Phosphorylation of the Yeast Rpb1 C-terminal Domain at Serines 2, 5, and 7*

The C-terminal domain (CTD) of Rpb1, the largest subunit of RNA polymerase II, acts as a binding platform for various mRNA processing and histone-modifying enzymes that act co-transcriptionally. These factors are targeted to specific phosphorylation states of the CTD that predominate at different stages of transcription. Within the repeating sequence YSPTSPS, serines 2 and 5 are major phosphorylation sites, but serine 7 phosphorylation was recently discovered in mammalian cells. Here we show that CTD serine 7 is also phosphorylated in yeast and that Ser-7(P) chromatin immunoprecipitation patterns resemble those of Ser-5(P). The basal factor TFIIH can phosphorylate Ser-7 in vitro and is necessary for Ser-7(P) in vivo. Interestingly, deletion of the CTD Ser-5(P) phosphatase Rtr1 leads to an increase in Ser-5(P) but not Ser-7(P).

Transcription by RNA polymerase II (RNA pol II) 3 is coupled to several other nuclear processes. RNA transcripts produced by RNA pol II are capped at their 5Ј ends, often spliced, and processed at their 3Ј ends. RNA pol II transcription also generates a stereotypical set of histone modifications, with high levels of methylation of histone H3 lysine 4 (H3K4) at active promoters and H3K36 methylation marking downstream transcribed regions. For both mRNA processing and histone methylation, the relevant enzymes directly bind to the C-terminal domain (CTD) of the RNA pol II largest subunit (1,2). The CTD consists of multiple repeats of the sequence YSPTSPS. Different phosphorylation sites predominate at different stages of transcription, with various accessory proteins binding to specific CTD phosphorylation patterns (3). Through this mechanism, information about the location of the RNA pol II molecule on the gene is encoded in the CTD phosphorylation pattern and used to recruit the appropriate mRNA and histone-modifying enzymes at the proper time during the transcription cycle.
It is important to note that the Ser-5(P)/Ser-2(P) model is based almost entirely on experiments using the monoclonal IgM antibodies H14 and H5 (3,23,24). Although these antibodies clearly recognize distinct epitopes, the Ser-2(P)-recognizing antibody H5 shows some cross-reactivity with Ser-5(P) (8,25). There have also been varying reports about whether Ser-5(P) is confined to promoters or persists throughout transcribed regions. Recently, Chapman et al. (26) generated a new set of monoclonal antibodies with strong specificity for Ser-5(P) (3E8) and Ser-2(P) (3E10), as well as an antibody that recognizes Ser-7(P) (4E12). Here we use these antibodies to confirm that Ser-5(P) levels are highest at promoters, whereas Ser-2(P) levels rise with increasing distance from the promoter. We also help explain why Ser-5(P) levels are usually seen to be highest at promoters but sometimes reported to remain high throughout elongation. Finally, in agreement with a recent report (27), we find that yeast CTD is also phosphorylated at Ser-7 and that this phosphorylation is dependent upon the kinase activity of basal transcription factor TFIIH. Ser-7(P) patterns are similar to those of Ser-5(P), and inhibition of Kin28 leads to loss of both phosphorylations. In contrast, deletion of the Ser-5(P) phosphatase Rtr1 (4) increases Ser-5(P) but not Ser-7(P).

EXPERIMENTAL PROCEDURES
Chromatin Immunoprecipitation (ChIP)-Chromatin solutions were prepared as described previously (28). Twenty or ten microliters of 4E12, 3E8, and 3E10 rat monoclonal antibodies (cell culture supernatant, a generous gift from Dirk Eick, Munich Center for Integrated Protein Science) were prebound to 15 l of protein G-Sepharose (GE Healthcare) at room temperature for 2 h and then incubated overnight with 400 l of chromatin solution (ϳ800 g of protein) at 4°C. Beads were then washed sequentially with FA lysis buffer (50 mM HEPES-KOH (pH 7.5), 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS) plus 275 mM NaCl, FA lysis buffer plus 500 mM NaCl, ChIP wash buffer (10 mM Tris-HCl (pH 8.0), 0.25 M LiCl, 1 mM EDTA, 0.5% Nonidet P-40, 0.5% sodium deoxycholate), and TE (10 mM Tris-HCl (pH 8.0), 1 mM EDTA). Immunoprecipitated chromatin was eluted from the beads by heating for 10 min at 65°C in the presence of 50 mM Tris-HCl (pH 7.5), 10 mM EDTA, 1% SDS and then incubated with Pronase (Roche Applied Science, 1 mg/ml final concentration) for 1 h at 42°C. Samples were then heated for 5 h at 65°C to reverse the cross-links. For H14 and H5 ChIP, antibodies and antimouse IgM agarose (Sigma A4540) were added simultaneously to chromatin without preincubation. For monoclonal antibody H14, the first two washes were substituted with FA lysis buffer plus 750 mM NaCl. In contrast, H5-precipitated beads were washed with FA lysis buffer plus 150 mM NaCl. After cross-link reversal, samples were processed and assayed by PCR as described previously (28).
In Vitro CTD Kinase Assay-Native TFIIH and recombinant GST⅐CTD were purified as described previously (29,30). To prepare dephosphorylated substrate, 600 g of GST⅐CTD were dephosphorylated with 60 units of Antarctic phosphatase (New England Biolabs) in 500 l of 1ϫ Antarctic phosphatase buffer for 1 h at 37°C. The protein was then repurified with glutathione-agarose (Sigma), dialyzed overnight in 20 mM HEPES-KOH (pH 7.0), 150 mM NaCl, 10% glycerol, 1 mM dithiothreitol at 4°C, and stored at Ϫ80°C until use.

RESULTS
Chromatin Immunoprecipitation Using New CTD Ser-5(P) and Ser-2(P) Antibodies-To verify our earlier results using the H5 and H14 antibodies, ChIP experiments were carried out using the new monoclonal antibodies created by Chapman et al. (26). On both the PMA1 and the ADH1 mRNA genes, the anti-Ser-2(P) antibody 3E10 showed a pattern very similar to that seen with H5, confirming that Ser-2(P) levels are low near the 5Ј end of genes and progressively increase with distance from the promoter (Fig. 1, A  and B). On the snR33 snoRNA gene, neither 3E10 nor H5 showed appreciable reactivity, consistent with termination only a few hundred nucleotides downstream of the promoter, before Ser-2(P) reaches detectable levels (Fig. 1C).
The anti-Ser-5(P) antibodies H14 and 3E8 both showed strong reactivity at the 5Ј ends of genes ( Fig. 1, A-C). However, in initial experiments, the 3E8 pattern showed a much stronger signal persisting throughout the transcribed region of longer genes, essentially matching total RNA pol II (not shown). In contrast, H14 reactivity dropped soon after the promoter (Fig. 1, A and B). Although this could indicate that the two antibodies were recognizing different patterns of Ser-5(P), we instead suspected that the 3E8 was more sensitive to lower levels of Ser-5(P) that persist during elongation (3). To test this idea, the ChIPs were repeated with less 3E8 antibody (Fig. 2). As predicted, the 3E8 pattern now closely resembled that of H14, with a distinct peak of Ser-5(P) apparent at the promoter. It is also notable that the normalized signal was much higher due to a significant reduction in the non-transcribed background signal. The variation in Ser-5(P) pattern is readily explained by the multi-epitope nature of the CTD. At saturating levels of antibody, any RNA pol II molecules with at least one accessible Ser-5(P) epitope will be precipitated. The results under these conditions correctly lead to the conclusion that some level of Ser-5(P) persists throughout elongation. In contrast, at lower antibody levels, the RNA pol II molecules with the most Ser-5-phosphorylated repeats will be preferentially recognized, leading to the equally correct conclusion that Ser-5(P) levels are highest at promoters. Therefore, the different Ser-5(P) patterns reported are unlikely to be due to the antibodies specifically but rather to the affinity or titer of that specific antibody lot. CTD Serine 7 Phosphorylation in S. cerevisiae Is Dependent upon TFIIH-ChIP of mammalian chromatin using the 4E12 antibody, which reacts with Ser-7(P), indicated that this modification predominated in downstream transcribed regions (26). Given that this modification was linked specifically to recruitment of snRNA processing factors that are not present in yeast (32), it was unclear whether to expect a Ser-7(P) signal in S. cerevisiae. In fact, ChIP in yeast cells with 4E12 gave a very clear signal (Figs. 1 and 2). Surprisingly, the 4E12 signal was strongest at the 5Ј ends of genes and decreased with distance from the promoter, similar to the Ser-5(P) pattern. Ser-7(P) was also observed on the short, non-polyadenylated snR33 gene (Fig. 1C). As observed with 3E8, using lower levels of 4E12 antibody accentuated the postpromoter decrease in Ser-7(P). However, a peak of reactivity was also now observed in the polyadenylation site of the PMA1 (but not ADH1) gene (Fig. 2). The source for this reactivity is unclear but suggests that Ser-5(P) and Ser-7(P) are not necessarily tightly linked.
Given that the Ser-7(P) pattern resembles that of Ser-5(P), we tested whether the kinase activity of TFIIH subunit Kin28, which phosphorylates CTD Ser-5, might also be necessary for the Ser-7(P) signal. This was done using an altered substrate mutant of Kin28 (Kin28-as) that can be specifically inhibited by the chemical NA-PP1 (31). Treatment of cells with NA-PP1 followed by ChIP showed that Ser-5(P) levels were greatly reduced as expected. Furthermore, Ser-7(P) levels were also decreased to near background levels upon Kin28 inhibition (Fig. 3). In contrast, Ser-2(P) and Rpb3 levels were only decreased by about 50 -70%, consistent with previous reports by us and others that loss of Kin28 kinase activity, as opposed to Kin28 protein, does not lead to loss of transcription (3,6,27,33). Therefore, the drop in Ser-7(P) is not an indirect effect caused by cessation of transcription.
TFIIH Can Directly Phosphorylate CTD Ser-7-Because the dependence of Ser-7(P) on Kin28 could be direct or indirect, purified TFIIH was tested for the ability to phosphorylate the CTD on Ser-7 in vitro. TFIIH was purified from yeast extracts as described (29,30) and incubated with recombinant GST⅐CTD produced in bacteria. After incubating the proteins with ATP, the reactions were probed with the new CTD monoclonal antibodies (Fig. 4). As expected, TFIIH produced a strong Ser-5(P) signal but little or no Ser-2(P). It was difficult to interpret the Ser-7(P) result because the GST⅐CTD produced a positive signal with 4E12 in the absence of any kinase, although a new band appeared that co-migrated with the shifted Ser-5(P) band (Fig. 4A). This result could mean either that 4E12 cross-reacted with unphosphorylated CTD or that some bacterial kinase was phosphorylating the recombinant GST⅐CTD. To distinguish between these possibilities, the GST⅐CTD was incubated with phosphatase and the experiment was repeated (Fig. 4B). This treatment led to loss of the kinase-independent Ser-7(P) signal, indicating that GST⅐CTD produced in bacteria is already phosphorylated at Ser-7. However, dephosphorylated GST⅐CTD incubated with purified TFIIH and ATP clearly reacted with anti-Ser-7(P) antibody, indicating that TFIIH can directly phosphorylate this residue.
Loss of Rtr1 Increases Ser-5(P) but Not Ser-7(P)-The Rtr1 protein was recently shown to be responsible for dephosphorylation of Ser-5(P) during elongation, with Ser-5(P) being assayed by reactivity with the H14 antibody (4). An rtr1⌬ strain was tested with 3E8 and 4E12 antibodies (Fig. 2). As reported previously, a large increase in Ser-5(P) levels was observed, particularly in downstream transcribed regions. In contrast, levels of Ser-7(P) did not increase. This suggests that Rtr1 may be specific for Ser-5(P) and that there could be a distinct phosphatase that targets Ser-7(P).

DISCUSSION
The proposal that Ser-5(P) marks active promoters, whereas Ser-2(P) gradually increases with distance from the promoter, was based on ChIP experiments using the H14 and H5 monoclonal antibodies (3). The Ser-5(P)/Ser-2(P) pattern has now been confirmed using an independently derived set of monoclonal antibodies (26). In the process of characterizing the new antibodies, some insight was gained into the variability that has been reported as to whether Ser-5(P) is limited to promoters or persists throughout elongation. We believe that this variability is due to the repetitive nature of the CTD. At saturating antibody titers, a single phosphorylated repeat should be sufficient to precipitate RNA pol II. Under these conditions, it becomes clear that some level of Ser-5(P) persists throughout elongation. However, because there is no distinction between RNA pol II molecules that are highly phosphorylated and those with low levels, the immunoprecipitation signal is not Chromatin from wild-type (WT) cells was immunoprecipitated with 20 or 10 l of 3E8 (␣-Ser-5(P)) or 4E12 (␣-Ser-7(P)) followed by PCR amplification with primers depicted in Fig. 1. The higher level of antibody shows Ser-5(P) and Ser-7(P) evenly throughout the gene, whereas reducing the amount of antibody accentuates higher phosphorylation levels near the promoter. In parallel, chromatin from an rtr1⌬ mutant was immunoprecipitated with 10 l of each antibody to test its effect on Ser-5(P) and Ser-7(P). A, ChIP analysis on the PMA1 gene. Gels are shown in the top panels, and quantification of ChIP data appears below. B, ChIP analysis on ADH1 gene as in panel A. linearly responsive to the number of epitopes. In contrast, at subsaturating antibody titers, there is clearly greater reactivity at promoters. This is presumably due to the increased probability of an antibody molecule reacting with a CTD that contains many phosphorylated repeats.
This hypothesis helps to explain several earlier observations. First, we have noted variability between different lots of H14 antibody, with only some batches showing appreciable signals in internal transcribed regions (3). Inconsistent antibody titers are likely to contribute to variability. H14 is an IgM, a class of multivalent antibodies, which may result in H14 binding being highly cooperative when multiple Ser-5(P) sites are encountered, further enhancing non-linearity of reactivity. Second, ChIPs in mammalian and Drosophila cells, which have a greater number of CTD repeats per polymerase molecule, often show a robust H14 signal in downstream transcribed regions. Third, deletion of the CTD phosphatase Rtr1 in yeast leads to a high H14 signal extending throughout the transcribed region, consistent with an increased number of Ser-5(P) epitopes being generated (4). In any case, the variation in ChIP patterns seen with different amounts of antibodies should be taken into consideration when comparing different reports and serves to underscore the need for additional approaches to characterize CTD phosphorylation patterns.
The existence of Ser-7(P) in S. cerevisiae is interesting and unexpected. The only role assigned to Ser-7(P) so far is in recruiting the mammalian snRNA 3Ј-processing complex Integrator (32). Given that S. cerevisiae does not contain an Integrator-like complex, it remains to be seen whether Ser-7(P) has an undiscovered function. It is interesting that the Ser-7(P) ChIP pattern is similar but not always identical to Ser-5(P), and Ser-7(P) does not appear to be targeted by the Ser-5(P) phosphatase Rtr1. We also observed Ser-7 phosphorylation on recombinant GST⅐CTD made in Escherichia coli, a finding that must be taken into consideration when using bacterially produced CTD for binding experiments. . Chemical inhibition of Kin28 kinase results in a decrease of both Ser-5(P) and Ser-7(P). An analog-sensitive Kin28 mutant strain (kin28-as) and its isogenic wild-type (WT) counterpart were treated with DMSO (the solvent for NA-PP1) or 10 M NA-PP1 for 2 h at 30°C. Cells were then subjected to ChIP assays with 3E8 (␣-Ser-5(P) (␣-Ser5P)), 4E12 (␣-Ser-7(P) (␣-Ser7P)), 3E10 (␣-Ser-2(P) (␣-Ser2P)), and ␣-Rpb3 antibodies. A, ChIP analysis of the ADH1 gene (left panel) and its quantification (right panels). B, similar ChIP analysis of the snR33 gene. Serine 7 is the most degenerate position of the CTD (34). Only about half of the human repeats, mostly in the N-terminal half of the CTD, has a serine at position 7. In Drosophila, the number is even less, with only 7 out of 44 repeats containing a Ser-7. The divergence at this position has led to speculation that the nonconsensus repeats may have evolved specific functions not shared with consensus repeats. Further experiments will be required to discern its function, but the fact that Ser-7 phosphorylation appears widely conserved over evolution suggests that this modification could be important at some stage of the CTD cycle.