|Home | About | Journals | Submit | Contact Us | Français|
RNA binding proteins often contain multiple arginine glycine repeats, a sequence that is frequently methylated by protein arginine methyltransferases. The role of this posttranslational modification in the life cycle of RNA binding proteins is not well understood. Herein, we report that Sam68, a heteronuclear ribonucleoprotein K homology domain containing RNA binding protein, associates with and is methylated in vivo by the protein arginine N-methyltransferase 1 (PRMT1). Sam68 contains asymmetrical dimethylarginines near its proline motif P3 as assessed by using a novel asymmetrical dimethylarginine-specific antibody and mass spectrometry. Deletion of the methylation sites and the use of methylase inhibitors resulted in Sam68 accumulation in the cytoplasm. Sam68 was also detected in the cytoplasm of PRMT1-deficient embryonic stem cells. Although the cellular function of Sam68 is unknown, it has been shown to export unspliced human immunodeficiency virus RNAs. Cells treated with methylase inhibitors prevented the ability of Sam68 to export unspliced human immunodeficiency virus RNAs. Other K homology domain RNA binding proteins, including SLM-1, SLM-2, QKI-5, GRP33, and heteronuclear ribonucleoprotein K were also methylated in vivo. These findings demonstrate that RNA binding proteins are in vivo substrates for PRMT1, and their methylation is essential for their proper localization and function.
Protein arginine methylation is a common posttranslational modification in higher eukaryotes, but its precise role is not well understood. Arginine methylation has been shown to affect several cellular processes, including intracellular localization, protein–protein interactions as well as transcription (Gary and Clarke, 1998 ; McBride and Silver, 2001 ; Stallcup, 2001 ). In yeast, Npl3p, Hrp1p, and Nab2p are shuttling proteins involved in the transport of mRNAs across the nuclear membrane (Anderson et al., 1993 ; Lee et al., 1996 ). Arginine methylation of these two heteronuclear ribonucleoproteins (hnRNPs) is required for their nuclear export (Shen et al., 1998 ; McBride et al., 2000 ; Green et al., 2002 ). We have recently shown that arginine methylation negatively affects the binding of proline-rich ligands to Src homology 3 (SH3), but not WW, domain protein modules (Bedford et al., 2000 ). The methylation of STAT1 is required for transcriptional activation by IFNα/β by displacing PIAS1 (Mowen et al., 2001 ). Arginine methylation can also have a positive effect on protein–protein interactions. The spliceosomal snRNP proteins SmD1, D3 and B/B′ contain symmetrically dimethylated arginines (sDMAs), and this modification is necessary for enhanced interaction with SMN (Brahms et al., 2000 , 2001 ; Friesen et al., 2001a ,b ), the product of the spinal muscular atrophy gene (Lefebvre et al., 1995 ). Transcription can also be regulated by arginine methylation through the modification of histones and cofactors (Chen et al., 1999a ; Wang et al., 2001 ; Xu et al., 2001 ).
The enzymes responsible for protein arginine methylation have been classified in three groups (Gary and Clarke, 1998 ). Type I enzymes promote the formation of both NG-monomethylated and asymmetric ω-NG,NG-dimethylated arginines (aDMAs). Type II enzymes catalyze the formation of monomethylated and sDMAs. The type III enzyme found in yeast catalyzes the monomethylation of the δ-guanidino nitrogen atom of the arginine residue. The known mammalian type I enzymes are protein arginine N-methyltransferase (PRMT)1, PRMT3, coactivator-associated arginine methyltransferase 1 (CARM1), and PRMT6 (Lin et al., 1996 ; Tang et al., 1998 ; Schurter et al., 2001 ; Frankel et al., 2002 ). These PRMTs are ubiquitously expressed and may serve multiple functions. PRMT1 is thought to be the major PRMT accounting for >90% of the generation of aDMAs (Tang et al., 2000 ) and has been shown to be required for early mouse postimplantation development (Pawlak et al., 2000 ). PRMT1 was first isolated through its interaction with BTG1 and TIS21, two genes associated with cell quiescence (Lin et al., 1996 ). Most of the proteins identified as being targets of PRMT1 are methylated within glycine-arginine rich (GAR) domains (Najbauer et al., 1993 ). Many proteins involved in RNA metabolism contain such regions with clustered arginine residues in an Arg-Gly-Gly motif (RGG box) or RG repeats (Burd and Dreyfuss, 1994 ). Heteronuclear ribonucleoproteins (hnRNPs) represent a major protein family that contains methylated arginines (Liu and Dreyfuss, 1995 ).
The K homology (KH) motif is a type of RNA binding domain found in a large family of proteins associated with RNA metabolism (Gibson et al., 1993 ; Siomi et al., 1993 ). A subfamily of KH domain proteins contains an extended KH domain in a larger protein module called the GRP33, Sam68, and GLD-1 (GSD) domain (Jones and Schedl, 1995 ). This family of proteins includes the Src substrate associated in mitosis of 68 kDa (Sam68; Wong et al., 1992 ; Fumagalli et al., 1994 ; Taylor and Shalloway, 1994 ), Sam68 like mammalian protein 1 and 2 (SLM-1 and SLM-2; Di Fruscio et al., 1999 ), Artemia salina glycine rich protein (GRP33), Caenorhabditis elegans germ-line defective 1 (GLD-1; Jones and Schedl, 1995 ), and QUAKING (QKI; Ebersole et al., 1996 ). Although the cellular function of Sam68 is unknown, it can functionally substitute for Rev in the export of unspliced human immunodeficiency virus (HIV) RNAs (Reddy et al., 1999 ). The majority of KH domain RNA binding proteins contain RGG boxes and RG-rich regions, making them potential arginine methyltransferase targets.
Herein, we report that Sam68, SLM-1, SLM-2, GRP33, QKI-5, and hnRNP K are methylated in vivo. Sam68 is asymmetrically methylated by PRMT1 in vivo, as assessed by using a novel antiasymmetrically dimethylated arginine antibody and ES cells harboring a prmt1 null mutation. At steady state, Sam68 is predominantly nuclear, with no cytoplasmic staining (Sam68/SLM nuclear bodies [SNBs]; Chen et al., 1999b ). In PRMT1-deficient cells or cells treated with methylase inhibitors a significant fraction of Sam68 resided in the cytoplasm. The methylation of Sam68 is required to enhance the export of unspliced HIV RNAs. Our findings demonstrate the necessity of arginine methylation in the localization and function of KH domain RNA binding proteins.
Peptides were synthesized at W.M. Keck Biotech Resource Center (New Haven, CT). Polyclonal antibodies were generated by using New Zealand White rabbits injected with peptides coupled to keyhole limpet hemocyanin (Sigma-Aldrich, St. Louis, MO). The peptides used to generate antibodies against asymmetrically dimethylated arginines (UBI, Upstate USA, Waltham, MA) and PRMT1 were as follows: peptide 24 (KGRGRGRGRGPPPPPRGRGRGRG), where all arginine residues are asymmetrically dimethylated, and PRMT1 (KTGEEIFGTIGMRPNAKNNRD). The 9E10 anti-myc monoclonal antibody (mAb) was from the American Type Culture Collection (Manassas, VA). The polyclonal antibodies were affinity purified over the antigenic peptide coupled to Affigel beads, eluted with 100 mM glycine pH 2.5, and concentrated by using Centricon columns (Millipore, Bedford, MA). The anti-Sam68 polyclonal antibody AD-1 has been described previously (Chen et al., 1999b ). The anti-SMN mAb was purchased from Transduction Laboratories (Lexington, KY).
ELISA plates (Costar, Cambridge, MA) were coated with the indicated quantity of peptide in 50 μl of 50 mM carbonate buffer, incubated at 37°C for 30 min, and blocked with blocking buffer (1% bovine serum albumin, 5% sucrose in phosphate-buffered saline [PBS]). Primary antibodies were diluted at 1:1000 in dilution buffer (1% bovine serum albumin, 0.5% ovalbumin, 10 mM Tris pH 7.4, 150 mM NaCl) and added to the corresponding well followed by incubation at 37°C for 30 min. The plate was washed extensively with PBS containing 0.1% Tween. Goat anti-rabbit antibodies covalently coupled to horseradish peroxidase (Cappel Laboratories, Durham, NC) was incubated at 1:1000 in dilution buffer at 37°C for 15 min. The plate was washed and developed using BM Blue POD substrate (Roche Diagnostics, Indianapolis, IN) and quantitated by using spectrophotometry at 405 nm.
PRMT1 was amplified from a mouse cDNA library by using EcoRI oligonucleotides (5′-GGA GAA TTC CTG TGG CCA GGC GGA AAG-3′) and (5′-CCG GAA TTC AGC GCA TCC GGT AGT CGG-3′), subcloned into myc-pcDNA (Chen et al., 1997 ) and verified by DNA sequencing. The T7 promoter expression vectors encoding Myc-epitope–tagged Sam68, SΔN, SΔN:Δ280–339, SΔN:Δ294–339, SΔN:ΔC, QKI-5, GRP33, SLM-1, SLM-2, and hnRNP K have been described previously (Chen et al., 1997 ; Di Fruscio et al., 1999 ). Mammalian expression vectors for SΔN:Δ280–339, SΔN:Δ294–339, and SΔN:ΔC were generated by subcloning the EcoRI fragment in myc-pcDNA. The Sam68- and PRMT1-GST fusion protein were expressed in bacteria by subcloning the full-length EcoRI fragment from myc-Sam68 into pGEX-KG.
For Figures Figures1,1, B and C, 2B, and 3E, proteins were expressed in HeLa cells by using the vaccinia virus T7 expression system and lysed in a 1% Triton-based lysis buffer as described previously (Chen et al., 1997 ). In other cases, HeLa cells were transfected with LipofectAMINE PLUS (Invitrogen, Carlsbad, CA) according to manufacturer protocol. For immunoprecipitations, cell lysates were incubated on ice with the primary antibody for 1 h. Then 20 μl of a 50% protein A-Sepharose slurry was added and incubated at 4°C for 30 min with constant end-over-end mixing. The beads were washed twice with lysis buffer and once with PBS. Protein samples were analyzed on SDS-polyacrylamide gels and transferred to nitrocellulose membranes. Immunoblotting was performed using the anti-myc (9E10) or anti-PRMT1 antibodies. The designated primary antibody was followed by goat anti-mouse or goat anti-rabbit antibodies conjugated to horseradish peroxidase (ICN Pharmaceuticals, Costa Mesa, CA) and chemiluminescence was used for protein detection (DuPont, Wilmington, DE).
Endogenous Sam68 was immunopurified from 5 × 108 HeLa cells by using 1 mg of polyclonal anti-Sam68 antibody (AD1) coupled to 1 g of protein A-Sepharose (Sigma-Aldrich). After extensive washings with lysis buffer, 1× PBS, the bound proteins were eluted with 500 μl of 1× PBS containing 250 μM of the immunogenic Sam68 peptide. Eluted proteins were resolved by SDS-PAGE and revealed by Coomassie Blue R-250 staining. The apparent band at 68 kDa was excised, in-gel digested with trypsin, and sent to the University of Calgary Mass Spectrometry Proteomics Facility for MALDI-TOF analysis on a Voyager DE-STR mass spectrometer.
Myc-tagged proteins (PRMT1, Sam68, SLM-1, SLM-2, GRP33, hnRNP K, or QKI-5) were expressed in HeLa cells and immunoprecipitated with anti-myc antibodies. The immune complex was incubated with 0.55 μCi of [methyl-3H]S-adenosyl-l-methionine (3H-SAM; PerkinElmer Life Sciences, Boston, MA) in the presence of 25 mM Tris-HCl at pH 7.5 for 1 h at 37°C in a final volume of 30 μl. Reactions were stopped by adding 20 μl of 2× SDS-PAGE sample buffer, followed by heating at 100°C for 10 min. Samples were loaded on 10% SDS-polyacrylamide gels and stained with Coomassie Blue. The destained gels were soaked in EN3HANCE (PerkinElmer Life Sciences) according to manufacturer instructions and visualized by fluorography. In vivo methylation labeling was performed by metabolically labeling cells with l-[methyl-3H]methionine directly in methionine-free medium for 3 h in the presence of cycloheximide and chloramphenicol as described previously (Liu and Dreyfuss, 1995 ). l-[35S]methionine was also used as a control under the same conditions. The cell lysates were immunoprecipitated and the 3H-labeled proteins were visualized by fluorography. ES cells were obtained from Earl H. Ruley (Vanderbilt University, Nashville, TN) and were cultured in the absence of a feeding layer in DMEM (no. 11960-044; Invitrogen) supplemented with 15% fetal bovine serum, 2% l-glutamine, 1.2% minimal essential medium nonessential amino acids (no. 11140-050; Invitrogen), 1.2% Pen/Strep, 1× 2-β-mercaptoethanol, and 0.5 ng/ml leukemia inhibitory factor.
HeLa cells were cultured directly on coverslips in a six-well dish. Cells plated on coverslips were incubate for 24 h with the vehicle (DMSO) or with the methyltransferase inhibitor adenosine-2′,3′-dialdehyde (AdOx) (Sigma-Adrich) at a final concentration of 1 mM. Transfection of HeLa cells for immunofluorescence was achieved using LipofectAMINE PLUS (Invitrogen) according to the manufacturer protocol by using 2 μg of DNA. Cells were fixed with 1% paraformaldehyde in 1× PBS at pH 7.4 for 5 min at room temperature (RT) and permeabilized with 0.5% Triton X-100 in PBS for 5 min at RT. The cells were incubated with the primary antibodies at RT for 1 h in PBS. The cells were washed with 0.1% Triton X-100 in PBS and incubated with the appropriate secondary antibodies in PBS. Goat anti-rabbit coupled to Alexa 488 (Molecular Probes, Eugene, OR) and goat anti-mouse coupled to Alexa 546 (Molecular Probes) were used as secondary antibodies. The cells were washed again, mounted onto glass slides, and visualized with a confocal microscope (Carl Zeiss, Thornwood, NY).
COS cells were transfected with a total of 1 μg of DNA containing 0.2 μg of Rev response element (RRE) chloramphenicol acetyltransferase (CAT) reporter plasmid pDM128, 0.7 μg of Rev expression vector B1-SVH6Rev, 0.7 μg of Rev mutant B1-SVH6RevM10, or 0.7 μg of pcDNA-Sam68, along with increasing concentrations of AdOx (0, 100, and 250 μM). The β-galactosidase expression vector pCH110 (0.1 μg; Amersham Biosciences, Piscataway, NJ) was included in all transfections for measuring the efficiency of transfection. Forty-eight hours after transfection, the cells were collected and resuspended in 150 μl of 0.25 M Tris-HCl, pH 7.8. The cell extracts were prepared by three freeze-thaw cycles, followed by a brief centrifugation to remove cell debris. CAT activity was normalized to the β-galactosidase activity and did not exceed twofold.
Metabolic labeling of HeLa cells by using l-[methyl-3H]methionine was performed in the presence of translation inhibitors to examine whether Sam68 was methylated in vivo. The presence of translation inhibitors ensured that the labeling was due to posttranslational methylation, not a low level of translational incorporation of l-[methyl-3H]methionine in newly synthesized proteins (Liu and Dreyfuss, 1995 ). Sam68 and a control known not to be methylated, SMN, were immunoprecipitated from cells that were metabolically labeled. The immunoprecipitated proteins were resolved by SDS-PAGE and visualized by fluorography. The fluorograph revealed that Sam68 was labeled in HeLa cells and that SMN was not (Figure (Figure1A,1A, lanes 1–4). Immunoblotting with anti-Sam68 and anti-SMN antibodies confirm that Sam68 and SMN were immunoprecipitated, respectively (Figure (Figure1A,1A, lane 5–8). Performing the metabolic labeling with l-[35S]methionine in the presence of protein synthesis inhibitors revealed that virtually no label is incorporated under these conditions (Figure (Figure1A,1A, lane 10). This control verifies that the protein synthesis inhibitors are functioning properly. These findings demonstrate that Sam68 is methylated in vivo.
Many other KH domain RNA binding proteins contain RG-rich sequences that are potential sites of methylation by PRMTs. The ability of other KH domain RNA binding proteins to become methylated in HeLa cells was examined by a transient transfection assay followed by metabolic labeling with l-[methyl-3H]methionine. Expression vectors encoding myc-epitope–tagged Sam68 (positive control), SLM-1, SLM-2, GRP33, the nuclear QKI-5 isoform, and hnRNP K were transfected in HeLa cells. Anti-myc immunoprecipitates were examined for incorporation of 3H-methyl groups (Figure (Figure1B,1B, lanes 1–6) and protein expression was assessed by immunoblotting total cell lysates (Figure (Figure1B,1B, lanes 7–12). Sam68, SLM-1, SLM-2, GRP33, QKI-5, and hnRNP K were methylated (Figure (Figure1B,1B, lanes 1–6). These findings suggest that many KH domain proteins are methylated in vivo.
PRMTs are known to frequently associate with their substrates (Gary and Clarke, 1998 ). The ability of KH domain proteins to coimmunoprecipitate PRMT activity was examined in myc immunoprecipitates of cells transiently transfected with myc-epitope–tagged KH domain RNA binding proteins. Anti-myc immunoprecipitates were incubated in the presence of [methyl-3H]S-adenosyl-l-methionine and the ability of the KH domain RNA binding proteins to serve as an exogenous substrates was examined. The proteins were resolved by SDS-PAGE, stained with Coomassie Blue for protein expression (Figure (Figure1C,1C, lanes 7–12), and the methylated proteins were visualized by fluorography (Figure (Figure1C,1C, lanes 1–6). Sam68, GRP33, hnRNP K, and SLM-2 coimmunoprecipitated protein methyltransferase activity, whereas SLM-1 and QKI-5 did not (Figure (Figure1C,1C, lanes 1–6). The reason for the absence of protein methyltransferase activity in SLM-1 and QKI-5 is unknown. Our findings demonstrate that Sam68, GRP33, hnRNP K, and SLM-2 are methylated in vitro by associated protein methyltransferases.
We have previously reported that the RG repeats flanking the P3 and P4 domains in Sam68 are asymmetrically dimethylated by glutathione S-transferase (GST)-PRMT1 in vitro (Bedford et al., 2000 ). To determine whether the protein methyltransferase activity associated with KH-containing RNA binding proteins could be attributed to endogenous PRMT1, we carried out coimmunoprecipitation experiments. A polyclonal PRMT1 antibody was generated using a peptide derived from a unique sequence of PRMT1. The antibody recognized recombinant GST-PRMT1, but not GST by immunoblotting (Figure (Figure2A).2A). The antibody recognized a major protein species of 45 kDa in HeLa cells, corresponding to the predicted molecular mass of human PRMT1 (Figure (Figure2A,2A, lane 4). The antibody also recognized proteins with molecular masses of 120, 48, and 42 kDa that were competed away using the immunogenic peptide (Figure (Figure2A,2A, lane 5). The p48 and p42 may represent PRMT1 isoforms, however, the identity of p120 is unknown.
Myc-tagged Sam68, SLM-1, SLM-2, QKI-5, hnRNP K, and GRP33 were expressed in HeLa cells and cell lysates were immunoprecipitated with control mouse IgG or anti-myc antibodies. The presence of coimmunoprecipitating PRMT1 was detected by immunoblotting with anti-PRMT1 antibodies (Figure (Figure2B).2B). Endogenous PRMT1 coimmunoprecipitated with Sam68, SLM-2, hnRNP K, and GRP33 (Figure (Figure2B,2B, lanes 3, 9, 15, and 18). P120, p48, and p42 were visible in total cell lysates (TCL) and anti-myc immunoprecipitations upon longer exposures (our unpublished data). Little or no PRMT1 was detected in SLM-1 and QKI-5 immunoprecipitates (Figure (Figure2B,2B, lanes 6 and 12), consistent with Figure Figure1C.1C. Equivalent expression of the myc-tagged proteins was assessed by immunoblotting with anti-myc antibodies (Figure (Figure2C).2C). These findings show that the KH domain RNA binding proteins Sam68, SLM-2, hnRNP K, and GRP33 interact with endogenous PRMT1.
The association of Sam68 with PRMT1, a protein methyltransferase that mediates the generation of aDMAs (Lin et al., 1996 ), suggests that Sam68 may contain aDMA in vivo. To examine whether Sam68 contained aDMAs in vivo, we generated an antibody that recognizes aDMA-containing polypeptides (see MATERIALS AND METHODS). Rabbits were immunized with peptide #24aDMA covalently coupled to keyhole limpet hemocyanin. The affinity-purified polyclonal antibody, named ASYM24, recognized strongly the aDMA-containing peptide #24aDMA, but showed no reactivity to the unmethylated peptide #24 or to a symmetrical dimethylarginine-containing peptide #24sDMA (Figure (Figure3,3, A and B; peptides 24aDMA, 24 and 24sDMA, respectively). We examined whether ASYM24 could also recognize any peptide containing aDMA. ASYM24 was generated using a peptide with neighboring proline motifs. Three proteins known to have this type of arrangement are Sam68, hnRNP K, and Wiscott-Aldrich Syndrome protein (WASP). Peptides from these proteins were generated containing aDMA and tested by ELISA with ASYM24. ASYM24 only recognized a peptide containing aDMAs corresponding to the proline motif P3 of Sam68 (Figure (Figure3B).3B). The unmethylated Sam68 P3 peptide was not recognized (Figure (Figure3B).3B). These results demonstrate that ASYM24 is an aDMA-specific antibody that has selectivity toward some polypeptides containing aDMA and recognizes a methylated sequence neighboring the Sam68 proline motif P3.
To determine whether full-length Sam68 is recognized by ASYM24, myc-tagged Sam68 was expressed in HeLa cells. Proteins from cell lysates were immunoprecipitated with control IgG or anti-myc antibodies. Immunoprecipitated Sam68 was indeed recognized by ASYM24 but not by normal rabbit serum (NRS) or ASYM24 preadsorbed with the immunogenic peptide (+24 pept.; Figure Figure3C,3C, lanes 4–12). ASYM24 also recognized unknown endogenous proteins in HeLa cells with molecular masses of 110 and 75 kDa (lane 7). Immunoblotting with anti-myc antibodies was carried out to control for protein expression and efficiency of the immunoprecipitation procedure (Figure (Figure3C,3C, lanes 1–3). To examine whether PRMT1 could increase the ASYM24 immunoreactivity of Sam68, recombinant GST-Sam68 was purified from bacteria, an organism that does not contain arginine methyltransferase activity (Gary and Clarke, 1998 ). The GST-Sam68 fusion protein was left untreated (mock) or methylated in vitro with purified GST-PRMT1 in the presence of unlabeled methyl donor S-adenosyl-l-methionine. The level of methylation was examined by immunoblotting with the aDMA-specific antibody ASYM24 (Figure (Figure3D).3D). ASYM24 recognized the methylated GST-Sam68, but not the unmethylated GST-Sam68 (Figure (Figure3D,3D, lanes 3 and 4). Anti-Sam68 AD1 antibody confirmed equal levels of GST-Sam68 (Figure (Figure3D,3D, lanes 1 and 2). To verify that this was also true in vivo, PRMT1 was coexpressed with Sam68 in HeLa cells. Again, the expression of PRMT1 dramatically increased the ability of Sam68 to be recognized by ASYM24 (Figure (Figure3E,3E, compare the 68-kDa protein in lanes 3 and 4), further suggesting that PRMT1 is able to methylate Sam68 in vivo. Finally, we wanted to address whether endogenous Sam68 contained aDMAs. Lysates were prepared from HeLa and HEK293 cells and immunoprecipitations were performed with three distinct anti-Sam68 antibodies (polyclonal sc333, AD1, and monoclonal 7-1). Immunoblotting revealed that endogenous Sam68 from HeLa and human embryonic kidney (HEK)293 is also recognized by ASYM24 (Figure (Figure3F,3F, lanes 1–9), suggesting that the arginines neighboring the proline motif P3 of Sam68 are asymmetrically dimethylated in vivo. The endogenous bands of 110 and 75 kDa recognized by ASYM24 did not coimmunoprecipitate with Sam68 and are likely to represent other asymmetrically dimethylated proteins in the cell (Figure (Figure33F).
To determine whether other regions of Sam68 were methylated, matrix-assisted laser desorption ionization/time of flight (MALDI-TOF) mass spectrometry was carried out with endogenous Sam68 purified from HeLa cells. Three domains of Sam68 harbor arginine-glycine motifs that could potentially be methylated (Table (Table1;1; P0, P3, and P4). By knowing that methylation of arginines prevents cleavage by trypsin (Baldwin and Carnegie, 1971 ), predicted peptide sizes that would result from monomethylated or dimethylated arginines were calculated and compared with the peptide masses identified by MALDI-TOF mass spectrometry. Arginines at position 45 and 52 were found to be dimethylated (Table (Table1).1). However, some peptides contained dimethylated R45 but unmodified R52, perhaps indicating that R45 is methylated first, followed by R52. No peptides containing only monomethylated arginines were identified for this region. These two arginines are the only arginines followed by two glycines, which is characteristic of the RGG motif found in hnRNPs. Arginines 310, 315, 320, and 325, which are located in between the proline motifs P3 and P4 were found to be monomethylated (Table (Table1).1). No peptides containing dimethylated arginines were identified for this region, indicating that those arginines are mostly found monomethylated endogenously. Interestingly, these arginines are each distanced by exactly four amino acids and are part of a distinct RG(A/T)XV motif. Finally, arginine 304, the second arginine in the RGRG motif downstream of the proline motif P3, was found to be dimethylated (but never monomethylated), whereas arginine 302 was unmodified. The absence of peptides containing methylated arginines from the region upstream of proline motif P3 is likely due to technical limitations (i.e., inability to detect very small or very large peptides). Overall, this analysis has revealed that Sam68 is endogenously methylated at least on seven of the 14 arginines found in an arginine-glycine context.
To address whether the RG repeats upstream of the proline motif P3 are methylated in vivo, we performed the methylation labeling assay by using Sam68 deletion mutants (Figure (Figure4A).4A). A Sam68 protein was engineered that removed all the RG repeats surrounding P3 except one (SΔN:Δ280–339). An expression vector encoding SΔN:Δ280–339 was transfected in HeLa cells, the cells were metabolically labeled and the level of in vivo methylation assessed by immunoprecipitating with anti-myc antibodies followed by SDS-PAGE and fluorography. SΔN:Δ280–339 was unable to incorporate 3H-methyl groups (Figure (Figure4B,4B, lane 2). Note that all arginines found to be methylated in the analysis by mass spectrometry are missing in the deletion mutant SΔN:Δ280–339, which is not methylated at all upon transfection. Two additional constructs were generated: one that added four RG repeats N-terminal to the proline motif P3 (SΔN:Δ294–339) and the other that restored the entire RG-rich region between proline motifs P3 and P4 (SΔN:ΔC). Both SΔN:Δ294–339 and SΔN:ΔC proteins were methylated in vivo, indicating that at least one of the RGs upstream of P3 is methylated (Figure (Figure4B,4B, lanes 3 and 4). Equivalent expression of myc-tagged proteins was confirmed by anti-myc immunoblotting of HeLa total cell lysates (Figure (Figure4B,4B, lanes 5–8). The deletion analysis has demonstrated that the RGs upstream of P3 (aa 280–294) are sufficient to observe in vivo methylation.
To determine whether methylation regulates Sam68 localization, HeLa cells were transfected with the Sam68 expression vectors encoding the various Sam68 deletion proteins described above. The cells were fixed, permeabilized, and the localization of myc-epitope–tagged Sam68 was detected by indirect immunofluorescence. The unmethylated Sam68 protein (SΔN:Δ280–339) had a whole cell distribution and behaved somewhat like a Sam68 devoid of its C-terminal nuclear localization signal (SΔN:ΔC; Figure Figure4C).4C). Interestingly, the addition of four RG repeats that results in the methylation of Sam68 (SΔN:Δ294–339) was sufficient to concentrate Sam68 in the nucleus, as observed with the wild-type protein (Figure (Figure4C).4C). These results uncover a direct correlation between the methylation of Sam68 (amino acids 280–294) and the cellular localization of the protein. Moreover, the fact that the cytoplasmic Sam68 protein (SΔN:ΔC) was methylated suggests that methylation of Sam68 can occur in the cytoplasm.
PRMT1 is thought to reside predominantly in the nucleus (Tang et al., 1998 ); however, it has been identified through purification from a cytoplasmic macromolecular complex (Lin et al., 1996 ). To further examine the localization of PRMT1, we transfected HeLa cells and performed indirect immunofluorescence by using anti-myc antibodies (Figure (Figure4D).4D). PRMT1 was localized mostly in the cytoplasm, although some of the cells showed both cytoplasmic and nuclear staining. The anti-PRMT1 antibodies were also used for immunofluorescence studies and were unable to stain endogenous PRMT1 (our unpublished data), perhaps indicating that our peptide antibody does not recognize the native folded protein or that PRMT1 is present in large complexes inaccessible to our antibody. A GFP-PRMT1 was recently reported to reside in both cellular compartments (Frankel et al., 2002 ).
To confirm that methylation of endogenous Sam68 is required for its nuclear localization, we examined whether methylase inhibitors could affect Sam68 cellular distribution. HeLa cells were treated with the drug vehicle dimethyl sulfoxide (DMSO, Figure Figure5,5, A and C) or with 1 mM AdOx (Figure (Figure5,5, B and D) for 24 h. The cells were fixed, permeabilized, and immunofluorescence was performed using the anti-Sam68 AD1 antibody (Figure (Figure5,5, C and D). The treatment of Sam68 with DMSO did not alter the localization of Sam68 in the nucleoplasm and in Sam68/SLM nuclear bodies (SNBs; Chen et al., 1999b ). The methylase inhibitor AdOx increased the cytoplasmic staining of Sam68 and SNBs were not visible (Figure (Figure5D).5D). Strikingly, the distribution of Sam68 in both the cytoplasm and the nucleus was comparable with the nonmethylated mutant SΔN:Δ280–339. To show that Sam68 methylation was effectively reduced by the AdOx treatment, we performed metabolic labeling to assess the levels of protein methylation (Figure (Figure5E).5E). A drastic reduction of Sam68 methylation was observed in the presence of AdOx and no change in protein level was observed (Figure (Figure5E).5E). Comparable results were obtained using another general methylation inhibitor 5′-deoxy-5′-methylthioadenosine (our unpublished data). These results suggest that arginine methylation is essential for the normal localization of Sam68.
To provide definitive proof for the involvement of PRMT1 in the methylation of Sam68 in vivo, we made use of embryonic stem (ES) cells in which insertion of a gene trap retrovirus into the second intron of the prmt1 gene has created essentially a null mutation (Pawlak et al., 2000 ). Ruley and coworkers reported that low levels of prmt1 transcripts (~1% of wild-type levels) can be detected in −/− ES cells (Pawlak et al., 2000 ). Consistent with this observation, our PRMT1 antibody permitted the detection of very low residual levels of PRMT1 in −/− ES cells (Figure (Figure6A,6A, compare lanes 1 and 3). Among the additional polypeptides that our antibody recognizes, p42 and p48 are also reduced, but p120 levels remain identical (Figure (Figure6A,6A, lanes 1–3), suggesting it represents a cross-reacting band that is not related to PRMT1. Numerous proteins can be detected in wild-type ES cells by immunoblotting with our ASYM24 aDMA-specific antibody (Figure (Figure6A,6A, lane 4). As expected if PRMT1 activity is greatly reduced, the majority of these polypeptides are no longer recognized by ASYM24 in PRMT1 −/− ES cells (Figure (Figure6A,6A, lane 6).
Metabolic labeling using l-[methyl-3H]methionine was performed in the presence of translation inhibitors to examine the methylation state of Sam68 in these cells where PRMT1 activity was greatly reduced. Immunoprecipitations were carried out with normal rabbit serum (CTRL), anti-Sam68, or anti-SMN antibodies. The immunoprecipitated proteins were resolved by SDS-PAGE and visualized by fluorography. The fluorograph revealed that Sam68 methylation was severely abrogated in PRMT1 −/− ES cells (Figure (Figure6B,6B, compare lanes 2 and 10). As expected and shown in Figure Figure1A,1A, SMN did not incorporate any [3H]methionine (Figure (Figure6B,6B, lanes 4, 8, and 12). Immunoblotting with anti-Sam68 and anti-SMN antibodies confirmed that Sam68 and SMN were present in equal amounts in each cell extracts (Figure (Figure6B,6B, lanes 13–15 and 16–18, respectively). These findings clearly establish PRMT1 as an enzyme responsible for the methylation of Sam68 in vivo.
We next examined the cellular localization of Sam68 in PRMT1-deficient ES cells because Sam68 is no longer methylated in these cells (Figure (Figure6).6). Indirect immunofluorescence was performed as described above by using the anti-Sam68 AD1 antibody and the cells were visualized using a confocal microscope (Carl Zeiss). ES cells carrying both wild-type alleles of PRMT1 (+/+) showed a normal nuclear localization of Sam68 (Figure (Figure7,7, A and B). In contrast, some cytoplasmic staining of Sam68 was detectable in PRMT1 −/− cells (Figure (Figure7,7, C and D). These results confirm that Sam68 arginine methylation by PRMT1 influences its cellular localization.
Sam68 has been shown to function as a cellular homologue of Rev in accelerating the transport of HIV RNAs from the nucleus to the cytoplasm (Reddy et al., 1999 ). We examined whether arginine methylation could regulate this function of Sam68 by determining whether methylase inhibitors could modulate RRE-directed reporter gene expression (Figure (Figure8).8). Transport to the cytoplasm and expression of the intronic CAT gene is Rev and RRE dependent, as described previously (Hope et al., 1990 ). COS7 cells were transfected with the RRE-CAT reporter plasmid in the presence of either an empty plasmid (pcDNA), expression vectors for Rev, an inactive mutant of Rev (Rev M10), and Sam68. After transfection, cells were treated with increasing concentrations of AdOx or DMSO and harvested for CAT activity (Figure (Figure8).8). The transfection of Sam68 or Rev with the RRE-CAT reporter resulted in an approximately fivefold increase in CAT activity relative to an empty plasmid or RevM10 (Figure (Figure8).8). The presence of the methylase inhibitor AdOx decreased CAT activity in a dose-dependent manner in Sam68, but not Rev-transfected cells (Figure (Figure8).8). Sam68 deletion mutants (SΔN:Δ280–339 and SΔN:Δ294–339) were unable to substitute for Rev in this assay, although they did not act as dominant negative as was reported for the SΔN:ΔC mutant (our unpublished data; Reddy et al., 1999 ). These findings suggest that arginine methylation of Sam68 regulates its ability to export RRE-containing HIV RNAs.
Herein, we report that the KH domain containing Sam68 harbors asymmetric dimethylarginines near its proline motif P3 in vivo by using a novel anti-aDMA–specific antibody. PRMT1 interacts with Sam68 and is a major PRMT responsible for Sam68 methylation in vivo. Arginine methylation was necessary to localize Sam68 to the nucleoplasm, where it normally resides. The methylation of Sam68 was required for its ability to facilitate the export of unspliced HIV RNAs into the cytoplasm. We demonstrate that other KH domain RNA binding proteins, including SLM-1, SLM-2, QKI-5, hnRNP K, and GRP33 are also methylated in vivo. Our findings suggest that arginine methylation may be an essential maturation step for the proper localization and function of RG-rich–containing RNA binding proteins.
The association of KH domain RNA binding proteins with PRMT1 demonstrates a link between RNA and PRMT1. The disruption of ribonucleoprotein complexes by RNase treatment increases the accessibility of proteins to in vitro methylation by PRMT1 (Frankel and Clarke, 1999 ). Moreover, the yeast Hmt1p/Rmt1p was identified in a screen for interactors of poly-(A)–associated proteins (Henry and Silver, 1996 ). These observations suggest that PRMTs are present in ribonucleoprotein complexes with their substrates.
SLM-1 and QKI-5 are methylated in vivo, but did not coimmunoprecipitate a protein methyltransferase activity and did not interact with PRMT1. Thus, SLM-1 and QKI-5 may be the target for a different PRMT that may not require a tight interaction with its substrates for activity. The fact that SLM-1 did not associate with PRMT1 is surprising because it shares a high degree of homology with Sam68 and SLM-2 and thus has a comparable number of RG repeats. Examining the RG sequences in SLM-1, compared with those of Sam68 and SLM-2 might provide new insights about the preferred motif of PRMT1 vs. other PRMTs. We have found that QKI-5 is weakly methylated in vivo. But what seems to be weak in vivo methylation could in fact be the lower amount of incorporated labeled methyl groups due to the few RG repeats found in QKI-5.
By using biochemical methods, it has been previously shown in insect cells that Sam68 contained methylated arginines in its sequence (Wong et al., 1992 ). We now provide evidence by using a novel aDMA-specific antibody that mammalian Sam68 harbors aDMA in vivo. The sites of arginine methylation are two hnRNP-like RGG motifs at position 45 and 52 in the N-terminal region as well as multiple RG repeats surrounding proline motif P3 as determined by mass spectrometry. Furthermore, the deletion analysis revealed that at least one arginine is methylated between amino acids 280–294 upstream of proline motif P3. All these methylation sites fit the PRMT1 GAR and RxR consensus sequence (Gary and Clarke, 1998 ; Smith et al., 1999 ). Interestingly, four RG repeats (310, 315, 320, and 325) located between proline motifs P3 and P4 were found to be monomethylated. It remains to be determined whether these represent precursors or intermediates leading to dimethylation. Arginine 3 of histone H4 has been shown to contain monomethylated arginines at steady state, and PRMT1 is thought to be the major, if not exclusive, methyltransferase responsible for this modification (Strahl et al., 2001 ).
The present study shows that arginine methylation of Sam68 is a prerequisite for its exclusive nuclear localization. HnRNP A2, another RNA binding protein methylated by PRMT1, accumulates in the cytoplasm in the presence of general methylation inhibitors and its nuclear localization requires an intact RGG domain (Nichols et al., 2000 ). Similarly, nuclear accumulation of high-molecular-weight fibroblast growth factor-2 could also be linked to methylation (Pintucci et al., 1996 ). These observations suggest that arginine methylation of RNA binding proteins regulates their nucleocytoplasmic transport. Indeed, arginine methylation facilitates the nuclear export of Npl3p, Nab2p, and Hrp1p in yeast and the methyltransferase activity of Hmt1p is required for this effect (Shen et al., 1998 ; McBride et al., 2000 ; Green et al., 2002 ). Moreover, arginine methylation of Npl3p RGG motif interferes with its Sky1p-mediated phosphorylation, which in turn prevents the interaction of Npl3p with the nuclear import receptor Mtr10p (Yun and Fu, 2000 ). This provides an indirect mechanism by which arginine methylation can affect nuclear import. The reason for Sam68 cytoplasmic accumulation is unknown, but it can be reasoned that 1) arginine methylation is cotranslational and newly synthesized Sam68 accumulates in the cytoplasm, or 2) unmethylated Sam68 is imported at a slower rate than it is exported. The use of protein synthesis inhibitors with methylase inhibitors rule out the first possibility. Thus, the likely possibility is that the nucleocytoplasmic transport of Sam68 is altered by methylation. This is consistent with the fact that Sam68 cannot functionally export HIV RNAs in the presence of methylase inhibitors.
The presence of conserved RGG boxes and RG-rich repeats in RNA binding proteins is thought to be necessary to mediate interactions with RNA. Indeed, this has been demonstrated for several RNA binding proteins, including Sam68, FMRP, hnRNP A1, and U (Kiledjian and Dreyfuss, 1992 ; Chen et al., 2001 ; Darnell et al., 2001 ). A major family of proteins containing aDMAs is the hnRNP RNA binding proteins (Liu and Dreyfuss, 1995 ) and the presence of aDMAs in other RNA binding proteins is rapidly being discovered. What may be the role of arginine methylation in the function and/or turnover of RNA binding proteins? It is conceivable that arginine methylation could modulate the contribution of RNA binding by the arginine-rich regions. Methylation does not change the positive charge of the arginines, but increases hydrophobicity and might prevent hydrogen bonding and/or introduce steric constraints that should reduce charge-induced interactions with RNA (Calnan et al., 1991 ). In agreement with this prediction, a reduction in hnRNP A1 binding to nonspecific single-stranded nucleic acid was observed subsequent to its arginine methylation (Rajpurohit et al., 1994 ). However, other studies have reported no significant effect of arginine methylation on RNA binding (Valentini et al., 1999 ; Raman et al., 2001 ). We have observed a modest effect of arginine methylation on Sam68 RNA binding activity in vitro, by using gel mobility shift assays and poly (U) homopolymer binding assays (our unpublished data). However, no difference in Sam68 binding to poly (U)-Sepharose could be detected between PRMT1 +/+ and PRMT1 −/− ES cell extracts (our unpublished data). Nonetheless, arginine methylation might affect interactions with specific RNA targets or structures that would have evolved to better accommodate the bulkier methyl groups. It is becoming increasingly clear that arginine methylation will not be a quick and rapid mode of regulation such as phosphorylation. Our data with Sam68 support the hypothesis that arginine methylation may be a maturation step for certain proteins such as RNA binding proteins. We define maturation as a step necessary for the proper localization of a protein and a step that is required to enhance its function. This maturation step may be either cotranslational or coupled to the nuclear import machinery. Because arginine methylation is most likely irreversible (Gary and Clarke, 1998 ), it may be a constitutive pathway for the maturation of proteins. Interestingly, four of the 12 neighboring Sam68 RG repeats were sufficient to properly localize Sam68 in the nucleus, implying that partial methylation may be sufficient to target the protein to its proper localization. The methylation of the RG repeats of Sam68 may be regulated and occur initially in the cytoplasm and then in the nucleus. Moreover, the level of RG methylation of certain proteins may reflect the “age” of the protein. The presence of multiple neighboring RG repeats also implies that methylation may be a processive event, especially because PRMTs associate with their substrates. The maturation hypothesis our data supports may apply to histones and Sm proteins.
In summary, this study demonstrates that Sam68 is asymmetrically dimethylated on several arginine residues by PRMT1 in vivo. This modification is required for proper Sam68 localization in the nucleus and for one of its functions, the ability to substitute for Rev in exporting unspliced HIV RNAs. These results suggest a general role for arginine methylation in the maturation and function of RNA binding proteins.
We thank Dave Schriemer (University of Calgary, Alberta, Calgary, Canada) for the mass spectrometry analysis and Yi Zhang and Earl H. Ruley for kindly providing PRMT1 −/− ES cells. This work was supported by grant 011291 from the National Cancer Institute of Canada with funds from the Canadian Cancer Society of Canada and grant MT13377 from the Canadian Institutes of Health Research. F.-M.B. and J.C. are recipients of a studentship and a postdoctoral fellowship from the National Cancer Institute of Canada, respectively. M.T.B. is supported by a Welch Foundation grant (G-1495). S.R. is a recipient of a Canadian Institutes of Health Research Investigator award.
Article published online ahead of print. Mol. Biol. Cell 10.1091/mbc.E02–08–0484. Article and publication date are at www.molbiolcell.org/cgi/doi/10.1091/mbc.E02–08–0484.