Yeast Carboxyl-terminal Domain Kinase I Positively and Negatively Regulates RNA Polymerase II Carboxyl-terminal Domain Phosphorylation*

Monoclonal antibodies that recognize specific carboxyl-terminal domain (CTD) phosphoepitopes were used to examine CTD phosphorylation in yeast cells lacking carboxyl-terminal domain kinase I (CTDK-I). We show that deletion of the kinase subunitCTK1 results in an increase in phosphorylation of serine in position 5 (Ser5) of the CTD repeat (Tyr1-Ser2-Pro3-Thr4-Ser5-Pro6-Ser7) during logarithmic growth. This result indicates that CTDK-I negatively regulates CTD Ser5 phosphorylation. We also show thatCTK1 deletion (ctk1Δ) eliminates the transient increase in CTD serine 2 (Ser2) phosphorylation observed during the diauxic shift. This result suggests that CTDK-I may play a direct role in phosphorylating CTD Ser2 in response to nutrient depletion. Northern blot analysis was used to show that genes normally induced during the diauxic shift are not properly induced in a ctk1Δ strain. Glycogen synthase (GSY2) and cytosolic catalase (CTT1) mRNA levels increase about 10-fold in wild-type cells, but this increase is not observed in ctk1Δ cells suggesting that increased message levels may require Ser2 phosphorylation. Heat shock also induces Ser2 phosphorylation, but we show here that this change in CTD modification and an accompanying induction of heat shock gene expression is independent of CTDK-I. The observation that SSA3/SSA4 expression is increased in ctk1Δ cells grown at normal temperature suggests a possible role for CTDK-I in transcription repression. We discuss several possible positive and negative roles for CTDK-I in regulating CTD phosphorylation and gene expression.

The CTD is heavily phosphorylated in vivo (11), and this is an essential modification for vegetative growth in yeast (12). Phosphorylation of the CTD is temporally linked to the transition between initiation and elongation in vitro (13)(14)(15) leading to a model in which pol II with an unphosphorylated CTD (pol IIA) participates in forming the initiation complex, and pol II with a phosphorylated CTD (pol II0) is engaged in transcript elongation (16). Among the best characterized CTD kinases are those present in pol II preinitiation and/or elongation complexes. The TFIIHassociated Cdk7/cyclin H (Kin28/Ccl1 in yeast) (17)(18)(19)(20)(21)(22) and Cdk8/cyclin C (Srb11/Srb10 in yeast) (23)(24)(25) can both phosphorylate the CTD in vitro and are well positioned for participation in promoter clearance. Much attention has also been paid to P-TEFb, an elongation factor (26,27) that contains an in vitro CTD kinase activity (28) comprised of Cdk9/cyclin T (29 -32). Consistent with its requirement for productive elongation (28), P-TEFb remains with pol II during elongation (33).
Yeast CTDK-I is a Cdk-cyclin kinase that is closely related to P-TEFb, although whether these kinases perform the same function in vivo has not been determined. CTDK-I was isolated as a complex that specifically phosphorylates the CTD in vitro (34). CTDK-I is comprised of three subunits encoded by CTK1, CTK2, and CTK3 (35). CTK1 encodes a kinase catalytic subunit closely related to the P-TEFb Cdk9 subunit (29). The Ctk2 protein resembles cyclin T, whereas Ctk3 shows no homology to known proteins (35). Like P-TEFb, CTDK-I has been shown to stimulate pol II elongation in vitro (36). This function is not essential, however, as deletion of any single or all of the genes encoding CTDK-I is not lethal in yeast (35,37). CTDK-I-deficient strains grow more slowly than wild type at normal temperatures and are unable to grow at low temperature (35,37) indicating that some cellular processes are impaired. Examining the phosphorylation state of the CTD in CTK1-deleted (ctk1⌬) cells initially indicated that the CTD was not phosphorylated in the normal fashion (37). In particular, the abundance of a slower mobility form of the largest subunit (Rpb1p) was reduced suggesting that the CTD was under-phosphorylated. Reactivity with anti-phospho-CTD antiserum remained, however, suggesting the existence of multiple CTD kinases in vivo. These studies did not address the identity of the sites phosphorylated by CTDK-I nor the sites that remain phosphorylated in its absence.
Standard approaches to mapping in vivo CTD phosphorylation sites are impractical due to the repetitive nature of the amino acid sequence. Less direct in vitro approaches have yielded important information about potential phosphorylation sites. For example, we showed that both serine 2 (Ser 2 ) and serine 5 (Ser 5 ) of the consensus heptapeptide can be phosphorylated in vitro by Cdc2 kinase (38), and mutation of these sites to alanine in the yeast CTD is lethal (12). Together these results suggest that these serine residues act as phosphoacceptors for one or more CTD kinases in vivo.
More recently, we mapped critical elements of the CTD phosphoepitopes recognized by a set of monoclonal antibodies (39). By using in vitro phosphorylated CTD fusion proteins, we showed that mAb H14 specifically binds CTDs phosphorylated at Ser 5 , whereas mAb H5 specifically binds CTD fusion proteins phosphorylated at Ser 2 . These antibodies bind in vivo phosphorylated CTD indicating that both Ser 2 and Ser 5 are phosphorylated in vivo. We further showed that Ser 2 and Ser 5 phosphorylation is independently regulated during yeast growth (39).
We now extend the characterization of CTD phosphorylation by showing that deletion of CTK1 results in a profound increase in phosphorylation of Ser 5 . In addition, the increase in Ser 2 phosphorylation previously observed during the diauxic shift is not observed in ctk1⌬ cells. Taken together, these results suggest that CTDK-I participates in both positive and negative regulation of CTD phosphorylation.
Heat Shock-Yeast strains ADH6 -1a and ADH6 -1b were grown in YPD in a water bath at 30°C to an A 600 of 1.0. The culture (75 ml in a 500-ml flask) was then shifted to a 42°C water bath for 30 min. Control cells were maintained at 30°C for the same time. The heat shock was terminated by diluting the culture with ice-cold water, and cells were harvested by centrifugation and extracts prepared as described earlier (39).
Antibodies-mAbs 8WG16, H5, and H14 have been described earlier (39,41). Anti-JC20 antibodies were raised in rabbit against a nine amino acid peptide (MVGQQYSSA) corresponding to the amino terminus of the largest subunit (Rpb1p) of yeast pol II. These antibodies were subsequently affinity purified on a peptide column (Affi-Gel, Bio-Rad) and were used at a dilution of 1:200 for Western blot analysis. Anti-Ssa1p (42) and anti-Ssa3p/Ssa4p (43) antibodies were kind gifts from Dr. David Meyer (University of California, Los Angeles) and Dr. Elizabeth Craig (University of Wisconsin Medical School), respectively. These antibodies were used at a dilution of 1:5000. Anti-Ssa3p,4p detects both Ssa3p and Ssa4p, and anti-Ssa1p recognizes all four Ssa proteins (42,43). The mAb anti-Nab3p antibody (44) was kind gift from Dr. Maurice Swanson (University of Florida Medical School).
Cloning CTK1-The wild-type CTK1 gene was amplified by PCR from a wild-type yeast genomic DNA template using primers starting 400 base pairs 5Ј and 250 bases 3Ј of the coding sequence. Vent DNA polymerase (New England Biolabs) was used in the amplification to minimize errors. The amplified gene was cloned into pRS415 (45) to produce pRSCTK1. This plasmid was transformed (46) into ctk1⌬ yeast (ADH6-1a) and selected for growth on plates lacking leucine. The resulting transformants did not display the slow growth and cold-sensitive phenotypes of ctk1⌬ cells indicating that a functional copy of CTK1 was cloned.
Western Blotting-Yeast extracts (50 -100 g of total protein) were prepared by grinding with glass beads as described previously (39) and subjected to SDS-polyacrylamide gel electrophoresis (5%) followed by electrophoretic transfer to nitrocellulose paper (Protran, Schleicher & Schuell). Blots were probed with various antibodies, and the immunoreactive proteins were detected using either anti-mouse Ig (Amersham Pharmacia Biotech) or anti-mouse IgM (Kirkegaard & Perry Laboratories) at a dilution of 1:3000. The reactive bands were illuminated using ECL (Amersham Pharmacia Biotech).
Northern Blotting-Total RNA from yeast cells were prepared by the standard acid phenol method (47). RNA (20 g) was separated on formaldehyde-agarose gels and transferred to nylon membrane (Nϩ Hybond paper, Amersham Pharmacia Biotech). Blots were hybridized with 32 P-RNA probes transcribed using T7 polymerase, and radioactive bands were detected using a PhosphorImager. The template DNA used in transcription reaction for making RNA probe was synthesized by PCR using yeast genomic DNA as template. Each downstream primer contains a T7 promoter sequence such that the PCR reaction can be used as template to synthesize the antisense probe. The primers used for the PCR are as follows: SSA4, 5Ј GAATCAGCTA GAATCGTACG CG  3Ј and 5Ј TAATACGACT CACTATAGGG CCTCTTCAAC CGTT-GGGCCG 3Ј; ENO1, 5Ј GCTAGATCCG TCTACGACTC 3Ј and 5Ј TA-ATACGACT CACTATAGGG GTCACCGTGG TGGAAGTTTTЈ; GSY2,  5Ј TCCCGTGACC TACAAAACCA 3Ј and 5Ј TAATACGACT CACTAT-AGGG TATTGGGGGT AACTGTCCCT 3Ј; CTT1, 5Ј CCAATAAGAT  CAATCAGCTC 3Ј and 5Ј TAATACGACT CACTATAGGG GGAGTAT-GGA CATCCCAAGT 3Ј; ACT1, 5Ј GTAAAGCCGG TTTTGCCGGT 3Ј and 5Ј TAATACGACT CACTATAGGG GAAGCCAAGA TAGAACCACC 3Ј. The SSA4 probe also recognizes SSA3, and as the sizes of these two RNA are the same, we are unable to distinguish the two RNA in the Northern blot.

Changes in CTD Phosphorylation in CTK1 Null Strains-
Earlier studies showed that deletion of CTK1 results in an apparent decrease in CTD phosphorylation as determined by an increase in electrophoretic mobility of the largest subunit and a decreased reactivity with polyclonal antibodies raised against in vitro phosphorylated CTD (37). However, this study did not address which of the known phosphoepitopes were affected. We have used a set of monoclonal antibodies that recognize different CTD phosphoepitopes to characterize the CTD phosphorylation patterns in ctk1⌬ cells relative to wildtype cells. The antibodies employed include the following: mAb 8WG16 which recognizes unphosphorylated Ser 2 ; mAb H5 which recognizes phosphorylated Ser 2 ; and mAb H14 which recognizes phosphorylated Ser 5 (39). These antibodies were used to examine the phosphorylation state of the CTD in total cell extracts prepared from growing yeast cells.
The results presented in Fig. 1A clearly indicate an increase in the mAb H14-reactive epitope in the ctk1⌬ strain suggesting an increase in Ser 5 phosphorylation. In multiple experiments we have consistently observed a 3-5-fold increase in H14 reactivity when equal amount of protein is loaded. This increase in H14 reactivity is not due to an increase in the amount of pol II as determined from the reactivity of other pol II-specific antibodies. We also see a similar increase in mAb H14 reactivity in two other independently derived ctk1⌬ strains MCY3664 (40) and YJC1169 2 indicating that the increase in H14 reactivity is not specific to the ADH6 background (35).
Consistent with an increase in CTD phosphorylation, we see a decrease in the mobility of Rpb1p detected with mAb 8WG16. This antibody recognizes the Ser 2 site in unphosphorylated repeats. It thus reacts both with the hypo-phosphorylated IIa species of Rpb1p and with Rpb1p that is phosphorylated on some, but not all, repeats. The decrease in mobility of the 8WG16-reactive species observed in Fig. 1A is indicative of an increase in overall phosphorylation of the CTD. Taken together, the results presented in Fig. 1A indicate that deletion of the CTK1 gene leads to an increase in serine 5 phosphorylation. Fig. 1A also indicates a decrease in the mAb H5-reactive epitope in the ctk1⌬ strain. In this figure we observe an approximately 2-fold reduction, but the magnitude of the decrease is dependent on growth state (see below). This result suggests that CTDK-I may be involved in phosphorylation of Ser 2 .
To determine whether all of the RNA polymerase II is released from the cells during grinding with glass beads, we compared the protein in the cell extracts with protein remaining in the pellet containing cell debris. Fig. 1B shows that the H14-and H5-reactive forms of pol II are completely extracted while some of the heat shock protein Ssa1p remains in the pellet.
To control for the possibility that changes in CTD phosphorylation were due to secondary genetic changes in the ctk1⌬ cells, we re-transformed these cells with a plasmid expressing wild-type CTK1. In Fig. 1C we see that this plasmid restores wild-type levels of reactivity with both mAb H14 and H5. Thus, the effects seen are specific to CTK1. We also observe a marked decrease in H5 reactivity in ctk1⌬ cells in this experiment compared with the experiment shown in Fig. 1A. In Fig. 1C the cells were grown in minimal media lacking leucine to select for the CTK1-containing plasmid. Growth in minimal media enhances the effect of ctk1⌬ on the H5-reactive epitope.
Growth-related Changes in CTD Phosphorylation in ctk1⌬ Cells-In our previous studies we showed that phosphorylation of different CTD phosphoacceptor sites is independently regulated. Both nutritional limitation and heat shock result in higher levels of Ser 2 phosphorylation with little change in Ser 5 phosphorylation (39). To examine growth-related changes in ctk1⌬ cells, we made protein extracts from CTK1 (ADH6 -1a) and ctk1⌬ (ADH6 -1b) cells at different stages of growth. A typical growth curve was obtained from cells grown in rich media ( Fig. 2A). As described earlier (37) ctk1⌬ cultures display slow growth but eventually reach stationary phase. Fig. 2B shows Western blots of protein extracts derived from the cultures described in Fig. 2A. As we described earlier, there is a marked increase in reactivity with mAb H5 as cells approach stationary phase (39). The timing of this increase coincides with the beginning of the "diauxic shift" that occurs upon depletion of glucose and involves major reprogramming of the pattern of gene expression (48). This transient increase in mAb H5 reactivity is not seen in ctk1⌬ cells when they reach the same point in the growth curve (Fig. 2B). However, CTK1 deletion does not entirely eliminate Ser 2 phosphorylation; some reactivity with mAb H5 remains, and this reactivity declines in a similar fashion as both wild-type and mutant cells enter stationary phase.
As shown earlier in Fig. 1, there is a severalfold increase in reactivity with mAb H14 at time points prior to the beginning of stationary phase. This increase is not due to changes in the concentration of Rpb1p as can be seen from immunoreactivity with antibody raised against a peptide corresponding to the amino-terminal nine amino acids of Rpb1p (Anti-JC20). This antibody detects approximately the same amount of Rpb1p in CTK1 and ctk1⌬ cells at similar points in the growth curve. In both strains the total amount of pol II subunit declines as cells reach stationary phase, an observation consistent with the known reduction in overall transcription as cells approach stationary phase (49). With both anti-JC20 and 8WG16 we observe an increase in the slower mobility form of Rpb1p in ctk1⌬ cells. This is most obvious for 8WG16 where the reactive band in ctk1⌬ cells is well above the 200-kDa marker, whereas that in the CTK1 cells is even with the marker. Antibody against heat shock protein Ssa1p was used to show that equal amounts Immunostaining with anti-NAB3 demonstrated that equal amounts of sample were loaded in each lane. The position of the IIa and II0 markers were determined by reprobing the blots with each of the other antibodies. B, yeast extract (100 g) prepared as above was compared with an equivalent amount of protein solubilized from the pellet with SDS sample buffer after cells were lysed with glass beads and centrifuged to remove insoluble material ("Experimental Procedures"). Antibodies and Western blotting was a described under "Experimental Procedures." C, yeast extracts (100 g) made from CTK1 and ctk1⌬ strains transformed with plasmid pRS415 (lanes 1 and 2) and pRSCTK1 (lane 3) were subject to Western blot analysis and probed with antibodies shown on the left. of protein were extracted from cells at various points in the growth curve.
CTK1-dependent Expression of Diauxic Phase Genes-The absence of inducible H5-reactive CTD epitope during diauxic shift in ctk1⌬ cells suggests that this form of pol II may be involved in regulating genes that are induced in this phase of the growth cycle. We have examined expression of glycogen synthase encoded by the GSY2 gene (50) and cytosolic catalase encoded by the CTT1 gene (51). Both of these genes have been shown to be induced during the diauxic shift (48). In Fig. 3 we show that in wild-type cells, expression of GSY2 and CTT1 is induced early in the diauxic shift, decreases, and then gradually increases as cells reach stationary phase. In contrast, in ctk1⌬ cells expression of neither GSY2 nor CTT1 is induced but rather gradually increases during growth with maximum expression reached after the diauxic shift. This maximum level is markedly lower than that seen in CTK1 cells during the diauxic shift.
ctk1⌬ Effects on CTD Phosphorylation and Heat Shock Gene Expression-As we have previously shown, heat shock leads to an increase in the mAb H5-reactive CTD epitope (39). In Fig.  4A we show that this same increase in mAb H5 reactivity is maintained in the ctk1⌬ background. We also observe that induction of heat shock proteins Ssa3p and Ssa4p occurs in both the mutant and wild-type backgrounds. Together, these results indicate that CTDK-I is not required for the heat shock response. Interestingly, some expression of Ssa3p/Ssa4p is observed even in the absence of heat shock (Fig. 4A), and Northern blot analysis (Fig. 4B) confirms that SSA3/SSA4 transcripts are observed during growth at 30°C only in the ctk1⌬ background. Thus, CTDK-I would appear to be involved in repression of heat shock gene expression under normal growth conditions. The effect of ctk1⌬ on SSA4 expression is specific; ACT1 expression is unchanged in the ctk1⌬ background, whereas expression of ENO1, which has previously been shown to be sensitive to CTD truncation (52), is reduced. DISCUSSION The pattern of CTD phosphorylation is a product of combined action of both CTD kinases and CTD phosphatases. Deleting a CTD kinase gene would be expected to upset the dynamic balance between phosphorylation and dephosphorylation and lead to changes in the CTD phosphorylation pattern. In the present study we have used anti-phospho-CTD monoclonal antibodies to show that deletion of CTK1 changes the pattern but does not eliminate phosphorylation of the yeast pol II CTD. This result is consistent with the existence of multiple CTD kinases. The changes we observe in the ctk1⌬ background are in part dependent on growth state and suggest diverse roles for CTDK-I in regulation of CTD phosphorylation.
CTDK-I and CTD Phosphorylation during Logarithmic Growth-The most unexpected result presented here is that the phosphorylation state of the CTD in logarithmically growing cells is increased in the ctk1⌬ background. This result suggests that CTDK-I plays a role in negatively regulating CTD phosphorylation during log phase growth. The log phase increase in CTD phosphorylation is specific for Ser 5 with little change observed in Ser 2 phosphorylation. Both Cdk7 (Kin28) and Cdk8 (Srb 10) kinases have been shown to preferentially phosphorylate Ser 5 (20,(53)(54)(55)(56). One possibility is that CTDK-I may negatively regulate one or both of these kinases, perhaps through phosphorylation. An alternative explanation for the increase in Ser 5 phosphorylation in ctk1⌬ cells is that CTDK-I may positively regulate a CTD phosphatase. Fcp1 is the only known CTD phosphatase (58,59), but it is not known if its activity is controlled by phosphorylation. In addition, we do not know whether Fcp1 phosphatase is selective for either phosphorylated Ser 2 , Ser 5 , or both.
CTK1 deletion could also lead to an increase in Ser 5 phosphorylation by a mechanism involving CTD phosphorylation. For example, if CTDK-I were to phosphorylate the CTD in such a fashion that the phosphoacceptors for other CTD kinases were blocked, then we would observe an increase in CTD phosphorylation upon CTK1 deletion. In this scenario CTDK-I would have to phosphorylate the CTD to low density, whereas CTD kinases that function in the absence of CTDK-I would need to phosphorylate the CTD to high density.
CTD Phosphorylation and Growth-We have previously shown that phosphorylation of Ser 2 transiently increases late in logarithmic growth (39). The timing of this change in CTD phosphorylation corresponds to the beginning of the diauxic shift that occurs when cells growing in glucose-based medium deplete the glucose and shift from fermentation to respiratory metabolism (60). In the present study we show that this increase in Ser 2 phosphorylation does not occur in ctk1⌬ cells. The most straightforward explanation is that CTDK-I is responsible for phosphorylating Ser 2 during the diauxic shift. We cannot, however, rule out less direct mechanisms in which CTDK-I may positively regulate an Ser 2 -specific kinase or negatively regulate an Ser 2 -specific phosphatase. Clearly, CTDK-I is not the only kinase capable of phosphorylating Ser 2 as some reactivity with mAb H5 remains in the ctk1⌬ strain, and this reactivity increases during heat shock.
The diauxic shift is accompanied by widespread change in expression of genes involved in carbon metabolism, protein synthesis, and carbohydrate storage (48,60). Several classes of genes co-regulated during the diauxic shift were identified in a DNA microarray survey (48). One particularly interesting group displays an average 10-fold increase in message levels during early diauxic shift. This class includes the predominantly expressed glycogen synthase gene GSY2 (50) and the cytosolic catalase gene CTT1 (51). In this paper we show that the expression pattern for GSY2 and CTT1 in wild-type cells correlates with the expression of H5-reactive epitope. Maximum expression is observed during early diauxic shift, and steady-state levels of RNA decline as cells approach stationary phase. In contrast, in ctk1⌬ cells GSY2 and CTT1 expression is not induced, and only modest increases are observed well after the diauxic shift. This is the first example of a gene that is dependent on CTDK-I for correct regulation. The similar timing of appearance of the H5-reactive epitope and the induction of GSY2 and CTT1 mRNA accumulation further suggests a possible role for Ser 2 phosphorylation in the increase in steadystate mRNA levels observed during the diauxic shift.
A Gene-specific Role for CTDK-I-The results presented here suggest that CTDK-I is not a general elongation factor but rather functions in a gene-specific fashion. We have identified a class of yeast genes expressed in late log phase as potential targets of regulation. The absence of diauxic phase Ser 2 phosphorylation and the coincident failure to increase expression of GSY2 and CTT1 suggest that expression of these genes is controlled through specific changes in phosphorylation of the CTD by CTDK-I. This is an apparently specific function of CTDK-I as heat shock-induced expression, which is also accompanied by an increase in Ser 2 phosphorylation, is not affected by CTK1 deletion (Fig. 4).
Our observations together with published observations about the role of CTDK-I in elongation favor a model in which Ser 2 phosphorylation regulates transcription elongation on specific genes. In this model CTDK-I is attracted to specific transcription complexes in response to regulatory signals. Phosphorylation of the CTD at Ser 2 by CTDK-I either at the time of promoter clearance or later in elongation establishes an efficient transcription elongation complex leading to accumulation of mRNA.
Involvement of CTDK-I could be triggered by factors bound at the promoter or by cis elements present in the transcribed regions of responsive genes. The promoters of GSY2 and CTT1 share several cis-acting promoter elements including the stress response element (61). If factors that bind to the promoter in response to impending glucose depletion can attract CTDK-I, the ensuing CTD phosphorylation could allow for more efficient clearance from the promoter. This may involve breaking contacts between the CTD and components of the preinitiation complex or could be due to the establishment of contacts between the newly phosphorylated CTD and components of the elongation complex. Alternatively, genes regulated by CTDK-I at the diauxic shift may share common cis-acting RNA elements similar to the human immunodeficiency virus transactivation-responsive region RNA (TAR) site. TAR binds the activator Tat which in turn recruits P-TEFb which is thought to phosphorylate the CTD leading to productive elongation (30,(62)(63)(64). Whether such cis-acting elements are present in yeast diauxic-specific genes is not known.
The results presented here show that CTDK-I plays both positive and negative roles in CTD phosphorylation. In the absence of CTDK-I activity the phosphorylation of Ser 5 during logarithmic growth increases, but the diauxic phase mAb H5reactive species is reduced. The coincident lack of accumulation of GSY2 and CTT1 mRNAs at the diauxic shift suggests that these genes are regulated by the H5-reactive form of pol II. The mechanism of this regulatory process is currently under study.