|Home | About | Journals | Submit | Contact Us | Français|
The RNA polymerase II (RNAP II) transcription cycle is accompanied by changes in the phosphorylation status of the C-terminal domain (CTD), a reiterated heptapeptide sequence (Y1S2P3T4S5P6S7) present at the C terminus of the largest RNAP II subunit. One of the enzymes involved in this process is Ssu72, a CTD phosphatase with specificity for serine-5-P. Here we report that the ssu72-2-encoded Ssu72-R129A protein is catalytically impaired in vitro and that the ssu72-2 mutant accumulates the serine-5-P form of RNAP II in vivo. An in vitro transcription system derived from the ssu72-2 mutant exhibits impaired elongation efficiency. Mutations in RPB1 and RPB2, the genes encoding the two largest subunits of RNAP II, were identified as suppressors of ssu72-2. The rpb1-1001 suppressor encodes an R1281A replacement, whereas rpb2-1001 encodes an R983G replacement. This information led us to identify the previously defined rpb2-4 and rpb2-10 alleles, which encode catalytically slow forms of RNAP II, as additional suppressors of ssu72-2. Furthermore, deletion of SPT4, which encodes a subunit of the Spt4-Spt5 early elongation complex, also suppresses ssu72-2, whereas the spt5-242 allele is suppressed by rpb2-1001. These results define Ssu72 as a transcription elongation factor. We propose a model in which Ssu72 catalyzes serine-5-P dephosphorylation subsequent to addition of the 7-methylguanosine cap on pre-mRNA in a manner that facilitates the RNAP II transition into the elongation stage of the transcription cycle.
Transcription by RNA polymerase II (RNAP II) occurs in distinct stages that include assembly of the preinitiation complex, promoter melting, initiation, promoter clearance, elongation, and termination (19). Progression through the transcription cycle is accompanied by changes in the phosphorylation status of the C-terminal domain (CTD), a reiterated heptapeptide sequence (YSPTSPS) present at the C terminus of Rpb1, the largest RNAP II subunit (27). RNAP II is recruited to the promoter in an unphosphorylated form (RNAP IIA) that is extensively phosphorylated (RNAP IIO) during transcription. Recycling of RNAP IIO following termination requires dephosphorylation to the IIA form. However, CTD phosphorylation/dephosphorylation is not simply a binary switch that correlates with initiation/termination. Rather, serine-2 and serine-5 of the CTD are phosphorylated and dephosphorylated at different stages of the transcription cycle. Moreover, the phosphorylation status of the CTD plays important and complex roles in RNAP II progression through the transcription cycle and in coordinating cotranscriptional RNA processing events (2, 5, 25, 41, 45, 53, 65).
Serine-5 and serine-2 of the CTD are phosphorylated by cyclin-dependent kinases. In yeast, serine-5 is phosphorylated by the Kin28 subunit of TFIIH prior to promoter clearance (36, 55), whereas serine-2 is phosphorylated by the Ctk1 subunit of the CTDK-I complex during elongation (7, 48). Two additional cyclin-dependent kinases, Srb10 and Bur1, have been reported to catalyze CTD phosphorylation, although the physiological relevance of these activities is less clear (52). Indeed, Bur1 catalyzes Rad6 phosphorylation as a requisite step in Rad6-mediated ubiquitylation of histone H2B during the transcription cycle (69). CTD phosphatases have also been defined. Some are specific for either serine-2-P or serine-5-P, whereas others fail to discriminate between the two substrates (discussed in reference 22). In yeast, the Ssu72 phosphatase is specific for serine-5-P dephosphorylation (22, 32), whereas the Fcp1 phosphatases from budding and fission yeast exhibit preference for serine-2-P (7, 23). Ssu72 and Fcp1 are phylogenetically conserved proteins, and human counterparts of both enzymes have been characterized (38, 67). Other CTD phosphatases, including the human small CTD phosphatases (71) and plant CTD phosphatase-like proteins (22, 28-30), have been identified. The small CTD and CTD phosphatase-like protein phosphatases exhibit specificity for serine-5-P, although their roles in transcription remain to be elucidated.
It is not clear where in the transcription cycle the CTD phosphatases act. Chromatin immunoprecipitation (ChIP) experiments revealed that serine-2 phosphorylation is accompanied by serine-5-P dephosphorylation during elongation, suggesting that serine-5-P dephosphorylation is a prerequisite for Ctk1-mediated serine-2 phosphorylation (7). Ctk1 is likely to require only partial serine-5-P dephosphorylation, however, as ChIP experiments show retention of serine-5-P at the terminator (46). Whether Ssu72 catalyzes serine-5-P dephosphorylation at more than one stage of the transcription cycle has not been determined, nor is it known if another serine-5-P phosphatase might also be involved. Human and yeast Fcp1 have been more thoroughly characterized, although their specific roles in the transcription cycle are also unresolved. Yeast Fcp1 localizes to the promoter and coding regions of active genes (7) and genetically interacts with the Spt4-Spt5 and Paf1 elongation complexes (10, 34, 38). Human Fcp1 stimulates elongation and facilitates recycling of RNAP II, although its effect on elongation in vitro was reported to be independent of catalytic activity (9, 38).
The role of the Ssu72 CTD phosphatase in the transcription cycle is especially enigmatic. Ssu72 was first identified based on genetic interaction with the general transcription factor TFIIB, an interaction that affects the accuracy of start site selection (68). Ssu72 physically associates with TFIIB (12, 70), the Rpb2 subunit of RNAP II (12), the Taf2 subunit of TFIID (58), and the Kin28 subunit of TFIIH (14). These interactions implicate Ssu72 in initiation. Yet Ssu72 is an integral component of the 3′ end cleavage-polyadenylation factor (CPF) complex (12, 16, 24, 44). Consistent with its presence in the CPF complex, ssu72 mutations adversely affect 3′ end processing (24) and termination (12, 14, 66). By ChIP, Ssu72 localizes predominantly to the 3′ ends of genes (44) but also occupies the promoter region (1). One possibility that might reconcile the interaction of Ssu72 as a component of the CPF complex with the transcription initiation machinery is suggested by the recent discovery of gene loops in yeast (1, 46). Juxtaposition of the promoter and terminator regions of the SEN1 and BUD3 genes results in formation of transient DNA loops in a manner dependent upon Ssu72 and its partner in the CPF complex, Pta1 (1). Conceivably, gene loops might facilitate recycling of RNAP II from the terminator to the promoter, with Ssu72 catalyzing conversion of RNAP IIO to the IIA form. There is no evidence, however, that gene loops actually stimulate transcription.
As part of our efforts to determine the role of Ssu72 in the transcription cycle, we are working with the temperature-sensitive (Tsm−) ssu72-2 mutant, which encodes the Ssu72-R129A form of the protein (47). Here we show that Ssu72-R129A is catalytically impaired, resulting in accumulation of the serine-5-P form of RNAP II in vivo. Suppressors of the ssu72-2 Tsm− phenotype overcome the CTD phosphatase deficiency by slowing the rate of RNAP II transcription. Whereas earlier studies defined a role for Ssu72 in the elongation-termination transition (12, 14, 66), our genetic and biochemical results suggest that Ssu72 also acts earlier in the transcription cycle. We present a model in which Ssu72 affects progression through the initiation-elongation and elongation-termination transitions by catalyzing incremental dephosphorylation of serine-5-P, in effect facilitating passage of RNAP II through checkpoints that monitor CTD phosphorylation status.
The Saccharomyces cerevisiae strains used in this study are listed in Table Table1.1. Strains LRB535 (wild type [WT]), YZS84 (MATa ssu72-2), and YDP87 (MATα ssu72-2) were described previously (47). Strains YMH930 (supA−), YMH931 (rpb1-1001), YMH932 (supC−), YMH933 (supD−), and YMH934 (rpb2-1001) are spontaneous Tsm+ revertants of YZS84. The RPB2 plasmid shuffle strain YMH922 (MATa his3Δ200 leu2-3,112 ura3-52 ssu72-2 rpb2::kanMX [pN1002:RPB2-URA3]) was created by introducing plasmid pN1002 (RPB2-CEN-URA3) into strain YZS84, followed by one-step disruption of the chromosomal RPB2 locus using the kanMX marker (37). Strains YMH935 (RPB2), YMH936 (rpb2-4), and YMH937 (rpb2-10) were derived from YMH922 by transformation with plasmids pN1867, pN1868, and pN1870, respectively, followed by counterselection of pN1002 on 5-fluoroorotic acid medium. Plasmids pN1867 (RPB2), pN1868 (rpb2-4), and pN1870 (rpb2-7) are low-copy-number CEN LEU2 plasmids that harbor the indicated RPB2 alleles. Strains YMH938 (dst1::his5+), YMH939 (ssu72-2 dst1::his5+), YMH940 (spt4::his5+), and YMH941 (ssu72-2 spt4::his5+) were created from strain LRB535 or YZS84 by one-step disruption of the DST1 or SPT4 chromosomal genes using the S. pombe his5+ marker, which complements his3Δ200 (37). Strain YMH942 (spt5-242 rpb2-1001) was created from GHY339 (spt5-242) (obtained from G. Hartzog) by plasmid shuffle using pN1002.
All growth media were prepared according to standard recipes (63). −Ino medium is synthetic complete medium lacking inositol; +Ino control medium is synthetic complete containing 55 μM inositol. Standard yeast genetic methods were used for making crosses, selecting diploids, inducing sporulation, and dissecting tetrads (64). Yeast transformations were done by the lithium acetate method (17). The following symbols are used to denote phenotypes: Tsm− and Csm−, heat and cold sensitivity, respectively, defined by impaired growth on yeast extract-peptone-dextrose (YPD) medium at 37°C and 16°C; Ino−, impaired growth on −Ino medium at 30°C relative to growth on +Ino medium. Plasmids carrying the URA3 marker were counterselected on synthetic medium containing 5-fluoroorotic acid (4). 6-Azauracil (6-AU) was added to YPD medium at the indicated concentrations.
Recombinant glutathione S-transferase (GST)-Ssu72 and GST-Ssu72-R129A were expressed in Escherichia coli strain BL21(DE3) transformed with pGEX-2TK expression plasmids pN1799 and pM1894, respectively, and purified as described previously (24). Phosphatase activity was measured by production of p-nitrophenol (spectroscopic absorbance at 410 nm) from p-nitrophenylphosphate (pNPP) as described by Ganem et al. (14).
The rpb1 and rpb2 suppressor alleles were recovered by gap repair (57). Plasmid pM243 (RPB1-URA3) was digested to completion with EcoNI and SnaBI, thereby deleting most of the RPB1 open reading frame. Vector DNA flanked by RPB1 sequences was purified by agarose gel electrophoresis and introduced into strain YMH931 (rpb1-1001 ura3) by transformation. Ura+ colonies were selected and screened for retention of the Tsm+ and Ino− suppressor phenotypes. Plasmid DNA was recovered, amplified in Escherichia coli, and analyzed by restriction digestion to confirm the presence of the rpb1 open reading frame (ORF). The resulting plasmid failed to complement the Tsm+ and Ino− phenotypes when introduced into strain YMH931, thereby confirming recovery of rpb1-1001. The DNA sequence of the entire rpb1-1001 ORF was determined using an ABI Prism Automated DNA sequencer and a set of RPB1-specific primers (3). The suppressor mutation was identified through BLAST alignments of the rpb1-1001 and RPB1 sequences. The rpb2-1001 allele was recovered using a similar strategy, as described previously (47).
Strains LRB535 (WT) and YZS84 (ssu72-2) were grown to an A600 of 3.0. Cells were collected by centrifugation, washed, and resuspended in disruption buffer [200 mM Tris-HCl, pH 7.9, 390 mM (NH4)2SO4, 10 mM MgSO4, 20% glycerol, and 1 mM EDTA supplemented with protease inhibitors]. Cells were frozen drop by drop into liquid nitrogen and lysed in a Waring blender, yielding a fine powder indicative of cell lysis. The powder was thawed at 4°C and centrifuged at 10,000 rpm for 10 min. The supernatant was recovered and centrifuged at 45,000 rpm in a 50.2 Ti Beckman rotor for 1 h. Protein was precipitated by 75% saturation with (NH4)2SO4, and the pH was adjusted using 10 μl of 1 M KOH per gram of (NH4)2SO4. Protein was recovered by centrifugation at 15,000 rpm for 30 min and resuspended in buffer B (20 mM HEPES, pH 7.5, 20% glycerol, 10 mM EGTA, 10 mM MgSO4 supplemented with protease inhibitors). The extract was then dialyzed against buffer B supplemented with 5 mM dithiothreitol and 1 mM phenylmethylsulfonyl fluoride until a conductivity of ≤20 μS was reached. Transcription reactions were carried out using 100 μg of whole-cell extract and 300 ng of template DNA, as described previously (6, 32). The double G-less cassette DNA template, pSLCYC-L, was derived from pSLG402 (33) by replacement of the AdML promoter with the CYC1 promoter (32).
Strains LRB535 (WT) and YZS84 (ssu72-2) were grown in YPD medium to an A600 of 0.2 to 0.5, collected by centrifugation and resuspended YPD medium prewarmed to 37°C. Following incubation for the indicated periods of times, cells were harvested by centrifugation. Proteins were extracted and prepared for Western blotting as described previously (49). Antibodies used to detect the RNAP IIA (8WG16) and the serine-5-P (H14) forms of RNAP II were obtained from Covance (USA). Polyclonal rabbit anti-Ssu72 antibody was generated using purified, recombinant protein (32). Rpa1 antibody was a gift from Steve Brill (Rutgers University).
Ssu72 is a protein phosphatase with specificity for serine-5-P of the RNAP II CTD both in vivo and in vitro (22, 32). The ssu72-2 allele encodes an arginine-129 to alanine (R129A) replacement and confers a marked temperature-sensitive growth defect (47). To determine whether the R129A replacement affects catalytic activity, we assayed purified GST-Ssu72 and GST-Ssu72-R129A proteins using pNPP as the substrate. Results showed that Ssu72-R129A has less than 40% of the phosphatase activity of normal Ssu72 (Fig. (Fig.1A).1A). We next sought to determine whether Ssu72-R129A affects CTD phosphatase in vivo. Western blot analysis showed that the serine-5-P form of RNAP II accumulates in the ssu72-2 mutant following a 60-min shift to the nonpermissive temperature of 37°C (Fig. (Fig.1B,1B, lanes 3 and 4), whereas no effect of the temperature shift was observed in the isogenic wild-type strain (lanes 1 and 2). Accumulation of the serine-5-P form of RNAP II is not due to protein instability, as no effect of the temperature shift on the steady-state level of the Ssu72-R129A protein was observed (Fig. (Fig.1B,1B, lanes 1 to 4). Thus, ssu72-2 encodes a catalytically impaired form of the RNAP II CTD phosphatase that results in accumulation of the serine-5-P form of RNAP II at the nonpermissive temperature.
Earlier studies indicated that Ssu72 affects the elongation-termination transition in vivo (12). To determine whether Ssu72 affects elongation in vitro, transcription reactions were performed using a double G-less cassette DNA template (pSLCYC-L) and whole-cell extracts prepared from isogenic wild-type and ssu72-2 strains. This system yields RNase T1-resistant transcripts as a measure of promoter-proximal (G-less I) and promoter-distal (G-less II) transcription (Fig. (Fig.2A).2A). Initiation occurs at two sites, +1 and +20, such that G-less I transcripts are represented by two products, 110 and 130 nucleotides (nt) in length, whereas G-less II transcripts are represented by 377-nt products regardless of whether initiation occurs at +1 or +20. Accordingly, this system allows for determination of the efficiency of transcription elongation by kinetic analysis of the accumulation of the G-less II transcript (377 nt) as a percentage of G-less I transcripts (110 plus 130 nt). Results showed that the ssu72-2 extract transcribed the G-less II cassette but with clearly reduced efficiency relative to the wild-type extract (Fig. 2B and C). Thus, Ssu72 plays a positive role in transcription elongation in a manner dependent upon its catalytic activity.
As a genetic approach to define the role of the Ssu72 CTD phosphatase in the transcription cycle, we sought suppressors of the Tsm− growth phenotype of the ssu72-2 mutant. Five independent, spontaneous Tsm+ revertants were isolated that were also partially Ino−, a phenotype often associated with defects in the RNAP II transcriptional machinery. The phenotypes of these strains are depicted in Fig. Fig.3.3. Genetic analysis revealed that the Tsm+ and Ino− phenotypes of each revertant are the result of recessive mutations unlinked to ssu72-2. Furthermore, the Tsm+ and Ino− phenotypes cosegregated through meiosis, indicating that Ino− is a pleiotropic phenotype associated with the Tsm+ suppressors. We designated the suppressor genes supA− through supE−.
Our initial approach to identify the suppressors was to ask if the supA− through supE− mutations are complemented by plasmid-borne wild-type RPB1 or RPB2, the genes encoding the two largest subunits of RNAP II. We found that supB− is fully complemented by RPB1 carried on either a CEN or 2μm plasmid, restoring the Tsm− phenotype (data not shown). Similarly, the supE− suppressor is fully complemented by RPB2 carried on a CEN plasmid and partially complemented by RPB2 on a 2μm plasmid. Subsequent allelism tests confirmed that supB− is allelic to RPB1 and that supE− is allelic to RPB2 (data not shown). Accordingly, we renamed the supB− suppressor rpb1-1001 and the supE− suppressor rpb2-1001. The other three suppressors were not complemented by either RPB1 or RPB2 and have not yet been identified.
The rpb1-1001 allele was cloned by gap repair, and the entire coding region was sequenced using a collection of primers that span the RPB1 ORF (3). A single-base-pair substitution was identified that encodes replacement of arginine-1281 by leucine (R1281L) (Fig. (Fig.4A).4A). R1281 is phylogenetically conserved (R or K) and forms the N-terminal residue of the β32 strand that comprises part of the DNA-binding “cleft” of Rpb1 (Fig. (Fig.4C)4C) (11). The rpb2-1001 allele was also cloned by gap repair, and the entire coding region was sequenced using primers that span its ORF (47). A single-base-pair substitution was identified that encodes replacement of arginine-983 by glycine (R983G) (Fig. (Fig.4B).4B). R983 is phylogenetically invariant and lies within the RNAP II DNA-RNA “hybrid binding” domain of Rpb2 (Fig. (Fig.4D)4D) (11). In an earlier study from our laboratory, we reported that the rpb2-100 suppressor of ssu72-2 encodes an R512C replacement within the “fork loop 2” domain of RNAP II (47). Thus, we have identified three structurally disparate amino acid replacements in either Rpb1 (R1281L) or Rpb2 (R512C, R983G) that suppress the Tsm− phenotype of the ssu72-2 mutation. These results define a functional relationship between the Ssu72 CTD phosphatase and regions of RNAP II distinct from the Rpb1 CTD domain.
How do the Rpb1 and Rpb2 amino acid replacements compensate for the ssu72-2 Tsm− phenotype and what does this tell us about the normal function of Ssu72? Rpb2-R983G is especially informative, as this replacement lies adjacent to two previously characterized Rpb2 replacements, also within the hybrid binding domain (Fig. (Fig.4B).4B). These are A1016T and P1018S encoded by rpb2-4 and rpb2-10, respectively. These two replacements slow the rate of RNAP II elongation in vitro (51) and in vivo (39). To determine whether suppression of ssu72-2 is due to slowing the rate of RNAP II elongation, we sought to determine whether the rpb2-4 and rpb2-10 alleles would suppress ssu72-2. Double ssu72-2 rpb2 mutants were generated by plasmid shuffle as described in Materials and Methods. The resulting strains were scored for growth on rich medium at 30°C and 37°C (Fig. (Fig.5A).5A). We found that the rpb2-4 and rpb2-10 alleles are very effective suppressors of ssu72-2 (cf., rows 2, 4, 6), comparable to the rpb2-1001 suppressor (row 7). In contrast, the rpb2-7 allele, which does not affect elongation, failed to suppress ssu72-2 (cf. rows 2 and 5), indicating that suppression of ssu72-2 is rpb2 allele specific. These results demonstrate that the growth defect associated with Ssu72-R129A is overcome by slowing the rate of transcription elongation.
The compound 6-AU adversely affects transcription elongation by reducing the intracellular pools of UTP and GTP (13, 39, 62). If rpb1-1001 and rpb2-1001 suppress ssu72-2 by reducing the rate of elongation, then 6-AU should mimic the effects of these suppressors. Indeed, 6-AU partially suppressed the ssu72-2 Tsm− phenotype at 50 μg/ml and 100 μg/ml (Fig. (Fig.5B),5B), a result consistent with earlier results (12). Thus, inhibiting the rate of transcription, either genetically by altered forms of RNAP II or phenotypically by depletion of NTP pools, effectively compensates for diminished Ssu72 CTD phosphatase activity. We suggest that reducing the rate of transcription elongation by RNAP II provides adequate time for a catalytically impaired Ssu72 CTD phosphatase to dephosphorylate serine-5-P during the transcription cycle. This idea is consistent with (i) partial retention of phosphatase activity by Ssu72-R129A (Fig. (Fig.1A)1A) and (ii) enhanced serine-5-P dephosphosphorylation associated with the rpb1-1001 suppressor (Fig. (Fig.1B,1B, cf. lanes 4 and 6).
Suppression of the growth defect associated with ssu72-2 by slowing the rate of RNAP II elongation is reminiscent of suppression of the Csm− growth defect of the spt5-242 mutation by the rpb2-10 allele and by 6-AU (20). The Spt4-Spt5 complex acts as a positive transcription elongation factor in yeast (20, 56) and has been proposed to coordinate events in the transcription cycle by acting as part of a promoter-proximal elongation checkpoint (42). Spt4, presumably as part of the Spt4-Spt5 complex, is required for recruitment of the Paf1 elongation complex following serine-5 phosphorylation (54). Interestingly, an spt4Δ deletion and the rpb2-10 allele were found to adversely affect RNAP II processivity (39). We therefore sought to determine whether an spt4Δ deletion would suppress the Tsm− phenotype of the ssu72-2 mutant. Results are shown in Fig. Fig.6A.6A. The spt4Δ deletion clearly suppressed the ssu72-2 Tsm− phenotype (cf. rows 4 and 6), comparable to suppression of ssu72-2 by the rpb1 and rpb2 mutations (Fig. (Fig.33 and and5A),5A), with no effect on growth of the wild-type strain at 37°C (cf. rows 1 and 3). In contrast, deletion of the DST1 gene, which encodes the transcription elongation factor TFIIS, had little or no effect (Fig. (Fig.6A,6A, cf. lanes 4 and 5). Furthermore, the Csm− phenotype of the spt5-242 mutant is suppressed by rpb2-1001 (Fig. (Fig.6B).6B). As such, rpb2-1001 suppresses both the ssu72-2 Tsm− growth phenotype as well as the spt5-242 Csm− phenotype. These results define a functional relationship between Ssu72 and the Spt4-Spt5 complex and underscore our conclusion that Ssu72 affects RNAP II elongation. We propose that, in addition to its role in the elongation-termination transition, the Ssu72 CTD phosphatase also affects the initiation-elongation transition (see Discussion).
This report addresses the role of the Ssu72 CTD phosphatase in the RNAP II transcription cycle. Our results demonstrate that the cell growth defect associated with accumulation of the serine-5-P form of the RNAP II CTD can be overcome by slowing the rate of transcription. This conclusion is supported by four lines of evidence. First, the rpb2-4 and rpb2-10 alleles, which encode slow forms of RNAP II (39, 51), suppress the Tsm− growth defect of the phosphatase-defective ssu72-2 mutant. Conversely, the rpb2-7 allele, which confers a growth defect similar to that of rpb2-4 and rpb2-10 (59), but does not affect elongation rate (51), does not suppress ssu72-2. One explanation for these results is that a slow RNAP II might allow more time to complete a checkpoint requiring serine-5-P dephosphorylation. Second, the Rpb2-R512C form of RNAP II, which was identified previously in our laboratory as a suppressor (rpb2-100) of ssu72-2 (47), has been found to be a very slow RNAP II in vitro (Z. F. Burton, personal communication). Third, 6-AU, which slows the rate of RNAP II transcription in vivo by depleting NTP substrates, suppresses the ssu72-2 Tsm− growth defect, a result consistent with an earlier report (12). Finally, an spt4Δ deletion that disrupts the Spt4-Spt5 positive regulator of transcription elongation (56) and reduces RNAP II processivity (39) rescues growth of ssu72-2 at the nonpermissive temperature, and the growth defect associated with the spt5-242 allele is alleviated by the rpb1-1001 suppressor of ssu72-2.
How does slowing the rate of transcription suppress ssu72-2 and what does this tell us about the normal role of Ssu72 in the transcription cycle? We propose that Ssu72 acts as a switch at multiple stages of the transcription cycle, including the initiation-elongation and elongation-termination transitions, by catalyzing incremental dephosphorylation of serine-5-P. Our model is depicted in Fig. Fig.7.7. Hyperphosphorylation of serine-5 is catalyzed by the Kin28 subunit of TFIIH, coincident with transcription initiation and as a requirement for capping enzyme recruitment (8, 31, 40, 55, 60, 72). In a manner dependent upon the Spt4-Spt5 complex (35), the capping machinery then modifies the 5′ end of the nascent transcript. We propose that Ssu72 acts subsequent to capping, catalyzing partial serine-5-P dephosphorylation, which in turn facilitates the initiation-elongation transition. Perhaps serine-5-P dephosphorylation displaces the capping machinery, which has been reported to repress transcription (43). Indeed, persistent association of the capping machinery with RNAP II could account, at least in part, for the impaired transcription associated with the ssu72-2 mutant (Fig. (Fig.2).2). This idea is also consistent with the observation that inactivation of the cap methyltransferase, Abd1, is associated with serine-5-P hyperphosphorylation (61). Additional serine-5-P dephosphorylation might occur during early elongation as a prerequisite to Ctk1-catalyzed serine-2 phosphorylation (31). Ssu72 also affects the elongation-termination transition (12, 14, 24, 66) and, along with the Fcp1 serine-2-P phosphatase, restores the initiation-competent, hypophosphorylated form of RNAP II (reviewed in reference 65). Each of these CTD phosphorylation and dephosphorylation events could facilitate the exchange of transcription and RNA processing factors that has been observed at the initiation-elongation and elongation-termination transitions (26, 50). We suggest that in the absence of normal Ssu72 catalytic activity slowing the rate of transcription would allow more time to complete transitions that are dependent upon serine-5-P dephosphorylation.
Several independent studies support a role for Ssu72 in the initiation-elongation transition. First, Ssu72 was identified based on functional interaction with the general transcription factor TFIIB (68). The sua7-1 allele, which encodes the TFIIB E62K replacement, shifts start site selection downstream of normal, and this effect is enhanced by the ssu72-1 allele. Furthermore, Ssu72 physically interacts with TFIIB (12, 70), TFIID (58), TFIIH (14), and RNAP II (12, 47). Second, the Tsm− growth defect of the ssu72-2 mutant is suppressed by an spt4Δ deletion (Fig. (Fig.6A).6A). This result defines a functional interaction between Ssu72 and the Spt4-Spt5 complex, a relationship that is underscored by suppression of the cold-sensitive growth defect of the spt5-242 mutant by the rpb2-1001 suppressor of ssu72-2 (Fig. (Fig.6B).6B). The Spt4-Spt5 complex has been proposed to act as part of a promoter-proximal elongation checkpoint to coordinate events in the transcription cycle (42). The nucleosome encompassing the initiator region of MET16 fails to undergo remodeling in an spt4Δ mutant and RNAP II accumulates at the promoter (42). This effect might be comparable to a slow form of RNAP II: failure to clear the promoter efficiently could allow more time for the catalytically impaired Ssu72 phosphatase to convert RNAP II to an elongation-competent form. Alternatively, spt4Δ might bypass the normal checkpoints that precede Ctk1-mediated serine-2 phosphorylation, a possibility suggested by elevated levels of serine-2-P at the MET16 promoter in an spt4Δ mutant (42). Finally, genetic interactions of ssu72 with other components of the elongation machinery were interpreted to reflect a role for Ssu72 at the capping checkpoint (15).
The decline in serine-5-P levels that coincides with the increase in serine-2 phosphorylation in the coding region of actively transcribed genes indicates that a serine-5-P phosphatase functions during the elongation stage of transcription (31). Our finding that the efficiency of elongation in vitro is adversely affected by the ssu72-2 mutation (Fig. (Fig.2)2) suggests that Ssu72 might be the phosphatase responsible for this transition. This possibility is also supported by genetic interactions between Ssu72 and the Ctk1 kinase and the Fcp1 phosphatase (14). ChIP experiments, however, indicate that Ssu72 occupies the terminator region and, to a lesser extent, the promoter region but not the coding region of genes (1, 44). Perhaps the architecture of the RNAP II elongation complex is such that Ssu72 cannot be detected by ChIP. Alternatively, a phosphatase other than Ssu72 might facilitate elongation, although Fcp1 is the only other CTD phosphatase that has been identified in yeast and Fcp1 appears to be specific for serine-2-P (7, 21).
Whereas rpb2-10 suppressed the growth defect of the ssu72-2 mutant (Fig. (Fig.5),5), rpb2-10 is lethal in combination with fcp1-As42, which encodes an altered form of Fcp1 (38). Other alleles of fcp1, however, suppress the rpb2-10 slow-growth defect, and it has not been reported whether any of these alleles affects Fcp1 catalytic activity or CTD phoshophorylation (38). Similar to the results reported here, spt4Δ genetically interacts with fcp1-As42: the slow-growth phenotype of the fcp1-As42 mutant is suppressed by spt4Δ, whereas the 6-AU sensitivity of the spt4Δ mutant is suppressed by fcp1-As42. These results underscore the functional relationship between the Ssu72 and Fcp1 CTD phosphatases and point to a common role for the Spt4-Spt5 complex in coordinating their activities during the transcription cycle.
Ssu72 also appears to affect the elongation-termination stage of the transcription cycle. Ssu72 is an integral component of the CPF 3′-end processing complex, where it physically interacts with the Pta1 subunit of CPF (12, 24). Ssu72 is required for 3′-end processing, although this function is independent of catalytic activity (24). Other studies have shown that Ssu72 is required for transcription termination of snoRNAs and specific mRNAs (12, 14, 44, 66). ChIP experiments revealed that serine-5 of the CTD remains at least partially phosphorylated at the terminator region of the FMP27 gene (46). Based on this observation and the presence of Ssu72 as an integral component of the CPF complex, we propose that Ssu72 affects the elongation-termination transition, in part, by catalyzing serine-5-P dephosphorylation (Fig. (Fig.7).7). In light of the recent evidence for gene looping (1, 46) and its dependence upon the catalytic activity of Ssu72 (1), it is conceivable that Ssu72 facilitates transcription reinitiation by catalyzing removal of the remaining serine-5-P residues to restore the initiation-competent, hypophosphorylated form of RNAP II.
Our discovery that the cell growth defect associated with the ssu72-2 allele can be suppressed by mutations that slow the rate of RNAP II elongation can be exploited to generate novel, elongation-defective forms of RNAP II. In this regard, the Rpb2-R512C derivative has proved to be particularly informative. Position R512 lies within the “fork loop 2” domain of RNAP II (11, 47) and has been proposed to interact with the DNA template strand at the i + 2 position just downstream of the active center (i + 1) (18). Gong and colleagues proposed that NTP substrates bind to template DNA at i + 2 and i + 3 sites prior to translocation into the active site, a mechanism that would facilitate both the efficiency and fidelity of transcription (18). The kinetic characterization of the R512C derivative as a slow form of RNAP II is consistent with this model. Characterization of additional suppressors of ssu72-2 offers an opportunity to further test the NTP-driven translocation hypothesis: novel amino acid replacements should turn up at positions predicted to interact with NTP substrates at the i + 2 and i + 3 positions. Other amino acid replacements, like Rbp1-R1281L (rpb1-1001), Rpb2-R983G (rpb2-1001), Rpb2-A1016T (rpb2-4), and Rpb2-P1018S (rpb2-10), which in the three-dimensional structure of RNAP II lie along the proposed trajectory of the DNA, could refine the path of template DNA as it approaches the RNAP II active center.
In summary, our results define the Ssu72 CTD phosphatase as a positive transcription elongation factor. Ssu72 appears to affect multiple stages of the transcription cycle, including the initiation-elongation and elongation-termination transitions, supporting the idea that exchange of transcription and processing factors is mediated by differential CTD phosphorylation. We are now interested in knowing how ssu72 mutants affect recruitment and exchange of transcription and RNA processing factors and how the activity of Ssu72 is regulated.
We are grateful to Claire Moore (Tufts University Medical School) for critical comments on the manuscript, to Zach Burton (Michigan State University) for communicating results prior to publication, and to the members of our laboratory for many insightful discussions during the course of this work. We also thank Krishnamurthy Shankarling (RWJMS) for GST-Ssu72 expression plasmids, Grant Hartzog (UC—Santa Cruz) for strains, and Steve Brill (Rutgers University) for Rpa1 antiserum.
This work was supported by the NIH Bridge to Doctoral Degree Program (GM58389), by the NIH IMSD Award-UMDNJ/Rutgers University Pipeline Program (GM55145), by NIH Graduate Training in Cellular and Molecular Biology (GM008360), and by NIH RO1 grants GM39484 and GM68887.
Published ahead of print on 13 November 2006.