Chemical inhibition of the TFIIH-associated kinase Cdk7/Kin28 does not impair global mRNA synthesis

Specific Inhibition of Kin28 Activity in Vivo.

The 1-NA-PP1 effectively decreased the growth of the strain harboring the analog-sensitive Kin28as allele while displaying no detectable effect on the isogenic WT strain [Fig. 1A and supporting information (SI) Fig. 6]. Our data are consistent with previous observations that this analog dramatically inhibits kinase activity in vivo when added at 50-fold molar excess over IC₅₀ (38). The 1-NA-PP1 was previously shown to inhibit kinase activity in vitro and decrease the bulk CTD phosphorylation in vivo (41); however, to determine whether the analog-inhibited Kin28as associated with highly active promoters, we measured the levels of CTD (Ser-5) phosphorylation at ACT1 and PDR5 by chromatin immunoprecipitation (ChIP). We find that Ser-5 phosphorylation was significantly diminished in the presence of 5 μM inhibitor (Fig. 1B). This inhibition was achieved within 20 min of adding 1-NA-PP1 and was sustained for at least 72 hours (data not shown). This analog is known to inactivate similarly engineered yeast kinases within 10 min (38, 40, 42). To compare the efficiency of kinase inhibition, we also performed ChIP experiments with Kin28ts3, a widely used temperature-sensitive allele. The results indicate that at the nonpermissive temperature, Ser-5 levels at PDR5 and ACT1 decrease in the Kin28ts3 strain; however, the inhibitory analog (1-NA-PP1) is far more effective at blocking Ser-5 phosphorylation by Kin28as (Fig. 1B). The residual Ser-5 phosphorylation in both Kin28ts3 and Kin28as strains may arise because of the action of other kinases (41) or because of incomplete inhibition. Moreover, the ability of 1-NA-PP1 to completely block Kin28as from phosphorylating a different target (Gal4) in vivo has been independently demonstrated (43). Thus, multiple independent lines of evidence show that 1-NA-PP1 is capable of rapidly and potently inhibiting Kin28as function in vivo.

Inhibition of Kin28 Does Not Inhibit Transcription.

Having demonstrated that 1-NA-PP1 is a potent inhibitor of Kin28as kinase activity even when it is associated with active promoters, we examined the consequences of kinase inhibition on transcription of specific genes as well as on global transcriptome profiles. First, we determined the effect of either chemical inhibition (Kin28as) or heat inactivation (Kin28ts3) on steady state expression of ACT1 and PDR5. Quantitative RT-PCR (qRT-PCR) measurements show a 5-fold decrease in mRNA levels of both genes in a Kin28ts3 strain after one hour at nonpermissive temperature (Fig. 2A). However, potent chemical inhibition of the kinase activity has no consequence on the transcript abundance of either gene (Fig. 2A).

To rule out indirect effects of the inhibitor on mRNA stability we tested the ability of the inhibited Kin28 strains to rapidly induce gene expression in response to extrinsic signals, like galactose and heat shock. The galactose-responsive expression of GAL1 was examined because this robust response is abolished in heat inactivated Kin28ts3 strain (35, 44, 45). Furthermore, as noted above, the inhibitor eliminates galactose-responsive phosphorylation of Gal4 by Kin28as (43). In agreement with previous reports, quantitative RT-PCR measurements show that GAL1 mRNA is not detected in heat-inactivated Kin28ts3 (Fig. 2B). In contrast, inhibition of Kin28as affects GAL1 induction only modestly (Fig. 2B). The modest decrease occurs at a relatively early stage in transcription because PCR probes centered at ≈234, 422, and 841 nucleotides downstream of the start codon show similar levels of transcript abundance.

To determine the consequences of chemical inhibition of Kin28 on global mRNA synthesis, we first measured total RNA content of cells treated for 2 h with 1-NA-PP1 (Fig. 3A). The effect of inhibitor on cellular growth is first apparent at this time interval (Fig. 1A). The subtle reduction in total RNA levels is not surprising, because the majority of cellular RNA consists of ribosomal RNA and is not synthesized by Pol II. However, contrary to global shutdown of transcription in thermally inactivated Kin28ts strains (35, 36), bulk polyA-bearing RNA showed only a modest decrease in abundance upon chemical inhibition of the kinase (Fig. 3A Center). The identity and abundance of the polyadenylated RNA was examined by using yeast arrays (Affymetrix, Santa Clara, CA). The levels of all mRNAs in the transcriptome are displayed in a box plot wherein each circle represents a specific gene transcript and the intensity level is indicative of transcript abundance. As shown in Fig. 3A Right, the global distribution of transcript abundance is not significantly altered upon treatment with the inhibitor. Moreover, consistent with previously reported modest decrease in Pol II occupancy in the coding regions of ADH1 and PMA1 (41), we find that the mRNA abundance of these two genes also is proportionately down-regulated upon Kin28 inhibition (P < 0.005). This correlation provides independent validation of the transcript levels reported by the array. Importantly, these results compellingly argue against a global defect in mRNA synthesis upon inhibition of Kin28 kinase activity.

Finally, we tested whether cells with significantly impaired Kin28 activity can mount the necessary global transcriptional response to survive a major environmental perturbation. The genome-wide heat-shock response of the Kin28as strains that were grown in the presence of 1-NA-PP1 for 2 h was examined (Fig. 3B). In contrast to the near-complete shutdown of gene expression reported for Kin28ts3 under heat shock (36) genome-wide mRNA synthesis is not impaired in the chemically inhibited Kin28as strain upon heat shock (Fig. 3B). This experiment eliminates the possibility that elevated temperature, in addition to kinase inhibition, is required for the global shutdown of transcription observed in the Kin28ts strain. Furthermore, the heat shock profile of the chemically inhibited Kin28as strain closely matched the gene expression signatures of WT strains that were subjected to heat shock (46, 47) (Fig. 3 C and D and SI Data Set 1).

In essence, chemical inhibition of Kin28 kinase activity resulted in a dramatic decrease of CTD phosphorylation but showed no apparent defect in transcription initiation, promoter clearance, transcript elongation, no global reduction in mRNA steady-state levels and no global defect in the induced gene expression.

Kin28 Inhibition Correlates with Reduction in Capped Transcripts.

CTD phosphorylation by Kin28 enhances 5′ capping of nascent transcripts (17, 21, 48). Similarly, we find that chemical inhibition of Kin28 kinase function leads to a severe reduction in 5′-capping of transcripts (Fig. 4). The capped transcripts from Kin28as cultures treated with the inhibitor (or DMSO) were immunoprecipitated with H20, a monoclonal antibody that binds the m⁷G and trimethylG caps (see SI Methods for details). The abundance of m⁷G capped ACT1 and most cellular transcripts decreases dramatically by 60 min after Kin28 inhibition. The residual capped transcripts in Fig. 4B are most likely the relatively long-lived snRNA that bear trimethyl caps. The reduction in 5′ capped transcripts further validates the robust inhibition of Kin28as by 1-NA-PP1.

Kin28 Inactivation and TFIIH Destabilization.

A possible explanation for the discrepancy between the transcriptional responses of a temperature-sensitive mutant and the analog-sensitive mutant of Kin28 is that upon heat inactivation, Kin28ts alleles are unable to retain protein interactions with other subunits of the TFIIH complex. The Tfb3 subunit of the Kin28-Ccl1-Tfb3 trimeric subcomplex is known to dissociate upon heat inactivation of Kin28ts (28, 49). The absence of this trimeric kinase subcomplex does not impair the ability of the residual seven-subunit core TFIIH complex to promote open complex formation and productive transcription in vitro (5, 28, 50). Similarly, inactivation of the Tfb3 homolog in Schizosaccharomyces pombe and Schwann cells does not impair global transcription in vivo (37, 51). By comparison, removal of the residual seven-subunit core complex results in cessation of transcription of most genes tested in vitro (5, 30). Inactivation of a helicase within this core subcomplex also leads to global inhibition of gene expression in vivo (52, 53).

To investigate whether the critical seven-subunit core complex remains associated with the PIC at the ACT1 and PDR5 promoters, we performed ChIP experiments using an antibody against an integral subunit of the core complex, the Rad3/XPD helicase. In parallel, the level of Pol II retained at both genes was also monitored by ChIP (α-Rpb3 mAb). In the Kin28as strain, Rad3 and Pol II levels at PDR5 or ACT1 were comparable irrespective of the addition of the inhibitor (Fig. 5A Top and SI Fig. 7). In contrast, in the Kin28ts3 strain, heat inactivation led to a dramatic reduction of Rad3 occupancy and a decrease of Pol II at the promoters of PDR5 (Fig. 5A Middle) as well as ACT1 (SI Fig. 7). Furthermore, Pol II was not detected in the coding region (ORF) of either gene (Fig. 5A and SI Fig. 7). To further confirm these observations, TFIIH retention was examined in strains harboring Kin28ts16, another commonly used temperature-sensitive allele of Kin28 (Fig. 5B). In this case, Kin28 is epitope-tagged (HA) and ChIP experiments were performed by using the α-HA monoclonal antibody as well as the commercially available α-Rad3 antibody. Consistent with results of Fig. 5A, upon heat inactivation, both Rad3 and Kin28 dissociated from the PDR5 promoter (Fig. 5B) as well as from the ACT1 promoter (SI Fig. 7B).

We also performed the reciprocal experiment using a Rad3ts strain. At nonpermissive temperatures the Rad3ts mutant leads to disassembly of the TFIIH complex in vitro and rapidly eliminates global transcription in vivo (52, 54). Using ChIP analysis, we find that Rad3ts dissociates from both PDR5 and ACT1 promoters under nonpermissive conditions in vivo (Fig. 5A Bottom and SI Fig. 7). Moreover, the reduction of Rpb3 occupancy and Ser-5 phosphorylation (Fig. 5C) is remarkably similar to that observed with heat inactivated Kin28ts alleles, suggesting that the near-identical kinetics of transcriptional arrest exhibited by Rad3ts and Kin28ts strains result from the loss of TFIIH complex integrity. Taken together, the data strongly suggest that the defects in transcription seen in both Kin28ts strains as well as Rad3ts strain are attributable to the destabilization of TFIIH complex rather than the inactivation of the kinase activity.

The chemical–genetic approach described here has been used with great success to elucidate cellular functions of several different classes of kinases (38, 55), including the role of Cdc28 in GAL1 induction (56). We used this approach to probe the conflicting reports on the role of Kin28 in mRNA synthesis. Our results strongly indicate that the kinase activity of Kin28 is not essential for promoter clearance, transcript elongation, or global mRNA synthesis. This conclusion is inconsistent with the widely held view that CTD phosphorylation by Kin28 is essential for mRNA synthesis. However, our results are in close agreement with a few genetic studies that arrived at a similar conclusion (21, 37, 51). Our data provide compelling evidence, because the arguments of long-term adaptive changes do not apply to the rapid and reversible chemical inhibition of Kin28 in living cells. Thus, the role of Kin28 as a CTD kinase is probably important for 5′ capping of transcripts and for enhancing the exchange of complexes that associate with Pol II during different stages of transcription. It is important to clarify however, that we focus on the role of Kin28 kinase activity rather than CTD phosphorylation in mRNA synthesis. It is possible that low-level phosphorylation of the CTD by Srb10 may suffice for the promoter release and transcript elongation by Pol II (41).

Another intriguing finding is that steady state mRNA levels do not diminish despite the dramatic reduction of capped RNA (Fig. 3A). It is noteworthy that previous observations of instability of uncapped transcripts are based on studies with temperature-sensitive capping mutants (57). As in the case of TFIIH, it is possible thermal inactivation of capping enzymes perturbs protein complexes and enhances mRNA instability

Finally, our data suggest that widely accepted models for the role of Kin28 kinase activity in transcription initiation and elongation need to be revised. Moreover, as new substrates for Kin28 are identified, additional roles of Kin28 in cellular function should be investigated.

Strains and Growth Conditions.

Kin28as, Kin28ts16 (54), Kin28ts3 (35), Rad3ts (52), and their respective isogenic WT strains were grown in rich medium with raffinose or dextrose as a carbon source. GAL1 induction experiments were performed by addition of galactose (2% final concentration) to cultures grown in yeast peptone raffinose medium.

RNA Preparation, Microarray Analysis, and Quantitative PCR.

RNA was isolated by using the hot phenol method. Microarray experiments and quantitative RT-PCR were performed by using standard methods as described in SI Methods. Affymetrix GCOS version 1.2 was used to perform a local normalization based on Poly-A spiked in controls. Additional statistical analysis was performed with R and Bioconductor.

ChIP and RNA Immunoprecipitation.

ChIP was performed as described (58) with modifications detailed in SI Methods. RNA immunoprecipitation was performed as in (59) with modifications detailed in SI Methods. H14 antibody (MMS-134R) was purchased from Covance, Rad 3 (sc-11963) and HA (sc-7392) antibodies were purchased from Santa Cruz Biotechnology, and remaining antibodies were gifted as described in Acknowledgments.