Termination Control by an RNA Polymerase II CTD R1810me2s-SMN Interaction One-sentence summary: Symmetric dimethylation of the human RNA polymerase II C-terminal domain (CTD) residue R1810 by the Protein Arginine Methyltransferase 5 directly recruits the protein Survival of Motor Neuron (SMN) and indirectly recruits the helicase Senataxin to resolve R-loops in transcription termination regions and possibly contribute to preventing the neurodegenerative disorder spinal muscular atrophy (SMA). Dorothy Yanling Zhao1,2,3, Gerald Gish2ǂ, Ulrich Braunschweig1ǂ, Yue Li4, Zuyao Ni1, Frank W. Schmitges1, Guoqing Zhong1, Ke Liu5, Weiguo Li5, Jason Moffat1,3, Masoud Vedadi5, Jinrong Min5, Tony J. Pawson2,3, Ben J. Blencowe1,3, Jack F. Greenblatt1,3 ,6 1. Donnelly Centre and Banting & Best Department of Medical Research, University of Toronto, Toronto, ON M5S 3E1, Canada 2. Samuel Lunenfeld Research Institute, Mount Sinai Hospital, 600 University Avenue, Toronto, ON M5G 1X5, Canada 3. Department of Molecular Genetics, University of Toronto, Toronto, ON M5S 1A8, Canada 4. Department of Computer Science, The Donnelly Centre, University of Toronto, Toronto, ON M5S 3G4, Canada 5. Structural Genomics Consortium, University of Toronto, Toronto, ON M5G 1L7, Canada 6. To whom correspondence should be addressed. E-mail: [email protected] ǂ Equal contribution 1 ABSTRACT The C-terminal domain (CTD) of the RNA polymerase II subunit POLR2A is a platform for modifications specifying the recruitment of factors that regulate transcription, messenger RNA processing, and chromatin remodeling. We now find that a CTD arginine residue (R1810 in human) that is conserved across vertebrates is symmetrically dimethylated (me2s). This R1810me2s modification requires Protein Arginine Methyltransferase 5 and recruits the Tudor domain of the Survival of Motor Neuron (SMN) protein, which is mutated in spinal muscular atrophy (SMA). SMN interacts with Senataxin, which is sometimes mutated in Ataxia Oculomotor Apraxia 2 (AOA2) and Amyotrophic Lateral Sclerosis (ALS4). Because R1810me2s and SMN, like Senataxin, are required for resolving R-loops created by RNA polymerase II in transcription termination regions, we propose that R1810me2s, SMN, and Senataxin are components of a pathway for R-loop resolution in which defects can influence transcription termination and may contribute to neurodegenerative disorders like SMA, AOA, and ALS. INTRODUCTION The CTD of the mammalian RNA polymerase II (RNAPII) subunit POLR2A contains 52 heptapeptide repeats, the N-terminal half containing mostly consensus heptads (Tyr1-Ser2-Pro3-Thr4- Ser5-Pro6-Ser7) and the C-terminal half many more heterogeneous repeats 1. These repeats can be phosphorylated on Tyr1, Thr4, and all three Ser residues, and specific phosphorylation patterns are important for various aspects of transcription, as well as co-transcriptional RNA processing and histone modifications 2-6. Two non-consensus human CTD Arg residues, R1603 and R1810, are conserved in vertebrate species. It was recently found that asymmetric dimethylation (me2a) of R1810 by the 2 CARM1/PRMT4 methyltransferase in human inhibits expression of snRNA and snoRNA genes7. It was also shown that this R1810me2a mark can be bound by the Tudor domain of TDRD3 in vitro7. We now show that the CTD residue R1810 can also be symmetrically dimethylated (me2s), a modification that requires the Protein Arginine Methyltransferase 5 (PRMT5). This R1810me2s modification recruits SMN, which then interacts with Senataxin, a helicase needed for resolving R-loops in transcription regions. RESULTS Identification of symmetric dimethylation of R1810 on the RNAPII CTD We performed immunoprecipitations using tagged TDRD3 and the RNAPII POLR2D subunit and observed that both tagged proteins could co-immunoprecipitate POLR2A with the me2a modification as detected by western blotting with the ASYMM24 antibody specific for Arg-me2a, whereas only precipitation of POLR2D, and not TDRD3, co-immunoprecipitated a form of POLR2A with an me2s modification that could be detected by the SYMM10 or Y12 antibodies specific for Arg- me2s (Extended Data Fig. 1a). We thus generated polyclonal antibodies against a CTD R1810me2s- containing 7mer peptide and found that immunoprecipitated POLR2A is recognized by the R1810me2s antibody (Fig.1a). To further determine whether the Arg-me2s modification indeed involved R1810, Raji cells stably expressing α-amanitin-resistant, HA-tagged, wild type or R1810A mutant POLR2A were generated7. After treatment with α-amanitin to deplete endogenous α-amanitin-sensitive RNAPII, followed by immunoprecipitation of RNAPII with anti-HA antibody, western blotting with antibodies recognizing R1810me2a7 or R1810me2s, as well as Y12 antibody, revealed that the R1810A mutation causes loss of both modifications, R1810me2a and R1810me2s (Fig. 1b, Extended Data Fig. 1b). The precipitated RNAPII was dephosphorylated prior to western blotting to enable more sensitive detection 3 of R1810me2s (Extended Data Fig. 1c). The specificities of the Y12 and R1810me2s antibodies are illustrated in the slot blots of Extended Data Fig.1d: they recognize an Arg-me2s peptide bracketing CTD R1810 much better than peptides with no Arg modification or R1810me2a. Therefore, R1810 is symmetrically dimethylated in cell extracts. The experiment of Extended Data Fig. 1a also indicates that TDRD3 recognizes R1810me2a in cell extracts, as well as in vitro7, although TDRD3 does not mediate inhibition of snRNA and snoRNA gene expression by R1810me2a8. Symmetric dimethylation of CTD residue R1810 requires PRMT5 The PRMT5/WDR77 complex is known to associate with RNAPII through the CTD phosphatase FCP19, which is consistent with our observation that PRMT5 co-purifies with RNAPII (Fig. 1a), so we tested whether PRMT5 might be needed to symmetrically dimethylate R1810. HEK293 cell lines stably expressing shRNAs for CARM1, PRMT5, and GFP were generated, and endogenous RNAPII was precipitated. Western blotting revealed that CARM1 knock-down causes loss of the R1810me2a mark but not the R1810me2s mark on RNAPII, whereas PRMT5 knock-down causes loss of the R1810me2s mark but not the R1810me2a mark (Fig. 1c, Extended Data Fig. 1e-f). Consistently, transient siRNA-mediated knock-down of PRMT5 also reduced R1810me2s, as detected by R1810me2s and Y12 antibodies, whereas over-expression of FLAG-tagged PRMT5 increased R1810me2s (Fig. 1d, Extended Data Figs. 1g, 3a, 3c). These experiments indicated that PRMT5 is required in vivo for the R1810me2s modification on the RNAPII CTD. PRMT5 is the catalytic subunit of the methylosome, which also contains WDR77 (MEP50)10,11. To test whether PRMT5 can directly methylate CTD arginine residues, we incubated recombinant PRMT5/WDR77 with tritiated S-adenosyl methionine (SAM [3H] ) and recombinant GST-N-CTD containing CTD repeats 1-29, which includes R1603, or GST-C-CTD containing CTD repeats 24-52, which includes R1810. Scintillation counting then monitored SAM [3H] labeling following 4 glutathione-agarose pull-down of the GST fusion proteins. This revealed that GST-C-CTD and GST- N-CTD were both methylated above the background defined by GST alone (which contains 14 arginines) (Fig. 1e). When the PRMT5 methylation assays were repeated with biotinylated 13-mer peptides containing R1810 or R1603, methylation was again observed on both R1603 and R1810 (Fig. 1f). These experiments indicated that PRMT5/MEP50 may require an additional co-factor9 and/or appropriate CTD phosphorylation to specifically methylate R1810. Recognition of R1810me2s by SMN Modified CTD residues and dimethylated Arg residues usually mediate their biological effects via interacting proteins, in the latter case by Tudor domains12,13. Because the Tudor domains of SMN and SPF30, as well as TDRD1, 2, 9, and 11, specifically bind Arg-me2s12, we used them in fluorescence polarization assays to identify whether any could bind an FITC-tagged 13mer CTD peptide containing R1810me2s. Of these, only SMN’s exhibited binding, with much lower affinity for R1810me2a and R1603me2s peptides and no detectable affinity for the unmodified peptides (Fig. 2a, Extended Data Fig. 2a, 2c). In contrast, the TDRD3 Tudor domain showed weak affinity (not shown) only for R1810me2a > R1603me2a above background (no modification or Arg-me2s), in line with published data7,14. Compared to a CTD peptide with R1810me2s alone, the presence of additional phospho-Tyr1 or –Ser2 modifications, or both, on the peptide only slightly enhanced its binding to SMN in fluorescence polarization assays and had no significant effect in isothermal titration calorimetry assays (Fig 2b, Extended Data Fig 2b, 2d ). Other phosphorylations near R1810 also had no effect on SMN binding in vitro (data not shown), indicating SMN association with R1810me2s is not greatly influenced by CTD phosphorylation. We also found, using co-immunoprecipitation, that SMN and POLR2A interact in cell extracts (Fig. 1a). Consistent with specific recognition of the R1810me2s modification by SMN, 5 immunoprecipitation of SMN from HEK293 cell extracts co-precipitated endogenous RNAPII with the R1810me2s modification but not R1810me2a (Fig. 1a, Extended Data Fig. 3b). To test whether R1810 is important for the association of RNAPII with SMN in vivo, HA-tagged wild type or mutant (R1810A) RNAPII were immunoprecipitated with anti-HA antibody. Western blotting showed that the R1810A mutation causes loss of the association of RNAPII with both SMN and TDRD3 (Fig. 2c). As expected,