Partial truncation of the yeast RNA polymerase II carboxyl-terminal domain preferentially reduces expression of glycolytic genes.

The largest subunit of RNA polymerase II contains an essential carboxyl-terminal domain (CTD) that consists of highly conserved heptapeptide repeats with the consensus sequence Tyr-Ser-Pro-Thr-Ser-Pro-Ser. Yeast cells with a partially truncated CTD grow slowly, are temperature- and cold-sensitive, and are unable to fully activate transcription of some genes. Screening a yeast wild-type cDNA library by means of comparative hybridization we find that CTD truncation preferentially reduces transcription of genes encoding glycolytic enzymes. Using a newly developed dual reporter assay we demonstrate that sensitivity to CTD truncation is conferred by the glycolytic gene promoters. Expression driven by glycolytic gene promoters is reduced, on average, about 3-fold in strains with the shortest CTD growing on either fermentable or nonfermentable carbon sources. Sensitivity to CTD truncation is particularly acute for the constitutively expressed ENO1 gene, which is reduced 10-fold in a strain with only eight CTD repeats. The sensitivity of constitutive ENO1 expression argues that CTD truncation can cause defects in uninduced as well as induced transcription.

The carboxyl-terminal domain (CTD) 1 of the RNA polymerase II largest subunit consists of tandem repeats of nearly identical amino acid heptamers (1,2). Although the consensus sequence of these heptamer repeats is highly conserved among different organisms, the number of repeats varies. The CTD of the yeast Saccharomyces cerevisiae contains 26 or 27 repeats (1,3), whereas that of mammals contains 52 (2,4,5). Yeast and mouse cells with partially truncated CTDs are viable, indicating a degree of functional redundancy within the repeat domain (3,4,6). The minimum CTD length required for viability of yeast cells is eight repeats, although these cells grow slowly and are both cold-and temperature-sensitive (7). Activated transcription driven by UAS elements in the INO1 and GAL10 promoters is defective in yeast CTD truncation strains (8 -10). In the shortest CTD strain tested (11 repeats) INO1-driven expression is reduced more than 10-fold, whereas GAL10-driven expression is reduced 2-fold (9). Uninduced transcription of these two genes was found to be unaffected by CTD truncation. Replacing the GAL10 UAS with Gal4p-binding sites yields similarly CTD truncation-sensitive transcription activation (9), whereas lengthening the CTD can suppress mutations that weaken activation by GAL4-VP16 (10). Together, these experiments suggest an interaction between transcription activators and the CTD. In contrast, the HIS4 UAS directs transcription at levels that remain unchanged in strains with a truncated CTD (9).
CTD truncation in yeast also produces a cold-sensitive lethal phenotype that can be suppressed by mutations in SRB genes (11). Nine of these genes have now been identified, and their products have been shown to form a complex with RNA polymerase II (14). This holoenzyme complex has been shown to respond to transcription activator proteins in vitro, suggesting that the CTD and SRB proteins are involved in transcription activation (12)(13)(14). Involvement of the CTD in activated transcription is also suggested by the inability of in vitro transcription extracts derived from CTD truncation strains to carry out activated transcription (15).
One problem in interpreting the effect of CTD deletion on transcription is that most experiments have focused on activated transcription driven by the GAL10 promoter, which is only moderately affected by CTD truncation (9). Based on the observation that CTD truncation mutants grow slowly in rich medium, a condition under which INO1 and GAL10 are repressed, we searched for genes that are poorly transcribed in complete medium. We report here the results of differential screening experiments, which reveal that many glycolytic genes depend on a full-length CTD for efficient transcription. We have designed a novel plasmid for comparing promoter strengths in yeast and have used this approach to show that the reduced transcription of glycolytic genes in CTD truncation strains is a property of the glycolytic promoters. The ENO1 promoter, in particular, is very sensitive to CTD truncation. Expression driven by this constitutively active promoter is reduced 10-fold in a strain containing eight CTD repeats. We propose that reduced transcription of highly expressed genes like ENO1 and ADH1 could lead to the growth defects associated with CTD truncation. were introduced into strain Z26 (MAT␣ his3⌬200 ura3-52 leu2-3, 112 rpb1⌬187::HIS3 GAL ϩ (pRP112)) (3), and the plasmid carrying the wild-type copy of the RPB1 gene was removed by 5-fluoro-orotic acid counterselection (16). Transformation of yeast cells was performed with the lithium acetate technique (18). In this paper we employ rpb1 genes with 8, 9, 10, 11, 16, or 26 consensus CTD repeats. The strains are named after the number of CTD repeats (i.e. CTD8). Our shortest viable strains are phenotypically equivalent to the CTD truncation mutants described previously (8,9).

Yeast Strains and Growth
Cells containing rpb1 genes were maintained on synthetic complete medium lacking leucine (17), and their derivatives containing pDES (see below) were maintained on synthetic complete medium lacking leucine and uracil (17). Liquid cultures used for RNA extraction or for enzymatic assays were grown on rich medium (YPD or YPLG) for one to two cell divisions prior to harvesting. YPD medium contained 2% glucose, 1% Bacto-yeast extract, and 2% Bacto-peptone (17). In YPLG medium, glucose was replaced by 2% glycerol and 2% lactate.
RNA Preparation-Yeast cells were inoculated in synthetic complete medium lacking the appropriate nutrients. Overnight cultures were diluted into YPD and grown for 1.5-2 generations to early log phase (A 600 ϭ 0.6 -0.8). Cells were collected, washed once in water, and stored at Ϫ80°C. Thawed cells were washed twice in RNA buffer (100 mM Tris-HCl, pH 7.5, 100 mM NaCl, 30 mM EDTA, and 1% Sarkosyl) and disrupted with glass beads in the presence of phenol. The supernatant was extracted once more in phenol and twice with chloroform/isoamyl alcohol. RNA was precipitated overnight at Ϫ20°C and resuspended in water.
Yeast cDNA Library Screening-A yeast wild-type gt10 cDNA library produced in Ron Davis's laboratory was kindly provided by Dr. Jef Boeke (Johns Hopkins University). Poly(A) ϩ RNA from cells containing an RPB1 gene with 8 or 26 CTD repeats (CTD8 and CTD26 cells, respectively) was isolated on an oligo(dT)-cellulose column (type 3, Collaborative Biomedical Products, Bedford, MA) according to the manufacturer's protocol. Radioactively labeled cDNA probes were synthesized with oligo(dT) primers using Moloney murine leukemia virus reverse transcriptase (RNase H Minus, Promega). Phage plaques (30,000) containing wild-type yeast cDNA inserts were transferred to duplicate nitrocellulose filters. One replica of the library was hybridized with the CTD8 cDNA probe, and the other replica was hybridized with the CTD26 cDNA probe. The intensity of hybridization of the two probes to individual plaques of the library was compared visually. Phage DNA was prepared from differentially hybridizing clones, digested with restriction enzymes, and separated on agarose gels. Southern blots of these digests were hybridized with the same CTD26 and CTD8 cDNA probes used in the plaque screen. Differential inserts were subcloned into pBluescript (Promega).
Nucleic Acid Analysis-Northern blots were prepared from total RNA isolated from CTD8 and CTD26 cells. Labeled probes complementary to the eight selected inserts were synthesized by oligo priming and hybridized to Northern blots (19) containing 20 g of total RNA from CTD8 or CTD26 cells. A single band corresponding to the expected mRNA was observed in each hybridization reaction. The intensity of hybridization was quantified by a PhosphorImager (Molecular Dynamics). Sequencing was performed with the Sequenase 2.0 kit (U. S. Biochemical Corp.). Sequences of differentially expressed clones were compared with the GenBank data base using the program BLAST (20) via the NCBI network. The actin control probe was derived from a 1.1kilobase pair BamHI-HindIII fragment from p10-AHX3, a genomic clone containing the yeast ACT1 gene (21) (kindly provided by Karen Chapman, Johns Hopkins University). Other probes for blot hybridizations were synthesized by PCR and were homologous to parts of the coding sequences of the analyzed genes. Primer extension analysis was performed according to standard protocols (19), and reaction products were separated on 8% polyacrylamide/8 M urea sequencing gels and analyzed by PhosphorImager (Molecular Dynamics).
Construction of pDES and Derivatives-To compare the relative strengths of different promoters, we constructed a plasmid with two different reporter genes each driven by different promoters. pLG669-z (22), which contains a ␤-galactosidase (GAL) gene, served as backbone for the construction of pDES (Fig. 1A). The coding sequence of ␤-glucuronidase (GUS) (23) was synthesized by PCR from a genomic clone on plasmid pVIT51 (kindly provided by Dr. Vit Lauermann, Johns Hopkins University). The 5Ј primer, GACGTCGAGCTCGAGATGACAA-GATCTGTAGAAACC, changed the amino terminus of the gene to conform with the N-end rule (24) and thus create a more stable reporter enzyme. This PCR primer also introduced a BglII site preceded by restriction sites for AatII and XhoI (restriction sites are underlined). The 3Ј primer added NotI and AatII sites downstream of the termina-tion codon. After digestion with AatII, the fragment containing the ␤-glucuronidase gene was ligated into pLG669-z at the AatII site located between the 2 sequence and the ampicillin resistance gene (Fig.  1A). The transcription terminator sequence of the ADH1 gene (25) was synthesized by PCR using the primers GCGGCCGCCACACTTCTA-AATAAGCG and GCGGCCGCCGGTGTGGTCAATAAGAGCG adding NotI sites (underlined) at both ends, and the amplified and NotI-digested fragment was ligated into the NotI site added 3Ј to the ␤-glucuronidase gene. The UAS of the CYC1 promoter, situated in front of ␤-galactosidase (GAL), was replaced by a linker that restored the SalI site and eliminated the XhoI site. The proximal promoter (TATA box) of CYC1 was replaced by the TEF1 promoter (26), synthesized by PCR with primers that created at its 3Ј end a BamHI site in frame with the start of the ␤-galactosidase coding sequence (see Fig. 1A and Ref. 22) and a SalI site at the 5Ј end. In this study the TEF1 promoter, because it was shown to be insensitive to CTD truncation, served as reference promoter. Test promoters were synthesized by PCR, with primers creating either SalI or XhoI sites at the 5Ј end of the fragment and either BamHI or BglII sites at the 3Ј end. Promoter fragments were ligated upstream of either the ␤-galactosidase or the ␤-glucuronidase gene. Due to the compatible ends created with BamHI and BglII, the same fragment could be introduced in front of either of the two reporter genes. The oligonucleotides used in the PCR synthesis of promoter sequences are listed in Table I, and the promoters driving the different reporter genes in different constructs are shown in Fig. 1B. Throughout this study the TEF1 promoter (5Ј to ␤-galactosidase) served as reference promoter.
Dual Reporter Gene Assay-pDES and its derivatives were maintained in cells by growth in synthetic complete medium lacking leucine and uracil. Overnight cultures were diluted into 8 ml of YPD or YPLG and grown for 1.5-2 generations to early log phase (A 600 ϭ 0.6 -0.8). Cells were collected, washed in water, and stored frozen at Ϫ80°C. Thawed cells were disrupted by vortexing (three times for 30 s at 4°C, interrupted for 90 s each time) in 250 l of DES extraction buffer (100 mM NaPO 4 , 10 mM EDTA, 20% glycerol, 10 mM ␤-mercaptoethanol, and 5 mM phenylmethylsulfonyl fluoride, pH 7.0) and an equal volume of glass beads. DES extraction buffer (250 l) containing 0.2% Triton X-100 was added, and cells were vortexed for an additional 30 s at 4°C. The extract was centrifuged at 16,000 ϫ g at 4°C for 10 min, and the supernatant was stored frozen at Ϫ80°C. Enzymatic reactions were initiated by adding aliquots of the protein extracts to DES assay buffer (50 mM NaPO 4 , 10 mM KCl, 2 mM MgSO 4 , and 10 mM ␤-mercaptoethanol, pH 7.0) containing either 1.5 mM p-nitrophenyl-␤-D-galactopyranoside or 1 mM p-nitrophenyl-␤-D-glucuronide (Sigma). Absorbance at 414 nm was monitored every 30 s over a period of at least 10 min in an HP8452 spectrophotometer with an automatic transport device. Enzyme activity was determined from the slope of the absorbance versus time plot. Alternatively, reactions were performed in microtiter plates and read in a microtiter plate reader. In initial experiments the protein concentration of the extract was determined, and the specific activities of the reporters were calculated in Miller units (27) expressed as nmol of product/mg of protein/min.
Promoter strength is defined here as the ratio of the reporter enzyme activities driven by the test and the control promoters. The activity of ␤-glucuronidase (driven by the test promoter) is normalized to ␤-galactosidase activity (directed by the CTD-insensitive TEF1 promoter). This ratio, expressed as arbitrary TEF1 units, is independent of protein concentration. Relative promoter strengths were obtained by comparing the promoter strengths obtained in CTD truncation strains to the value in wild-type cells (set arbitrarily at 100%). For each pDES construct, enzyme activities from three transformants were measured. The standard deviations of these measurements were within 15%.

Identification of Genes Sensitive to CTD Truncation-Al-
though previous studies have shown that INO1 and GAL10 are not fully induced in short CTD strains (8 -10), this defect cannot explain the slow growth of these strains in rich medium. To identify genes that are normally expressed in rich medium but have reduced expression levels in cells with a truncated CTD, we devised a differential screening approach. A wild-type yeast cDNA library was hybridized with cDNA probes prepared from CTD26 and CTD8 cells. The two probes hybridized with equal intensity to the majority of plaques, indicating that most mRNAs are equally abundant in CTD26 and CTD8 cells. Examining 30,000 plaques, we identified 84 that hybridized more efficiently to the CTD26 than to the CTD8 probe. Fig. 2 shows an example of hybridized filters in which the CTD8 filter has been purposely overexposed to emphasize reduced hybridization to some plaques (indicated by arrows in Fig. 2).
Eight inserts with the highest degree of differential hybridization were subcloned for further analysis. Sequencing revealed that four of the eight highly differential cDNA clones encoded the glycolytic enzyme alcohol dehydrogenase (ADH1) (28). Three other differential clones also encoded glycolytic enzymes: glyceraldehyde 3-phosphate dehydrogenase (TDH3) (29), phosphoglycerate mutase (GPM1) (30), and pyruvate decarboxylase (PDC1) (31). One clone contained an unknown yeast DNA sequence. Two highly expressed clones with equal hybridization intensities were also selected as controls. Sequencing revealed that both of these clones encoded the translation elongation factor 1␣ (TEF1) (26).
Differential abundance of glycolytic mRNAs was confirmed on Northern blots of equal amounts of RNA isolated from CTD26 and CTD8 cells. A representative result of the Northern blot analysis (Fig. 3) demonstrates that ADH1 and PDC1 transcript levels are reduced in CTD8 cells, whereas TEF1 and ACT1 mRNA levels are the same in these two strains. This experiment confirms the differential bacteriophage plaque hybridization result and shows that the steady state level of ADH1 and PDC1 mRNAs are reduced 3-5-fold in the CTD8 strain. ADH1 and PDC1 are expressed to about the same level in CTD26, whereas TEF is expressed to a greater and ACT1 to a lesser extent, indicating that sensitivity to CTD truncation does not correlate with the level of expression. Equal hybridization to the control mRNAs confirms our plaque hybridization in the case of TEF1 and is consistent with earlier characterization of the CTD truncation insensitivity of ACT1 The first sequence listed is the primer used at the 5Ј end of the promoter (the distance from the ATG codon is indicated), and the second is the primer used at the 3Ј end of the promoter. Uppercase letters indicate genomic sequences. Linker sequences containing restriction sites or modifying for in frame ligation are in lowercase letters. Restriction sites are underlined. Except for TPI1 (53) and ENO1 (45), all sequences were retrieved from the Genbank database.

Gene
Position Primer sequences 5Ј/3Ј Reference Ϫ455 5Ј-gtcgaCCTCCGTACATTCATGT-3Ј 26 5Ј-ggatccggtCATTGTTTAGTTAA-3Ј  2. Differential screening of a wild-type cDNA library. Duplicates of filters carrying a wild-type cDNA library were hybridized with the cDNA probes prepared from CTD8 and CTD26 cells, respectively. Differential plaques (arrows) are shown surrounded by plaques that hybridize equally strong to both cDNA probes. The CTD8 filter has been overexposed to emphasize the reduced hybridization to the indicated plaques.

Sensitivity to CTD Truncation Is a Property of Glycolytic Gene Promoters-Previous studies have demonstrated that the induction of INO1
and GAL1 transcription is sensitive to CTD truncation, and this sensitivity is conferred by promoter sequence elements (8 -10). To test whether other glycolytic genes are differentially expressed, we developed a sensitive system for the analysis of relative promoter strengths in yeast. We constructed a plasmid in which two reporter genes, ␤-galactosidase and ␤-glucuronidase, are driven by different promoters (Fig. 1). As a control, TEF1 and TEF2 promoters were used to drive GUS as well as GAL expression (pDES10, pDES10 -2). In these strains the GUS activity was consistently ϳ5-10% higher than GAL, reflecting a slight bias in favor of expression from the GUS cassette. This bias results in a slight underestimation of the CTD truncation sensitivity of the glycolytic promoters using the dual reporter system. The dual reporter system eliminates concerns about relative copy number of different plasmids and obviates the need to normalize reporter gene activities to protein concentration or to cell number. This is an important consideration when comparing expression of genes under different growth conditions.
We have compared a series of different glycolytic promoters to a TEF promoter control. In one test of the dual reporter system the ADH1 and TEF1 promoters were introduced into pDES, and the levels of GUS and GAL reporter enzymes and their mRNAs were determined as described under "Materials and Methods." The specific activities of GUS and GAL in CTD26 were 6,408 and 2,815 Miller units (27), respectively. In the CTD8 strain the activities of GUS and GAL were 1,461 and 2,501, respectively. The ratio of these pairs of values, 2.29 and 0.59 we define as the promoter strengths (with respect to the TEF1 promoter) of the ADH1 promoter in CTD26 and CTD8, respectively. The results of this analysis and a similar analysis of the ENO2 promoter are summarized in Table II, which shows that ADH1 promoter-driven expression in CTD8 is reduced to about 25% of the CTD26 level, a result that is consistent with the difference observed in ADH1 mRNA abundance seen in these two strains. ENO2 promoter-driven expression is reduced to about 30% of that of wild type.
To verify that the reporter enzyme assays accurately reflect transcript levels, we measured reporter gene mRNA levels in cells containing pDES9. Fig. 4 (lanes 5 and 6) shows that ␤-glucuronidase mRNA expressed from the ADH1 promoter is reduced to about 25% of that of wild type in the CTD8 strain.
By comparison, the mRNA levels of the TEF1-driven ␤-galactosidase (lanes 5 and 6) and TEF1 and TEF2 (32) (lanes 3 and  4), and ACT1 (lanes 1 and 2) cellular mRNAs are unchanged in the CTD26 and CTD8 strains. Together with the reporter enzyme assay, this result indicates that sensitivity to CTD truncation is conferred by the ADH1 promoter, and this sensitivity is maintained despite the high copy number of the DES plasmid. These results further show that the dual reporter system is a valid means of testing relative reporter strengths in yeast.
We have used the dual promoter system to test a set of glycolytic promoters, including some that were not identified in the plaque hybridization screen. Fig. 5 shows that expression driven by promoters of glycolytic genes participating in the Embden-Meyerhof pathway are reduced in the CTD8 strain. The least sensitive promoters are TDH3 and GPM1, which are expressed at 73 and 53% of normal values. Expression of all other glycolytic promoters (TPI1, PGK1, ENO2, PYK1, PDC1, and ADH1) is reduced more than 2-fold, with ADH1 and ENO2 the most severely affected. Northern blot analysis of mRNA levels of the respective genes ( Fig. 3 and not shown) confirmed the enzyme data. All messages were present in CTD8 cells at less than 55% relative to wild-type cells, GPM1 (67%) being the one exception.
Two experiments were performed to determine whether transcription factors regulating expression of glycolytic genes are sufficiently abundant in CTD8 cells. The transcription factors Gcr1p (33)(34)(35)(36), Rap1p (37)(38)(39), Abf1p (40,41), and Reb1p (42) bind to a large number of glycolytic promoters. The promoters of these four genes were assayed in the double enzyme system and found to be unaffected by CTD truncation (not shown). In a second experiment, the UAS of the TPI1 and ENO1 promoters were subjected to gel retardation assays employing extracts from CTD8 and CTD26 cells. Both extracts resulted in identical shift patterns, indicating the presence of the same transcription factors in CTD8 cells (data not shown). These two experiments argue that reduction in glycolytic gene expression is not due to a lack of transcription factors.
Expression of ENO1 Is Sensitive to CTD Truncation under Different Growth Conditions-The yeast genome contains two enolase genes, ENO1 and ENO2, with nearly identical coding  Miller (27) and are expressed in nmol of product/mg/min. b To calculate promoter strength the specific activity of GUS directed by the indicated test promoters (ADH1 or ENO2) was divided by the specific activity of GAL driven by the CTD length-insensitive TEF1 promoter. The result, expressed in arbitrary TEF1 units, is the average of measurements from three independent transformants and is presented with the standard deviation of the three measurements.
c Relative promoter strength is expressed in percent of promoter strength in strain CTD26 (wild-type length CTD).
FIG. 4. mRNA levels of pDES reporter genes. Differential transcription from various promoters in cells with truncated and wild-type length CTD was analyzed by primer extension. Total RNA was prepared from CTD26 and CTD8 cells and hybridized to oligonucleotides priming toward the 5Ј end of the transcripts (see "Materials and Methods"). After reverse transcription, the synthesized DNA fragments were separated on 8% polyacrylamide/8 M urea sequencing gels and exposed for PhosphorImager analysis. Lanes 1 and 2 show cellular ACT1 RNA levels. A single primer was used to quantitate cellular TEF1 and TEF2 transcripts (lanes 3 and 4). Primers homologous to ␤-glucuronidase (driven by ADH1) and ␤-galactosidase (driven by TEF1) transcripts were mixed and incubated with RNA extracted from CTD26 and CTD8 cells. CTD26 RNA is present in lanes 1, 3, and 5; whereas CTD8 RNA was used for lanes 2, 4, and 6. The positions of the extended fragments corresponding to the different transcripts is indicated on each side. sequences (43). ENO2 expression is elevated when cells are shifted from gluconeogenic to glycolytic growth, whereas ENO1 transcription levels remain unchanged in response to changes in carbon source (44,45). To test for the role of glucose in CTD truncation sensitivity, we examined the effect of CTD truncation on the constitutively expressed ENO1 promoter during gluconeogenesis and glycolysis. Fig. 6A shows that expression of a reporter gene driven from the ENO1 promoter was reduced to approximately 10% of that of wild-type levels when grown on glucose and to approximately 30% of that of wild-type levels in gluconeogenic cells. Similar reductions were also seen for ADH1-and ENO2-driven expression in cells growing in nonfermentable (YPLG) media (data not shown). Primer extension experiments confirmed that expression of reporter gene mRNA is also substantially reduced in the CTD8 strain in nonfermentative growth (not shown).
The sensitivity of ENO1 expression to CTD truncation is most evident in the CTD8 strain. To assess the effect of less severe CTD truncations we examined the reduction of ENO1 promoter-driven expression in different CTD truncation strains. Fig. 6B shows that strains with 9, 10, or 11 consensus repeats are also deficient in ENO1-driven expression, although not to the same extent as CTD8. Strains with 10 or 11 repeats grow at a similar rate to that of the wild-type parent (7), indicating that the defect in ENO1-driven expression is not a result of the slow growth of the CTD8 strain. DISCUSSION Yeast CTD truncation mutants growing in rich medium have progressively more severe growth defects as the length of the CTD approaches the minimum length required for viability (3,7,11). Previous experiments indicated that CTD truncation reduces activated transcription of INO1 and GAL10 (8 -10) but not their uninduced transcription levels. Defective activation of these genes cannot, however, explain the slow growth of CTD truncation strains growing in rich medium because these genes are not required for growth in glucose. Triggered by the observation that CTD truncation mutants have severe growth defects in rich medium, we have screened for genes that are poorly expressed in cells bearing eight CTD heptamer repeats, the shortest CTD sufficient for viability.
We report here that expression of genes encoding glycolytic enzymes is reduced by CTD truncation. This reduction is not simply a decrease in the maximum level of transcription of highly expressed genes as the promoter strengths, and mRNA levels we measured have been compared with the highly expressed TEF genes. In addition, expression driven by the moderately strong ENO1 promoter is very sensitive to CTD truncation. Although the effect of CTD truncation on most glycolytic genes is modest (2-3-fold), as is the effect on GAL10 expression (9), the cumulative effect of this transcriptional deficiency in a set of highly expressed genes could well contribute to the growth defects characteristic of CTD truncation mutants. Poor expression of pyruvate decarboxylase in particular may explain the failure of short CTD strains to grow on pyruvate medium (46). We cannot, however, rule out the possible contribution of CTD truncation-induced defects in expression of other growthrelated genes. The sensitivity of glycolytic gene expression to CTD truncation suggests that the step in the transcription cycle that requires the CTD may be a prominent feature of the regulation of these genes. In the case of GAL10 expression, the sensitivity to CTD truncation seems to involve the binding site for Gal4p (9). Consistent with this explanation, mutations that suppress CTD truncation defects define a set of genes whose products assemble with polymerase II to form a high molecular weight complex termed the holoenzyme that is involved in transcription activation (12)(13)(14). The nature of the interaction between activators and the holoenzyme complex is unclear, although both Srb2p and TBP have been shown to interact with the CTD (47,48).
Glycolytic gene expression requires a common set of transcription activating factors including Gcr1p (33)(34)(35)(36), Rap1p (37-39), Abf1p (40,41), and Reb1p (42). Although glycolytic promoters typically contain binding sites for several of these proteins, the binding sites for these factors are not restricted to FIG. 5. CTD truncation reduces glycolytic gene expression. Glycolytic promoters were studied using pDES plasmids carrying the indicated promoter in front of ␤-glucuronidase. pDES constructs were transformed into CTD26 and CTD8 cells, and enzyme assays were performed as described under "Materials and Methods." The relative promoter strength (in percent of activity of wild-type cells) is indicated. FIG. 6. Response of the constitutively transcribed ENO1 promoter to CTD truncation. A, analysis of ENO1 promoter strength in cells grown under fermenting and nonfermenting conditions. Double enzyme assays were performed on CTD26 and CTD8 cells transformed with pDES1-ENO1 grown in glucose (YPD) or glycerol plus lactate (YPLG) containing medium as described under "Materials and Methods." The relative promoter strength (in percent of activity of wild-type cells) is indicated. B, relative promoter strength as a function of decreasing CTD length was studied in cells transformed with pDES11 grown in glucose and is expressed in percent of activity in CTD26 cells. glycolytic promoters (37, 38, 49 -51), indicating that these proteins play other roles in addition to regulating glycolysis (52). Co-regulation of glycolytic genes probably involves the concerted activities of multiple factors. For example, the binding sites for Rap1p and Gcr1p are in close proximity in several glycolytic promoters, and studies indicate that these factors cooperate to activate transcription (36,53,54). Reduced expression of glycolytic genes in the CTD8 strain could be due to a failure of the polymerase II holoenzyme to recognize or respond to an otherwise synergistic pairing of these factors. Such an interaction between Rap1p/Gcr1p and polymerase II could be direct or could involve co-activators such as GAL11/SPT13 (55), GCR2 (56), and/or GCR3 (57), which are necessary for full expression of some or all glycolytic genes. These activators and co-activators are required not only for fermentative but also nonfermentative growth (58), an observation consistent with our observation of CTD-sensitive expression from glycolytic promoters in glycerol-lactate medium. Whether these factors interact with the CTD is not known.
CTD truncation sensitivity of glycolytic promoters might also be due to unique core promoter elements. The dihydrofolate reductase (dhfr) promoter requires the CTD for transcription in vitro (59 -62) and the promoter elements that confer this requirement map to the TATA box and transcription start site region (62). Glycolytic promoters could have similar core promoter elements. The observation that GAL10 sensitivity to CTD truncation maps to upstream sites (9) suggests that core promoter elements are not involved in the CTD sensitivity of this promoter.
In vitro transcription with yeast extracts derived from CTD truncation strains has indicated that Gal4p-activated transcription is reduced in comparison with transcription by wildtype polymerase II (15). Uninduced transcription in these experiments was very low, however, making quantification difficult. We have recently repeated these experiments using a more sensitive assay, and we find that uninduced transcription is reduced to the same extent as activated transcription. 2 Using the sensitive dual enzyme system, we have also found that uninduced INO1 expression is also reduced in the CTD8 strain. 3 Thus, the defect caused by CTD truncation may be in establishment of the uninduced level of transcription.
Uninduced transcription levels are determined by several parameters; among them are promoter core and activation sequences and the factors and co-factors that bind them. In addition, repressor elements play an important role in determining the uninduced transcription level (63). The sensitivity of glycolytic promoters to CTD truncation could also be dictated by alterations in the response of the CTD8 polymerase to repressor activities. The cold-sensitive phenotype of a yeast CTD truncation mutant has previously been shown to be suppressed by deletion of the gene encoding the repressor SIN1 (64). In addition, two recently identified components of the holoenzyme complex, SRB10 and SRB11 (65), have been shown to be identical to genes involved in glucose repression (66) and ␣2 repression (67). Taken together these results argue that the CTD may play a role in repression of some genes. Because SRB10/SRB11 encode a Cdk/cyclin pair, it would not be surprising if this form of regulation involves CTD phosphorylation. The ENO1 gene has a repressor element (45), and it is possible that this element also plays a role in the enhanced CTD sensitivity of the ENO1 promoter.
Our observation that expression of glycolytic genes is sensitive to CTD truncation provides us with several new clues about the role of the CTD in transcription. The effect on these genes is seen in growth on different carbon source and is very strong on the constitutively expressed ENO1 promoter. This promoter will be extremely valuable in devising in vivo and in vitro experiments aimed at determining the mechanism of CTD activity in yeast transcription activation and repression.