CDKs (cyclin-dependent kinases) ensure directionality and fidelity of the eukaryotic cell division cycle. In a similar fashion, the transcription cycle is governed by a conserved subfamily of CDKs that phosphorylate Pol II (RNA polymerase II) and other substrates. A genetic model organism, the fission yeast Schizosaccharomyces pombe, has yielded robust models of cell-cycle control, applicable to higher eukaryotes. From a similar approach combining classical and chemical genetics, fundamental principles of transcriptional regulation by CDKs are now emerging. In the present paper, we review the current knowledge of each transcriptional CDK with respect to its substrate specificity, function in transcription and effects on chromatin modifications, highlighting the important roles of CDKs in ensuring quantity and quality control over gene expression in eukaryotes.
A conserved network of CDKs (cyclin-dependent kinases) controls cell division and gene expression from yeast to metazoans. Protein phosphorylation by CDKs plays essential roles in the transcription cycle of Pol II (RNA polymerase II). In eukaryotes, multiple CDKs are conserved components of the Pol II transcription machinery: the Cdk7–cyclin H–Mat1 complex, part of the general transcription initiation factor TFIIH (transcription factor IIH); Cdk8–cyclin C, which associates with the Mediator; Cdk9–cyclin T, also known as P-TEFb (positive transcription elongation factor b); and Cdk12 and Cdk13, which form complexes with cyclin K and also participate in elongation [1–4].
Progression of Pol II transcription correlates with changes in phosphorylation state of the CTD (C-terminal domain) of the Pol II large subunit Rpb1, a common target of transcriptional CDKs with distinct, but overlapping, specificities in vitro . The CTD consists of tandem heptad repeats of consensus sequence Y1S2P3T4S5P6S7 that, when phosphorylated, generate binding sites for transcription regulators and RNA-processing and chromatin-modifying enzymes. Studies in yeast by ChIP suggested a ‘CTD cycle’ of dynamic, reversible phosphorylation that marks progression through initiation, elongation and termination. Around the TSS (transcription start site), Ser5 phosphorylation predominates over Ser2 phosphorylation, coincident with formation of the PIC (pre-initiation complex) and early elongation. The relative abundance of Ser2 phosphorylation increases towards the 3′-end of the gene body. Ser7 phosphorylation roughly coincides with Ser5 phosphorylation and may be especially important for expression of small nuclear RNAs [6,7]. Thr4 phosphorylation is the most recently characterized phosphorylation of the CTD. It facilitates histone mRNA 3′-end processing, and requires Cdk9 in higher eukaryotes . In accordance with a role for Thr4 phosphorylation in transcription termination, it peaks towards the 3′-ends of human genes, slightly downstream of Ser2 phosphorylation . Recruitment of different capping enzymes, chromatin modifiers and elongation factors depends on the phosphorylation state of Pol II . The Pol II CTD is not the only target of transcriptional CDKs; other important substrates have been identified, through which CDKs may control other functions in transcription and RNA processing.
Because of their overlapping and interdependent functions in transcription, assigning precise roles to individual CDKs has been difficult. Chemical genetics provided a solution to this problem: the engineering of inhibitor-sensitivity into individual kinases by expansion of the ATP-binding pocket to accommodate bulky adenine analogues that bind poorly to wild-type kinases . Our laboratory and others have introduced such as (analogue-sensitive) mutant CDKs into both human and fission yeast (Schizosaccharomyces pombe) models, to dissect specific functions of individual CDKs involved in the Pol II transcription cycle (Figure 1).
Transcriptional CDKs in fission yeast
CDKs that act in transcription
TFIIH is an essential component of the PIC formed at the promoters of genes transcribed by Pol II. In humans, Cdk7 regulates pausing of Pol II in the promoter-proximal region; the inhibition of its kinase activity prevents recruitment of DSIF [DRB (5,6-dichloro-1-β-D-ribofuranosylbenzimidazole)-sensitivity-inducing factor], a conserved heterodimer of Spt4 and Spt5 that binds to Pol II, and NELF (negative elongation factor) [12,13]. Cdk7 inhibition also causes retention of the general transcription factor TFIIE, suggesting a mechanism whereby Cdk7 promotes exchange of TFIIE for DSIF to facilitate pausing and ensure an orderly transition from initiation to elongation, possibly regulated by phosphorylation of Spt5 and both α and β subunits of TFIIE [13–15].
In metazoans, Cdk7 is also the CAK (CDK-activating kinase), which is needed to activate CDKs involved in cell-cycle progression [16–18]. We have shown recently that Cdk7 also phosphorylates Cdk9 to promote its activation on chromatin and, presumably, the release of paused Pol II into productive elongation . In S. pombe, the orthologue of the Cdk7–cyclin H–Mat1 complex is the Mcs6–Mcs2–Pmh1 complex. Mcs6 and its orthologues [Kin28 in budding yeast (Saccharomyces cerevisiae) and Cdk7 in humans] phosphorylate the Pol II CTD preferentially on Ser5 and Ser7in vitro [12,19–21]. Phosphorylation of the CTD by Mcs6, probably on Ser7, primes subsequent phosphorylation by Cdk9 [19,22]; a similar priming mechanism may link metazoan Cdk7 and Cdk9 functions . Finally, recruitment of the Cdk9 orthologue to transcribed genes depends on TFIIH activity in both budding and fission yeast [22,24], but not in human cells .
P-TEFb was identified as a complex that stimulates transcription elongation in vitro, where it antagonizes pausing enforced by DSIF and NELF . In eukaryotes, Cdk9 phosphorylates the CTR (C-terminal repeat) region of the DSIF subunit Spt5 [22,25–28]. Yeast do not have NELF (which is also a substrate of human Cdk9), and the lack of this factor may explain the absence of Pol II promoter-proximal pausing . One of the roles for pausing in metazoans is to facilitate the recruitment of pre-mRNA processing machinery before elongation [29,30]. Conservation of most of the pause establishment and release machinery suggests that yeast also link RNA processing to the initiation–elongation transition, despite the lack of a discernible pause. Consistent with this conjecture, S. pombe Cdk9 binds to Pcm1, the mRNA 5′-cap methyltransferase, in order to ensure the coupling of capping to transcript elongation [22,31,32]. Initially, Cdk9 was thought to be the major CTD Ser2 kinase, but in vitro, it phosphorylates both Ser5 and Ser2 with approximately equal efficiency . In vivo, Ser2 phosphorylation is only modestly decreased by inhibition of Cdk9 alone; selective inhibition of Lsk1, the orthologue of metazoan Cdk12/Cdk13, or combined inhibition of Mcs6 and Cdk9, is needed to abolish Ser2 phosphorylation . Both Cdk9 and Mcs6, are essential, as is Ser5 of the CTD. Tethering mRNA-capping enzymes to the CTD can circumvent the requirement for Ser5 for vegetative growth, however, implying that capping enzyme recruitment is the essential function of the Ser5 phosphorylation mark in fission yeast .
Until recently, the prevalent view was that the separate functions of fission yeast Cdk9 and Lsk1 (and of budding yeast Bur1 and Ctk1) were combined in metazoan Cdk9. However, the recent discovery of metazoan Cdk12 and Cdk13 as Ser2 kinases orthologous to budding yeast Ctk1 seems to refute this notion, and to align yeast and metazoan CDK networks more closely from an evolutionary standpoint [4,34,35]. Inhibition of Lsk1 or Ctk1 or their deletion severely affects Ser2 phosphorylation in vivo [22,36,37]. Whereas both Bur1 and Cdk9 are essential in their respective organisms, Ctk1 and Lsk1 are both dispensable for vegetative growth [31,38–41]. Lsk1 binds to genes just after the TSS and peaks at the 3′-ends of coding regions, in correlation with the distribution of Ser2 phosphorylation .
In eukaryotes, Cdk8, also known as Srb10 in yeast, together with cyclin C, Med12 and Med13, forms a kinase module that competes with Pol II for binding to the Mediator complex. In mammals, Cdk8 was reported to phosphorylate cyclin H and inhibit the ability of TFIIH to phosphorylate the CTD of Pol II, thereby impeding transcription initiation . This modification does not occur in yeast (or in Drosophila), however, and more recent studies have uncovered mechanisms whereby the Cdk8 subcomplex instead promotes TFIIH function, either by enhancing its recruitment or through allosteric mechanisms. An alternative inhibitory mechanism appears to function in yeast whereby Cdk8 itself phosphorylates the Pol II CTD before, and in opposition to, PIC assembly . In agreement with these data, the Cdk8 module has been proposed to work mainly as a negative regulator of transcription. However, it has also been shown to play an important positive role in activation of p53, Wnt/β-catenin, serum response and hypoxia transcriptional programmes in higher eukaryotes [43–45]. In S. pombe, the Mediator and Cdk8 module share a similar chromatin-binding pattern, even on highly transcribed genes. On the basis of these data, it was hypothesized that the kinase module and the core Mediator remain closely associated during transcription . Cdk8 has been shown to phosphorylate specific transcription factors to regulate their activity [47–49]. Mammalian Cdk8 has been reported to phosphorylate histone H3 on Ser10 (H3S10P), a modification that fluctuates during the cell cycle and peaks in mitosis [50,51]. In S. pombe, Cdk8 phosphorylates the transcription factor Fkh2 (forkhead 2) in a periodic manner that correlates with mitotic gene activation. That phosphorylation controls mitotic entry, which is delayed by inhibition of Cdk8 activity or mutation of phosphorylation sites within Fkh2 to non-phosphorylatable residues . These findings support the idea that the Mediator might couple mitotic commitment and the mitosis-specific transcription programme.
Not all CDKs involved in transcription phosphorylate the CTD of Pol II. The most recently identified transcriptional CDK in fission yeast, Cdk11, forms a complex with a cyclin called Lcp1, and regulates the association of the Cdk8 module with the Mediator through phosphorylation of the Mediator subunits Med24 and Med7. A Cdk11–Lcp1 complex was unable to phosphorylate the CTD of Pol II in vitro, and its ablation did not affect Ser5 and Ser2 phosphorylation levels in vivo .
Transcriptional CDKs and chromatin marks
Among transcriptional CDKs, Cdk9 is the one most clearly implicated in maintenance and regulation of covalent histone modification patterns in chromatin. Histone H2B mono-ubiquitylation (H2Bub1) is a conserved mark associated with transcribed chromatin, which is necessary for transcription termination of both polyadenylated and non-polyadenylated mRNAs [12,13,54,55]. H2Bub1 was nearly abolished by selective Cdk9 inhibition in fission yeast, and was also diminished upon mutations of Spt5 that deleted the entire CTD or changed all Thr1 positions (the site phosphorylated by Cdk9 within the Spt5 nonapeptide repeat) to alanine . H2Bub1 likewise depends on Bur1 and Spt5 in S. cerevisiae [57–59], and on P-TEFb in human cells . As in budding yeast and metazoans, inhibition of Cdk9 activity in fission yeast impaired recruitment of the PAF (polymerase-associated factor) complex, which facilitates H2Bub1 [27,28,56]. In S. pombe, the link between Cdk9 and H2Bub1 is a positive-feedback loop: mutation of the ubiquitin-acceptor residue of H2B impeded the recruitment of Cdk9 to chromatin and diminished phosphorylation of Spt5-Thr1 .
H2Bub1 is necessary for co-transcriptional generation of histone H3 Lys4 methylation (H3K4me) by the methyltransferase Set1 in metazoans . Indeed, selective inhibition of Cdk9 in cdk9as cells diminished not only H2Bub1, but also histone H3 Lys4 di- and tri-methylation (H3K4me2 and H3K4me3 respectively) and Set2-dependent histone H3 Lys36 trimethylation (H3K36me3). Dependent on H3K36me, the HDAC (histone deacetylase) complex Rpd3S is recruited to chromatin in budding yeast, binding to phosphorylated Ser2 and Ser5 in order to deacetylate histones H3 and H4. Rpd3S reduces transcriptional elongation, but also prevents cryptic transcription initiation in coding regions [61,62]. Treatment of mcs6as or lsk1as cells with the AS kinase-specific inhibitor 3-MB-PP1 [1-(t-butyl)-3-(3-methylbenzyl)-1H-pyrazolo[3,4-d]pyrimidin-4-amine] has little or no effect on H2Bub1, H3K4me or H3K36me3 . The requirements for Mcs6, Cdk9 or Lsk1 in placing other histone marks remain to be elucidated. In mammalian cells, the allele-specific inhibition of TFIIH in cdk7as/as cells produced the same decrease in H2Bub1 as inhibition of Cdk9 with the moderately selective small molecule flavopiridol; we hypothesized that this is due to the Cdk7-dependent activation of Cdk9 through T-loop phosphorylation .
As mentioned above, mammalian Cdk8 phosphorylates Ser10 of histone H3 and acts synergistically with the histone acetyltransferase Gcn5, enhancing acetylation of histone H3 Lys14 (H3K14Ac). Consistent with this model, Cdk8 knockdown causes a significant decrease in H3S10P and H3K14Ac marks . To date, this mechanism has not been tested in yeast.
Consequences of CDK function in transcription
Impairment of transcriptional CDK function has various effects on gene expression depending on the CDK, the organism and the type of mutant analysed. Inhibition of an AS version of the Cdk7 orthologue in budding yeast, Kin28, had no major effects on global transcript levels, but reduced 5′-capped mRNA species . In contrast, a TS (temperature-sensitive) mutant of Kin28 virtually abolished Pol II-dependent transcription , apparently by destabilization of the PIC . Both TS and AS alleles of mcs6 caused specific (and similar), rather than global, decreases in transcription, as well as induction of genes involved in stress responses [22,65]. Similarly in mice, knockout of Mat1 affected CTD phosphorylation, but gross defects in transcription were not detected . These studies, together with the fact that CTD phosphorylation is necessary for co-transcriptional recruitment of factors involved in RNA processing, suggest that TFIIH activity may be needed more for mRNA maturation than for transcription.
Similarly, allele-specific inhibition of Cdk9 in cdk9as fission yeast cells leads to repression of ~10–20% of transcripts. Our laboratory found extensive overlap between the transcripts affected by inhibition of Mcs6 or Cdk9. Moreover, the simultaneous inhibition of both kinases increases the amplitude of changes, but does not significantly expand the set of affected genes. Among the repressed genes were ones involved in cytokinesis and cell separation, which are periodically expressed under the control of Sep1 and Ace2 transcription factors, and this derangement of transcription is accompanied by septation and cell-separation defects . Interestingly, similar transcriptional and morphological phenotypes have been reported for different Mediator subunit mutants .
Intriguingly, loss or inhibition of Lsk1 does not grossly affect transcript levels in vegetative growing cells (, and M. Sansó and R.P. Fisher). Only the expression programme for mating and sporulation is compromised upon lsk1 deletion, owing to the lack of transcription of ste11, the master regulator of the sexual pathway in S. pombe. Accordingly, ablation of CTD Ser2 phosphorylation by mutation to alanine is not lethal, but blocks initiation of the sexual differentiation programme in response to nitrogen starvation . In S. cerevisiae, deletion of Ctk1 impairs 3′-end RNA processing, but does not cause defects in expression levels [37,68]. In contrast, Bur1 is required for transcriptional elongation, but is apparently dispensable for most Pol II CTD phosphorylation .
Finally, in S. pombe, both Cdk8 and Cdk11 have similar transcriptional effects: repression or induction of the same small subset of genes. Deletion of cdk8 had a more pronounced effect, but the double cdk8 cdk11 mutant does not accumulate expression defects, suggesting that both kinases work in the same pathway .
CDK regulation of gene expression: a matter of quantity or quality, or both?
The CTD of Pol II and the CTR of Spt5, when modified by phosphorylation, provide platforms to recruit nucleosome remodellers, histone modifiers, spliceosome subunits and termination factors to influence mRNA synthesis and maturation [10,70]. Both are prominent substrates of transcriptional CDKs, along with other proteins also implicated in proper co-ordination of transcription and RNA processing. Moreover, perturbations of CDK network function cause severe disruptions in genome-wide patterns of covalent histone modification. The changes in total transcript levels caused by allele-specific inhibition of essential transcriptional CDKs, or loss of non-essential ones, are often gene-specific. On the other hand, numerous reports implicate these kinases in genome-wide regulation of co-transcriptional events , mediated in part through Pol II- and Spt5-dependent recruitment of factors needed for mRNA maturation. An emerging view is that CDKs might mainly regulate the quality rather than the quantity of Pol II-dependent transcripts, although they might also govern the steady-state levels of specific sets of transcripts. However, even in the absence of large changes in mRNA levels upon CDK inactivation, inaccurate or sloppy processing of primary transcripts may have dramatic consequences for the cell. This class of molecular defect might not be detectable by conventional methods of RNA measurement, but the application of newer methods such as RNA ultra-sequencing, ChIP-Seq or GRO-Seq (global run-on sequencing) should help to elucidate the fundamental roles of CDKs in shaping the transcriptome.
The 7th International Fission Yeast Meeting: Pombe 2013: An Independent Meeting/EMBO Conference held at University College London, London, U.K., 24–29 June 2013. Organized and Edited by Jürg Bähler (University College London, U.K.) and Jacqueline Hayles (Cancer Research UK London Research Institute, U.K.).
DRB (5,6-dichloro-1-β-D-ribofuranosylbenzimidazole)-sensitivity-inducing factor
negative elongation factor
- Pol II
RNA polymerase II
positive transcription elongation factor b
transcription start site
We thank past and present members of the Fisher laboratory for helpful discussions. Work reviewed was supported in part by the National Institutes of Health [grant numbers GM056985 and GM076021 (to R.P.F.)].