Hyperphosphorylation of the C-terminal Repeat Domain of RNA Polymerase II Facilitates Dissociation of Its Complex with Mediator*

The Mediator complex associates with RNA polymerase II (RNAPII) at least partly via the RNAPII C-terminal repeat domain (CTD). This association greatly stimulates the CTD kinase activity of general transcription factor TFIIH, and subsequent CTD phosphorylation is involved in triggering promoter clearance. Here, highly purified proteins and a protein dissociation assay were used to investigate whether the RNAPII·Mediator complex (holo-RNAPII) can be disrupted by CTD phosphorylation, thereby severing one of the bonds that stabilize promoter-associated initiation complexes. We report that CTD phosphorylation by the serine 5-specific TFIIH complex, or its kinase module TFIIK, is indeed sufficient to dissociate holo-RNAPII. Surprisingly, phosphorylation by the CTD serine 2-specific kinase CTDK1 also results in dissociation. Moreover, the Mediator-induced stimulation of CTD phosphorylation previously reported for TFIIH is also observed with CTDK1 kinase. An unrelated CTD-binding protein, Rsp5, is capable of stimulating this CTD kinase activity as well. These data shed new light on mechanisms that drive the RNAPII transcription cycle and suggest a mechanism for the enhancement of CTD kinase activity by the Mediator complex.

The Mediator complex transduces signals from sequencespecific transcriptional regulators to the general transcription machinery (16 -18). The association of Mediator with RNAPII, and its function in transcription, depends on the RNAPII CTD (16). Mediator is able to bind an unphosphorylated glutathione S-transferase-CTD fusion protein in vitro (16) and can be displaced from RNAPII using the monoclonal antibody 8WG16, which specifically recognizes the CTD repeat (17,19). Apart from the CTD-interacting surfaces, Mediator also interacts with RNAPII domains outside of the CTD (20,21).
The observation that RNAPII enters the initiation complex in the unphosphorylated (RNAPIIA) form and leaves it in the hyperphosphorylated (RNAPIIO) form gave rise to the idea that the initiation to elongation transition is somehow facilitated by hyperphosphorylation of the CTD (3). The association with Mediator might be one of the mechanisms by which RNA-PIIA is tied to promoters. In crude yeast extracts, Mediator is almost entirely absent from the fraction containing chromatinbound RNAPIIO engaged in transcription, whereas it is present in the soluble fraction where only RNAPIIA is detected (19). Furthermore, purified Mediator in solution is able to dynamically exchange with Mediator in immobilized holo-RNAPII (19). The idea that phosphorylation might somehow result in the dissociation of the Mediator⅐RNAPII complex was further supported by experiments using immobilized transcription templates in transcription-proficient yeast extracts (22). In these extracts, the Mediator complex and other factors formed an activator-dependent "scaffold" for the binding of RNAPII, enabling efficient transcription initiation and re-initiation. Interestingly, TFIIH and ATP (but not AMP-PNP) were found to stimulate the dissociation of RNAPII from these scaffolds, indicating that phosphorylation, but not transcription, is important for dissociation (22). Further investigation of the stability of pre-initiation complexes assembled in crude yeast extract indicated that phosphorylation of RNAPII (either by TFIIH or Srb10/Med12-Srb11/Med13) caused some ATP-dependent dissociation of the pre-initiation complex (22,23). These experiments also showed that TFIIH has two protein targets in Mediator (Med4 and Med14). However, given that numerous protein-protein interactions exist in the pre-initiation complex, no clear conclusion on the precise functional effect of these multiple phosphorylation events could be made.
The mechanism that results in the release of the Mediator complex from holo-RNAPII has so far not been studied in a defined system containing only RNAPII holoenzyme and CTD kinase. Here, we have investigated the stability of the holo-RNAPII complex in relation to CTD phosphorylation using such a system. We also investigated the mechanism underlying stimulation of CTD kinase activity by the Mediator complex.
Dissociation Assay-Purified holo-RNAPII (incorporating Med14-TAP and Rpb8 -3HA affinity-tagged subunits) was immobilized on either anti-HA affinity resin or IgG resin, respectively. TFIIH, TFIIK, or CTDK1 was added, with or without ATP, to allow CTD phosphorylation. Beads and superna-tants were separated by centrifugation, and dissociation was indicated by the ATP-dependent appearance in the supernatant of Mediator subunits, which was monitored by immunoblotting with antibodies to detect Mediator and RNAPII, respectively. The holo-RNAPII peak from hydroxyapatite (17) was used as source of holo-RNAPII. For the dissociation assay where holo-RNAPII was immobilized via Med14/Rgr1 (Fig.  5B), 300 l of this fraction was added to IgG beads equilibrated in the same buffer and incubated on a roller table for 2 h at 4°C. Beads were pelleted (500 rpm, 1 min, room temperature) and the supernatant was removed, leaving the resin undisturbed. Unbound protein was removed from the beads by three washes with 2 ml of binding buffer (40 mM Hepes, pH 7.6, 0.25 mg/ml bovine serum albumin, 0.2% Tween 20, 0.01% Nonidet P-40, 200 mM KOAc, 15 mM imidazole, 20% glycerol, 5 mM ␤-mercaptoethanol, 1ϫ protease inhibitor mixture) (see above), and the supernatant was removed as before.
Holo-RNAPII carried a protein-A tag as part of the TAP tag on Med14. Because Protein-A binds firmly to the constant region of most antibodies, we removed it by tobacco etch virus (TEV) protease treatment: after IgG immobilization, 300 l of binding buffer containing 0.5-1 unit/l TEV protease (TEV protease or AcTEV protease; Invitrogen) was added and incubated for 30 min at 16°C or overnight at 4°C. Eluted and highly purified holo-RNAPII in the supernatant was pooled with two further washes of 300 l of binding buffer. For use in dissociation assays, this version of holo-RNAPII was then immobilized via the HA tag on Rpb8. For this purpose, the monoclonal antibody 12CA5 was immobilized on Affi-Gel-15 resin (Bio-Rad) according to the manufacturer's instruction. 12CA5 beads were equilibrated by repeated (3ϫ) pelleting (500 rpm, room temperature) and resuspension in 2 ml of binding buffer prior to incubation with the pool of holo-RNAPII elutions from IgG-Sepharose. Binding was at 4°C for 2-4 h with gentle shaking. Unbound proteins were removed by two washes with 2 ml of binding buffer followed by two washes of 2 ml of phosphorylation buffer (40 mM Hepes, pH 7.6 (at 25°C), 75 mM KOAc, 0.1 mg/ml bovine serum albumin, 0.06% Tween 20, 0.003% Nonidet P-40, 7.5 mM magnesium acetate, 3 mM imidazole, 10% glycerol, 5 mM ␤-mercaptoethanol, 1ϫ protease inhibitor mixture). Finally, the resin was resuspended in an appropriate volume of phosphorylation buffer prior to addition of CTD kinases and, where appropriate, ATP. The reactions were incubated at 30°C for 60 min before being returned to ice. Supernatants and pellets were separated as above. Two washing steps of 500 l of binding buffer each were pooled with the initial supernatant sample. Protein in the supernatants was then concentrated by extraction with 10 l of Strataclean resin (Stratagene). Protein from supernatants and beads was analyzed by immunoblotting.
Dissociation Assay Using IgG Beads and [␥-32 P]ATP-Holo-RNAPII (100 g) from hydroxyapatite was bound to 100 l of IgG-Sepharose for 2 h at 4°C and washed as described above for binding of holo-RNAPII to 12CA5 beads. Kinases were added, and salt was adjusted to 100 mM KOAc in a volume of 10 l/reaction. Dissociation assays were started by adding 10 l 2ϫ reaction mix (80 mM Hepes, pH 7.6, 20 mM MgCl 2 , 10 mM ␤-mercaptoethanol, 10% glycerol, 200 M ATP, 10 Ci of [␥-32 P]ATP) to each reaction and incubated at 30°C for 30 min. Phosphorylation was stopped by adding 10 l of 30 mM EDTA. After separation of supernatant and beads as above, sample buffer was added and reactions were analyzed by 10% SDS-PAGE. The gel was dried and exposed to a PhosphorImager plate (Storm 800; GE Healthcare). The results were processed and quantified by densitometry using the ImageJ program (rsb.info.nih.gov/ij/).
CTD Kinase Assays-Kinase assays were performed in 20 l of final volume of reaction buffer (40 mM Hepes, pH 7.5, 50 mM KOAc, 10 mM MgCl 2 , 5 mM ␤-mercaptoethanol, 5% glycerol, 62.5 M ATP, 10 Ci of [␥-32 P]ATP). Each reaction contained 300 ng of RNAPII as well as kinases and kinase-stimulating proteins. Incubation was at 30°C for 60 min. Reactions were terminated by addition of 5 l of 5ϫ SDS-PAGE sample buffer and incubation at 95°C for 3-5 min. Reactions were analyzed by SDS-PAGE, and the dried gel was exposed to film or Phos-phorImager plate.

RESULTS
To investigate whether CTD phosphorylation per se is sufficient to disrupt the association between Mediator and RNAPII, we used highly purified proteins in a defined biochemical assay system. The experimental design used holo-RNAPII purified from a yeast strain expressing an epitope-tagged version of one of the RNAPII subunits. Holo-RNAPII was then immobilized on an affinity resin, and a dissociation assay was performed ( Fig.  1). This assay entailed adding highly purified kinase (TFIIH, TFIIK, or CTDK1) to phosphorylate the CTD of Rpb1, with ATP-dependent appearance in the supernatant of Mediator subunits indicating that phosphorylation of the CTD caused holo-RNAPII dissociation.
The inhibition of hyperphosphorylation of the immobilized holo-polymerase might be due to steric hindrance, as the tag on the C-terminal of Rpb2 is close to the Rpb1 domain, which links the CTD to RNAPII (29). To try and overcome this problem, the Rpb8 subunit was tagged instead. This was done in a strain expressing a tagged Mediator subunit (Med14 (Rgr1), which had previously been used for Mediator purification) (30), creating an RPB8 -3HA MED14-TAP strain.
Holo-RNAPII from this strain would thus enable us to investigate dissociation of Mediator from immobilized RNAPII or, alternatively, dissociation of RNAPII from immobilized Mediator. In contrast to what had been observed previously with Rpb2-tagged holo-RNAPII, immobilization of Rpb8-tagged holo-RNAPII complex via the triple HA tag on Rpb8 (or on IgG via Med14-TAP) allowed TFIIH to quantitatively shift Rpb1 to the position typical for hyperphosphorylated RNAPII (Fig. 2B, compare lanes 4 and 6; data not shown).
Under some circumstances, the Srb10/Srb11 kinase is found associated with holo-RNAPII (31). However, phosphorylation was dependent on ATP and TFIIH, suggesting that little, if any, kinase activity was associated with our holo-RNAPII preparation (Fig. 2C, lane 1, and data not shown). Together, these results indicated that the Rpb8 -3HA Med14-TAP holoenzyme could be used to study phosphorylation-dependent Mediator⅐RNAPII dissociation.
Phosphorylation by TFIIH Causes Dissociation of Mediator and RNAPII-Holo-RNAPII was immobilized on 12CA5 antibody beads via the tag on Rpb8. TFIIH kinase, with or without ATP, was then added to study dissociation. Successful phosphorylation in the presence of ATP was indicated by the quantitative mobility shift of Rpb1 on a 5% SDS-PAGE gel that was silver-stained (Fig. 3A). The result of the coincident dissociation experiment is shown in Fig. 3B. Substantial ATP-dependent dissociation of Med4 from immobilized RNAPII was seen by following the Med4 band: in the presence of ATP, Med4 dissociated from the resin (compare lanes 2 and 3) and appeared in the supernatant (compare lanes 5 and 6). Negligible dissociation was observed in the presence of ATP but absence of kinase (data not shown). Interestingly, when adding the FIGURE 1. Dissociation assay. A, schematic representation of the experimental approach used. Purified holo-RNAPII, epitope-tagged on RNAPII (Rpb8-HA) or on Mediator (Med14-TAP), was immobilized on affinity resin. CTD kinase was then added, resulting in hyperphosphorylation of the RNAPII CTD (indicated by minuses (negative charge)) on the CTD in the lower panel. If phosphorylation results in dissociation of the Mediator from RNAPII, Mediator should appear in the soluble fraction (with RNAPII remaining resin-associated). B, silver-stained gel of holo-RNAPII and core RNAPII used in this study. Please note that the holo-RNAPII shown here was actually further purified through its immobilization (and wash) on 12CA5 resin prior to incubation with kinase. Identification of bands was inferred from published purifications of holo-RNAPII (17,50,51) and from immunoblots using antibodies against Med1, Med2, Med7, and Med8, as well as the TAP tag on Med14 and the HA tag on Rpb8 (data not shown).
slowly hydrolyzing ATP analogue ATP-␥S, a small but detectable level of dissociation above background was observed (compares lane 4, 5, and 6). The basis and relevance of this observation are unknown. More importantly, no significant holo-RNAPII dissociation was observed in the presence of TFIIH but absence of ATP (compare lanes 5 and 6), strongly indicating that TFIIH-mediated phosphorylation of the CTD, not the presence of TFIIH per se, causes holo-RNAPII dissociation.
The TFIIK Kinase Subcomplex Is Sufficient for Dissociation-The observed ATP-dependent dissociation of holo-RNAPII fits well with the change in phosphorylation status of the Rpb1 CTD during promoter clearance (1). Serine 5 is the primary target for phosphorylation at this stage, with serine 2 phosphorylation only becoming enriched later in the transcription cycle (1,32). Although CTD phosphorylation appeared to be sufficient for dissociation in the above experiments, a possible (active) role mediated by other subunits of TFIIH, such as the Rad3 (XPD in man) and Ssl2 (XPB) ATP-dependent helicases, could also be envisioned. TFIIK (Kin28/Ccl1/Tfb3) is the submodule of TFIIH that harbors the CTD kinase Kin28, and Kin28 interacts with Rpb1 and Rpb2 in vitro (33). To establish whether phosphorylation by TFIIK is sufficient to elicit holo-RNAPII dissociation, the TFIIK complex was purified as previously described (27). Moreover, having observed that TFIIH phosphorylation leads to holo-RNAPII dissociation, it was also interesting to investigate whether dissociation requires phosphorylation at the residue specifically targeted by this kinase, serine 5, or whether serine 2 phosphorylation might also dissociate holo-RNAPII. We therefore also purified the kinase that predominantly phosphorylates serine 2, CTDK1.
Two preliminary experiments were carried out to verify the function of TFIIK and CTDK1 prior to their use in the dissociation assay. First, a titration was performed to establish how much of each kinase was needed to quantitatively shift the Rpb1 band from holo-RNAPIIA immobilized on 12CA5 beads (Fig.  4A). Second, because previous experiments have shown that the target specificity of TFIIK and CTDK1 differ, we tested the activity of CTDK1 on a CTD substrate that was first treated with a saturating amount of TFIIK to fully phosphorylate serine 5. As expected, CTDK1 caused a supershift of the polymerase, indicating that the two kinases indeed hit different targets in the CTD (Fig. 4B).
The purified kinases were then used in the dissociation assay. The effect of holo-TFIIH and TFIIK is compared in Fig. 5A. Although a somewhat higher background level of ATP-independent Mediator release is evident in this experiment (lane 3), FIGURE 2. The position of the epitope tag affects CTD phosphorylation upon RNAPII immobilization. A, comparison of the phosphorylation efficiency using soluble RNAPII and RNAPII immobilized via a tag on the C terminus of the Rpb2 subunit. B, tagging RNAPII on the C terminus of Rpb8 alleviates the inhibition of phosphorylation observed with immobilized Rpb2-tagged RNAPII. C, phosphorylation of RNAPII is dependent on the addition of TFIIH kinase and ATP. Silverstained SDS-PAGE gels are shown. To maximize the resolution of phosphorylated and unphosphorylated RNAPII, the gels were run until the 100-kDa marker was at the bottom of the gel. For simplicity, only the migration of the Rpb1 subunit is shown. Migration of other RNAPII subunits is unchanged by phosphorylation (not shown).  7 and 9). This indicates that the activity of the TFIIK kinase complex is sufficient to dissociate holo-RNAPII in an ATP-dependent manner and suggests that the other ATPases (helicases) in TFIIH do not play an important role in this event. Indeed, several independent dissociation assays failed to detect any significant difference in the ability of TFIIH and TFIIK to dissociate holo-RNAPII (data not shown).
The dissociation experiment was repeated, this time by immobilizing holo-RNAPII via Mediator rather than RNAPII and using TFIIK, CTDK1, or both as the kinase source (Fig. 5B). For technical reasons, the dissociating RNAPII was in this experiment detected by performing the assay with labeled ATP and following the fate of the phosphorylated polymerase by autoradiography. As can be seen, increasing amounts of kinase indeed led to increasing dissociation of RNAPII (compare lanes 1-2 with lanes 3-4 and 5-6 for TFIIK and lanes 7-8 with lanes 9 -10 for CTDK1), indicating that phosphorylation by both TFIIK and CTDK1 promotes holo-RNAPII dissociation. When the CTD kinases were combined (lanes 11-12 and 13-14), somewhat more phosphorylated RNAPII was recovered in the supernatant, indicating an additional effect on dissociation when both serine 2 and serine 5 are phosphorylated.
Previous work by others showed that immobilized holo-RNAPII can be separated into RNAPII and Mediator by washing with buffer containing 1 M urea. In contrast, the Mediator complex itself sustains its integrity in up to 2 M urea (34). In an attempt to exacerbate the destabilizing effects of CTD phosphorylation, the dissociation assay was also performed using buffers containing 300 mM urea (Fig. 5C). Inclusion of 0.3 M urea in the (post-phosphorylation) wash buffer did not destabilize the holo-RNAPII complex on its own (Fig. 5C, lane 2).
However, when used after incubation with TFIIK, significant ATP-dependent dissociation of Mediator was observed (lane 6). The effect was even clearer when using purified CTDK1 (lane 10). Taken together, these results indicate that CTD phosphorylation of not only serine 5 but also serine 2 can result in significant dissociation of the Mediator⅐RNAPII complex.
CTD Phosphorylation and CTD-binding Proteins-Previous experiments by others showed that Mediator stimulates the CTD kinase activity of TFIIH manyfold (17). The finding that TFIIH phosphorylation dissociates the RNAPII holoenzyme suggests that Mediator might stimulate its own dissociation from RNAPII during the transition from initiation to elongation. In this light, the ability of CTDK1 to dissociate the RNAPII holoenzyme was interesting. In order to look more closely at the mechanism of CTD phosphorylation stimulation, we purified free Mediator complex via the TAP tag on Med14 and used it in RNAPII phosphorylation assays. An autoradiogram showing  5) and CTDK1 (6 -10) required to achieve RNAPII hyperphosphorylation. The concentration of TFIIK added was 56 ng/l and that of CTDK1 was 67 ng/l. A silver-stained SDS-PAGE gel is shown. Unless otherwise stated, the amount of kinase required for quantitative CTD phosphorylation was used in subsequent experiments. B, silver-stained SDS-PAGE gel showing sequential phosphorylation of RNAPII by first TFIIK and then CTDK1. . Asterisks denote nonspecific cross-reacting bands (light and heavy chain antibody). B, different amounts of TFIIK and CTDK1 were used with radioactively labeled ATP in the dissociation assay, this time immobilizing holo-RNAPII via the Med14 subunit and detecting the dissociated, phosphorylated polymerase by autoradiography. The relative amounts of phosphorylated RNAPII in the supernatant (S) compared with the beads (B) are indicated below. The gel was not run far enough to allow a distinction between RNAPIIA and RNAPIIO forms. Some kinase contamination was present in the holo-RNAPII preparation used for these experiments but not enough to result in substantial dissociation (lanes 1 and 2). C, Western blots examining the presence of Mediator (Med4 and Med1) and RNAPII (Rpb8) on the affinity beads (B) or in the dissociated material (S) after addition of ATP, and TFIIK (60 ng), or CTDK1 (70 ng) in the presence of 300 mM urea.
the stimulation of TFIIH kinase activity is shown in Fig. 6A. CTD phosphorylation increases in proportion to the amount of holo-TFIIH, indicating that TFIIH activity was within the linear range (compare lane 2 to 4). Mediator on its own had no detectable CTD kinase activity (see lane 5), but CTD phosphorylation by TFIIH kinase was greatly stimulated by Mediator complex as expected (compare lanes 2 and 6). The stimulation of kinase activity by Mediator was also observed with the TFIIK submodule of TFIIH (Fig. 6B). Others have presented similar results (35).
To investigate whether purified Mediator could stimulate CTDK1 activity as well, the stimulation assay was repeated using this kinase (Fig. 6C). In these experiments, phosphorylation of RNAPII was measured by cross-reactivity with the ser-ine 2 phosphorylation-specific H5 antibody. Lane 1 shows that the substrate was largely unphosphorylated in the absence of CTDK1 kinase. As expected, increasing amounts of CTDK1 resulted in increasing RNAPII phosphorylation (lanes 3 and 4). Interestingly, adding purified Mediator to the lowest amount of CTDK1 kinase stimulated kinase activity significantly (com- pare lanes 3 and 5). These and other data (not shown) indicate that Mediator can also stimulate CTD serine 2 phosphorylation by CTDK1.
Rsp5, Another CTD-binding Protein, Can Stimulate CTD Phosphorylation-The ability of Mediator to stimulate the kinase activity of both TFIIK and CTDK1 prompted further experiments to investigate the basis for the stimulation. In theory, interactions between Mediator subunits and TFIIK and CTDK1 could underlie the stimulation of kinase activity, but no such interactions have been reported. Numerous proteins, (general transcription factors (GTFs), isomerases, the capping machinery, splicing factors, etc.) are known to interact with the CTD. We speculated that the CTD might be less accessible for kinases in its free unstructured form compared with when it is in contact with certain binding partners. To test the hypothesis that enhancement of kinase activity might also be conferred by other CTD-binding proteins, the CTD kinase stimulation experiment was repeated with the ubiquitin ligase Rsp5. Rsp5 (Reverses SPt mutation 5) interacts directly with the CTD (36), and this interaction is required for ubiquitylation of RNAPII in response to DNA damage (27,37). The result of a CTD kinase assay using CTDK1 and Rsp5 is shown in Fig. 6D. Interestingly, although the effect was not as dramatic as that observed with Mediator, phosphorylation of the CTD was clearly stimulated by Rsp5. This result demonstrates that not only Mediator, but also an unrelated CTD-binding protein, Rsp5, can stimulate CTD phosphorylation.

DISCUSSION
Phosphorylation of the CTD by TFIIH Dissociates Holo-RNAPII-Dissociation of Mediator from holo-RNAPII can be triggered by the addition of TFIIH and ATP. Others have previously reported that the Srb10/Srb11 kinase is found associated with holo-RNAPII (31). However, in our experiments neither phosphorylation nor dissociation of holo-RNAPII occurred in the presence of ATP but absence of kinase, showing that no CTD kinase activity was associated with our holo-RNA-PII preparations. Nevertheless, holo-RNAPII purified from an srb10⌬ strain was also tested, with results identical to those presented here (data not shown).
It is important to note that the interaction of Mediator with RNAPII is governed by both CTD and non-CTD interactions (21). Interestingly, the affinity of a recombinant Mediator head module for RNAPII in a ternary RNA⅐DNA⅐RNAPII complex is diminished by an order of magnitude by the presence of the RNA transcript (38), suggesting that the initiation of transcription itself might lead to some destabilization of Mediator-RNA-PII interactions. Thus, the release of RNAPII from the Mediator interactions that recruit and bind it to promoters might well be stimulated by both transcript initiation/early elongation (severing non-CTD interactions) and CTD phosphorylation (severing CTD-mediated interactions). Dissociation of Holo-RNAPII Does Not Require Holo-TFIIH, Only the Kinase Submodule TFIIK-The core TFIIH subunit Tfb1 has been reported to interact physically with the Mediator Med15/Gal11 subunit in vitro (39). Thus, a priori one might assume that the dissociation mechanism of holo-RNAPII would be dependent on a combination of protein-protein interactions and CTD phosphorylation. However, ATP-dependent holo-RNAPII dissociation can be effected by TFIIK kinase alone, indicating that the core TFIIH subunits are dispensable for the dissociation mechanism. The TFIIK subunit Kin28 was shown by Far Western to interact with Rpb1 and Rpb2 (33). This raised the question whether a combination of Kin28-RNAPII contacts and CTD hyperphosphorylation was needed for dissociation and whether dissociation was dependent on the site of CTD phosphorylation. However, we found that phosphorylation of the CTD by CTDK1 also dissociates holo-RNA-PII. This supports the idea that phosphorylation per se is sufficient for holo-RNAPII dissociation. The fact that both TFIIK and CTDK1 can dissociate holo-RNAPII also strongly argues that CTD phosphorylation rather than phosphorylation of, for example, a Mediator subunit is responsible. Indeed, characterization of the holo-RNAPII phosphorylation patterns using TFIIK and CTDK1 shows that only the RNAPII CTD is targeted by both kinases (data not shown). Although we cannot completely rule out that the kinases dissociate holo-RNAPII by distinct mechanisms involving phosphorylation of different Mediator subunits, we find it much more likely that it occurs as a direct consequence of simple charge repulsion brought about via TFIIK-(serine 5) and CTDK1-mediated (serine 2) phosphorylation of the CTD.
The dissociation of holo-RNAPII upon CTDK1-mediated phosphorylation of serine 2 is surprising because promoter clearance has been associated primarily with serine 5 hyperphosphorylation (40). In contrast, serine 2 phosphorylation is only enriched later in the transcription cycle (1,14). Interestingly, lack of TFIIH kinase activity has a surprisingly modest effect on transcription (41)(42)(43), which might argue that CTD phosphorylation is not required for promoter clearance. Our biochemical data, however, suggest that CTDK1-mediated phosphorylation could to a certain extent overcome the need for TFIIH kinase activity in this process. This idea is supported by recent genetic data that indicate a functional overlap between TFIIH and CTDK1 in Saccharomyces pompe (44).
Interestingly, a direct interaction of human Mediator complexes with acidic activator domains can elicit Mediator conformational change, and Mediator adopts a specific CTDdependent conformation (45). The negatively charged hyperphosphorylated CTD may thus change Mediator conformation in the holo-RNAPII complex to effect release. Although the physiological relevance of the observed CTDK1-mediated dissociation of holo-RNAPII still needs to be determined, it is also worth noting that an association of Mediator with the transcribed region of genes has recently been reported (46,47). This led to the proposal that Mediator somehow coordinates transcription initiation with transcriptional events in the coding region of eukaryotic genes. Our data could indicate a role for CTDK1 kinase in this function, or perhaps polymerase restart after arrest in certain instances involves Mediator association, followed by CTDK1-mediated release.
Although CTDK1 is the main serine 2 kinase in vivo (14), it was reported that CTDK1 is also able to phosphorylate serine 5 of the yeast CTD fused to glutathione S-transferase (10). In our hands, however, CTDK1 specifically targets serine 2 when RNAPII is the substrate. CTDK1-mediated phosphorylation of RNAPII that was prephosphorylated on serine 5 by a saturating amount of TFIIK led to a supershift in RNAPII electrophoretic mobility. Furthermore, we have used TFIIK and CTDK1 separately to create RNAPIIO (15). The serine 5-specific phosphatase Ssu72 was then used to dephosphorylate these substrates; RNAPIIO phosphorylated by TFIIK was efficiently dephosphorylated, whereas CTDK1-phosphorylated RNAPIIO was not. This indicates that few, if any, serine 5 residues are phosphorylated by CTDK1 under our assay conditions. The kinetics of CTDK1 activity was previously reported to depend critically on reaction conditions (48). Differences in the CTDK1 purification procedure, the assay conditions, or the substrate used may explain the discrepancy between our results and those of others (10).
Mediator and Rsp5 Stimulate CTD Kinase Activity-In the course of the work on Mediator⅐RNAPII dissociation, the effect of Mediator (and Rsp5) on the activity of CTD kinases was also studied, resulting in some interesting and surprising observations. Previous results had shown that the Mediator complex strongly stimulates the CTD kinase activity of TFIIH (16,17) and that it also stimulates the activity of TFIIK (35). Our results confirm and extend these findings, as we also observed stimulation of CTDK1 kinase activity by both Mediator and Rsp5. Using TFIIK, we also noted a weak stimulation of CTD kinase activity by Rsp5 (data not shown). Based on these data, it is tempting to speculate that binding of the CTD by Mediator complex serves to "present" a more accessible target to kinases. According to this model, the free CTD has a structure that is suboptimal for phosphorylation and Mediator interaction disrupts this structure or somehow orders the CTD, allowing processive phosphorylation to occur. Several results support the idea that altered "CTD presentation" underlies kinase stimulation. First, Guidi et al. (35) previously showed that Mediator can stimulate phosphorylation of glutathione S-transferase-CTD as well as of RNAPII, suggesting that TFIIK interactions with the non-CTD parts of the RNAPII complex are not absolutely required for stimulation. Second, the reported interaction between Mediator and TFIIH (via Tfb1) does not seem to play an important role in kinase stimulation, as the activity of the TFIIK submodule of TFIIH (which lacks the Tfb1 subunit) is also stimulated. Third, purified Mediator also stimulates the kinase activity of CTDK1, and no physical interaction between CTDK1 and Mediator has been reported. Fourth, Rsp5 protein can also (to a lower degree) stimulate CTDK1 activity, suggesting that stimulation of CTD phosphorylation is not even intrinsic to the Mediator complex or the kinase. Rather, stimulation of CTD phosphorylation might be a consequence of a change in CTD presentation upon the binding of certain partner proteins.
The effect of Rsp5 on CTD phosphorylation is intriguing. Rsp5 is the E3 ubiquitin ligase for DNA damage-induced ubiquitylation of RNAPII and binds the polymerase strongly, exclu-sively via the CTD (27,37,49). The physiological role, if any, of the Rsp5 stimulation of CTDK1 kinase activity remains to be determined.