosphodiester bonds (23). One of the two substitutions (N509Y) affects a residue of the invariant Mg²⁺-binding motif and was found to be lethal unless suppressed by the second region D substitution (T506I) (23). This is reminiscent of the sit1-8D substitution (N479Y), which changes the analogous residue in the invariant motif of RNAPII (Fig. 1) and can also be lethal unless suppressed by the sit1-8G substitution (A1076V).

Region F, the other region altered by sit1 mutations, forms a part of the enzyme active site which is targeted by certain inhibitors. Region F can be mutated to confer resistance to -amanitin (reviewed in reference 6), which inhibits the translocation step of eukaryotic RNAPII (21), and to the antibiotic streptolydigin (44, 51), which inhibits the elongation phase of prokaryotic RNAP (18, 32, 48). A role for region F in transcriptional elongation and translocation is also supported by studies of mutant RNAPs. One Drosophila melanogaster RNAPII mutant (C4) bearing an -amanitin resistance substitution in region F (R741H) (Fig. 1) was found to elongate more slowly in vitro and have an increased apparent K_m for UTP (20). In yeast, a mutant RNAPIII (C160-270) that carried a double substitution in region F (D829A R830A) (Fig. 1) and was defective in the transition from abortive initiation to elongation and in its ability to exit from pause sites was isolated (47). The mutant enzyme also showed an increased RNA cleavage activity in halted ternary complexes. Finally, in E. coli, several region F mutations were isolated that increase termination in the trp operon leader region or decrease termination at a -independent terminator in vivo (49). These mutations also affected termination efficiency at other -independent terminators in vitro, although the magnitude and direction of the effect appeared to be dependent on the type of terminator used (49). The last of these observations raise the possibility that the effect of sit1 mutations on transcription varies from one gene to another. Indeed sit1 mutations can lead to both over- and underexpression of genes such as HIS4 and CYC1 (Table 2), respectively. These profound effects on gene expression most likely underlie the many pleiotropic phenotypes of sit1 mutants.

Further support for the notion that sit1 mutations alter the active site of RNAP comes from experiments in which the Mg²⁺ ion of the enzyme was replaced by Fe²⁺. Hydroxyl radicals generated by Fe²⁺ cleave protein and DNA in a 1-nm radius of the ion. In these experiments, DNA is cleaved immediately upstream of the transcriptional start site (54), indicating that the Mg²⁺ ion is near or at the catalytic center of the enzyme. In a similar fashion, cut sites in the RNAP peptide chain also identified residues that are in the active-site area (35). ' was cleaved in three regions: conserved regions D, F, and G. These regions correspond to those in which the sit1 mutants and a suppressor to the growth defect of a sit1 mutation (sit1-8G) were identified.

Two lines of evidence suggest that the pleiotropic effects of sit1 alleles can vary according to the genetic background of the strain in which they are expressed. First, in at least one instance, we have shown that the Sit phenotype can be semidominant, whereas in other genetic backgrounds it is recessive. Second, we found that the sit1-8D substitution is lethal in one genetic background, whereas in another it is viable. Although we do not yet fully understand which set of genes is involved in these background differences, we nevertheless have gathered evidence to suggest that at least one gene, PPR2, which encodes the transcription elongation factor TFIIS, can affect the phenotype imposed by the sit1-8D substitution. We found that in a genetic background in which it is normally viable, the sit1-8D allele can be synthetically lethal in combination with a disruption of the nonessential gene PPR2 and that this synthetic phenotype can be suppressed by the sit1-8G substitution which lies in, or close to, a TFIIS-binding site in RNAPII (5, 50). These findings have two implications. First, they raise the possibility that some of the differences we observed between genetic backgrounds are due to differences in genes which encode proteins that can affect transcriptional elongation, either directly or indirectly. Further analysis of these genetic differences should reveal whether this hypothesis is correct. Second, and most importantly, the genetic connection between sit1-8D, sit1-8G, and a known transcription elongation factor lends further support to the hypothesis that sit1-8D affects the elongation properties of RNAPII in vivo.

Other than mutations that truncate the C-terminal domain of the largest subunit of RNAPII (2, 42), the sit1 mutations are the only mutations that implicate RNAPII directly in the process of transcriptional activation. Our finding that sit1 substitutions affect start site selection and occur in regions of RNAPII that form the enzyme active center implies that there is an intimate relationship between initiation and elongation and has implications for the mechanism of transcriptional activation. To our knowledge, our findings provide the first evidence in eukaryotes that alteration of the enzyme catalytic center can affect transcriptional activation. Previous studies of the mechanism of transcriptional activation have suggested that activation domains stimulate transcription by several mechanisms, including counteracting the repressing effect of chromatin, recruiting the RNAPII holoenzyme complex to the promoter through multiple contacts between activation domains and components of the RNAPII holoenzyme, and/or increasing the processivity of elongating RNAPII (reviewed in references 12 and 46). The results presented in this study provide genetic evidence that partial activation of transcription can also occur by alteration of the RNAPII active site and suggest that some transcription-regulatory proteins exert their effect by modifying, either directly or indirectly, the RNAPII active center. In E. coli, the conformation of the RNAP active center and in particular the way in which it contacts the nascent RNA and the DNA template are major determinants of the processivity of the enzyme (reviewed in reference 29), which can be regulated by antitermination factors such as the N and Q gene products of phage lambda (reviewed in reference 24). By analogy, eukaryotic regulatory proteins that are known to increase the processivity of RNAPII, such as human immunodeficiency virus type 1 Tat, yeast GAL4, human p53 and E2F1, and herpes simplex virus VP16 (14, 28, 52), could exert their effect by altering, either directly or indirectly, the structure or conformation of the enzyme active site during elongation. sit1 mutations, by directly affecting the RNAPII active center, could help stabilize a more processive form of the enzyme and in doing so partially bypass the need for promoter-bound transcriptional activators at HIS4.