Phosphorylation of the RNA Polymerase II Carboxy-Terminal Domain by the Bur1 Cyclin-Dependent Kinase

Strains and growth conditions.

The S. cerevisiae strains used in this study are listed in Table 1. All media, including rich yeast-peptone-dextrose (YPD) synthetic complete (SC) drop-out medium (e.g., SC-Ura) and minimal and sporulation media, were made as described previously (46). 6-Azauracil (6AU) plates contained SC-Ura drop-out mix and 50 μM 6AU. Standard genetic methods for mating, sporulation, and tetrad analysis (46) were used throughout.

Plasmids.

pGP112 contains a 3.3-kb Sau3A-EcoRI Bur1 fragment in pRS426. pGP211 is identical to pGP112, except it contains bur1-3, a D213A mutant allele created by oligonucleotide-directed mutagenesis. pSM21 contains BUR1 FLAG tagged at the portion corresponding to the N terminus in a pRS426 plasmid background. pSM14 contains FLAG–bur1-3 in a pRS426 plasmid background. The RPB1⁺ and rpb1 truncation plasmid series has been described in Nonet et al. (36). pRU8 contains both FLAG-BUR1 and His₆-BUR2 in pRS426, and pRU9 contains FLAG–bur1-3 and His₆-BUR2 in pRS426.

Immunoprecipitation and Western blots.

Yeast transformants were grown to an A₆₀₀ of 1.0 in selective SC drop-out media. Extracts were made by bead beating in lysis buffer containing 450 mM NaCl essentially as described previously (2). All protein manipulations were carried out at 4°C in the presence of protease inhibitors (aprotinin [100 μg/ml], pepstatin A [70 μg/ml], leupeptin [50 μg/ml], and phenylmethylsulfonyl fluoride [1 mM]). For some experiments, 10 mM NaF, 20 nM okadaic acid, and 1 mM EGTA were added to inhibit phosphatase activity. Immunoprecipitations were performed essentially as described previously (2), except 1 mg of protein extract in lysis buffer was incubated with 50 μl of an M2 anti-FLAG antibody-bead conjugate slurry (Sigma) in a final volume of 1 ml. We have also used a HEPES-acetate immunoprecipitate-/wash buffer (50 mM HEPES, 150 mM potassium acetate, 1 mM magnesium acetate, 1 mM EDTA, 0.05% Tween, 1 mM dithiothreitol) to eliminate background bands for some experiments as described in the figure legends. Following immunoprecipitation, beads were either resuspended in lysis buffer for subsequent kinase assays or in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer. Western analysis of immunoprecipitated samples was performed as described previously (42). Primary antibodies used were either M2 anti-FLAG (Sigma) or anti-Rpb1 (antibody E2, which was raised against the second exon of Drosophila Rpb1, was a gift of A. Greenleaf, Duke University).

Immunoprecipitation-kinase assay.

A small aliquot (10 μl) of immunoprecipitated sample was added to an equal volume of kinase buffer (25 mM Tris-HCl [pH 7.5], 0.5 mM EDTA, 1 mM dithiothreitol, 10 mM MgCl₂), and the reaction was initiated by the addition of 1 μCi of [γ-³²P]ATP. The reaction mixtures were incubated at 30°C for 30 min, after which they were stopped by the addition of SDS-PAGE sample buffer. Products were resolved by SDS-PAGE and visualized by autoradiography. For the kinase assay using unlabeled ATP, ATP was added to a final concentration of 2.5 mM.

CTD kinase assay.

FLAG-Bur1 protein was immunoprecipitated as described above. A small aliquot (10 μl) of immunoprecipitate was added to 34 μl of kinase buffer (25 mM Tris-HCl [pH 7.8], 10 mM MgCl₂, 0.1% Tween 20, 0.1% nonfat milk) containing 5 μl of β-galactosidase (Gal)–CTD substrate (kind gift of A. Greenleaf, Duke University); the reaction was initiated by the addition of 1 μCi of [γ-³²P]ATP, and the mixture was incubated at 30°C for 30 min. Reactions were stopped by the addition of SDS-PAGE sample buffer. Products were resolved by SDS-PAGE and visualized by autoradiography.

Rpb1 coimmunoprecipitates with, and is phosphorylated by, Bur1 in vitro.

BUR1 and BUR2 encode a Cdk-cyclin protein kinase complex proposed to have a general role in Pol II-dependent transcription. To understand the function of this complex and identify potential substrates in as unbiased a manner as possible, we established a Bur1- and Bur2-dependent immunoprecipitation-kinase assay. Briefly, Bur1 was tagged at its amino terminus with the FLAG epitope and expressed in yeast on a high-copy-number plasmid from its own promoter. Neither the FLAG epitope nor the overexpression interfered with BUR1 function, since FLAG-BUR1 complemented bur1 mutations, including bur1Δ, as efficiently as did BUR1⁺, and overexpression of BUR1⁺ or FLAG-BUR1⁺ did not cause any mutant phenotypes. Extracts were prepared, FLAG-Bur1 was immunoprecipitated with FLAG-specific M2 monoclonal antibody beads, and kinase activity towards coimmunoprecipitated proteins was assayed by incubation with [γ-³²P]ATP and subsequent gel electrophoresis and autoradiography. Two controls were utilized to determine whether the phosphorylated products were Bur1 dependent: extracts were prepared from a strain that expressed untagged BUR1 and from a strain that expressed FLAG–bur1-3, an allele that is predicted to be catalytically inactive, based on analogous mutations that inactivate other protein kinases (14).

Using this assay, two Bur1-specific substrates were observed, with molecular masses of ∼80 and ∼210 kDa (Fig. 1B); generation of these phosphorylated products required the presence of both the FLAG epitope and the active Bur1. Western blots of the immunoprecipitated material demonstrated that FLAG-BUR1 and FLAG–bur1-3 were expressed and immunoprecipitated to equivalent levels (Fig. 1A); reduced phosphorylation of the ∼80- and ∼210-kDa proteins in the bur1-3 lane (Fig. 1B, compare lanes 2 and 3) is therefore due to loss of kinase activity and not simply to a physical absence of the kinase. Phosphorylation of both substrates was also dependent upon the Bur2 cyclin (58). Because Rpb1 migrates at ∼210 kDa in SDS-polyacrylamide gels, similar to our larger candidate substrate, we first determined whether the ∼210-kDa substrate was Rpb1. FLAG-Bur1 was immunoprecipitated with anti-FLAG agarose beads, and the immunoprecipitated material was probed in Western blots using an Rpb1-specific antibody. The result (Fig. 1C) demonstrated that Rpb1 coimmunoprecipitates with FLAG-Bur1 and that coimmunoprecipitation requires the FLAG epitope, indicating that it is not due to nonspecific binding of Rpb1 to the antibody beads. Coprecipitation of Bur1 and Rpb1 does not require Bur1 kinase activity, as Rpb1 coimmunoprecipitates with the catalytically inactive Bur1-3. Bur1 and Rpb1 coimmunoprecipitated even under the stringent conditions of 0.6 M NaCl (data not shown), indicating that this interaction was highly specific.

The preceding experiment demonstrates that Rpb1 coimmunoprecipitates with Bur1. To determine whether the ∼210-kDa protein phosphorylated in the in vitro kinase assay was Rpb1, immunoprecipitation-kinase assays were performed using extracts prepared from a strain that contained an rpb1 allele (rpb1Δ103) in which the CTD is truncated from 26 to repeats (36). Strains carrying the rpb1Δ103 CTD truncation are viable, and the Rpb1Δ103 protein migrates faster than Rpb1⁺ in SDS-polyacrylamide gels. We therefore expressed Bur1 and FLAG-Bur1 in Rpb1⁺ and rpb1Δ103 strains. When the immunoprecipitation-kinase assay was performed using extracts prepared from an rpb1Δ103 strain, the ∼210-kDa phosphorylated protein was not observed; instead, a protein with slightly greater mobility that comigrates with Rpb1Δ103 was phosphorylated (Fig. 2A), demonstrating that the ∼210-kDa substrate is Rpb1. The mobility of the ∼80-kDa substrate was unchanged in the RPB1⁺ and rpb1Δ103 lanes (data not shown). Combined, these results demonstrate that Rpb1 coimmunoprecipitates with Bur1 and is a substrate for Bur1 in vitro.

Bur1 phosphorylates a CTD fusion protein in vitro.

The immunoprecipitation-kinase assays described above demonstrate that Rpb1 is phosphorylated by Bur1 in vitro. To determine whether Bur1 phosphorylates Rpb1 within the CTD, we utilized a β-Gal–CTD fusion that has previously been used to investigate CTD phosphorylation (29). When FLAG-Bur1 and FLAG–Bur1-3 were immunoprecipitated and assayed for activity on this model CTD substrate, phosphorylation of β-Gal–CTD was observed using FLAG-Bur1 (Fig. 2B). Phosphorylation required the CTD (Fig. 2B, compare lanes 2 and 5) and epitope-tagged Bur1 (Fig. 2B, compare lanes 4 and 5) and was greatly reduced using the catalytically impaired FLAG–Bur1-3 (Fig. 2B, compare lanes 5 and 6), indicating that it was due to Bur1 activity and not coimmunoprecipitation of another CTD kinase.

Phosphorylation site specificity.

Rpb1 is differentially phosphorylated in vivo, primarily on serine 2 and serine 5 of the YSPTSPS CTD consensus repeat. To determine whether Bur1 associates with a specific phosphorylated subset of Rpb1, FLAG-Bur1 was immunoprecipitated with anti-FLAG beads and the immunoprecipitated material was probed with monoclonal antibodies that recognize unphosphorylated CTD repeats (antibody 8WG16), CTD phosphorylated on serine 2 (antibody H5), and CTD phosphorylated on serine 5 (antibody H14). The Rpb1 population that coimmunoprecipitated with FLAG-Bur1 and FLAG–Bur1-3 cross-reacted with 8WG16 but not with H14 or H5 (Fig. 3, lanes 3 and 4), suggesting that Bur1 associates primarily with Rpb1-containing unphosphorylated CTD repeats. The H5 and H14 antibodies readily recognized Rpb1 in the crude extract (Fig. 3, lanes 1 and 2), indicating that the lack of a signal in the immunoprecipitates was not simply due to the absence of serine 2- and serine 5-phosphorylated forms in the crude extract.

To determine whether Bur1 phosphorylates the CTD on serine 2 or serine 5 of the consensus repeat, FLAG-Bur1 and its associated proteins were immunoprecipitated, kinase assays were performed using nonradioactive ATP, and the reaction products were then probed with the phosphorylation state-specific antibodies. Although the coimmunoprecipitated Rpb1 was not recognized by H14, after the kinase assay, H14 reactivity was readily detected (Fig. 3, compare lanes 3 and 5). H14 reactivity was dependent upon Bur1 activity, since it was not observed in reaction mixtures containing the inactive FLAG–Bur1-3 (Fig. 3, compare lanes 5 and 6). Bur1-dependent phosphorylation in vitro reduced reactivity with 8WG16 (Fig. 3, compare lanes 5 and 6), as expected, since 8WG16 recognizes only the unphosphorylated repeat. No reactivity with H5 was observed either before or after the kinase reaction, indicating that Bur1 neither associates with nor phosphorylates serine 2 of the CTD repeat to any detectable level. Combined, these results suggest that Bur1 primarily associates with Rpb1 containing unphosphorylated CTD repeats and then phosphorylates Rpb1 on serine 5.

Genetic links between BUR1 and CTD function.

The results described above demonstrate that Rpb1 coimmunoprecipitates with, and is a substrate for, Bur1 in vitro. To determine whether these activities are physiologically relevant, we performed the following genetic tests examining whether the functions of Bur1 and the Rpb1 CTD overlap in vivo. First, bur1 mutations were crossed with rpb1 CTD truncation mutants to allow examination of the double-mutant phenotype. The bur1-2 allele was chosen for the following tests because it causes several clear phenotypes yet does not cause a severe growth defect by itself that might complicate interpretation of double-mutant phenotypes. The bur1-2 rpb1Δ103 double mutants grew extremely poorly compared to the individual mutants (Fig. 4A), suggesting that BUR1 and the Rpb1 CTD participate in the same process. Second, when the catalytically impaired bur1-3 allele is overexpressed from a high-copy-number plasmid, it causes increased transcription from the lys2-128δ promoter insertion allele, resulting in lysine-independent growth (Fig. 4B). This dominant negative Spt⁻ phenotype (57) is suppressed by rpb1 CTD truncations (Fig. 4B), indicating that the bur1-3 dominant negative Spt⁻ phenotype requires an intact CTD. Third, if Bur1 is a physiological CTD kinase, then we would predict genetic interactions between bur1 mutations and mutations in genes that encode other CTD kinases or phosphatases. We therefore crossed bur1-2 with deletions of SRB10 and CTK1, two temperature-sensitive alleles of KIN28, and an fcp1 allele (10). FCP1 encodes an essential conserved CTD phosphatase (1, 5) that is broadly required for transcription in vivo (25) and stimulates both initiation and elongation by Pol II under specific assay conditions in vitro (7). Interestingly, bur1-2 was lethal in combination with ctk1Δ, but no effects were observed in combination with the two kin28 temperature-sensitive alleles or with srb10Δ (Table 2). These results link the functions of BUR1 and CTK1 and distinguish these two kinases functionally from KIN28 and SRB10. We also detected a genetic interaction between bur1 and fcp1; the bur1-2 fcp1-110 double mutants grew extremely slowly and were very tightly Ts⁻ and Ino⁻, unlike either of the single mutants (Fig. 4C). To determine the extent of functional overlap with BUR1, we also tested whether kin28, srb10Δ or ctk1Δ mutations cause a Bur⁻ phenotype. In agreement with the synthetic lethality results that suggest partial redundancy between CTK1 and BUR1, ctk1Δ is weakly Bur⁻, while srb10Δ and kin28 temperature-sensitive alleles are Bur⁺ (data not shown). In summary, we have documented genetic interactions between Bur1, the Rpb1 CTD, another CTD kinase, and a CTD phosphatase that strongly link Bur1 and CTD functions in vivo.

Genetic evidence implicating Ctk1 and Bur1 in elongation.

The synthetic lethality observed between bur1-2 and ctk1Δ mutations suggested that Bur1 and Ctk1 have overlapping functions. Although the role of CTK1 in vivo is not completely understood, Ctk1 has been shown to stimulate transcription elongation in vitro (30). To determine whether Ctk1 and Bur1 might function during elongation in vivo, we examined their response to 6AU. Sensitivity to 6AU is a strong indicator of a role during elongation, as mutations in the genes that encode the elongation factors TFIIS, the Spt4-Spt5 complex (DSIF), and the elongation-defective forms of Pol II all are 6AU sensitive (45). The bur1-2, bur2-1, and ctk1Δ alleles each conferred sensitivity to 6AU (Fig. 5A), similar to the deletion alleles of PPR2 (TFIIS) or SPT4 (DSIF subunit), suggesting that they participate in elongation in vivo. In contrast, mutations in KIN28 or SRB10, which each encode CTD kinases, are 6AU resistant (data not shown).

To further test whether Bur1 has a role during elongation in vivo, bur1-2 mutants were crossed with strains containing mutations in genes encoding other known elongation factors. Several combinatorial effects were observed. In particular, bur1-2 was lethal in combination with spt4Δ and spt5-4 (Table 2), indicating that in the presence of a defective Spt4-Spt5 complex (DSIF), normal BUR1 function is essential for viability. Synthetic growth defects were also observed in combination with spt6 and ppr2 mutations, and bur1-2 ppr2Δ mutants were strongly temperature sensitive, unlike either of the single mutants (Fig. 5B). The bur1 synthetic double-mutant phenotypes were specific for elongation factors, since no combinatorial defects were observed when bur1-2 was combined with mutations in the genes encoding the Srb4 subunit of the Pol II holoenzyme, the Srb10 and Kin28 CTD kinases, or deletion of one of the histone H2A-H2B loci (Table 2). An enhancement of the bur1-2 growth defect and other bur1 mutant phenotypes was also observed in combination with the elongation-deficient rpb2-7 and rpb2-10 alleles. These effects were not very strong, however, and might simply be due to additive effects between these bur1 and rpb2 alleles. Interestingly, bur1-2 also showed no synthetic phenotype when combined with a mutation in the gene encoding the Spt16 subunit of the FACT complex, which is also involved in elongation. Although this result stands in contrast to the interactions of bur1-2 with the SPT4, SPT5, SPT6, and PPR2 genes, it is consistent with recent observations that FACT and DSIF likely have different, and perhaps even antagonistic, modes of action (37, 54).

Is Bur1 or Ctk1 a functional homolog of P-TEFb?

P-TEFb is a cyclin-dependent protein kinase that stimulates elongation in Drosophila and mammalian cell extracts (44). Bur1 and Ctk1 were the best candidates for yeast P-TEFb catalytic subunits, as they were originally cited as being equally similar to Cdk9, the P-TEFb catalytic subunit in metazoans (59). A more recent phylogenetic sequence comparison (32) using the recently completed Drosophila, Caenorhabditis elegans, and available human genomic sequences indicates that Bur1 is the likely yeast ortholog of Cdk9, whereas Ctk1 is more closely related to a human kinase, CHED (15). Four results described here now provide experimental support for this idea, extending the similarities between Bur1 and P-TEFb beyond sequence alignments, into the functional level. First, like P-TEFb, Bur1 is a cyclin-dependent kinase that phosphorylates the Rpb1 CTD. Second, the 6AU sensitivity of bur1 mutations and synthetic lethality with ppr2Δ (TFIIS) mutations suggest that, like P-TEFb, Bur1 is also involved in elongation. Third, P-TEFb interacts biochemically with DSIF (17, 55), composed of Spt4 and Spt5 subunits, while bur1 mutations interact genetically with spt4 and spt5 mutations (Table 2) (G. Hartzog, unpublished observations). Fourth, like P-TEFb (39, 41), Bur1 physically associates with Pol II. Several of these characteristics are also shared with Ctk1. Thus, the functional similarities between Bur1, Ctk1, and P-TEFb, combined with the identification of CHED in humans, force us to consider the intriguing possibility that multiple P-TEFb-like CTD kinases might be involved in elongation in yeast and in other organisms.