Transcription by RNA polymerase II (pol II) requires the ordered binding of distinct protein complexes to catalyze initiation, elongation, termination, and coupled mRNA processing events. One or more proteins from each complex are known to bind pol II via the carboxy-terminal domain (CTD) of the largest subunit, Rpb1. How binding is coordinated is not known, but it might involve conformational changes in the CTD induced by the Ess1 peptidyl-prolyl cis/trans isomerase. Here, we examined the role of ESS1 in transcription by studying one of its multicopy suppressors, BYE1. We found that Bye1 is a negative regulator of transcription elongation. This led to the finding that Ess1 also inhibits elongation; Ess1 opposes elongation factors Dst1 and Spt4/5, and overexpression of ESS1 makes cells more sensitive to the elongation inhibitor 6-AU. In reporter gene assays, ess1 mutations reduce the ability of elongation-arrest sites to stall polymerase. We also show that Ess1 acts positively in transcription termination, independent of its role in elongation. We propose that Ess1-induced conformational changes attenuate pol II elongation and help coordinate the ordered assembly of protein complexes on the CTD. In this way, Ess1 might regulate the transition between multiple steps of transcription.
ESS1 encodes a highly conserved, peptidyl-prolyl isomerase (PPIase; Haneset al. 1989; Haniet al. 1995). Ess1 and its homologs (Dodo and Pin1) have been implicated in such diverse biological functions as checkpoint control in mitosis (Luet al. 1996), signaling during embryonic development (Hsuet al. 2001), and regulation of gene transcription (Wu et al. 2000, 2001; Shaw 2002). ESS1 was discovered in Saccharomyces cerevisiae and was shown to be essential in this organism (Haneset al. 1989) and in the human fungal pathogen Candida albicans (Devasahayamet al. 2002). It is present in all eukaryotic organisms that have been examined, but is not essential in some other fungi (Crenshawet al. 1998; Huanget al. 2001; P. Ren, A. Rossettini and S. D. Hanes, unpublished observations) or in metazoans (Maleszkaet al. 1996; Fujimoriet al. 1999). As do other PPIases, e.g., cyclophilins and FKBPs (Dolinski and Heitman 1997), Ess1 likely functions by the cis-trans conversion of the peptide bond preceding prolines within target proteins, thus altering protein conformation (Fischeret al. 1998; Haniet al. 1999). This might occur in newly synthesized proteins to aid in folding or in mature proteins to control their activity or association with other protein complexes (Dolinski and Heitman 1997). Ess1 and its homologs have a substrate specificity that is distinct from that of other PPIases and has been defined as Ser-Pro or Thr-Pro, where the residues preceding proline are phosphorylated (Yaffeet al. 1997; Fischeret al. 1998).
The first major clue that Ess1 is important for transcription came from results of a multicopy suppressor screen, in which five out of six genes identified were transcription related (Wuet al. 2000). Among those identified was FCP1, which encodes a phosphatase specific for the carboxy-terminal domain (CTD) of Rpb1, the largest subunit of RNA polymerase II (pol II; Archambaultet al. 1997; Koboret al. 1999). ESS1 was then shown to interact genetically with RPB1 and, in particular, with alleles that alter the CTD (Wu et al. 2000, 2001; C. B. Wilcox, A. Rossettini and S. D. Hanes, unpublished observations). ESS1 also interacts with genes involved in CTD function, such as SRB2 and SRB10, which suppress the phenotype of CTD truncations (Wu et al. 2000, 2001; C. B. Wilcox, A. Rossettini and S. D. Hanes, unpublished observations), and Ess1 binds the CTD in vitro (Morriset al. 1999; Wuet al. 2000). These and other data suggest that the interaction of Ess1 with the CTD is important for RNA pol II transcription in vivo (Wu et al. 2000, 2001).
The CTD consists of multiple copies of the heptad repeat, YSPTSPS, which undergoes reversible phosphorylation during transcription (Cordenet al. 1985; Dahmus 1996). This covalent modification regulates the binding of numerous transcriptional accessory proteins to the CTD and might, in part, control the transition from one step of transcription to the next (McCrackenet al. 1997a; Patturajanet al. 1998; Morris and Greenleaf 2000; Barillaet al. 2001; Dichtlet al. 2002b; Licatalosiet al. 2002). Phosphorylation of Ser2 and Ser5 in the heptad repeat also generates Ess1 target sites (phospho-SerPro) that might trigger the binding of Ess1 to the CTD (Morriset al. 1999). As previously proposed, the bound Ess1 would isomerize the Ser-Pro peptide bonds, altering the CTD conformation, and thus provide an additional, noncovalent means to regulate the binding and release of accessory proteins to the CTD (Wuet al. 2000).
Since Ess1 binds the CTD, it could potentially act during any step in which the CTD is involved. The CTD was initially shown to be required for initiation (Carlson 1997), and we have obtained genetic evidence that Ess1 is required at this step (C. B. Wilcox, A. Rossettini and S. D. Hanes, unpublished observations). The CTD also appears to play a role in elongation, termination, and mRNA processing (including 5′ capping, splicing, and 3′-end processing; Hirose and Manley 2000; Howe 2002; Proudfootet al. 2002). The CTD's involvement in elongation is suggested by the changes in the pattern of CTD phosphorylation that occur during elongation (Payneet al. 1989; O'Brienet al. 1994; Komarnitskyet al. 2000; Schroederet al. 2000) and by the genetic interactions observed between the CTD and the elongation factors Dst1 (yeast TFIIS) and Spt4-Spt5 (Lindstrom and Hartzog 2001). The CTD's role in termination may be direct, or it may be related to its role in 3′-end processing (McCrackenet al. 1997b; Licatalosiet al. 2002). The requirement for the CTD in elongation and termination raises the possibility that Ess1 may also be important for these postinitiation steps. In fact, work by Hani et al. (1995) has implicated Ess1 in termination.
To identify in which step(s) in transcription Ess1 functions, we first characterized a potent multicopy suppressor of ESS1, the yeast open reading frame (ORF) YKL005C, which in this article we have renamed BYE1 (Bypass of ESS1). We studied BYE1 because it encodes a protein that contains a TFIIS-like domain and was therefore suspected to be an elongation factor. Indeed, we found that Bye1 interacts with Rpb1 and acts negatively in elongation, a result that also implicated Ess1 in elongation, since high levels of Bye1 eliminate the requirement for Ess1. Using genetic interaction tests, chemical inhibitor experiments, and reporter gene assays, we showed that Ess1 is involved in elongation and acts to oppose the positive effects of known elongation factors Dst1 and the Spt4-Spt5 complex. The results also indicate that the termination readthrough defects previously reported in ess1 mutants (Haniet al. 1999) may be unrelated to the role of Ess1 in elongation, suggesting that Ess1 acts in multiple, sequential steps during transcription. Our results are consistent with the model that Ess1 coordinates the binding of accessory proteins to the CTD to regulate the transition between discrete steps of transcription.
MATERIALS AND METHODS
Yeast strains: Strains used in this article are listed in Table 1. W303 1A (wild type) and W303 1B were gifts from Rod Rothstein. ess1ts strains YGD-ts8W (ess1A144T) and YGD-ts22W (ess1H164R) were described previously (Wuet al. 2000). YXW21 and YXW20 are MATα versions of YGD-ts8W and YGD-ts22W, respectively, obtained in backcrosses with W303 1B. In all crosses involving ess1ts strains, here and below, ESS1 allele status was confirmed by allele-specific PCR. YXW85-a was generated by mating CBW1 (Wuet al. 2000) with YXW21 and selecting for G418R segregants (containing rpb1Δ) that were temperature sensitive (ts; containing ess1A144T). YXW86-a was derived in the same way, except that CBW1 was mated with YXW20. YXW84-a, which has the same genotype as CBW1, was regenerated in these crosses by selecting for G418R non-ts segregants. YXW95, YXW96, and YXW97 were made by plasmid shuffling in YXW84-a using plasmids pYF1866, pYF1869, and pYF1864, as appropriate.
A bye1Δ strain, YXW35-a, was made by transforming W303 1B with a bye1ΔLEU2 fragment derived from pXW13 (Wuet al. 2000) followed by backcrossing correct deletion isolates (as demonstrated by PCR) with W303 1A and identifying Leu+ segregants. A dst1Δ strain, YXW52-α, was made by transforming W303 1B with a dst1ΔURA3 fragment generated by PCR using URA3-specific primers containing 45 nucleotides homologous to DST1 flanking sequence. Deletions were confirmed by PCR, and two independent isolates were used. YXW56 and YXW57-α (ess1ts dst1Δ double-mutant strains) were generated by mating YXW52-α with YGD-ts8W and YGD-ts22W, respectively, and obtaining Ura+ (carrying dst1Δ) ts segregants. YAR1–YAR3 were derivatives of YXW52-α, YXW56-a, and YXW57-α in which the URA3 in dst1ΔURA3 was replaced by HIS3 using a dst1ΔHIS3 PCR fragment. Correct recombinants were confirmed by PCR. Two independent isolates were used in each study. A dst1Δ bye1Δ double-mutant strain, YXW103-α, was made by mating between YAR1 and YXW35-a and selecting His+ (carrying dst1Δ) Leu+ (carrying bye1Δ) segregants.
GHY180 (spt4Δ), OY96 (spt5-194), GHY92 (spt5-242), and a wild-type control strain OY100 were gifts from Grant Hartzog (Hartzoget al. 1998), as was OY175 (spt5Δ+ pSPT5; G. Hartzog, unpublished data). YXW77, sharing the same genotype as OY175, was a meiotic segregant derived from mating OY175 with YXW21. YXW82-a (spt5Δ+ pspt5-194) was made by plasmid shuffling in OY175 using pBM24 and then pspt5194. The final shuffle isolates were backcrossed to YXW21 and meiotic segregants with the correct genotype were obtained. YXW79-a and YXW80-a were ess1ts spt5Δ+ pSPT5 strains generated by mating YXW77 with YXW21 and YXW20, respectively, and selecting for Leu+ (bearing spt5Δ)Ura+ (bearing pSPT5) ts segregants from the resultant diploid.
Plasmids: Yeast expression vector pJGS-4 (2μ, TRP1) was used to express hemagglutinin (HA)-tagged wild-type and mutant Bye1 proteins. It contains an ADH1 promoter and an ADH1 terminator, between which there are unique EcoRI and XhoI sites followed by stop codons in all three reading frames (Zhu and Hanes 2000). pWTN, expressing N-terminally HA-tagged Bye1, was constructed in three steps. First, an EcoRI PCR fragment corresponding to the N-terminal portion of Bye1 (arbitrarily defined by the unique EcoRI site at the beginning of the TFIIS-like domain) was inserted into pJGS-4, giving rise to pA(untag). Second, a pair of annealed oligos encoding the HA epitope and SV40 nuclear localization sequence (NLS) was inserted into the BglII site at the beginning of Bye1. This intermediate construct encodes a deletion mutant of Bye1 (pA). Finally, the remainder of Bye1 was reconstituted into pA as an EcoRI-XhoI fragment from pXW11 (Wuet al. 2000). pWTC, expressing the C-terminally tagged Bye1, was made in a similar way, except that the annealed HA-NLS oligos were inserted into the BglII site at the C-terminal end of Bye1. pB was made by inserting an AvrII-XhoI PCR fragment encoding the TFIIS-like domain into the same sites in pA. pC was made by replacing the BglII-XhoI fragment in pA(untag) with a BglII-XhoI PCR fragment encoding the TFIIS-like domain, followed by insertion of the annealed HA-NLS oligos into the BglII site. pE was made by inserting an AvrII-XhoI PCR fragment encoding the tagged C-terminal region of Bye1 into the same sites in pA(untag). pD and pF were also derived from pA(untag) by replacing the BglII-XhoI fragment with a PCR fragment encoding either the TFIIS-like domain plus the tagged C-terminal region of Bye1 or just the tagged C-terminal region of Bye1, respectively. All plasmids containing PCR-generated fragments were sequenced and found to be free of mutations.
Bye1 point mutants were made using PCR overlap extension (Hortonet al. 1990). The PCR fragment containing the desired point mutations in the TFIIS-like domain was inserted into the AvrII and XhoI sites of pA. Subsequently, the C-terminal region of Bye1 was reconstituted as a KpnI fragment from pWTN. This KpnI fragment also contained an ADH1 terminator, replacing that in the intermediate plasmids. The presence of desired mutations was confirmed by DNA sequencing, which also verified the absence of undesired mutations.
pJGS-DST1, expressing Dst1, was made by insertion of the Dst1 coding sequence as a PCR fragment into the XhoI site in pJGS-4. pXW14 was made by insertion of an XhoI-BamHI restriction fragment expressing BYE1 from pXW11 (Wuet al. 2000) into the same sites in pRS423 (Christiansonet al. 1992). YEpESS1 and YEpHESS1 were previously described (Haneset al. 1989), as well as pRS413-ESS1 (Wuet al. 2000) and pRS424-ESS1 (Wuet al. 2001).
Plasmids expressing wild-type RPB1 (pYF1866) or mutant rpb1 (pYF1869 and pYF1864) were gifts from James Friesen (Archambaultet al. 1992). pRP112, which expresses wild-type RPB1 on a URA3 plasmid, was provided by Jeff Cordon (West and Corden 1995). Plasmids expressing wild-type SPT5 (pMS4 and pBM24) and mutant spt5 (pspt5-194 and pspt5-242) were provided by Grant Hartzog (Swansonet al. 1991; Hartzoget al. 1998; G. Hartzog, unpublished data). The artificial arrest site (ARTAR) elongation reporter and control plasmids, pDK12 and pDK1, were gifts from Kevin Struhl (Kulish and Struhl 2001). Termination reporter and control plasmids, pL101 and pD16, were obtained from Linda Hyman (Tulane University; Hymanet al. 1991).
Yeast media and manipulations: Rich medium (YPD) and selective complete synthetic medium (CSM) were prepared according to standard protocols (Adamset al. 1997). Yeast plates containing 5-fluoroorotic acid (5-FOA) or G418 were made according to standard protocol (Wachet al. 1994; Adamset al. 1997). Plates containing 6-azauracil (6-AU) were prepared by adding 6-AU to desired final concentrations into CSM minus ura medium or CSM medium lacking uracil and another amino acid for plasmid selection. For 6-AU sensitivity assays, pRS306 (CEN, URA3; Sikorski and Hieter 1989) was introduced into all ura– strains to confer ura prototroph. Spot-test growth assays were performed by growing yeast cells to midlog phase at 30° and then diluting each culture to ∼107 cells/ml (OD600 = 0.5). Serial dilutions were then carried out and 2–3 μl of each was spotted onto desired media. Each experiment was carried out several times, using multiple isolates, with relevant genotypes spotted on the same plate and incubated for the same time.
Structure modeling: The TFIIS-like domain of Bye1 was modeled using the nuclear magnetic resonance (NMR) structure of domain II of Dst1 (PDB ID: 1ENW) as a template. Modeling was done both through the SwissModel server (http://www.expasy.org/swissmod) and by the MolScript software (Kraulis 1991). These two gave similar results.
Western blotting: HA-tagged Bye1 proteins in whole cell lysates were detected using anti-HA mouse monoclonal antibody 12CA5 (Roche, Indianapolis) as primary antibody at 1 μg/ml, and HRP-conjugated goat anti-mouse Ig (Amersham, Arlington Heights, IL) as secondary antibody at 1:3000. Results were revealed using an enhanced chemiluminescence detection kit.
Reporter assays: Cells were grown in raffinose medium to midlog phase at 25°, prior to galactose induction. β-Gal activity was measured as previously described (Hanes and Brent 1989). For the ARTAR-based reporter assays, in agreement with Kulish and Struhl (2001), we found that ARTAR caused a decrease in β-gal activity only under low-galactose conditions (0.02%). However, in our assays, this decrease was subtle (about twofold after 2 hr of induction) and transient (not detectable after 4 hr of induction). Further reduction of the galactose concentration or the induction time resulted in β-gal activities that were too low to be quantitated. These differences may be due to the fact that we used a high-copy plasmid version of the reporter system, rather than the integrated version used previously (Kulish and Struhl 2001).
Positive interactions between ESS1 and its suppressor, BYE1: One approach to understanding the role of Ess1 in transcription is to study genes whose mutation or overexpression suppresses the lethality of ess1 mutations. BYE1 was identified as a multicopy suppressor of temperature-sensitive ess1 mutations, and it also suppresses an ess1Δ mutation (Wuet al. 2000). Whereas deletion of BYE1 had no obvious phenotype on its own, it showed a synthetic-enhancing interaction with ess1ts mutations. For example, bye1Δ ess1ts double mutants showed severe slow-growth phenotypes at 30°, a temperature at which both single mutants grow normally (data not shown). Both the suppression and the synthetic phenotype suggest that ESS1 and BYE1 have partially overlapping functions or act in the same or a parallel pathway.
The TFIIS-like domain of Bye1 is required for suppression of ess1 mutants: Bye1 contains two obvious sequence motifs that suggest a role in transcription (Figure 1A). The first is a PHD finger [amino acids (aa) 72–133], which has been found in proteins involved in chromatin remodeling (Aaslandet al. 1995). The second is a TFIIS-like region (aa 177–354), which shares 43% overall similarity to a conserved sequence in the Drosophila elongation factor, TFIIS (Wind and Reines 2000). The TFIIS-like region in Bye1 is only weakly similar to that of the yeast TFIIS homolog, Dst1. In addition, Bye1 contains two bipartite NLSs at its N terminus (aa 30–47; Dingwall and Laskey 1991) and was reported to localize to the nucleus (TRIPLES project; Kumaret al. 2000).
To determine which sequence motif within Bye1 is required for suppression of ess1 mutants, we generated a series of deletion constructs (Figure 1A). Each deletion was engineered to carry an SV40 NLS to ensure nuclear localization of the mutant protein and an HA epitope to allow protein levels to be monitored. Addition of the NLS and HA tag to full-length Bye1 at either the amino terminus (WTN) or the carboxy terminus (WTC) did not affect Bye1's ability to suppress ess1 mutants (Figure 1B). Each Bye1 deletion mutant was tested for suppression activity in both ess1A144T and ess1H164R mutants (Figure 1B and data not shown). Identical results were obtained using either allele.
The results are summarized in Figure 1A and demonstrate that the TFIIS-like domain is necessary for suppression (mutants A, E, and F), although it does not appear to be sufficient (mutant C). By contrast, the PHD finger and the C-terminal region are individually dispensable for suppression (mutants D and B). Western analyses confirm that differences in the ability of the deletion mutants to suppress were not due to differences in protein levels in either ess1A144T (Figure 1C, right) or ess1H164R mutants (data not shown), since levels of mutant proteins were equal to or greater than those of wild-type Bye1. The fact that the TFIIS-like domain is required suggests that an elongation-related activity may be important for the suppression of ess1 mutants.
We also observed that deletion mutants that retained the PHD finger but lacked the TFIIS-like domain (mutants A and E) exhibited dominant negativity, in which they enhanced the growth defects of ess1ts mutants (Figure 1B, 32°). This might be explained by the fact that PHD fingers are thought to mediate protein-protein interactions and dominant negativity might occur because the mutant proteins form inactive complexes that interfere with wild-type Bye1 function.
The TFIIS-like domain of Bye1 likely binds Rpb1 to suppress ess1 mutants: Although there is little overall sequence similarity between Bye1 and the elongation factor Dst1, a short region in the C-terminal portion of the TFIIS-like domain of Bye1 appears to be conserved (56 residues with 27% identity and 43% similarity; Figure 2A). This region in Dst1 mediates binding to the Rpb1 subunit of pol II and four residues in this region (K196, R198, R200, and K209) form a basic patch that is critical for Rpb1 binding (Awreyet al. 1998). Bye1 also contains several basic residues in this region, suggesting that it, too, might bind Rpb1 (Figure 2A).
To investigate this possibility, we first generated a structural model of the conserved region of Bye1 (residues 292–347) using the NMR structure of the TFIIS-signature domain of Dst1 as a template (Figure 2B). The two structures superimposed well, with an RMS deviation of only 0.16Å. The comparison clearly identifies side chains of three of the four key residues in Dst1 (R198, R200, and K209) as having spatial equivalents in Bye1 (K305, R307, and K316), with the possibility of a fourth (K196 in Dst1 similar to K299 in Bye1). Together, these residues in Bye1 might form a basic patch capable of binding Rpb1.
We tested this idea by generating alanine substitutions in three of the four conserved residues in Bye1 (K305, R307, and K316). Since BYE1 mutations (bye1Δ) have no discernible phenotype, we measured their ability to suppress ess1ts mutants. The R307A mutation lost its ability to suppress, while the K305A mutation was significantly reduced for suppression (Figure 2C). K316A, however, suppressed almost as well as the wild type. This could be due to the presence of other basic residues (K314 and K317) adjacent to K316, which might supply the requisite contact. The loss of suppression for K307A and K305A was not due to an insufficient expression of the mutant proteins (Figure 2D). These data suggest that Bye1 binds Rpb1 using its TFIIS-like domain and that this interaction is required for suppression.
To support this idea, we tested the converse, using two Rpb1 alleles (rpo21-18 and rpo21-24) that fail to bind Dst1 due to a small insertion or a substitution in the cognate binding site of Rpb1 (I1237TRARV and E1230K, respectively; Archambaultet al. 1992; Wuet al. 1996). If Bye1 also binds this site, then these mutations should abolish the Bye1-Rpb1 interaction, and suppression of ess1ts mutants should be lost. For this experiment, ess1H164R rpb1 double-mutant strains containing either vector alone or a multicopy BYE1 plasmid were tested for growth at 34° (nonpermissive for ess1ts mutants). As a control, we first showed that the rpo21 alleles, which are known to be ts at 37° (Archambaultet al. 1992), were, in fact, able to grow normally at 34° (Figure 3A, top). Once this was established, we showed that BYE1 overexpression could no longer suppress ess1H164R in the rpb1 mutant backgrounds (rpo21-18 and rpo21-24; Figure 3A, bottom), indicating that the Rpb1 residues required for interaction with Dst1 are also required for suppression by BYE1. These results are consistent with the idea that Bye1 binds to Rpb1. Biochemical experiments will be needed to confirm a physical interaction between Bye1 and Rpb1 and to demonstrate that this binding is direct.
The importance of the Bye1-Rpb1 interaction is revealed when Ess1 function is compromised. In the course of our experiments, we found that ess1A144T rpb1 double mutants (using rpo21-18 or rpo21-24 alleles) were synthetic lethal and that ess1H164R rpb1 mutants grew poorly. This was demonstrated by the inability to cure a pRPB1 (URA3) plasmid from ess1A144T rpb1 double-mutant cells on 5-FOA medium or, in the case of the ess1H164R rpb1, by a reduced curing rate (Figure 3B). Although ess1H164R rpb1 cells were viable, they were slow growing (Figure 3A, 25° plate). These phenotypes mimicked those observed for ess1ts bye1Δ mutants. Thus, it appears that loss of the Bye1 binding site on Rpb1 is functionally equivalent to deletion of BYE1. The simplest interpretation is that the Bye1-Rpb1 interaction is needed for Bye1 to augment Ess1's role in transcription, and this interaction becomes critical when Ess1 activity is compromised.
Finally, since our experiments suggested that Bye1 and Dst1 might bind the same site on Rpb1, it then seemed possible that a bye1 dst1 double mutant might have the same phenotype as the rpo21 mutants (ts for growth at 37°; 6-AU sensitive, see below). However, growth of the bye1 dst1 double mutant was not ts; growth was similar to that of the single mutants at all temperatures tested (25°,30°,34°, and 37°; data not shown). This result suggests that perhaps binding of other proteins to this site might also affect Rpb1 function.
BYE1 mutants are less sensitive to 6-AU, suggesting a negative role in elongation: The mutational and genetic analyses described above (Figures 2C and 3, A and B) suggest that Bye1 and Dst1 bind to the same site on Rpb1. Bye1 might therefore compete with Dst1 and antagonize its role in elongation. To determine whether BYE1 has a negative role in elongation, we employed the elongation inhibitor 6-AU, which is thought to promote pausing and arrest by limiting the intracellular pools of GTP and UTP (Exinger and Lacroute 1992). Mutations in genes that promote elongation are often associated with hypersensitivity to 6-AU (e.g., dst1Δ; Losson and Lacroute 1981; Wind and Reines 2000).
If Bye1 functions negatively in elongation, then deletion of BYE1 should render cells less sensitive to 6-AU. Indeed, we observed a reduced sensitivity to 6-AU in bye1Δ cells, especially at elevated temperature (32°), at which wild-type cells were 6-AU hypersensitive (Figure 4A). As controls for 6-AU efficacy, DST1 deletion and overexpression strains (dst1Δ; pDST1) were used. Details of the 6-AU experiments are described in materials and methods. Although overexpression of BYE1 had no effect on wild-type cells (data not shown), it rendered dst1Δ mutant cells more sensitive to 6-AU (Figure 4B). These results suggest that Bye1 has a negative role in elongation, opposite to that of Dst1, consistent with an Rpb1 binding-site competition model.
We also tested the 6-AU sensitivity of the bye1 dst1 double mutant (Figure 4A), thinking that perhaps the bye1 mutation might reverse the sensitivity caused by dst1Δ. It did not, however, as results showed that dst1Δ is epistatic to bye1Δ since the double mutant did not grow on 100 μg/ml 6-AU.
DST1-ESS1 genetic interactions suggest a negative role for Ess1 in elongation: Our original goal was to study the BYE1 suppressor as a means to understand ESS1 function. Thus far, we have provided evidence for a negative role of Bye1 in elongation. This implies a corresponding role for Ess1, because ess1 mutations are suppressed by overexpression of Bye1. If Ess1 functions negatively in elongation, then Ess1 and Dst1 should oppose one another, just as Bye1 and Dst1 do. In fact, deletion of DST1 partially rescued the defects in ess1ts mutant cells (ess1ts dst1Δ), allowing growth at nonpermissive temperature (35°; Figure 5A). Moreover, overexpression of DST1 enhanced the defects in ess1ts mutant cells, resulting in slow growth at permissive temperature (32°; Figure 5B). This effect was reversed by reintroducing wild-type ESS1 (Figure 5B). These results show that ESS1 and DST1 genetically oppose one another and, since Dst1 stimulates elongation, the implication is that Ess1 inhibits elongation.
ess1 mutants are less sensitive to 6-AU: To test whether Ess1 inhibits elongation, we measured the 6-AU sensitivity of ess1 mutant cells. We found that ess1H164R mutant cells were more resistant to 6-AU than wild-type cells were (Figure 6A), especially when compared to growth without the drug. Results with the ess1A144T mutant showed the same trend but were more subtle, perhaps because its growth is generally less robust (data not shown). As expected, overexpression of ESS1 had the opposite effect, rendering dst1Δ mutant cells more sensitive to 6-AU (Figure 6B). These two results suggest that Ess1 inhibits elongation. Accordingly, loss of Ess1 function may cause hyperelongation by pol II and, if so, this effect should be reversed by 6-AU. Remarkably, as predicted, low concentrations of 6-AU rescued growth of ess1H164R mutants at restrictive temperature (34°; Figure 6C).
ESS1 interacts genetically with elongation factors SPT4 and SPT5: Genetic interactions reported between SPT4 and SPT5 and elongation factors including DST1 indicate that SPT4 and SPT5 function positively in elongation and that elongation is compromised in spt4 and spt5 mutant cells (Hartzoget al. 1998). Consistent with a negative role for ESS1 in elongation, overexpression of ESS1, which does not affect the growth of wild-type cells (Haneset al. 1989; Figure 7A), enhanced the growth defects of spt4 and spt5 mutants, two of which are ts (spt4Δ and spt5-194) and one of which is cold sensitive (spt5-242; Figure 7A). The effects of ESS1 on spt mutants were most pronounced at semipermissive temperatures, at which spt mutants carrying vector alone are still viable. This effect of ESS1 overexpression resembled the effect previously reported for dst1Δ on these mutants (Hartzoget al. 1998), again indicating a negative role for ESS1 in elongation (summarized in Figure 7B).
Since overexpression of ESS1 enhanced spt5 mutant phenotypes, mutation of ESS1 might be expected to suppress them. However, we were unable to generate stable ess1ts spt5-194 or ess1ts spt5-242 double mutants by tetrad dissection (where the spt5 mutations were carried on plasmids), despite repeated attempts (Table 2 and data not shown). This suggests that mutation of ESS1 did not suppress spt5 alleles, but instead that ess1ts spt5 double mutants are synthetic lethal. This was confirmed for one spt5 allele, using a plasmid-shuffle assay in which ess1ts spt5Δ cells carrying an spt5-242 plasmid failed to lose a URA3-based SPT5 plasmid on 5-FOA (Figure 7C). This synthetic lethality was somewhat surprising, but might be explained by the fact that both ESS1 and SPT5 are essential genes thought to have multiple effects on transcription (Lindstromet al. 2003; C. B. Wilcox, A. Rossettini and S. D. Hanes, unpublished observations). Indeed, human Spt5 has been shown to have both positive and negative effects on transcription in vitro (Wadaet al. 1998).
Physical interactions between Ess1 and Spt5 may also occur as suggested by their common presence within protein complexes (Hoet al. 2002). In addition, the human orthologs (Pin1 and hSpt5) have been shown to interact in vitro via a Ser-Pro rich region (the CTR) within hSpt5 (Lavoieet al. 2001). However, the CTR is not found in yeast Spt5, so any potential interactions between Ess1 and Spt5 are likely to be different. Functional consequences of potential Ess1-Spt5 (or Pin1-hSpt5) interactions have yet to be demonstrated.
ess1ts causes a reduced sensitivity to an ARTAR: One way in which Ess1 and/or Bye1 could inhibit elongation is by promoting polymerase pausing or arrest. To test this idea, we employed a reporter gene system that has been reported to monitor elongation arrest (Kulish and Struhl 2001). In this system (Figure 8A), the ARTAR sequence (containing three artificial arrest sites) is inserted into the coding region of lacZ, driven by the GAL1-inducible promoter. Elongation arrest at the ARTAR would decrease lacZ expression, as measured by β-galactosidase activity.
If ESS1 or BYE1 promotes arrest, then mutation of either of them should result in a pol II that is less likely to arrest and thus result in a reduced ARTAR activity. ARTAR activity was determined as the ratio of β-gal activity of (–)ARTAR:(+)ARTAR reporters. In our assays, the ARTAR activity was rather low (see materials and methods), resulting in only a 2.1-fold decrease in β-gal activity in wild-type cells (Table 3A). In ess1A144T and ess1H164R mutant cells, ARTAR activity was reduced to 1.6- and 1.9-fold, respectively (Table 3A). Although this reduction was modest, the effect was reproducible and could be reversed by addition of plasmids expressing wild-type ESS1 (Table 3A). The results are consistent with the idea that ess1 mutants cause hyperelongation. No effect on ARTAR activity was detected in bye1Δ mutant cells.
A better demonstration of an effect of ess1ts on ARTAR activity was provided using dst1Δ cells. Consistent with the earlier work (Kulish and Struhl 2001), we found that reporter gene expression is generally lower in dst1Δ mutant cells. However, more important was the finding that deletion of DST1 increased ARTAR activity (to 3.5-fold; Table 3B), but that the increase in ARTAR activity was reversed (to 1.8-fold) by ess1A144T and ess1H164R in double mutants (Table 3C). Thus, ess1 is epistatic to dst1Δ with respect to ARTAR activity, suggesting that hyperelongation caused by ess1 mutations counteracts the inefficient elongation in dst1Δ cells. If the ARTAR does, in fact, cause elongation arrest in vivo, then our results suggest that Ess1 promotes pausing or arrest and that loss of this function causes hyperelongation, rendering pol II resistant to arrest.
Mutations in ESS1 cause defects in transcription termination: In addition to inhibiting elongation, ESS1 may also promote termination. This was first suggested by the fact that ESS1 was reisolated as PTF1 (processing/termination factor 1) in a screen for mutations that resulted in transcription readthrough and 3′-end processing defects using cryptic or weak terminators (Hani et al. 1995, 1999). To confirm that Ess1 functions in termination, we employed a reporter gene system that uses the ADH2 terminator (ADH2T; Hymanet al. 1991). In this system (Figure 8B), the ADH2T is inserted in the mini-intron of rp51, which is fused to lacZ. Correct termination at the ADH2T results in a truncated transcript and no β-gal activity. However, in the case of a termination defect, polymerase reads through the ADH2T, synthesizing the rp51-lacZ fusion transcript, which will be spliced and translated and give rise to β-gal activity. Note that this assay is an indirect measure of readthrough, since it actually monitors overall reporter gene expression.
We estimated transcription readthrough by comparing β-gal activity reporters with or without the terminator. In this assay, ess1H164R mutant cells showed a significant readthrough (almost 20%), consistent with earlier studies (Haniet al. 1999). Readthrough was not observed in either wild-type cells or ess1H164R cells expressing wild-type ESS1 on a plasmid (<3%; Table 4). These results suggest that Ess1 is important for termination.
The roles of ESS1 in elongation and termination may be independent. Or they may be interdependent so that changes in elongation might also affect termination; for example, in ess1 mutant cells, hyperelongating pol II might fail to recognize or respond to the cis-acting poly(A) and 3′ cleavage sequences and, thus, be unable to terminate. If the observed termination defect is a consequence of the elongation defect, then we expect it to be rescued by the elongation-related ess1 suppressors, BYE1 and dst1Δ. However, overexpression of BYE1 did not prevent transcription readthrough, nor did the dst1Δ mutation (Table 4). Moreover, overexpression of DST1 did not enhance this readthrough. These results indicate that the defects in termination and elongation in ess1 cells are not interdependent, suggesting that the functions of Ess1 in termination and elongation are separable. Similar results were obtained with ess1A144T mutant cells (data not shown). These results show that Ess1 functions in sequential postinitiation steps in transcription.
We have shown that Ess1 and Bye1 are likely negative regulators of elongation. This is the first described function for Bye1 (formerly known as YKL005C). Although Ess1 and Bye1 both seem to inhibit elongation, they probably act by different mechanisms. Ess1 binds phos-pho-Ser-Pro sites within the CTD of Rpb1 and is likely to isomerize the CTD, coordinating the exchange of proteins required for elongation and termination. Bye1 probably binds Rpb1, but elsewhere within the protein (∼aa 1230–1237), and might act by competing with Dst1 for binding to the elongating pol II complex. These results lead to the idea that, in yeast, as in mammals (Yamaguchiet al. 1999; Wadaet al. 2000; Ping and Rana 2001), elongation by RNA pol II proceeds as a result of the balance between positive and negative elongation factors.
Bye1 is a negative regulator of transcription elongation: The opposite effects of Bye1 and Dst1 on 6-AU sensitivity and on the growth of ess1 mutant cells suggest that these proteins function in the same process, but perform opposite functions. This might be explained by the fact that Bye1 and Dst1 share a similar domain and may bind to the same site on Rpb1. This would result in a competition for Rpb1 binding in which Bye1 displaces Dst1, reducing the ability of pol II to elongate.
However, two findings suggest that Bye1 does more than simply displace Dst1, but that it has an inherent negative activity. First, overexpression of BYE1 increased 6-AU sensitivity, suggesting an inhibition of elongation, even when no Dst1 was present (in dst1Δ mutants). Second, the TFIIS-like domain alone, which contains the Rpb1 binding site and should displace Dst1, failed to suppress ess1 mutants. The simplest explanation is that in addition to displacing Dst1, Bye1 actively inhibits elongation. For example, Bye1 might induce a change either in the structure of the elongating pol II complex or in the association of other factors with this complex. This negative elongation activity of Bye1 might depend on sequences outside the Rpb1 binding site, which differ from those in Dst1. Biochemical experiments are needed to test the negative role for Bye1 in elongation and to help elucidate the underlying mechanism.
Ess1 is a negative regulator of transcription elongation: Several lines of evidence indicated that Ess1 inhibits elongation. These include the genetic interactions observed among ESS1 and DST1, SPT4, SPT5, the effects of Ess1 on 6-AU sensitivity, and the effects of ess1 mutations on elongation through an artificial arrest site. Inhibition of elongation by Ess1 may be an important means to reduce the expression of certain genes. Alternatively, we favor the idea that Ess1 may be required for the proper expression of certain genes by attenuating pol II elongation to a degree sufficient for closely coupled events such as mRNA capping, splicing, and chromatin remodeling (Hartzoget al. 2002; Howe 2002; Proudfootet al. 2002) to occur in an orderly manner.
Ess1 may inhibit elongation by increasing the likelihood of pol II pausing or arrest rather than by reducing the overall rate of elongation. An effect of Ess1 on pausing or arrest is suggested by the fact that ess1 mutations increased reporter gene expression in the ARTAR assay (implying reduced pol II arrest; Table 3) and is consistent with genetic interactions observed between ESS1 and SPT5. Previous studies showed that while the growth defects of spt5-242 mutant cells are rescued by 6-AU (presumably by reducing the rate of elongation) and by an Rpb2 mutation that results in a “slow” polymerase (Hartzoget al. 1998), they are exacerbated by pol II pausing or arrest (Hartzoget al. 1998). We found that overexpression of ESS1 also exacerbated the growth defects of spt5-242 mutant cells (Figure 7A), consistent with a role for Ess1 in pol II pausing or arrest. An exact role of Ess1 in pausing or arrest, however, may be difficult to elucidate in vivo, given the artificial nature of ARTAR and the lack of information about natural pause or arrest sites in yeast. Future work will benefit from the use of in vitro elongation systems (Christieet al. 1994).
An additional role for Ess1 in termination and 3′-end processing: In addition to its role in elongation, ADH2T reporter assays strongly suggest a role for Ess1 in termination. Such a dual role has also been found for CHD1 and SSU72, which had been described as termination factors but also seem to affect elongation (Woodageet al. 1997; Alenet al. 2002; Dichtlet al. 2002a; Kroganet al. 2002). In the case of Ess1, we have shown the functions to be separable, because the termination defect in ess1 mutants was not rescued by the elongationrelated suppressors (BYE1 and dst1Δ). Thus, Ess1 seems to act independently in these two processes.
Are the functions of Ess1 in elongation and termination required for viability? Deletion of ESS1 is lethal, probably due to an essential role in transcription. The work of Hani et al. (1999) suggested that termination might be the critical step, since the ess1 mutants isolated in their transcription readthrough screen were also ts for growth. However, we do not think this is the case because neither BYE1 nor dst1Δ, which restore cell viability, rescued transcription readthrough of ADH2T (Table 4), nor did any of the previously described (Wuet al. 2000) multicopy suppressors of ess1 mutants (data not shown). In addition, ess1 mutations caused readthrough only at weak terminators (Haniet al. 1999), and we found that efficiency of termination by ADH2T was only reduced rather than abolished in ess1 mutants. For example, the levels of ADH2T reporter gene expression suggested that termination was still ∼80% effective in ess1H164R cells vs. ∼97% effective in wild-type cells (Table 4). Thus, it is possible that only a subset of genes will be affected by loss of Ess1 and that these effects on termination may be modest.
We suggest that instead, elongation may be the critical step since reducing elongation genetically or with 6-AU restores viability. However, for both elongation and termination, it is possible that suppression occurs as a result of a total bypass of the lethal defect, for instance by increasing the rates of other steps in transcription. Therefore, it remains a formal possibility that defects in either process may cause the lethality. Isolation of elongation- or termination-specific suppressors that do not rescue the growth defects may be useful to narrow the possibilities.
Ess1 may coordinate the binding of elongation and termination factors to the CTD: Coordinating the binding of accessory proteins to the pol II complex is crucial for the multistep process of transcription. For example, the exchange of initiation factors with elongation factors is needed for pol II to begin elongation (Pokholoket al. 2002). A similar exchange of elongation with termination factors is likely to be required. Since many of these factors bind to the CTD, Ess1 may play a direct role in coordinating protein exchange by altering the conformation of the CTD. The elongation and termination defects observed in ess1 mutant cells may result from a loss of this coordination.
Ess1 may also play an indirect role in coordinating protein exchange. Ess1-induced conformational changes in the CTD may alter its affinity for CTD-specific kinases and phosphatases. This will lead to changes in the CTD phosphorylation state, which would in turn affect the binding of other proteins. During elongation, the CTD has been observed to undergo changes in phosphorylation (Komarnitskyet al. 2000; Schroederet al. 2000), and it is possible that these changes are initiated by Ess1. Finally, Ess1 may have additional effects on gene regulation as a result of interactions with other transcription-related proteins, such as members of histone deacetylase complexes (e.g., Arévalo-Rodríguezet al. 2000).
Isomerization of the CTD may generate a diverse and dynamic platform for protein binding: The transition between discrete steps in transcription, such as that between elongation and termination, is likely to be a complicated process involving the exchange of many different proteins, perhaps multiple copies of each protein, and this exchange may not occur in an all-or-none fashion. Ess1-dependent conformational changes might be capable of coordinating such complicated exchanges, because there are multiple Ess1 target sites for isomerization within the CTD (two per repeat, 26 repeats), and individual repeats may be isomerized independently. As a result, Ess1 could generate a variety of CTD isomers, each attracting a distinct spectrum of pol II accessory proteins.
Moreover, CTD isomerization may act together with CTD phosphorylation to constitute an autoregulatory loop that governs CTD interactions. In this loop, Ess1-dependent isomerization would control the CTD phosphorylation state. Since Ess1 binds only to phosphorylated CTD, this would in turn control Ess1 binding and CTD isomerization. Such a phosphorylation-isomerization regulatory loop could control the binding and release of individual protein complexes to pol II and help drive the transcription cycle.
We thank Jeff Cordon, James Friesen, Grant Hartzog, Linda Hyman, and Kevin Struhl for gifts of plasmids and yeast strains. We thank Hongmin Li for help with structural modeling, the Wadsworth Center Molecular Genetics Core Facility for oligonucleotide synthesis and DNA sequencing, and Joan Curcio, Chuck Lowry, Randy Morse, Derek Scholes, and members of the Hanes lab for helpful discussions and/or critical comments on the manuscript. A.R. was supported by a National Science Foundation Research Experience for Undergraduates grant BIR-9987844. This work was supported by National Institutes of Health grant R01-GM55108 (S.D.H.).
Communicating editor: L. Pillus
- Received June 10, 2003.
- Accepted August 13, 2003.
- Copyright © 2003 by the Genetics Society of America