Synthetic Lethal Interactions Suggest a Role for the Saccharomyces cerevisiae Rtf1 Protein in Transcription Elongation
Patrick J. Costa, Karen M. Arndt

Abstract

Strong evidence indicates that transcription elongation by RNA polymerase II (pol II) is a highly regulated process. Here we present genetic results that indicate a role for the Saccharomyces cerevisiae Rtf1 protein in transcription elongation. A screen for synthetic lethal mutations was carried out with an rtf1 deletion mutation to identify factors that interact with Rtf1 or regulate the same process as Rtf1. The screen uncovered mutations in SRB5, CTK1, FCP1, and POB3. These genes encode an Srb/mediator component, a CTD kinase, a CTD phosphatase, and a protein involved in the regulation of transcription by chromatin structure, respectively. All of these gene products have been directly or indirectly implicated in transcription elongation, indicating that Rtf1 may also regulate this process. In support of this view, we show that RTF1 functionally interacts with genes that encode known elongation factors, including SPT4, SPT5, SPT16, and PPR2. We also show that a deletion of RTF1 causes sensitivity to 6-azauracil and mycophenolic acid, phenotypes correlated with a transcription elongation defect. Collectively, our results suggest that Rtf1 may function as a novel transcription elongation factor in yeast.

TRANSCRIPTION of mRNA by RNA polymerase (pol) II involves multiple steps, which include initiation, promoter clearance, elongation, and termination. Transcription regulatory factors could potentially target any of these steps to determine the level of transcript production. Recent evidence indicates that the transition from initiation to elongation is a highly regulated event in the transcription cycle. An important participant in this transition is the essential carboxyl-terminal domain (CTD) of the largest subunit of RNA pol II. The CTD contains highly conserved tandem repeats of the consensus sequence Tyr-Ser-Pro-Thr-Ser-Pro-Ser (Dahmus 1996). Yeast RNA pol II contains 26 or 27 repeats, while the mammalian enzyme contains 52 repeats. Phosphorylation of the CTD accompanies the transition from transcription initiation to elongation (Dahmus 1996). The hypophosphorylated form of RNA pol II preferentially enters the preinitiation complex (PIC), which assembles at the promoter (Luet al. 1991; Chesnutet al. 1992). Subsequently, the CTD is extensively phosphorylated. Several CTD kinases have been described. In yeast, these include CTDK-I, Srb10, and the essential TFIIH-associated kinase, Kin28 (Dahmus 1996). The coordinate regulation of these kinases is not well understood. However, as a component of the PIC, Kin28 appears to play a pivotal role in phosphorylating the CTD early in the transcription process (Dahmus 1996; Hampsey 1998). Transcription elongation is then executed by hyperphosphorylated RNA pol II (Cadena and Dahmus 1987; Payneet al. 1989; O'Brien et al. 1994). Upon completion of the transcript, the CTD must be dephosphorylated to reinitiate the transcription cycle. A CTD phosphatase, whose activity is stimulated by TFIIF, has been identified in human and yeast cells (Chamberset al. 1995; Archambaultet al. 1997; Choet al. 1999; Koboret al. 1999).

Several factors that affect initiation by RNA pol II also have roles in transcription elongation. Chromatin and chromatin remodeling factors are involved in the regulation of both processes, since nucleosomes provide a potent impediment to promoter recognition and mRNA chain elongation (Paranjapeet al. 1994; Kingstonet al. 1996; Uptainet al. 1997). The general transcription factors (TF) IIF and TFIIH, which are essential for PIC assembly and initiation, also regulate elongation. TFIIF interacts directly with RNA pol II and suppresses transient pausing by the enzyme (Uptainet al. 1997). The kinase activity of TFIIH has been shown by several studies to participate in elongation (Yankulov et al. 1995, 1996; Parada and Roeder 1996; Cujecet al. 1997; García-Martínezet al. 1997). Interestingly, phosphorylation of the CTD by TFIIH can be stimulated by the human immunodeficiency virus (HIV)-1 Tat protein, providing one mechanism by which Tat promotes transcription through an elongation block (Parada and Roeder 1996; Cujecet al. 1997; García-Martínezet al. 1997). Last, certain transcriptional activators facilitate elongation, possibly by recruiting elongation or chromatin remodeling factors to the polymerase (Yankulovet al. 1994; Brownet al. 1996).

Relative to the initiation step of transcription, much less is known about the factors that expressly control elongation. However, recent work has led to the characterization of several elongation factors, including TFIIS, P-TEFb, ELL, the elongator complex, and the Spt4-Spt5 complex (Marshall and Price 1995; Uptainet al. 1997; Hartzoget al. 1998; Wadaet al. 1998; Oteroet al. 1999; Wittschiebenet al. 1999). Of these proteins, TFIIS is the best characterized. TFIIS facilitates RNA pol II passage through arrest sites by stimulating an intrinsic ribonuclease activity of RNA pol II and causing cleavage of the nascent transcript near the 3′ end. In essence, this action resets RNA pol II and provides an additional opportunity to progress through an arrest site (Uptainet al. 1997). The Spt4 and Spt5 proteins form a complex that binds to RNA pol II and regulates elongation (Hartzoget al. 1998; Wadaet al. 1998). Interestingly, SPT4, SPT5, and a related gene, SPT6, were originally identified in a genetic selection for factors that regulate transcription initiation in yeast (Winston 1992). Considerable evidence suggests that these genes regulate transcription through an effect on chromatin structure (Swanson and Winston 1992; Bortvin and Winston 1996; Hartzoget al. 1998). Undoubtedly, the complexity of the RNA pol II transcription circuitry will require the involvement of additional initiation and elongation factors.

In accordance with this prediction, we previously reported the identification of a novel Saccharomyces cerevisiae gene, RTF1 (Restores TBP Function), whose product affects TATA-binding protein (TBP) function in vivo. RTF1 was uncovered in a genetic selection for extragenic suppressors of a TBP-altered specificity mutant, TBP-L205F (Arndtet al. 1994; Stolinskiet al. 1997). The altered DNA-binding specificity of TBP-L205F causes an Spt phenotype (Arndtet al. 1994). This phenotype reflects the ability of TBP-L205F to suppress the transcriptional defects caused by the insertion of the retrotransposon Ty or its long terminal repeat (δ) within the promoter of a gene. Because Ty elements contain several transcription signals, including a potent TATA box, their integration within a promoter establishes a competition between cis-acting transcription elements (Winston 1992). Mutations that confer an Spt phenotype are thought to affect this competition, and we have previously suggested that Rtf1 suppresses the Spt phenotype of TBP-L205F by directly or indirectly regulating TATA site selection by TBP (Stolinskiet al. 1997). Importantly, rtf1 deletion mutations (rtf1Δ) confer an Spt phenotype even in the presence of wild-type TBP (Stolinskiet al. 1997). RTF1 encodes a nuclear protein with a predicted mass of 65.8 kD (Stolinskiet al. 1997). The protein is rich in charged amino acids, a feature common to many transcription factors (Karlin 1993; Stolinskiet al. 1997), but lacks known functional motifs.

To clarify the role of Rtf1 in transcription, we have performed a genetic screen for mutations that cause lethality in combination with an rtf1 deletion mutation. The results of this screen, together with additional genetic interactions between Rtf1 and known elongation factors, suggest that Rtf1 is important for transcription elongation in yeast.

MATERIALS AND METHODS

Genetic methods and media: Rich (YPD), YPGlycerol (YPG), minimal (SD), synthetic complete (SC), 5-fluoro-orotic acid (5-FOA), and sporulation media were prepared as previously described (Roseet al. 1990). Galactose and sucrose media contained YEP (1% yeast extract, 2% Bacto-peptone), 1 μg/ml antimycin A, and either 2% galactose or 2% sucrose, respectively. Formamide, LiCl, and NaCl media contained YEP and the appropriate chemical (3% deionized formamide, 0.3 m LiCl, 1.2 m NaCl, or 1.4 m NaCl). SD media lacking (–Ino) or containing (+Ino) inositol were prepared as previously described (Shermanet al. 1981). Hydroxyurea media were prepared by supplementing SC media with 100 mm hydroxyurea (US Biological). 6-azauracil and mycophenolic acid media were prepared by supplementing SC-Ura media with 50 μg/ml 6-azauracil (Aldrich Chemical, Milwaukee) and 20 μg/ml mycophenolic acid (Sigma, St. Louis), respectively. All yeast strains used to test for 6-azauracil and mycophenolic acid sensitivity contained a URA3+ allele in the genome. Transformation of yeast cells was performed using the lithium acetate procedure and plasmids were recovered from yeast as described (Arndtet al. 1994).

Yeast strains: The S. cerevisiae strains used in this study appear in Table 1. Strains were constructed by standard methods (Roseet al. 1990). All FY, GHY, GY, and KY strains are isogenic with FY2, a GAL2+ derivative of S288C (Winstonet al. 1995). To introduce the ade2 and ade3 mutations into an rtf1Δ background, strain PSY137 (Koeppet al. 1996) was mated to KY409 (Stolinskiet al. 1997). This cross generated KA48, the original strain used for the synthetic lethal screen. With the exception of KA49, KA50, KA51, KA52, KA53, KA68, KA72, and KA76, all subsequently numbered KA strains were obtained from genetic crosses with KA48 derived mutants. The srb5Δ strain L937 is described in Roberts and Winston (1997).

Plasmids: Standard techniques were used for plasmid construction (Ausubelet al. 1998). pPC1, which harbors the RTF1, ADE3, and URA3 genes, was constructed by cloning the 3.1-kb SalI fragment from pKA61 (Stolinskiet al. 1997) into the SalI site of pPS719 (pRS426 + ADE3). pPC3, which contains the RTF1, ADE3, and TRP1 genes, was created by cloning the same insert into the SalI site of pPS793 (pRS424 + ADE3). pLS20, which contains RTF1 in pRS314, has been described (Stolinskiet al. 1997).

The following plasmids were created to verify the identity of the genes responsible for synthetic lethality with rtf1Δ and to determine linkage of the complementing genes to the synthetic lethal mutations. pPC13 (SRB5) and pPC14 (SRB5) were created by inserting a 1.9-kb BamHI-EcoRI fragment from pCT39 (Thompsonet al. 1993) into the corresponding sites of pRS314 and pRS304 (Sikorski and Hieter 1989), respectively. pPC20 (CTK1) and pPC19 (CTK1) were created by inserting a 3.7-kb PvuII insert from pPC15, one of three CTK1-containing library isolates, into the SmaI site of pRS314 and pRS304, respectively. pPC26 (FCP1) and pPC27 (FCP1) were constructed by subcloning a 2.7-kb SnaBI-XhoI insert from pPC25, the original FCP1-containing library isolate, into the SmaI and XhoI sites of pRS314 and pRS304, respectively. pPC29 (POB3) and pPC30 (POB3) are derived from pPC23. pPC23 was generated by inserting an 8.8-kb SalI-SacI fragment from pPC21, one of two POB3-containing library isolates, into the corresponding sites of pRS314. pPC29 and pPC30 were then created by subcloning a 2.4-kb ScaI-EcoRI fragment from pPC23 into the SmaI and EcoRI sites of pRS314 and pRS304, respectively.

View this table:
TABLE 1

Saccharomyces cerevisiae strains

The cloning of the pob3-272 and fcp1-110 mutations from strains KA58 and KA65, respectively, was achieved by gap repair (Orr-Weaveret al. 1983). pPC23 was digested with BstEII and BglII to delete a 4.0-kb fragment containing the POB3 gene. pPC25 was digested with StuI and SphI to excise a 3.7-kb fragment containing FCP1. The resulting vector fragments were transformed into the appropriate yeast strains. Plasmid DNA was recovered from Trp+ transformants, propagated in Escherichia coli, and retransformed into KA58 or KA65 to confirm by phenotypic analysis that the mutations had been cloned. The locations of the mutations were determined by subcloning and sequence analysis.

Synthetic lethal screen with rtf1Δ: To identify mutations that are synthetically lethal with rtf1Δ, we employed a red/white colony-sectoring assay (Kranz and Holm 1990). In brief, this assay is based on the observations that ade2 mutant colonies are red and that ade2 ade3 double mutant colonies are white, because ade3 mutations are epistatic to ade2 mutations (Kranz and Holm 1990). An ade2 ade3 strain carrying the wild-type ADE3 gene on a plasmid will form solid red colonies only when the plasmid is stably maintained. When grown under nonselective conditions, rapidly dividing cells can lose the ADE3-containing plasmid and generate colonies with red and white sectors. Therefore, this colony-sectoring assay can be used to detect mutants that require the ADE3-containing plasmid for life and thus appear solid red.

To find mutations that are lethal in combination with rtf1Δ, strain KA48 was transformed with plasmid pPC1. The transformed strain was plated on YPD and mutagenized by exposure to 7500 μJ/cm2 of UV light to ∼60% survival. Approximately 45,000 colonies were screened for those that appeared red and nonsectored (Sect phenotype). After purification, 235 colonies maintained the Sect phenotype. The Sect strains were then subjected to a second screen on plates containing 5-FOA, a drug that kills cells with a functional URA3 gene (Roseet al. 1990). Thirty-four Sect strains were 5-FOAS, and these strains were subjected to two additional tests to confirm that the synthetic lethality was specific to RTF1 and only one other gene. First, the 5-FOAS Sect strains were transformed individually with the centromeric plasmids pPC3 and pLS20 (Stolinskiet al. 1997). Plasmid pPC3 bears the wild-type RTF1, ADE3, and TRP1 genes, while pLS20 harbors only RTF1 and TRP1. If the synthetic lethality is not specific to URA3 or ADE3 expression, both pPC3 and pLS20 should confer 5-FOAR, but only pLS20 should allow the mutant strains to regain a sectored phenotype (Sect+). Second, 5-FOAS Sect strains that passed the above criteria were backcrossed to KA49, to test for dominance/recessivity and for 2:2 segregation of the 5-FOAS and Sect phenotypes. Fourteen mutants exhibited 2:2 segregation of these phenotypes, indicating that the synthetic lethal mutation in these mutants was due to a single gene. To determine if the synthetic lethal mutations conferred Spt and/or Bur phenotypes, the mutant strains were crossed to either KY616 or KY617.

Identification of synthetic lethal genes: The genes responsible for the synthetic lethality of complementation groups A (SRB5), B (CTK1), and C (FCP1) were cloned from a pRS200 (TRP1 CEN)-based S. cerevisiae genomic library (American Type Culture Collection, Rockville, MD; Sikorski and Hieter 1989) by complementing the 5-FOAS and Sect phenotypes of strains KA54, KA55, and KA56. Plasmid DNA was purified from 5-FOAR and Sect+ transformants that had lost plasmid pPC1 and was retransformed into the original Sect strain to confirm that the complementing activity was due to the library plasmid. Clones possessing complementing activity were subjected to DNA sequence analysis. In some instances, RTF1 clones were obtained, as established by restriction endonuclease analysis and/or DNA sequencing. The gene corresponding to complementation group D (POB3) was cloned from a YCp50-based S. cerevisiae genomic library (Roseet al. 1987) by complementing the Spt phenotype of strain KA59. Two complementing library plasmids that carried overlapping inserts were obtained. To confirm that a shared ORF, POB3, also complemented the 5-FOAS and Sect phenotypes, a POB3-containing fragment was inserted into plasmid pRS314 and transformed into strain KA57.

To determine if the cloned genes were allelic to the original synthetic lethal mutations, TRP1-marked integrating plasmids containing the cloned genes were transformed into yeast and linkage between TRP1 and the synthetic lethal mutations was examined. This analysis was performed using the following manipulations: (1) pPC14 was linearized by digestion with BstBI, transformed into KA50, and the resulting integrant was crossed to KA54; (2) pPC19 was linearized with NdeI, transformed into KA51, and the resulting integrant was crossed to KA55; (3) pPC27 was linearized with MscI, transformed into KA52, and the resulting integrant was crossed to KA56; and (4) pPC30 was linearized with BsmI, transformed into KA53, and the resulting integrant was crossed to KA57. Following tetrad analyses, all Trp segregants exhibited 5-FOAS and Sect phenotypes, demonstrating that the integration constructs were targeted to the genetically identified loci. To further demonstrate that we had cloned the correct gene responsible for complementing the Spt and Bur phenotypes of complementation group D, plasmid pPC30 was linearized by digestion with BsmI, transformed into KY571, and the resulting integrant was crossed to KA58. Following tetrad analysis, all Trp segregants were Spt and Bur, demonstrating that we had cloned the gene responsible for these phenotypes.

Identification of the chd1-52 suppressor mutation: A pob3-272 ura3 strain preferentially maintains a CEN URA3 plasmid harboring POB3 and exhibits weak 5-FOA sensitivity. This characteristic was used to clone the gene responsible for suppressing the extreme growth defect caused by the pob3-272 mutation. To test for dominance/recessivity and for 2:2 segregation of the growth suppression phenotype, strain KA61 was crossed to KA60. The resulting diploid exhibited weak 5-FOA sensitivity, indicating the suppressor mutation was recessive. Following tetrad analysis, the weak 5-FOA sensitivity segregated 2:2, demonstrating that this phenotype was due to a mutation in a single gene.

The gene responsible for suppressing the pob3-272 growth defect was determined as follows. Strain KA58 was transformed with plasmid pPC21. A Ura+ transformant was subsequently transformed with a pRS200 (TRP1 CEN)-based yeast genomic library. Double transformants that contained a library plasmid that complemented the suppressor mutation would strongly maintain pPC21, because plasmid loss would uncover the pob3-272 allele in an otherwise wild-type background. Ura+ Trp+ transformants that exhibited weak 5-FOA sensitivity (i.e., poor growth on 5-FOA media lacking tryptophan after 2 days at 30°) were identified by replica plating. The 5-FOAS transformants were mated to the wild-type strain FY23. Library plasmid DNA was obtained from selected diploids after causing the loss of plasmid pPC21 on 5-FOA media lacking tryptophan. Two different library plasmids, one of which contained CHD1, elicited a weak 5-FOAS phenotype upon retransformation into the initial strain used for cloning. To demonstrate that the suppressor mutation was linked to CHD1, GHY713 was crossed to KA60. All pob3-272 segregants from 19 complete four-spore tetrads exhibited wild-type growth, indicating that we had cloned the correct gene.

View this table:
TABLE 2

Mutations identified in the synthetic lethal screen with rtf1Δ

RESULTS

The rtf1Δ mutation is synthetically lethal with the loss of global transcription regulators: A synthetic lethal screen was performed with an rtf1Δ mutation to identify potential interactions with Rtf1 in vivo. Mutations that are lethal in combination with an rtf1Δ allele might reveal factors that regulate the same process as Rtf1 or factors that physically interact with Rtf1. By using a plasmid-sectoring assay (Kranz and Holm 1990), we screened for mutations that cause synthetic lethality with rtf1Δ. Ultimately, 14 synthetic lethal mutations were identified (see materials and methods for details). The mutations are all recessive and comprise nine complementation groups (Table 2; data not shown). This article describes the genes corresponding to four of these groups.

The genes responsible for the synthetic lethality were cloned by complementation, and their identities were verified by subcloning and linkage analysis. Gene and mutant allele names are listed in Table 2. Three of the genes, defined by mutations in SRB5, CTK1, and FCP1, have been directly implicated in the function and modification of RNA pol II. Srb5 is an important component of the SRB/mediator complex that associates with the CTD of RNA pol II, mediates the response to transcriptional activators, and stimulates phosphorylation of the CTD by TFIIH (Hampsey 1998). Ctk1 is the catalytic subunit of the CTDK-I kinase (Lee and Greenleaf 1991). This kinase has been shown to specifically phosphorylate the CTD and promote efficient elongation by RNA pol II in vitro (Lee and Greenleaf 1989, 1997; Sterneret al. 1995). Fcp1 is a recently described TFIIF-associated, CTD-specific protein phosphatase (Archambaultet al. 1997; Choet al. 1999; Koboret al. 1999). Fcp1 also possesses a positive elongation function independent of its phosphatase activity (Choet al. 1999). Our screen also uncovered a mutation in the POB3 gene. Pob3 shares similarity with HMG1-like proteins and forms a complex in yeast with Cdc68/Spt16 (Brewsteret al. 1998; Wittmeyeret al. 1999), a protein that has been implicated in the regulation of transcription by chromatin structure (Maloneet al. 1991; Rowleyet al. 1991; Brewsteret al. 1998). The human homologues of Pob3 and Cdc68/Spt16 form a complex known as FACT (facilitates chromatin transcription), which facilitates transcription elongation on chromatin templates in vitro (LeRoyet al. 1998; Orphanideset al. 1999). Together with additional data presented below, the identification of mutations in SRB5, CTK1, FCP1, and POB3 in our synthetic lethal screen suggests that Rtf1 regulates transcription elongation in vivo, perhaps at the initiation to elongation transition.

To confirm the synthetic lethal relationships by an approach distinct from the plasmid loss assay, we performed genetic crosses between an rtf1Δ strain and strains that carry the synthetic lethal mutations in an RTF1+ genomic background. Following tetrad analysis of the heterozygous diploid strains generated from these crosses, we observed no rtf1Δ ctk1-217 double mutant spores. The srb5-77 and fcp1-110 mutations in combination with rtf1Δ gave rise to microcolonies that were visible only after 3–4 days of growth at 30° (Figure 1A; data not shown). By this method, the synthetic growth defect involving the pob3-272 mutation was the least severe. Double mutant spores containing rtf1Δ and pob3-272 gave rise to small, visible colonies after 3–4 days of incubation at 30°. However, as described in a subsequent section, genetic analysis of the pob3-272 isolate was more complex, since we found that an additional mutation was present that affected the growth of our original strain.

Genetic analysis of mutations obtained in the synthetic lethal screen: To assist in our studies, strains harboring the synthetic lethal mutations were tested for several mutant phenotypes. As summarized in Table 2, the mutations cause a variety of phenotypes, many of which have been associated with defects in transcription. Spt and Bur phenotypes, inositol auxotrophy (Ino), and defects in galactose metabolism (Gal) are often correlated with mutations that affect the general transcription apparatus and/or chromatin factors (Winston 1992; Prelich and Winston 1993; Hampsey 1997). The Bur [Bypass upstream activation sequence (UAS) requirement] phenotype, a characteristic of strains mutant for histones or other transcriptional repressors, reflects the ability to bypass the requirement for a UAS within the SUC2 promoter (Prelich and Winston 1993). Salt and formamide sensitivity are also caused by mutations that affect transcription, including those that alter transcription elongation and chromatin structure (Oteroet al. 1999; Tsukiyamaet al. 1999). Cold sensitivity (Cs) is frequently associated with defects in protein complex assembly (Hampsey 1997). The Ino and Gal phenotypes caused by the srb5-77 mutation and the Cs phenotype caused by the ctk1-217 mutation are in agreement with phenotypes conferred by other mutations in these genes (Lee and Greenleaf 1991; P. J. Costa and K. M. Arndt, unpublished observations).

View this table:
TABLE 3

Genetic interactions between rtf1Δ and mutations in genes encoding RNA pol II holoenzyme components

Figure 1.

—The srb5-77 allele is distinct from the srb5Δ allele. (A) srb5-77 rtf1Δ strains, but not srb5Δ rtf1Δ strains, have a microcolony phenotype. Yeast strains FY23, KY404, L937, KA66, KA68, and KA67 were streaked to YPD media and grown for 4 days at 30° before photography. The sickness of the srb5Δ rtf1Δ strain is not clearly evident after 3 days of growth. (B) srb5-77 strains exhibit strong inositol auxotrophy. The yeast strains shown in A were transferred by replica plating to synthetic media lacking or containing inositol and grown for 2 days at 30°. As shown in the photograph, srb5-77 rtf1Δ strains do not grow on –inositol media and grow poorly on +inositol media. Strain orientations in B are the same as in A.

To determine if the synthetic lethality or extreme synthetic sickness between rtf1Δ and our srb5 and ctk1 mutations was allele specific, we examined the phenotypes of double mutant strains containing an rtf1Δ and either an srb5Δ or a ctk1Δ mutation. We found that the ctk1Δ rtf1Δ double mutant strains are inviable (data not shown), suggesting that our ctk1 allele, ctk1-217, is probably a null allele. In support of this view, ctk1-217 and ctk1Δ mutations confer the same mutant phenotypes (Lee and Greenleaf 1991; P. J. Costa and K. M. Arndt, unpublished observations). In contrast, the srb5Δ rtf1Δ double mutant strains are viable, but exhibit several synthetic phenotypes (Figure 1 and Table 3). The double mutant strains grow more slowly than either single mutant and exhibit an exacerbated Ino phenotype compared to srb5Δ strains. In addition, the srb5Δ mutation completely suppresses the Spt phenotype conferred by the rtf1Δ mutation. srb5Δ rtf1Δ double mutant strains are significantly healthier than srb5-77 rtf1Δ strains, which exhibit a microcolony phenotype (Figure 1A). This result suggests that the srb5-77 allele, although recessive for its interaction with rtf1Δ, is distinct from an srb5Δ allele. In accordance with this conclusion, the srb5-77 mutation, unlike the srb5Δ mutation, confers weak sensitivity to the compound 6-azauracil (6-AU; Table 2). As described in more detail below, sensitivity to 6-AU often indicates a defect in transcription elongation (Exinger and Lacroute 1992; Uptainet al. 1997).

Fcp1 and Pob3 are encoded by essential genes in yeast (Archambaultet al. 1997; Wittmeyer and Formosa 1997), suggesting that we have identified partial loss-of-function alleles of these genes. The human and yeast homologues of Fcp1 contain an essential phosphatase motif, two binding sites for the RAP74 subunit of the general transcription factor TFIIF, and a BRCA1 car-boxyl-terminal (BRCT) domain (Archambaultet al. 1997; Choet al. 1999; Koboret al. 1999). To identify the domain in Fcp1 that is altered by the fcp1-110 mutation, we cloned the mutant gene and determined its DNA sequence. The fcp1-110 mutation changes codon 615 in the open reading frame from a glutamine codon to a stop codon. The phosphatase and BRCT domains are amino-terminal to the Fcp1-110 stop codon. Previous studies showed that the two RAP74 binding sites in Fcp1 map to amino acids 457–666 and 667–732 (Archambaultet al. 1997). Therefore, the fcp1-110 mutation is predicted to eliminate one RAP74 interaction domain and truncate the remaining domain. Together, our findings suggest that the Fcp1-TFIIF interaction may be important for the elongation function of Fcp1 in vivo. To determine whether the fcp1-110 mutation compromises transcription elongation in vivo, we examined the phenotype of double mutant strains that contain the fcp1-110 mutation and a deletion of the nonessential gene PPR2. PPR2 encodes the well-characterized elongation factor TFIIS (Exinger and Lacroute 1992). Interestingly, fcp1-110 ppr2Δ double mutant strains exhibit a strong growth defect and enhanced inositol auxotrophy compared to strains harboring the fcp1-110 mutation alone (Figure 2; data not shown). In addition, the fcp1-110 mutation causes strains to be weakly sensitive to 6-AU (Table 2).

Figure 2.

—The fcp1-110 allele genetically interacts with a deletion of PPR2. Yeast strains KA79 and KY624 were mated, sporulated, and asci were dissected by tetrad analysis. A photograph of the dissection plate was taken after 4 days of growth at 30°.

Figure 3.

—The growth defect conferred by the pob3-272 mutation is suppressed by a mutation in CHD1. Yeast strains KA62, GHY713, KA63, and KY573 were streaked on YPD media and grown for 3 days at 30°.

Pob3 is similar to HMG1-like proteins found in a wide variety of organisms, including Arabidopsis thaliana, Schizosaccharomyces pombe, Drosophila melanogaster, Caenorhabditis elegans, mouse, and humans (Wittmeyer and Formosa 1997). However, unlike several other family members, Pob3 does not possess an HMG box, a DNA-binding motif found in the abundant chromatin-associated protein, HMG1 (Wittmeyer and Formosa 1997). We cloned and sequenced the pob3-272 mutation and found that it encodes a substitution of lysine for isoleucine at position 282. The analogous amino acid in the HMG1-like proteins of eleven other species is either an isoleucine or valine. The alteration of a highly conserved small, hydrophobic residue to an extended, charged amino acid is likely to cause a distortion in the Pob3-272 protein, possibly affecting its interaction with another protein.

A mutation in CHD1 suppresses the growth defect conferred by the pob3-272 mutation: During our genetic analysis, we found that the pob3-272 mutation causes extreme sickness in an otherwise wild-type background (Figure 3). The original pob3-272 mutant strain isolated in our synthetic lethal screen harbored one additional mutation that suppressed this growth defect. Double mutant strains containing the pob3-272 allele and the suppressor mutation exhibit nearly wild-type growth. We took advantage of these observations to clone the pob3-272 suppressor (see materials and methods) and determined, through linkage analysis, that the suppressor mutation was in the gene CHD1. Following its identification, we designated the suppressor mutation as chd1-52. We also found that a chd1Δ allele behaves similarly in suppressing the growth defect caused by the pob3-272 mutation (Figure 3; data not shown). CHD1 encodes a well-conserved protein with a domain structure that suggests a role in chromatin function (Woodageet al. 1997). Interestingly, the human homologue of yeast Chd1 has been shown to interact physically with SSRP1, the human homologue of yeast Pob3 (Kelleyet al. 1999). Since we identified a CHD1 allele as an outcome of our synthetic lethal screen, we also examined if rtf1Δ chd1 double mutant strains exhibit any genetic interaction. We observed no synthetic phenotypes for these strains (data not shown). However, as mentioned above, rtf1Δ pob3-272 chd1 triple mutant strains give rise to small, visible colonies only after 3–4 days of growth. Since pob3-272 chd1 double mutant strains exhibit nearly wild-type growth properties, the triple mutant combinations indicate a genetic interaction involving all three genes.

RTF1 exhibits genetic interactions with a small subset of genes encoding RNA pol II holoenzyme components: Because we identified an allele of SRB5 in our synthetic lethal screen, we asked whether RTF1 displays genetic interactions with mutations that affect other members of the RNA pol II holoenzyme. In addition to srb5Δ, we tested null mutations in the nonessential genes GAL11, SIN4, SRB2, and SRB10. We also tested a partial loss-of-function allele of the essential gene RGR1 (Sakaiet al. 1990) and two temperature-sensitive alleles of KIN28 (Valayet al. 1993). In contrast to our results with srb5-77, we did not observe synthetic lethality or severe synthetic sickness between the rtf1Δ mutation and mutations in these six other holoenzyme genes (Table 3). However, gal11Δ rtf1Δ double mutants do exhibit a slight growth defect, and the gal11Δ mutation completely suppresses the Spt phenotype associated with rtf1Δ. In addition, srb2Δ rtf1Δ double mutant strains exhibit an exacerbated Ino phenotype.

Like Ctk1, the holoenzyme-associated Kin28 and Srb10 proteins are cyclin-dependent kinases that phosphorylate the CTD of RNA pol II (Dahmus 1996). While Kin28 plays a positive role in transcription by facilitating the transition from initiation to elongation (Dahmus 1996; Hampsey 1998), Srb10 inhibits initiation by phosphorylating the CTD prior to PIC assembly (Hengartneret al. 1998). In striking contrast to the inviability of rtf1Δ ctk1Δ double mutant strains, rtf1Δ srb10Δ double mutant strains exhibit no synthetic phenotypes. For both kin28 alleles, the rtf1Δ kin28 double mutant strains showed synthetic Ino phenotypes but no significant defect in growth rate compared to the rtf1Δ and kin28 parents (Table 3). These findings further support the conclusion that the known CTD kinases have distinct roles in transcription and argue that the strong genetic interaction between rtf1Δ and srb5-77 is not a general property of mutations that affect holoenzyme components.

Figure 4.

rtf1Δ strains are sensitive to 6-azauracil and mycophenolic acid. Yeast strains GHY285, FY243, FY69, FY91, KY425, KY426, and KY459 were grown on YPD media and transferred by replica plating to SC-uracil media lacking or containing 50 μg/ml 6-azauracil or 20 μg/ml mycophenolic acid. Photographs were taken after 2 days of growth at 30°.

The rtf1Δ mutation confers sensitivity to 6-azauracil and mycophenolic acid: The results from our synthetic lethal screen indicate a role for Rtf1 in transcription elongation. To test this hypothesis further, we examined the sensitivity of rtf1Δ strains to 6-AU and mycophenolic acid (MPA). 6-AU and MPA decrease nucleotide levels in vivo and are thought to increase pausing and arrest by RNA pol II, thereby augmenting the need for factors that stimulate elongation (Exinger and Lacroute 1992; Uptainet al. 1997). Therefore, sensitivity to these compounds is often associated with mutations that inactivate transcription elongation factors (Exinger and Lacroute 1992; Uptainet al. 1997; Hartzoget al. 1998) or lower the elongation rate of RNA pol II (Powell and Reines 1996). Relative to isogenic wild-type strains, rtf1Δ strains are strongly sensitive to both 6-AU and MPA (Figure 4). The degree of sensitivity is comparable to that conferred by mutations in SPT4 and PPR2, which encode the elongation factors Spt4 and TFIIS, respectively (Exinger and Lacroute 1992; Hartzoget al. 1998).

RTF1 genetically interacts with known elongation factor genes: To further test the idea that Rtf1 functions during elongation, we investigated genetic interactions between RTF1 and several genes encoding transcription elongation factors. We observed several synthetic interactions with mutations in genes encoding Spt4, Spt5, Spt6, TFIIS, and Spt16. First, rtf1Δ spt4Δ double mutants are very sick, show strong temperature sensitivity (Ts) for growth, and are weakly Gly (inability to use glycerol as the sole carbon source; Figure 5; Table 4). In addition, these double mutant strains are Spt+, indicating a rare case of mutual suppression of Spt phenotypes. Likewise, mutations in the essential genes SPT5 and SPT6 (Swanson and Winston 1992), in combination with the rtf1Δ allele, cause a slight growth defect and a strong Ts phenotype (Stolinskiet al. 1997; Table 4).

Figure 5.

rtf1Δ spt4Δ double mutant strains are extremely sick and temperature sensitive for growth. (A) Yeast strains KY610, KY611, FY243, and KY405 were streaked on YPD media and grown for 3 days at 30° before photography. (B) The plate shown in A was allowed to grow an additional day at 30° and then replica plated to two YPD plates. These plates were incubated for 3 days at 37° or 30° before photography. The double mutant strains were constructed by crossing the indicated spt4Δ and rtf1Δ parents and performing tetrad analysis.

We also tested for a potential genetic interaction between rtf1Δ and a deletion of PPR2. For the rtf1Δ ppr2Δ double mutant, the only synthetic phenotype we observed was the ability of ppr2Δ to suppress the Spt phenotype associated with rtf1Δ. The absence of additional rtf1Δ ppr2Δ phenotypes may be due to functional redundancy with other elongation factors. Therefore, we examined if elimination of these other factors created a more critical situation for the cell. Indeed, we observed synthetic lethality for the rtf1Δ spt4Δ ppr2Δ triple mutant. Correspondingly, we found that rtf1Δ spt5-194 ppr2Δ mutants exhibit an exacerbated sickness compared to rtf1Δ spt5-194 strains (Table 4). Hartzog et al. (1998) have previously shown that spt4Δ ppr2Δ and spt5-194 ppr2Δ strains are viable, but are moderately Ts at 37°. Important for our results is our observation that spt4Δ ppr2Δ and spt5-194 ppr2Δ strains exhibit little or no growth defect at 30°. Last, we constructed the rtf1Δ spt4Δ spt16-197 triple mutant and found it to possess an extreme growth defect, growing much more slowly than the rtf1Δ spt4Δ double mutant (Table 4). In contrast, spt4Δ spt16-197 and rtf1Δ spt16-197 double mutants showed no synthetic phenotypes in our analysis (Table 4; data not shown). We also tested several of the double mutant combinations for 6-azauracil sensitivity. Strains carrying the rtf1Δ allele in combination with either spt4Δ, spt6-14, or ppr2Δ still exhibited sensitivity to 6-AU at the concentration tested (50 μg/ml). Finally, we examined the phenotype of an rtf1Δ rpb2-10 double mutant strain. The rpb2-10 mutation alters an amino acid in the second largest subunit of RNA pol II and encodes an enzyme with a decreased elongation rate in vitro (Powell and Reines 1996). rtf1Δ rpb2-10 double mutants exhibit a slight growth defect compared to either single mutant (Table 4), suggesting that the elongation rate of the Rpb2-10 enzyme may be further reduced in the absence of Rtf1. Collectively, our findings indicate that the requirement for Rtf1 is significantly increased by mutations that impair transcription elongation in yeast.

DISCUSSION

In this study, we provide evidence that Rtf1 has a role in transcription elongation in vivo. Through a genetic screen, we have shown that the function of Rtf1 is critical when the activities of four global regulators of RNA pol II transcription, Srb5, Ctk1, Fcp1, and Pob3, are eliminated or altered by mutation. Each of these proteins has been implicated in CTD phosphorylation and/or transcription elongation. Our genetic studies further indicate a functional redundancy between RTF1 and genes encoding several elongation factors. In addition, we have found that rtf1Δ mutations cause sensitivity to 6-AU and MPA, phenotypes often associated with defects in transcription elongation (Uptainet al. 1997).

Our results suggest several possible mechanisms for how Rtf1 may govern transcription elongation. In one model, Rtf1 may modulate the phosphorylation state of the CTD, perhaps in a gene-specific manner. In support of this idea, we uncovered mutations in CTK1 and SRB5 in our synthetic lethal screen. CTK1 encodes the cyclin-dependent kinase subunit of CTDK-I (Lee and Greenleaf 1991), a complex that specifically phosphorylates the CTD (Lee and Greenleaf 1989; Sterneret al. 1995) and promotes efficient elongation by RNA pol II in vitro (Lee and Greenleaf 1997). SRB5 encodes a component of the Srb/mediator complex that stimulates phosphorylation of the CTD in vitro (Hampsey 1998). Importantly, Srb5-deficient holoenzyme is significantly impaired in its ability to support CTD phosphorylation (Leeet al. 1999). If Rtf1 regulates CTD phosphorylation, a mutation in RTF1 together with a mutation in a gene encoding either a CTD kinase or a regulator of a CTD kinase could alter the extent or pattern of CTD phosphorylation in a way that prevents transcription of one or more essential genes.

View this table:
TABLE 4

Genetic interactions between rtf1Δ and mutations in genes involved in transcription elongation

Because the Srb/mediator complex plays a key role in transcriptional activation, an alternative explanation for the discovery of an srb5 mutation in our screen is that Rtf1 and Srb5 function in parallel pathways to facilitate holoenzyme recruitment. However, we do not favor this hypothesis for two reasons. First, we did not observe synthetic lethality or severe synthetic sickness between the rtf1Δ mutation and mutations in genes encoding five other Srb/mediator components, some of which have been directly implicated in activator-stimulated RNA pol II recruitment (Barberiset al. 1995; Hanet al. 1999; Leeet al. 1999). Second, previous work has shown that different subcomplexes of the Srb/mediator possess distinct functions and that Srb5 is required for a step in transcription that follows activator-mediated recruitment of the polymerase (Liet al. 1995; Leeet al. 1999).

Independent of any effect on CTD phosphorylation, Rtf1 may regulate transcription elongation in a more general fashion, such as by affecting chromatin structure or by altering the elongation properties of RNA pol II. Accordingly, we identified an fcp1 mutation and a pob3 mutation in our screen. In a recent study, Fcp1 has been shown to possess a positive elongation function independent of its CTD phosphatase activity (Choet al. 1999). This raises the possibility that Fcp1 remains associated with RNA pol II during elongation. We have shown that the fcp1-110 gene harbors a nonsense mutation that is predicted to remove one TFIIF interaction domain and truncate a second domain of this type. The mutation does not alter the phosphatase motif. Previous studies showed that the phosphatase activity of Fcp1 is stimulated by TFIIF in vitro (Chamberset al. 1995; Archambaultet al. 1997). Therefore, our results do not distinguish between an effect of the fcp1 mutation on CTD modification and a potentially more direct effect on the elongation properties of RNA pol II. Nevertheless, the isolation of an fcp1 allele in our synthetic lethal screen, together with the synthetic interaction between fcp1-110 and ppr2Δ, provides genetic support for a role of Fcp1 in transcription elongation and suggests that the interaction between Fcp1 and TFIIF is important for this function in vivo.

The human counterpart of the Pob3-Cdc68/Spt16 complex, FACT, has been shown to facilitate elongation specifically on nucleosomal templates in vitro (LeRoyet al. 1998; Orphanideset al. 1999). Since FACT interacts with histone H2A/H2B dimers, Orphanides et al. (1999) have proposed that FACT may function by promoting nucleosome disassembly upon transcription by RNA pol II. Importantly, the pob3-272 mutation isolated in our screen confers Spt and Bur phenotypes, both of which correlate well with a role for Pob3 in chromatin function. Whereas both phenotypes have been previously attributed to mutations in CDC68/SPT16 (Maloneet al. 1991; Prelich and Winston 1993), our results extend these phenotypes to a mutation in POB3. In addition, they provide support for the involvement of the Pob3-Cdc68/Spt16 complex in transcription elongation in vivo. Interestingly, this complex has been shown to interact with DNA polymerase α (Wittmeyer and Formosa 1997; Wittmeyeret al. 1999), suggesting that both DNA and RNA polymerases may employ this complex to move through chromatin.

The pob3-272 mutation alters a highly conserved amino acid. In addition to the Spt and Bur phenotypes, this alteration results in a severe growth defect. We found that a mutation in the CHD1 gene suppresses the growth defect, but not the Spt and Bur phenotypes (data not shown). Chd1 has a well-conserved tripartite structure, which includes chromo (chromatin organization modifier) domains, a Snf2-related helicase/ATPase domain, and a DNA-binding domain (Woodageet al. 1997). Chromo domains have been found in Polycomb and heterochromatin-binding protein 1, proteins that have important roles in chromatin compaction and transcriptional silencing (Paro 1993). Data from yeast suggest that Chd1 may be involved in the inhibition of transcription (Woodageet al. 1997). Our finding that a deletion of CHD1 can suppress the growth defect conferred by a mutation in POB3 also suggests that Chd1 has a negative role in transcription, possibly at the level of elongation. Kelley et al. (1999) have shown that the human homologues of Pob3 and Chd1 physically interact in vivo and in vitro. It will be of interest to determine if yeast Pob3 and Chd1 also physically associate, since such an interaction may have significance for both DNA replication and transcription.

In addition to the genes identified through the synthetic lethal screen, we uncovered a range of interactions between RTF1 and genes that encode Spt4, Spt5, Spt6, Spt16, and TFIIS. In most cases, the combination of the rtf1Δ mutation with mutations in these genes results in a more severe phenotype. Particularly noteworthy is the inviability of rtf1Δ spt4Δ ppr2Δ triple mutant strains, a suggestion that the complete loss of three elongation factors cannot be tolerated by the cell. We have found that RTF1 genetically interacts with genes encoding both components of the Spt4-Spt5 complex, both subunits of the Pob3-Cdc68/Spt16 complex, and TFIIS. We also detected an interaction between RTF1 and SPT6. Spt6 functionally interacts with elongation factors (Hartzoget al. 1998), physically interacts with histones, and assembles nucleosomes in vitro (Bortvin and Winston 1996). The synthetic and conditional phenotypes of the multiply mutated strains most likely reflect a functional redundancy among the RNA pol II elongation factors in yeast.

We initially reported that rtf1 mutations suppress the Spt phenotype of the TBP-altered specificity mutant TBP-L205F by altering transcription initiation (Stolinskiet al. 1997). Our current work indicates that RTF1 has a role in elongation and genetically interacts with SPT4, SPT5, SPT6, SPT16, and POB3, all genes implicated in the control of transcription by chromatin structure. Together, these findings suggest that Rtf1 may suppress the Spt phenotype of TBP-L205F by altering chromatin structure and controlling the accessibility of competing TATA boxes. Alternatively, Rtf1 may influence the productive elongation of transcripts that initiate from distinct start sites within a promoter. In support of these ideas, it should be noted that SPT4, SPT5, SPT6, and SPT16 were all initially identified by their ability to cause an Spt phenotype (Maloneet al. 1991; Winston 1992). This phenotype has been described as an effect on transcription initiation (Winston 1992). However, recent data have implicated all four genes in elongation (Hartzoget al. 1998; Orphanideset al. 1999). Further work is needed to determine whether Rtf1 directly regulates both the initiation and elongation stages of the transcription cycle.

In summary, by a combination of genetic approaches, we have obtained evidence that Rtf1 regulates transcription elongation in yeast. Further genetic studies coupled with a biochemical characterization of Rtf1 and its interacting partners should provide additional insights into its mode of action. Since we have recently recognized proteins with similar sequence in humans and C. elegans, our studies on the S. cerevisiae Rtf1 protein will also be applicable to an understanding of transcriptional regulation in other eukaryotes.

Acknowledgments

We are very grateful to the following individuals for the gifts of strains and plasmids: Grant Hartzog, Greg Prelich, Pamela Silver, and Fred Winston. We thank Michael Kobor for the suggestion of testing fcp1-110 strains for hydroxyurea sensitivity and Diana Cardona for analyzing the 6-AU sensitivity of srb5-77 strains. We are grateful to Grant Hartzog, Greg Prelich, and members of the Arndt laboratory, especially Margaret Shirra, for many helpful discussions and critical reading of the manuscript. This work was supported by National Institutes of Health grant GM52593 to K.M.A.

Footnotes

  • Communicating editor: M. Johnston

  • Received April 21, 2000.
  • Accepted June 14, 2000.

LITERATURE CITED

View Abstract