Report

CTD Tyrosine Phosphorylation Impairs Termination Factor Recruitment to RNA Polymerase II

See allHide authors and affiliations

Science  29 Jun 2012:
Vol. 336, Issue 6089, pp. 1723-1725
DOI: 10.1126/science.1219651

Don't Terminate Me!

DNA transcription progresses through three phases—initiation, elongation, and termination—of messenger RNA chains. The transcribing enzyme, RNA polymerase (Pol) II, recruits factors that assist in each of these phases. Mayer et al. (p. 1723) now show that the C-terminal domain (CTD) of actively elongating Pol II is phosphorylated at conserved tyrosine residues. This modification impairs recruitment of termination factors. Factor exchange on the transcribing polymerase enzyme may be explained by an extended CTD code that is based on differential phosphorylation of the tyrosines and two well-characterized serine residues in the CTD.

Abstract

In different phases of the transcription cycle, RNA polymerase (Pol) II recruits various factors via its C-terminal domain (CTD), which consists of conserved heptapeptide repeats with the sequence Tyr1-Ser2-Pro3-Thr4-Ser5-Pro6-Ser7. We show that the CTD of transcribing yeast Pol II is phosphorylated at Tyr1, in addition to Ser2, Thr4, Ser5, and Ser7. Tyr1 phosphorylation stimulates binding of elongation factor Spt6 and impairs recruitment of termination factors Nrd1, Pcf11, and Rtt103. Tyr1 phosphorylation levels rise downstream of the transcription start site and decrease before the polyadenylation site, largely excluding termination factors from gene bodies. These results show that CTD modifications trigger and block factor recruitment and lead to an extended CTD code that explains transcription cycle coordination on the basis of differential phosphorylation of Tyr1, Ser2, and Ser5.

The C-terminal domain (CTD) is a flexible, tail-like extension of RNA polymerase (Pol) II and consists of 26 (yeast) or 52 (human) highly conserved heptapeptide repeats of the consensus sequence Tyr1-Ser2-Pro3-Thr4-Ser5-Pro6-Ser7. During the transcription cycle, changes in CTD phosphorylation patterns coordinate the recruitment of transcription and mRNA processing factors to Pol II (13). During early transcription, Ser5 phosphorylation recruits the mRNA capping enzyme (4, 5). Ser2 phosphorylation occurs during transcription elongation and functions in recruitment of RNA 3′-processing and termination factors (6). Phosphorylations at Ser7 (79) and Thr4 (10) have roles in processing of specific RNAs. Tyr1 phosphorylation was described for human Pol II almost two decades ago (11), but whether this has a functional role and whether it exists in other species are unknown.

We generated a monoclonal antibody against a Tyr1-phosphorylated CTD peptide (3D12, Methods). Because the functional CTD unit is a pair of repeats (12), we determined antibody specificity by using di-heptapeptides bearing combinations of phosphorylations (Fig. 1A and fig. S1). This revealed a high affinity for the Tyr1-phosphorylated CTD that was not impaired by adjacent Ser2 phosphorylation and no affinity to other CTD peptides (Fig. 1A and fig. S1). The antibody immunoprecipitated Pol II from extracts of the yeast Saccharomyces cerevisiae (Fig. 1B), and the precipitated polymerases were also phosphorylated at Ser2, Ser5, and Ser7 (fig. S2). The antibody also recognized Pol II that was purified from human cells with antibody 1C7 (fig. S1) and phosphorylated in vitro by the Tyr1 kinase c-Abl (13) (Fig. 1C). Thus, antibody 3D12 specifically recognizes the Tyr1-phosphorylated CTD, and Tyr1 phosphorylation occurs in yeast.

Fig. 1

Pol II CTD is phosphorylated at Tyr1. (A) Part of the CTD sequence around phosphorylated Tyr1 (Y1). Residues Ser2, Pro3, Thr4, Ser5, Pro6, and Ser7 are denoted S2, P3, T4, S5, P6, and S7, respectively. CTD residues that interfere with 3D12 antibody binding upon phosphorylation are highlighted in black. (B) Western blot analysis of whole-cell extract from proliferating yeast (Input). Pol II was immunoprecipitated with antibodies 8WG16, 3D12, and 1C7 (IP Pol II) and probed with 8WG16 or 3D12. Isotype controls are shown. Ig, immunoglobulin. (C) Antibody 3D12 detects CTD Tyr1 phosphorylation in HeLa cells (Input). Pol II was immunoprecipitated with antibody 1C7 (IP 1C7, fig. S1) and incubated with cAbl kinase, leading to a 3D12 signal (IP+cAbl). The hyper- (II0) and hypophosphorylated forms (IIA) of Pol II are indicated.

To investigate whether genome-associated Pol II is phosphorylated at Tyr1, we used high-resolution chromatin immunoprecipitation (ChIP) profiling in proliferating yeast (14). Data from two biological replicates (R = 0.94) were averaged and revealed strong signals over protein-coding and small nucleolar RNA genes (fig. S3). To test whether Tyr1 phosphorylation occurs on all transcribed protein-coding genes, we measured covariation in ChIP data for other CTD phosphorylations by singular value decomposition (14). The first singular vector explained 83.8% of the variance (fig. S4), indicating a similar occurrence of phosphorylations at Tyr1, Ser2, and Ser5. A correlation between levels of Tyr1 phosphorylation and mRNA expression (15) (fig. S5A) further indicated that Tyr1 phosphorylation is functionally relevant.

Gene-averaging of ChIP profiles (14) revealed Tyr1 phosphorylation in the coding region (Fig. 2, A and B). Whereas Tyr1 phosphorylation signals were low at promoters, they increased downstream of the transcription start site (TSS). The gene-averaged profile resembled that for Ser2 phosphorylation, except that Ser2 phosphorylation signals persist downstream of the polyadenylation (pA) site for ~200 nucleotides (nt), whereas Tyr1 phosphorylation signals decrease already around 180 nt upstream of the pA site (Fig. 2, A, B, and D). The point of Tyr1 phosphorylation signal increase was dependent on the TSS, whereas the point of decrease was dependent on the pA site but not on gene length or expression level (Fig. 2D and figs. S5B and S6). These results indicate that Tyr1 phosphorylation marks are set and removed within the transcription cycle.

Fig. 2

Gene-averaged ChIP profiles for CTD phosphorylations and termination factors. (A) DNA frame with promoter, 5′-untranslated region (UTR), open reading frame (ORF), and 3′-UTR. Dashed black lines indicate the TSS and pA sites. The dashed gray line marks the position 180 nt upstream of the pA site. (B) Gene-averaged profiles for Ser5 (14), Ser2 (14), and Tyr1 phosphorylation for 339 genes of medium length (1238 ± 300 nt). (C) Gene-averaged profiles for Nrd1, Pcf11 (14), and Rtt103. ChIP-chip occupancy of Nrd1 and Rtt103 is on the left y axis, Pcf11 occupancy on the right y axis. (D) Gene-averaged Tyr1 phosphorylation ChIP profiles for small (S) (725 ± 213 nt, 266 genes), medium (M) (B), and long (L) (2217 ± 679 nt, 299 genes) gene-length classes, aligned at the pA site.

To investigate whether Tyr1 phosphorylation influences factor recruitment to Pol II, we determined genomic occupancy profiles for termination factors Nrd1, Rtt103, and Pcf11, which contain a CTD-interacting domain (CID). The gene-averaged Nrd1 occupancy peaked at the beginning of the transcribed region, 193 ± 44 nt downstream of the TSS (Fig. 2C). This region also showed maximum signals in Ser5 phosphorylation, and genomic Nrd1 and Ser5 phosphorylation profiles correlate (R = 0.6), consistent with Nrd1 binding to the Ser5-phosphorylated CTD (16). The general presence of Nrd1 at protein-coding genes extends previous results (17, 18) and befits a role of Nrd1 in early transcription termination (2, 1921). Rtt103 showed peak occupancy at the end of genes, 112 ± 27 nt downstream of the pA site, where peak levels of Pcf11 were also observed (14) (Fig. 2C and fig. S7). Because this region shows the maximum difference between Ser2 and Tyr1 phosphorylation signals, Tyr1 phosphorylation may impair recruitment of Rtt103 and Pcf11 upstream of the pA site. Consistent with this, genome-wide occupancies of Rtt103 and Pcf11 do not correlate well with Ser2 phosphorylation signals (R = 0.4, for both), although both proteins bind the Ser2-phosphorylated CTD (22, 23).

To test whether Tyr1 phosphorylation impairs CTD binding of termination factors, we determined the affinity of purified recombinant CIDs of yeast Nrd1, Pcf11, and Rtt103 for various CTD diheptad phosphopeptides (table S1) by using fluorescence anisotropy (Fig. 3, A to C, and fig. S8). None of the CIDs bound to an unphosphorylated CTD peptide. Consistent with previous results (16, 23), Pcf11-CID and Rtt103-CID bound to the Ser2-phosphorylated CTD peptide [dissociation constants (KD) = 54 ± 6 μM (±SD) and 12 ± 2 μM, respectively; Methods], whereas the Nrd1-CID preferentially bound to a Ser5-phosphorylated CTD peptide (KD = 85 ± 25 μM). In contrast, none of the CIDs bound Tyr1-phosphorylated CTD peptides, regardless of whether additional phosphorylations were present or not. Thus, Tyr1 phosphorylation blocks CID binding to the CTD in vitro, consistent with the hypothesis that it impairs termination factor recruitment in vivo.

Fig. 3

CTD Tyr1 phosphorylation blocks termination factor binding. Fluorescence anisotropy titration of CTD peptides with recombinant Pcf11-CID (A), Rtt103-CID (B), Nrd1-CID (C), and Spt6 tandem SH2 domain (D). When possible, binding affinity was determined as the protein concentration at half-maximum binding by nonlinear Hill fit (Origin). The remaining affinity of Pcf11-CID for the Tyr1/Ser2-phosphorylated CTD peptide was not observed at higher salt concentration (fig. S7). Tyr1-phosphorylated CTD peptides were not bound by CIDs [(A) to (C)], but by Spt6 tandem SH2 domain (for Tyr1-P, KD = 3.6 ± 0.15 μM; Tyr1-P+Ser2-P, KD = 1.9 ± 0.04 μM; Tyr1-P+Ser5-P, KD = 1.3 ± 0.06 μM; Ser2-P, KD = 8.4 ± 0.19 μM; Ser5-P, KD = 5.2 ± 0.09 μM).

Structural modeling reveals how Tyr1 phosphorylation blocks the CID-CTD interaction. Crystal structures of Ser2-phosphorylated CTD peptides bound to Pcf11-CID (22) and Rtt103-CID (23) are available, and the structure of the Nrd1-CID (16) is known. In the Pcf11-CTD structure, the Tyr1 hydroxyl group forms a hydrogen bond with the Asp68 side chain in the CID (22) (fig. S9A). This indicates that Tyr1 phosphorylation blocks CTD binding because of electrostatic repulsion of two negatively charged groups, the Asp68 side chain and the Tyr1 phosphate. The CTD-binding aspartate residue is conserved in the Nrd1 CID (16). In the Rtt103-CTD structure, the corresponding residue, Asn65, forms a hydrogen bond with the Tyr1 hydroxyl group (23) that is incompatible with Tyr1 phosphorylation. Generally, a Tyr1 phosphate group modeled onto CTD peptides in CID complex structures results in steric clashes (fig. S9B).

Structural considerations also indicated that Thr4 phosphorylation (10) interferes with CID binding by destabilizing the bound CTD conformation (22). This predicted that Thr4 phosphorylation levels at the pA site are low, to enable recruitment of Pcf11 and Rtt103. Indeed, ChIP profiling revealed that Thr4 phosphorylation is limited to the transcribed region (fig. S10). Modeling further indicated that Ser7 phosphorylation is unlikely to interfere with CTD-CID binding, consistent with Nrd1 recruitment in the 5′ region of genes where Ser7 phosphorylation levels are high (14). Thus, genome-wide signals of CTD phosphorylation at Thr4 and Ser7 are consistent with the function of Tyr1 phosphorylation in impairing termination factor recruitment.

To investigate whether Tyr1 phosphorylation also impairs CTD interactions of factors with other CTD-binding domains, we investigated the tandem Src homology 2 (SH2) domain of elongation factor Spt6. This domain binds the Ser2-phosphorylated CTD (2427) and is required for high Spt6 occupancy on transcribed genes (14), suggesting that Tyr1 phosphorylation does not interfere with its CTD binding. Indeed, the recombinant domain (residues 1250 to 1444) bound very well to CTD peptides phosphorylated at Tyr1, Tyr1 and Ser2, or Tyr1 and Ser5 but not to unphosphorylated CTD (Fig. 3D). These results were consistent with recent data (26, 27) and showed that interactions with Tyr1-phosphorylated CTD peptides were even stronger than for peptides with phosphorylations at Ser2 or Ser5 alone. This shows that Tyr1 phosphorylation stimulates CTD binding of a bona fide elongation factor.

We tested whether Tyr1 phosphorylation depends on one of the yeast CTD kinases, Kin28, Srb10, Bur1, or Ctk1, which correspond to human Cdk7, Cdk8, Cdk9, and Cdk12, respectively. Inhibition of these kinases in vivo did not significantly affect Tyr1 phosphorylation signals (table S2 and fig. S11). This indicates that Tyr1 phosphorylation of the yeast CTD depends on a kinase other than the known CTD kinases. Consistent with this, Tyr1 phosphorylation in human cells is achieved by c-Abl (13), a kinase that lacks a yeast homolog.

Our results extend the previously proposed CTD code (3, 28, 29), which was based on Ser2 and Ser5 phosphorylation, leading to an extended CTD code for the coordination of the transcription cycle with factor recruitment (fig. S12). During initiation and early elongation, the CTD is phosphorylated on Ser5, which facilitates recruitment of the capping enzyme and Nrd1. Peak occupancy levels are reached for Nrd1 and Pol II 150 to 200 nt downstream of the TSS (14), likely marking a decision point where Pol II transiently pauses and either terminates or continues elongation (2). When Tyr1 and Ser2 phosphorylation levels rise, Pol II binds elongation factors stably and continues elongation. Tyr1 phosphorylation releases Nrd1 and impairs recruitment of Rtt103 and Pcf11, suppressing termination during elongation. Before the pA site, Tyr1 phosphorylation levels drop, whereas Ser2 phosphorylation levels remain high. This enables recruitment of Rtt103 and Pcf11 that is enhanced by cooperative interactions between factors (23) and with nascent RNA (18), resulting in 3′-RNA processing and transcription termination. Our results indicate that Tyr1 CTD phosphorylation is a target to activate transcription by suppressing Pol II termination and explain why mutation of Tyr1 to phenylalanine, which lacks the oxygen atom required for phosphorylation, is lethal (30).

Supplementary Materials

www.sciencemag.org/cgi/content/full/336/6089/1723/DC1

Materials and Methods

Figs. S1 to S12

Tables S1 and S2

References (3135)

References and Notes

  1. Acknowledgments: We thank S. Etzold and K. Leike for help and S. Hahn for providing yeast strains. P.C. was supported by the Deutsche Forschungsgemeinschaft (DFG), SFB646, SFB960, TR5, NIM, the BioImaging Network, a European Research Council (ERC) Advanced Grant, and the Jung-Stiftung. D.E. was supported by DFG, TR5, and SFB684. Microrray data have been deposited in ArrayExpress under accession code E-MTAB-1060. A.M. carried out ChIP-chip and fluorescence anisotropy experiments. M.H. and C.H. validated antibodies. M.L. analyzed ChIP-chip data. M.S. carried out modeling. A.S. carried out additional ChIP assays. E.K. generated the 3D12 antibody. D.E. and P.C. designed and supervised research. A.M. and P.C. prepared the manuscript, with help from all authors.
View Abstract

Navigate This Article