CDK8

Molecular Cell, Vol. 3, 673–678, May, 1999, Copyright 1999 by Cell Press

GAL4 Is Regulated by the RNA Polymerase II Holoenzyme–Associated Cyclin-Dependent Protein Kinase SRB10/CDK8 Martin Hirst,* Michael S. Kobor,† Neena Kuriakose,* Jack Greenblatt,† and Ivan Sadowski*‡ * Department of Biochemistry and Molecular Biology University of British Columbia Vancouver, British Columbia V6T 1Z3 † Banting and Best Department of Medical Research University of Toronto Toronto, Ontario M5G 1L6 Canada

Summary Phosphorylation of the yeast transcription factor GAL4 at S699 is required for efficient galactose-inducible transcription. We demonstrate that this site is a substrate for the RNA polymerase holoenzyme–associated CDK SRB10. S699 phosphorylation requires SRB10 in vivo, and this site is phosphorylated by purified SRB10/ SRB11 CDK/cyclin in vitro. RNA Pol II holoenzymes purified from WT yeast phosphorylate GAL4 at sites observed in vivo whereas holoenzymes from srb10 yeast are incapable of phosphorylating GAL4 at S699. Mutations at GAL4 S699 and srb10 are epistatic for GAL induction, demonstrating that SRB10 regulates GAL4 activity through this phosphorylation in vivo. These results demonstrate a function for the SRB10/ CDK8 holoenzyme–associated CDK that involves regulation of transactivators by phosphorylation during transcriptional activation. Introduction Eukaryotic genes are regulated by interactions between transacting factors bound to DNA at enhancers or upstream activating sequences, and components of the general transcription factor (GTF) machinery that assemble near the site of transcriptional initiation. Transactivator proteins have been shown to interact with RNA polymerase II holoenzyme and TFIID components (for review, see Ptashne and Gann, 1997). These interactions are thought to catalyze assembly of complexes that are competent for transcriptional initiation or elongation. In this view of transcriptional regulation, one supposes that the GTFs are passively herded into position by activator proteins that are the primary targets for physiological signaling mechanisms. However, yeast and mammalian RNA polymerase II holoenzymes, and mammalian TFIID complexes, are known to contain protein kinases, whose presence implies that the GTF machines may themselves be targets for signals that modulate gene expression (for review, see Carlson, 1997). KIN28 and its associated cyclin CCL1 comprise the kinase component of yeast TFIIH (Feaver et al., 1994; Svejstrup et al., 1996), whose mammalian counterparts ‡ To whom correspondence should be addressed (e-mail: sadowski@

interchange.ubc.ca).

are represented by CDK7/cyclin H (Serizawa et al., 1995). TFIIH kinase activity is largely responsible for postinitiation phosphorylation of the C-terminal domain (CTD) of RNA Pol II, which is thought to be necessary for elongation (O’Brien et al., 1994). TFIIH kinase activity and CTD phosphorylation are required for transcription of most, but not all, class II genes (Cismowski et al., 1995; Lee and Lis, 1998). The function of the second holoenzyme-associated CDK/cyclin pair represented by SRB10/SRB11, respectively, in yeast, and CDK8/cyclin C in mammalian cells, is unknown (Hengartner et al., 1998). Unlike KIN28/ CCL1, SRB10 and SRB11 are dispensable for yeast growth but are required for regulation of specific genes, including those involved in sucrose and galactose utilization (Kuchin et al., 1995), meiosis (Strich et al., 1989), mating type (Wahi and Johnson, 1995), and stress response (Cooper et al., 1997). The GAL regulon requires SRB10/ SRB11 for both repression by glucose and activation by galactose (Kuchin et al., 1995; Liao et al., 1995; Serizawa et al., 1995). These differential effects suggest that this holoenzyme-associated CDK/cyclin pair may actively participate in gene regulatory decisions. However, the function of this kinase in regulating transcriptional responses remains unknown because its substrates and upstream effectors have not been identified. It has been suggested that preinitiation phosphorylation of the Pol II CTD by SRB10 might account for its general inhibitory role in transcription (Hengartner et al., 1998). Phosphorylation of GAL4 correlates precisely with activation of the GAL genes (Mylin et al., 1989; Sadowski et al., 1991). GAL4 is phosphorylated on at least four different sites (see Figure 1A), but only one phosphorylation at S699 is required for full GAL gene induction (Sadowski et al., 1996). Previous results indicate that GAL4 phosphorylation occurs as a consequence of its transcriptional activation function, and it has been suggested that phosphorylation is mediated by GTF components (Sadowski et al., 1991). We show here that GAL4 is phosphorylated by both RNA polymerase holoenzyme–associated CDKs; S699 is specifically phosphorylated by SRB10, while KIN28/TFIIH predominantly phosphorylates S837. These observations demonstrate that one function of the RNA Pol II holoenzyme–associated CDKs involves modulation of gene expression by phosphorylation of transcriptional activators during their interaction with the RNA Pol II holoenzyme. Results SRB10 Is Required for GAL4 S699 Phosphorylation In Vivo We have previously argued that GAL4’s multiple phosphorylations (see Figure 1A) occur during interaction with the GTFs (Sadowski et al., 1991, 1996). To directly examine this hypothesis, we determined whether the Pol II holoenzyme–associated CDKs are necessary for GAL4 phosphorylation in vivo. For these experiments, we used the GAL4D683 derivative (Sadowski et al.,

Molecular Cell 674

Figure 1. GAL4 Phosphorylation Requires the RNA Pol II Holoenzyme CDKs (A) Schematic representation of WT GAL4 (top) protein indicating the location of identified phosphorylation sites and organization of functional domains. The GAL4D683 derivative (Sadowski et al., 1996) is indicated below. (B) WT (lanes 1–6) or kin28ts (lanes 7–12) yeast transformed with a vector control (lanes 1, 4, 7, and 10), GAL4D683 (lanes 2, 5, 8, and 11), or GAL4D683T (lanes 3, 6, 9, and 12) expression plasmids were labeled with [32P]-orthophosphate at 258C (lanes 1–3 and 7–9) or 378C (lanes 4–6 and 10–12). GAL4 protein was recovered by immunoprecipitation and analyzed by SDS-PAGE and autoradiography ([32P]). Whole-cell extracts from parallel mock-labeled samples were analyzed for GAL4 protein by Western blotting with 8C1 a-GAL4 McAb (GAL4 WB). (C) WT (lanes 1–3) and srb102 (lanes 4–6) yeast transformed with a vector control (lanes 1 and 4), GAL4D683 (lanes 2 and 5), or GAL4D683T (lanes 3 and 6) expression plasmids were analyzed as in (B). (D) GAL4D683 protein expressed in WT (panel 1) or srb102 (panel 2) yeast was labeled with [32P]-orthophosphate in vivo. Labeled protein was digested with trypsin, and phosphopeptides were resolved by electrophoresis (horizontal dimension) and chromatography (vertical). A synthetic peptide corresponding to the predicted GAL4 S699– containing tryptic peptide (YVSPGSVGPSPVPLK, S699 underlined) was phosphorylated in vitro with HA-SRB10 and analyzed separately (panel 3), or comigrated with in vivo labeled GAL4D683 peptides (panel 4).

1996), which lacks the central inhibitory segment, activates transcription efficiently, retains the major binding site for GAL80 at the C terminus, and importantly, bears all of the known sites of GAL4 phosphorylation (Figure 1A) (Stone and Sadowski, 1993; Sadowski et al., 1996). GAL4D683T bears alanine substitutions at serines 691, 696, and 837, leaving S699 as the only phosphorylation that produces a slower migrating electrophoretic species

(Sadowski et al., 1996). Note that GAL4D683T also has additional phosphorylations that do not produce slower migrating species, including at least one in the DNAbinding domain (Figure 1A). GAL4D683 and GAL4D683T were efficiently phosphorylated in vivo when labeled with [32P]-orthophosphate, albeit more weakly at higher temperatures (Figure 1B). In cells bearing a kin28ts allele, we observed efficient phosphorylation at the permissive temperature (258C), but barely detectable phosphorylation of either derivative at the nonpermissive temperature (378C), demonstrating that KIN28 is required for most of the phosphorylations on GAL4, apparently including that in the DBD (Figure 1B). We also found that phosphorylation was impaired in cells bearing an srb10 disruption (Figure 1C). In srb10 cells, the slower migrating GAL4 species was absent as judged by 32P labeling and immunoblotting. These results demonstrate that each of the holoenzyme-associated CDKs are required for some of GAL4’s phosphorylations in vivo. To determine whether SRB10 is required for specific GAL4 phosphorylations, we subjected in vivo 32P-labeled GAL4D683 protein to tryptic phosphopeptide analysis. Protein from WT cells labeled in galactose generated eight phosphopeptides, designated 1–8 (Figure 1D, panel 1). S699 phosphorylation was demonstrated to produce phosphopeptide 1 because a synthetic peptide representing the predicted tryptic fragment containing GAL4 S699 phosphorylated in vitro (see below) resolved as a major spot (Figure 1D, panel 3), which comigrated with phosphopeptide 1 from in vivo labeled GAL4 protein (Figure 1D, panel 4). Phosphopeptide 5 represents S837 phosphorylation, while phosphopeptide 2 is derived from the DNA-binding domain (data not shown). We found that GAL4D683 protein labeled in srb10 cells lacked several phosphopeptides observed in WT cells, including 1, 4, 5, and 6 (compare Figure 1D, panels 1 and 2). The loss of phosphopeptides representing serines 699 and 837 is consistent with the observation that the srb10 disruption eliminates the slower migrating electrophoretic GAL4 species (Figure 1C). We conclude that mutations to SRB10 selectively eliminate a subset of GAL4 phosphorylations in vivo, including those at serines 699 and 837, whereas mutations to KIN28 globally inhibit GAL4 phosphorylation. SRB10, but Not KIN28, Phosphorylates GAL4 S699 In Vitro To determine whether the RNA Pol II holoenzyme–associated CDKs were capable of directly phosphorylating GAL4, we used recombinant WT GAL4 protein (Kang et al., 1993) as substrate for kinase reactions in vitro with immunopurified HA-tagged SRB10 and KIN28. We found that both HA-KIN28 and HA-SRB10 phosphorylated GAL4 in vitro (lanes 2 and 6, Figure 2A), but neither phosphorylated recombinant GAL80 (lanes 3 and 7). Tryptic phosphopeptide analysis of GAL4 phosphorylated in vitro with HA-SRB10 and HA-KIN28 generated predominantly phosphopeptide 1 for both kinases, indicating S699 phosphorylation (data not shown). Several observations indicate that GAL4 S699 is a direct substrate for SRB10 whereas KIN28 is not capable of directly phosphorylating this site. First, GAL4 protein phosphorylated in vitro with HA-KIN28 immunopurified from srb10 cells produces phosphorylation at S837 (peptide 5) but not at

SRB10/CDK8 Regulates GAL4 675

(Figure 2E, panel 2). These results demonstrate that the two RNA Pol II holoenzyme–associated CDKs, SRB10/ SRB11 and KIN28/CCL1, differ in their specificities toward GAL4 in vitro; SRB10 predominantly phosphorylates S699 whereas KIN28 exclusively phosphorylates S837.

Figure 2. SRB10 Phosphorylates GAL4 at S699 In Vitro (A) In vitro kinase reactions were performed with immunopurified HA-SRB10 (lanes 1–4) or HA-KIN28 (lanes 5–8). Reactions contained WT GAL protein (lanes 2, 4, 6, and 8), GAL80 protein (lanes 3 and 7), or no added substrate (lanes 1 and 5). Lanes 4 and 8 contain a control immunoprecipitate with an unrelated antibody (anti-vimentin). (B) In vitro kinase reactions were performed with purified either recombinant WT SRB10/SRB11 complexes (lanes 2 and 3), D290A mutant SRB10/SRB11 (lane 1), or purfied yeast TFIIH (lanes 4 and 5). Reactions contained either recombinant WT GAL4 (lanes 1, 3, and 5), or no added substrate (lanes 2 and 4). (C) HA-KIN28 (left) and HA-SRB10 (right) immunoprecipitates from (D) were used for in vitro kinase reactions with synthetic peptides. Reactions contained no peptide, WT S699 peptide, or S699A mutant peptide. Peptides were separated from the reaction and 32P incorporation was determined (CPM). (D) Immunoprecipitates with a-HA McAb from yeast bearing a vector control (lane 1), or expressing HA-KIN28 (lane 2) or HA-SRB10 (lane 3), were immunoblotted with the same antibody. (E) WT GAL4 protein phosphorylated in vitro with purified TFIIH (panel 1) or purified recombinant WT SRB10/SRB11 (panel 2) was subjected to tryptic phosphopeptide analysis. The identity of the numbered peptides was confirmed by comigration with peptides generated from in vivo labeled GAL4D683 protein (data not shown).

S699 (data not shown), indicating that S699 must be phosphorylated by an associated SRB10-dependent activity. Secondly, an S699-containing synthetic GAL4 tryptic peptide was phosphorylated approximately 40fold more efficiently in vitro by HA-SRB10 than by HAKIN28 (Figure 2C), despite the fact that the kinases were recovered in approximately equivalent amounts (Figure 2D). Finally, although purified yeast TFIIH containing KIN28/CCL1 (Feaver et al., 1994) and purified recombinant SRB10/SRB11 are both capable of phosphorylating GAL4 in vitro (Figure 2B), tryptic phosphopeptide analysis reveals that purified TFIIH phosphorylates GAL4 exclusively at S837 and not S699 (Figure 2E, panel 1) whereas purified recombinant SRB10/SRB11 complexes phosphorylate GAL4 on S699 and to a lesser extent at S837

GAL4 S699 Is Phosphorylated by SRB10-Containing RNA Pol II Holoenzymes In Vitro To determine whether RNA Pol II–associated SRB10 is capable of phosphorylating GAL4 S699, we used purified holoenzymes for in vitro kinase reactions. RNA Pol II holoenzymes were purified by affinity chromatography with GST-TFIIS from either WT or srb102 yeast as described (Pan et al., 1997). Western blotting demonstrated that GST-TFIIS-purified holoenzyme complexes from WT yeast contained all of the known holoenzyme components, including the largest core subunits, RPB1, TFIIB, TFIIE, as well as KIN28 and SRB10 (Figure 3A and data not shown). Many different proteins within the holoenzymes from both WT and srb10 yeast were found to be phosphorylated after incubation with [g-32P]ATP in vitro (data not shown). We found that GAL4 also was phosphorylated in reactions with holoenzymes from both WT and srb10 yeast (Figure 3B). Phosphopeptide mapping of recombinant WT GAL4 phosphorylated in vitro with WT holoenzymes produced all of the peptides normally observed with GAL4D683 in vivo (Figure 3C, panel 1) plus an additional peptide (N) that must occur within the central portion of GAL4 (see Figure 1A). Significantly, we found that GAL4 phosphorylated by holoenzymes purified from srb10 yeast was missing several phosphopeptides, including peptide 1 (Figure 3C, panel 3), which comigrates with a synthetic S699-containing tryptic peptide phosphorylated in vitro with recombinant SRB10 (Figure 3C, panel 2). These results demonstrate that GAL4 S699 is phosphorylated by RNA Pol II holoenzyme–associated SRB10. SRB10 Is Epistatic to GAL4 S699 Phosphorylation for GAL Induction Several groups have noted that SRB10 is required for efficient induction of the GAL genes (Kuchin et al., 1995; Liao et al., 1995; Serizawa et al., 1995). We observe a similar effect in strains expressing WT GAL4, where GAL induction is approximately 4-fold lower in srb10 cells (Figure 4). Significantly, however, we also find that the GAL4 S699A mutation does not cause an additional defect in GAL transcription in srb10 cells (Figure 4, srb102). In fact, disruption of srb10, or mutation of GAL4 S699 to alanine, produces approximately equivalent defects in GAL induction (Figure 4). This finding demonstrates that SRB10 and GAL4 S699 phosphorylation are genetically epistatic to one another and represent equivalent, rather than parallel, regulatory mechanisms. The GAL4 S699 phosphorylation is not required for GAL4 to activate transcription because the S699A mutation has little effect on transcriptional activation in strains lacking the negative regulator GAL80 (Sadowski et al., 1996; Figure 4, gal802). Consistent with the model that SRB10 regulates GAL induction through the GAL4 S699 phosphorylation, we find that srb10 mutations have a more severe effect in a GAL80 WT strain (Figure 4, WT) than in a

Molecular Cell 676

Figure 4. SRB10 Is Epistatic to GAL4 S699 (A) GAL1-LacZ expression was measured following 2 hr induction with 2.0% galactose in yeast strains YJMH4 (WT), YJMH5 (srb102), YJMH8 (gal80-), and YJMH9 (gal802srb102), bearing a control vector (2), or expressing WT GAL4 (WT), or GAL4 S699A (A699) from plasmids pJR006 and pJR007, respectively.

Figure 3. GAL4 Is Phosphorylated at S699 by Purified RNA Pol II Holoenzymes (A) RNA Pol II holoenzymes were purified from WT (W303) and srb10 (H617) yeast by chromatography with GST-yTFIIS. Control preparations contained proteins bound in chromatography with GST. Eluted proteins were analyzed by SDS-PAGE and immunoblotting with the antibodies indicated on the left. (B) In vitro kinase reactions were performed with affinity-purified RNA Pol II holoenzymes immobilized on GST-yTFIIS beads (even lanes), or GST control preparations (odd lanes) derived from a WT (lanes 1–6) or srb102 (lanes 7–12) yeast. Reactions contained WT GAL4 protein (lanes 3, 4, 9, and 10), GAL80 protein (lanes 5, 6, 11, and 12), or no added substrate (lanes 1, 2, 7, and 8). GAL4 and GAL80 were recovered from the reaction by immunoprecipitation and analyzed by SDS-PAGE and autoradiography [32P]. (C) WT GAL4 protein was phosphorylated in vitro with purified immobilized RNA Pol II holoenzymes derived from WT (panels 1 and 2) or srb10 (panel 3) yeast, recovered by immunoprecipation, and tryptic phosphopeptides were analyzed as above. The synthetic GAL4 S699–containing tryptic peptide was phosphorylated in vitro with purified recombinant WT SRB10/11 (data not shown) and comigrated with in vitro labeled recombinant GAL4 (panel 2).

strain lacking gal80 (Figure 4, gal802). Thus, although we observe a slight decrease in transcriptional activation by WT GAL4 when srb10 is deleted in gal80 yeast (compare WT GAL4, gal80, and gal80 srb10 [data not shown]), the predominant effect of SRB10 on GAL transcription appears to involve regulation of induction through the GAL4 S699 phosphorylation. Combined with the fact that SRB10 is required for phosphorylation of S699 in vivo, and that SRB10 phosphorylates GAL4 S699 in vitro, these results demonstrate that GAL4 is directly controlled by an SRB10-mediated phosphorylation. Discussion GAL4 was one of the first transcription factors demonstrated to be phosphorylated when it was initially shown

that phosphorylation correlated with its activation function (Mylin et al., 1989). The significance of these phosphorylations for gene regulation became enigmatic when it was realized that they likely occurred as a consequence of, rather than a contributing factor for, GAL4’s transcriptional activation function (Sadowski et al., 1991). Mutations to RNA Pol II holoenzyme components generally affect GAL4 phosphorylation in vivo (Mylin et al., 1989; Sadowski et al., 1991; data not shown), and we find that the mediator-associated CDK SRB10 is required for a specific subset of GAL4’s phosphorylations. Furthermore, GAL4 is specifically phosphorylated at S699 by recombinant SRB10 in vitro whereas KIN28 only phosphorylates GAL4 at S837. GAL4 is also phosphorylated at all of the sites normally observed in vivo by purified RNA Pol II holoenzymes in vitro. These results demonstrate that GAL4’s phosphorylations must occur during transcriptional activation by the holoenzyme-associated CDKs. Of the multiple phosphorylations on GAL4, only that at S699 appears to be necessary for induction of the GAL genes. SRB10 is also required for GAL induction, and srb10 disruptions are epistatic to the GAL4 S699A mutation. These results demonstrate a direct regulatory role for phosphorylation of GAL4 by SRB10. Function of the RNA Pol II Holoenzyme CDKs in Eukaryotic Transcription The C-terminal domain (CTD) of the largest RNA Pol II subunit becomes phosphorylated during transition to an elongating complex (O’Brien et al., 1994). Three protein kinases in yeast have been implicated in CTD phosphorylation, including the two holoenzyme-associated CDKs KIN28 and SRB10, and a related enzyme encoded by CTK1 (Liao et al., 1995; Sterner et al., 1995; Valay et al., 1995). Much work characterizing the function of these kinases in transcriptional regulation has focused on their role in CTD phosphorylation, which in many analyses is considered to be overlapping and at least partially redundant (Kuchin and Carlson, 1998; Lee and Lis, 1998). Our results demonstrate that GAL4 is a substrate for both SRB10 and KIN28, but each of these kinases phosphorylates separate specific sites. In contrast, we find that purified CTK1 does not phosphorylate WT GAL4 in vitro (data not shown). These results suggest that

SRB10/CDK8 Regulates GAL4 677

each of these CDKs must have separate preferred substrates in vivo. SRB10 was originally identified in screens for suppressors of constitutively repressed genes (Carlson et al., 1984; Strich et al., 1989; Balciunas and Ronne, 1995; Wahi and Johnson, 1995). Null mutations of srb10 suppress phenotypes associated with CTD truncations (Hengartner et al., 1998). It has been suggested that SRB10 causes transcriptional repression by phosphorylating the CTD prior to preinitiation complex formation (Hengartner et al., 1998). The role of this mechanism in regulating specific genes is unclear because it does not explain how SRB10 can function both in activation and repression, nor why some genes are completely unaffected by SRB10 mutations. SRB10 has also been shown to interact directly with the C-terminal activation domain of GAL4 (Ansari and Ptashne, submitted), implicating this CDK as a target for transcriptional activation. In combination with our results, these observations demonstrate that direct interaction between GAL4 and SRB10 contributes to holoenzyme recruitment and also provides the opportunity for a direct phosphorylation on GAL4, which is necessary for full GAL induction. Regulation of RNA Polymerase Holoenzyme CDK Function Several observations implicate the RNA polymerase holoenzyme as a downstream target for physiological signaling. CTD phosphorylation appears to be regulated during entry into a stationary phase (Patturajan et al., 1998), in response to heat shock in yeast (Cooper et al., 1997) and mammalian cells (Dubois et al., 1997), and by stress-activated MAP kinase pathways (Venetianer et al., 1995). A mouse mediator complex was also shown to be associated with a nuclear MAP kinase (Jiang et al., 1998). Finally, the SRB10-associated cyclin C (SRB11) has been shown to be degraded in response to environmental signals including heat and peroxide stress (Cooper et al., 1997), and SRB10 protein levels have been shown to decrease as nutrient levels become depleted in liquid cultures (Holstege et al., 1998). We have also shown that in gal80 cells, where the GAL genes are activated constitutively, GAL4 is unphosphorylated in cells growing in nonfermentable carbon but becomes rapidly phosphorylated in response to fermentable sugars (Sadowski et al., 1996). These observations suggest that environmental factors that influence holoenzyme CDK activity are also likely to modulate transcription of specific genes through phosphorylation of their cognate transcription factors. Therefore, we propose that a critical function of the RNA polymerase holoenzyme CDKs involves modulation of specific transactivators in response to general physiological signals to coordinate inducible transcription with the cellular environment.

equivalent LEU2, ARS-CEN plasmids, which express GAL4 and GAL4 S699A, respectively, created by cloning HindIII/BamH1 GAL4 fragments into pRS315. The GAL4D683 and GAL4D683T expression plasmids have been described previously (Stone and Sadowski, 1993; Sadowski et al., 1996). Plasmid pIS028 is a 2-micron, TRP1 plasmid, which expresses HA-tagged SRB10 from the ADH1 promoter. pKIN28Ha expressing HA-tagged Kin28 was as described (Cismowski et al., 1995). Cells for b-galactosidase assays were grown in SD-Leu containing nonfermentable carbon induced by adding galactose from a 40% sterile stock. Strains were lysed with glass beads, and b-gal activity was measured in extracts (Sadowski et al., 1996). All results are an average of at least three independent determinations. Antibodies and Recombinant Proteins Rabbit anti-GAL4 DBD polyclonal antibody was as described (Sadowski et al., 1991). The mouse 8C1 hybridoma producing a-GAL4 AR2 was isolated from a monoclonal panel produced from mice immunized with WT GAL4 protein. Monoclonal antibodies against HA and vimentin were obtained commercially. The mouse 8WG16 hybridoma producing a-RPB1 was a kind gift of R. Burgess. Polyclonal antibodies against SRB10, TFIIB, SRB5, and TFA2 were a kind gift of R. Young. Rabbit anti-TFB1, anti-TFA1, anti-GAL80, and anti-yTBP polyclonal anitbodies were produced from rabbits immunized with purified GST-TFB1 (aa 1–136), GST-TFA1, 6-His-GAL80, or yeast TBP, respectively. Rabbit anti-KIN28 polyclonal antibody was as described (Feaver et al., 1994). WT GAL4 protein was produced by expression in insect cells (Kang et al., 1993) and purified by extraction of insoluble material from infected cell lysates with RIPA buffer (Sadowski et al., 1991) modified to contain only 0.02% SDS. GAL80 protein was produced as a 6-His fusion in pRSET-A and purified by Ni-chelate affinity chromatography. Recombinant SRB10/SRB11 complexes were produced by coexpression in insect cells and purified as described (Hengartner et al., 1998). Purified RNA Pol II holoenzymes were produced from yeast extracts prepared as described (Schultz et al., 1991) and purified as described for human cells (Pan et al., 1997). Metabolic Labeling, Tryptic Phosphopeptide Analysis, and In Vitro Kinase Assays [32P]-orthophosphate labeling of yeast, immunoprecipitations, and tryptic phosphopeptide analysis was as described (Hung et al., 1997). Phosphopeptides were resolved by electrophoresis in the horizontal dimension at pH 2.1 and chromatography (butanol:acetic acid:dH2O:pyridine, 75:50:37.5:15.5) in the vertical dimension. Phosphopeptides were detected using a Molecular Dynamics phosphorimager. For in vitro kinase reactions, yeast expressing HAKIN28 and HA-SRB10 were grown in SD-Trp containing 2% raffinose to OD A600 5 1.0 and then were induced with 2% galactose for 2 hr. Cells were lysed in kinase lysis buffer (KLB) (20 mM Tris, 150 mM NaCl, 1 mM EDTA, 10% glycerol, O.3% NP-40, 1 mM PMSF, and protease inhibitors) by vortexing with glass beads. HA-tagged kinases were recovered by immunoprecipitation from 2 mg clarified lysate with anti-HA McAb and protein G–Sepharose beads. Immune complexes were washed three times with KLB and two times in kinase buffer (KB) (10 mM MgCl2, 50 mM Tris [pH 7.5], 1 mM DTT). Kinase reactions were performed in 5 ml KB with 2 pmol [g-32P]ATP for 30 min at 308C. Reactions were resolved on 10% SDS PAGE and visualized by exposure to Kodak Biomax film. Synthetic peptide (10 mg) was added per kinase reaction, and peptides were separated from the reaction by diluting with 400 ml dH2O and filtration through Micron-10 microconcentrators (Amicon). Peptides in the filtrate were separated from unincorporated label by chromatography on MicronSCX cation exchange columns (Amicon) and counted by liquid scintillation.

Experimental Procedures

Acknowledgments

Plasmids and Yeast Manipulations W303-1A (WT) and H617 (srb10) were used for all in vivo labeling and immunoprecipating experiments (Thomas and Rothstein, 1989; Balciunas and Ronne, 1995). All other yeast strains used in this study were derived from MCY3684 or MCY3686 (Carlson et al., 1984). Plasmid YCpG4 is TRP1, ARS-CEN, which expresses GAL4 from its own promoter (Sadowski et al., 1996). pJR006 and pJR007 are

We thank Marian Carlson, Hans Ronne, Jim Friesen, Mark Solomon, R. Burgess, and Richard Young for gifts of yeast strains, antibodies, plasmids, and recombinant virus, and Roger Kornberg for generously providing purified yeast TFIIH. This work was supported by MRC grants to I. S. and J. G. J. G. is a investigator of the HHMRI. Received August 21, 1998; revised March 5, 1999.

Molecular Cell 678

References Balciunas, D., and Ronne, H. (1995). Three subunits of the RNA polymerase II mediator complex are involved in glucose repression. Nucleic Acids Res. 23, 4421–4425.

M., Bensaude, O., Warren, S.L., and Corden, J.L. (1998). Growthrelated changes in phosphorylation of yeast RNA polymerase II. J. Biol. Chem. 273, 4689–4694. Ptashne, M., and Gann, A. (1997). Transcriptional activation by recruitment. Nature 386, 569–577.

Carlson, M. (1997). Genetics of transcriptional regulation in yeast: connections to the RNA polymerase II CTD. Annu. Rev. Cell Dev. Biol. 13, 1–23.

Sadowski, I., Niedbala, D., Wood, K., and Ptashne, M. (1991). GAL4 is phosphorylated as a consequence of transcriptional activation. Proc. Natl. Acad. Sci. USA 88, 10510–10514.

Carlson, M., Osmond, B.C., Neigeborn, L., and Botstein, D. (1984). A suppressor of SNF1 mutations causes constitutive high-level invertase synthesis in yeast. Genetics 107, 19–32.

Sadowski, I., Costa, C., and Dhanawansa, R. (1996). Phosphorylation of Ga14p at a single C-terminal residue is necessary for galactoseinducible transcription. Mol. Cell. Biol. 16, 4879–4887.

Cismowski, M.J., Laff, G.M., Solomon, M.J., and Reed, S.I. (1995). KIN28 encodes a C-terminal domain kinase that controls mRNA transcription in Saccharomyces cerevisiae but lacks cyclin-dependent kinase-activating kinase (CAK) activity. Mol. Cell. Biol. 15, 2983–2992.

Schultz, M., Choe, S., and Reeder, R. (1991). Specific initiation by RNA polymerase I in a whole-cell extract from yeast. Proc. Natl. Acad. Sci. USA 88, 1004–1008.

Cooper, K.F., Mallory, M.J., Smith, J.B., and Strich, R. (1997). Stress and developmental regulation of the yeast C-type cyclin Ume3p (Srb11p/Ssn8p). EMBO J. 16, 4665–4675. Dubois, M.F., Vincent, M., Vigneron, M., Adamczewski, J., Egly, J.M., and Bensaude, O. (1997). Heat-shock inactivation of the TFIIH-associated kinase and change in the phosphorylation sites on the C-terminal domain of RNA polymerase II. Nucleic Acids Res. 25, 694–700. Feaver, W.J., Svejstrup, J.Q., Henry, N.L., and Kornberg, R.D. (1994). Relationship of CDK-activating kinase and RNA polymerase II CTD kinase TFIIH/TFIIK. Cell 79, 1103–1109. Hengartner, C., Myer, V., Liao, S., Wilson, C., Koh, S., and Young, R. (1998). Temporal regulation of RNA polymerase II by Srb10 and Kin28 cyclin-dependent kinases. Mol. Cell 2, 43–53. Holstege F., Jennings E., Wyrick J., Lee T., Hengartner, C., Green, M., Golub, T., Lander, E., and Young, R. (1998). Dissecting the regulatory circuitry of a eukaryotic genome. Cell 95, 717–728. Hung, W., Olson, K.A., Breitkreutz, A., and Sadowski, I. (1997). Characterization of the basal and pheromone-stimulated phosphorylation states of Ste12p. Euro. J. Biochem. 245, 241–251. Jiang, Y.W., Veschambre, P., Erdjument-Bromage, H., Tempst, P., Conaway, J., Conaway, R., and Kornberg, R.D. (1998). Mammalian mediator of transcriptional regulation and its possible role as an end-point of signal transduction pathways. Proc. Natl. Acad. Sci. USA 95, 8538–8543. Kang, T., Martins, T., and Sadowski, I. (1993). Wild type GAL4 binds cooperatively to the GAL1–10 UASG in vitro. J. Biol. Chem. 268, 9629–9635. Kuchin, S., and Carlson, M. (1998). Functional relationships of Srb10Srb11 kinase, carboxy-terminal domain kinase CTDK-I, and transcriptional corepressor Ssn6-Tup1. Mol. Cell. Biol. 18, 1163–1171. Kuchin, S., Yeghiayan, P., and Carlson, M. (1995). Cyclin-dependent protein kinase and cyclin homologs SSN3 and SSN8 contribute to transcriptional control in yeast. Proc. Natl. Acad. Sci. USA 92, 4006– 4010. Lee, D., and Lis, J.T. (1998). Transcriptional activation independent of TFIIH kinase and the RNA polymerase II mediator in vivo. Nature 393, 389–392. Liao, S.-M., Zhang, J., Jeffery, D.A., Koleske, A.J., Thompson, C.M., Chao, D.M., Viljoen, M., van Vuuren, H.J.J., and Young, R.A. (1995). A kinase-cyclin pair in the RNA polymerase II holoenzyme. Nature 374, 193–196. Mylin, L.M., Bhat, J.P., and Hopper, J.E. (1989). Regulated phosphorylation and dephosphorylation of GAL4, a transcriptional activator. Genes Dev. 3, 1157–1165. O’Brien, T., Hardin, S., Greenleaf, A., and Lis, J. (1994). Phosphorylation of RNA polymerase II C-terminal domain and transcriptional elongation. Nature 370, 75–77. Pan, G., Aso, T., and Greenblatt, J. (1997). Interaction of elongation factors TFIIS and elongin A with a human RNA polymerase II holoenzyme capable of promoter-specific initiation and responsive to transcriptional activators. J. Biol. Chem. 272, 24563–24571. Patturajan, M., Schulte, R.J., Sefton, B.M., Berezney, R., Vincent,

Serizawa, H., Makela, T.P., Conaway, J.W., Conaway, R.C., Weinberg, R.A., and Young, R.A. (1995). Association of Cdk-activating kinase subunits with transcription factor TFIIH. Nature 374, 280–282. Sterner, D.E., Lee, J.M., Hardin, S.E., and Greenleaf, A.L. (1995). The yeast carboxyl-terminal repeat domain kinase CTDK-I is a divergent cyclin-cyclin-dependent kinase complex. Mol. Cell. Biol. 15, 5716– 5724. Stone, G., and Sadowski, I. (1993). GAL4 is regulated by a glucoseresponsive functional domain. EMBO J. 12, 1375–1385. Strich, R., Slater, M.R., and Esposito, R.E. (1989). Identification of negative regulatory genes that govern the expression of early meiotic genes in yeast. Proc. Natl. Acad. Sci. USA 86, 10018–10022. Svejstrup, J.Q., Feaver, W.J., and Kornberg, R.D. (1996). Subunits of yeast RNA polymerase II transcription factor TFIIH encoded by the CCL1 gene. J. Biol. Chem. 271, 643–645. Thomas, B.J., and Rothstein, R. (1989). Elevated recombination rates in transcriptionally active DNA. Cell 56, 619–630. Valay, J.G., Simon, M., Dubois, M.F., Bensaude, O., Facca, C., and Faye, G. (1995). The KIN28 gene is required both for RNA polymerase II mediated transcription and phosphorylation of the Rpb1p CTD. J. Mol. Biol. 249, 535–544. Venetianer, A., Dubois, M.F., Nguyen, V.T., Bellier, S., Seo, S.J., and Bensaude, O. (1995). Phosphorylation state of the RNA polymerase II C-terminal domain (CTD) in heat-shocked cells. Possible involvement of the stress-activated mitogen-activated protein (MAP) kinases. Euro. J. Biochem. 233, 83–92. Wahi, M., and Johnson, A.D. (1995). Identification of genes required for alpha 2 repression in Saccharomyces cerevisiae. Genetics 140, 79–90.