Regulation of the Nuclear Poly(A)-binding Protein by Arginine Methylation in Fission Yeast*

Two structurally different poly(A)-binding proteins (PABP) bind the poly(A) tract of mRNAs in most mammalian cells: PABPC in the cytoplasm and PABP2/PABPN1 in the nucleus. Whereas yeast orthologs of the cytoplasmic PABP are characterized, a gene product homologous to mammalian PABP2 has not been identified in yeast. We report here the identification of a homolog of PABP2 as an arginine methyltransferase 1 (RMT1)-associated protein in fission yeast. The product of the Schizosaccharomyces pombe pab2 gene encodes a nonessential nuclear protein and demonstrates specific poly(A) binding in vitro. Consistent with a functional role in poly(A) tail metabolism, mRNAs from pab2-null cells displayed hyperadenylated 3′-ends. We also show that arginine residues within the C-terminal arginine-rich domain of Pab2 are modified by RMT1-dependent methylation. Whereas the arginine methylated and unmethylated forms of Pab2 behaved similarly in terms of subcellular localization, poly(A) binding, and poly(A) tail length control; Pab2 oligomerization levels were markedly increased when Pab2 was not methylated. Significantly, Pab2 overexpression reduced growth rate, and this growth inhibitory effect was exacerbated in rmt1-null cells. Our results indicate that the main cellular function of Pab2 is in poly(A) tail length control and support a biological role for arginine methylation in the regulation of Pab2 oligomerization.

The 3Ј-end of nearly all eukaryotic mRNAs harbors a poly(A) tail. Whereas numerous studies indicate that the presence of a poly(A) tail on a eukaryotic mRNA is required for efficient nuclear export, RNA stability, and translational control (1), recent findings have questioned the direct role of the poly(A) tail in specific steps of mRNA metabolism (2)(3)(4). The length of this stretch of polyadenosines is controlled in a species-dependent manner with average lengths of 70 and 250 nucleotides in yeast and mammals, respectively (5,6). Yet, the exact mechanism that controls poly(A) tail length remains elusive. mRNA poly(A) tail synthesis involves at least 20 different proteins in the yeast Saccharomyces cerevisiae. The cleavage factors I and II (CF I & II) are first responsible for the cotranscriptional cleavage of the pre-mRNA, which is then followed by the polyadenylation of the upstream fragment by members of the cleav-age/polyadenylation factor (CPF) complex, including the poly(A) polymerase Pap1 (7)(8)(9). In the presence of a protein with poly(A) binding activity, CF I/II, CPF, and Pap1 are sufficient to reconstitute poly(A) tail synthesis de novo on an RNA in vitro (10 -12). Despite remarkable similarities in the polyadenylation machinery between yeast and mammals, species-specific factors have been described. Particularly, the product of the pabp2/pabpn1 gene is thought to be specific to metazoans (13), as no apparent yeast homolog has yet been identified.
Poly(A)-binding protein (PABP) 3 2 was originally identified through biochemical enrichment of a polyadenylation stimulatory factor from calf thymus extracts (14). Mammalian PABP2 is characterized by a putative coiled-coil region, a single RNA recognition motif (RRM), and a C-terminal arginine-rich domain (6,13). The affinity of PAPB2 to synthetic poly(A) tails is in the nanomolar range and requires both the RRM and the C-terminal arginine-rich domain for optimal and specific interaction with poly(A) (15,16). Experiments using in vitro polyadenylation assays led to a model in which PABP2 stimulates processive poly(A) synthesis by direct and simultaneous interactions with the growing poly(A) tail and the poly(A) polymerase (17). Although in vitro experiments have provided information about the biochemical properties of PABP2, little is known about the mechanism by which it regulates poly(A) tail synthesis in vivo.
A Drosophila system was recently established to address the function of PABP2 in vivo. These studies provided evidence for the critical role of PABP2 during embryonic development as well as an unsuspected role in the regulation of cytoplasmic polyadenylation (18). An understanding of the in vivo mechanism by which PABP2 regulates polyadenylation is significant, because the human genetic disorder oculopharyngeal muscular dystrophy (OPMD) is linked to mutations in the pabp2 gene (19). A physiological hallmark of this disorder is the accumulation of fibrous inclusions in the nuclei of skeletal muscle fibers that consist of PABP2 aggregates (20). Yet, the molecular defects in PABP2 function that lead to the establishment of this disease remain unclear. Mammalian PABP2 is also subject to a post-translational modification where specific arginine residues are methylated (21). How arginine methylation modulates the function of PABP2 as well as the cellular methyltransferase responsible for PABP2 methylation remains unknown.
Protein arginine methyltransferases (PRMT) catalyze the monomethylation of specific arginine residues in proteins using S-adenosyl-L-methionine as a methyl donor. PRMTs are divided into two major classes depending on the type of dimethylarginine they generate: type I PRMTs modify proteins by the catalysis of asymmetric N G -N G -dimethylarginine, whereas type II PRMTs catalyze the formation of symmetric N G -N G -dimethylarginine (22). prmt-encoding genes have been identified from the sequenced genomes of yeast, worms, flies, plants, and mammals; but not prokaryotes.
The biological role of PRMTs is likely mediated by the modification of substrate proteins. Accordingly, extensive large scale efforts have aimed to identify arginine methylated proteins (23)(24)(25). Based on these and other studies, arginine methylation appears to impact a variety of cellular processes including ribosome biosynthesis (26), T-cell activation (27), cytokine and interferon signaling (28), cell differentiation (29), and DNA repair (30). The role of arginine methylation in these cellular pathways is likely regulated by biochemical activities such as protein-protein interactions (31,32), protein subcellular localization (33,34), transcription and chromatin remodeling (35,36), mRNA metabolism (37,38), and translation (26,39). Not unexpectedly, mice engineered for deletion of the prmt1 (40), prmt4 (41), or prmt5 (42) gene are inviable. Yet, our understanding of the biological role of each PRMT and how arginine methylation alters the biological function of proteins is limited by the few physiological substrates identified to date.
In this study, we report the identification of a nuclear poly(A)-binding protein (Pab2) in fission yeast. Our results reveal that pab2-null cells produce hyperadenylated mRNAs and are cold-sensitive. We also demonstrate that the protein arginine methyltransferase 1 (RMT1) is the enzyme responsible for Pab2 arginine methylation. Experiments to determine the effect of methylation on Pab2 function indicate that the oligomerization levels of Pab2 are increased when it is not methylated. Accordingly, we found that the growth inhibitory effect caused by Pab2 overexpression is exacerbated in rmt1⌬ cells. Our results thus establish Pab2 as an important regulator of poly(A) tail length control and support a model in which the oligomerization-dependent cellular toxicity of Pab2 is modulated by arginine methylation.

MATERIALS AND METHODS
Strains, Growth Media, and Genetic Methods-The complete list of the Schizosaccharomyces pombe strains used in this study is given in Table 1. Cells were grown at 30°C in yeast extract medium with amino acid supplements (YES) and Edinburgh minimum medium (EMM) containing appropriate amino acid supplements. S. pombe cells were transformed with plasmids and PCR products by the lithium acetate method. Disruption of pab2 was performed by PCR-mediated gene targeting using 100-nt oligonucleotides with 80-nt from the appropriate regions of the pab2 genomic sequence (43). The oligonucleotide sequences used for the construction of these strains are available upon request. Following transformation of a diploid strain, G418-resistant colonies were screened for pab2 heterozygosity by colony-PCR. Meiosis and sporulation were induced in selected heterozygote diploids by plating on malt extract agar and tetrads were dissected with a micromanipulator (MSM 200, Singer Instruments). nmt1 ϩ -dependent gene expression was repressed by the addition of 15 M thiamine to the growth medium.
Antibodies-Rabbit polyclonal antibodies specific to fission yeast RMT1 and RMT3 were raised at Covance Research Products (Denver, PA) against GST fusion proteins purified from Escherichia coli. Rabbit polyclonal Myc antibody A-14 and the mouse monoclonal anti-GST B-14 were from Santa Cruz Bio- technology Inc. (Santa Cruz, CA). Asymmetric dimethylarginine-specific rabbit polyclonal antibody (ASYM24) was from Upstate (Charlottesville, VA) and the rabbit polyclonal anti-GFP was from Invitrogen (Burlington, ON, Canada). Immunoprecipitation Experiments-30 -50 ml of mid-log phase fission yeast cells grown in either YES or EMM were used for immunoprecipitation experiments. Cells were lysed in icecold PBS-MT (1ϫ phosphate-buffered saline supplemented with 2 mM MgCl 2 and 1% Triton-X-100) containing a mixture of protease inhibitors (Roche Applied Science) with a Fastprep FP120 (Thermo Electro Corp.) using 0.5-mm glass beads. Clarified lysates were normalized for total protein concentration using the Bradford protein assay (Bio-Rad), and 1 mg of total proteins was subjected to immunoprecipitation using agaroseconjugated anti-Myc (9E10; Santa Cruz Biotechnology). Immunoprecipitated proteins were eluted in sample buffer, separated on 12% SDS-PAGE, transferred to nitrocellulose membranes, and analyzed by immunoblotting.
Microscopy-For localization of GFP-Pab2, FBY13, and FBY1, cells that were previously transformed with pREPEGFP-N-Pab2 were grown to saturation in EMM containing 15 M thiamine. Cells were then washed twice to remove the thiamine and allowed to grow for 18 -20 h before direct fluorescence microscopy. The nuclei of cells were stained using Hoechst 33342 (Sigma).
Recombinant Protein Expression and in Vitro Pull-down Assays-GST, GST-Pab2, and GST-Pab2⌬C28 were expressed in E. coli BL21 DE3 cells (Invitrogen). Protein expression was induced by the addition of 0.5 mM isopropyl-1-␤-D-thiogalactopyranoside for 3 h at 37°C for GST and GST-Pab2⌬C28, and for 18 h at 18°C for GST-Pab2. Following centrifugation of the cells and subsequent washing in ice-cold phosphate-buffered saline, the bacterial cells were broken by sonication, and the cell membranes were solubilized by the addition of Triton X-100 to a final concentration of 1%. The cell debris was sedimented by a 10-min centrifugation at 12,000 rpm at 4°C, and the clarified lysates were subjected to glutathione-Sepharose resin (Amersham Biosciences). After extensive washing, the proteins were eluted from the resin by incubating with 10 mM reduced glutathione resuspended in Tris-HCl pH 8.6; 250 mM NaCl; 0.1% Triton X-100. The proteins were aliquoted and stored at Ϫ80°C.
For poly(A) pull-down experiments, 2 mg of recombinant protein or 0.5 mg of total cell extracts were incubated for 1 h at 4°C with 15 l of poly(A)-Sepharose 4B (Sigma) that had been previously pre-equilibrated in buffer A (50 mM Tris-HCl, pH 8.0; 150 mM NaCl; 0.1% Triton-X-100; 2 mM MgCl 2 ; 1 mM dithiothreitol; 1 mM EDTA). The beads were then washed three times with 1 ml of buffer A, and the bound proteins eluted by incubating for 5 min at 95°C in SDS-PAGE sample buffer. For the oligonucleotide polymers competition experiment, 10-fold excess of polyadenylic or polycytidylic acids (Sigma) were added to poly(A)-Sepharose simultaneously with GST-Pab2. The copurification of the different proteins with poly(A)-Sepharose was analyzed by immunoblotting.
Poly(A) tail length analyses were based on a previously described method (47). Briefly, 1 g of total fission yeast RNA was 3Ј-end-labeled at 4°C for 18 -20 h with 25 Ci of [ 32 P]cytidine biphosphate using T4 RNA ligase (Ambion). Following digestion of RNAs with RNases T1 and A, the remaining poly(A) tails were ethanol precipitated after proteinase K and phenol-choloform treatments. Poly(A) tracts were separated on 8% acrylamide-7 M urea gels and analyzed using a STORM 860 instrument (Molecular Dynamics).

A Gene Product Similar to Nuclear Poly(A)-binding Proteins
Copurifies with the Fission Yeast Protein Arginine Methyltransferase 1 (RMT1)-An affinity purification approach was used to identify novel RMT1-binding proteins in the fission yeast S. pombe. Homologous recombination-mediated gene tagging generated a strain that expresses C-terminal TAP-tagged RMT1 from its endogenous promoter. Following tandem purification and analysis of the eluted protein mixture by mass spectrometry, peptides corresponding to several RNA recognition motif (RRM)-containing proteins were identified ( Table 2). Inspection of the amino acid sequence of the uncharacterized spbc16e6.12c gene revealed extensive similarity to nuclear poly(A)-binding proteins: a predicted coiled-coil domain consisting of regularly spaced aliphatic residues, a single RNP-type RRM, and a C-terminal arginine-rich domain (Fig. 1). Similar to the Drosophila nuclear poly(A)-binding protein (PABP2), the S. pombe spbc16e6.12c gene product is considerably shorter than metazoan PABP2 proteins because of the lack of an N-terminal region rich in alanine, glycine, glutamic acid, and proline residues. Fission yeast SPBC16E6.12c shares 47% identity and 66% similarity with human PABP2. We therefore named the S. pombe spbc16e6.12c gene pab2 on the basis of this homology. Sequence searches did not find any pab2 ortholog in the S. cerevisiae genome.
A construct expressing Myc-tagged Pab2 was generated to validate our large-scale affinity purification results and examine whether RMT1 could coimmunoprecipitate with S. pombe Pab2. The Myc-Pab2 construct was used to transform wild-type

PAB2 Regulation by Arginine Methylation
as well as cells in which the genomic copy of the rmt1, rmt3, or rmt5 arginine methyltransferase-encoding gene was deleted. Total cells extracts prepared from the different yeast cultures were subjected to immunoprecipitation using a monoclonal Myc antibody and analyzed by immunoblotting using various antibodies. As can be seen in Fig. 2, RMT1, but not RMT3, was found to coprecipitate with Myc-Pab2 in extracts prepared from wild-type cells (lane 7). As controls, Myc immunoprecipitates prepared from extracts of vector-transformed (lane 6) and rmt1-null cells that expressed Myc-tagged Pab2 (lane 8) did not recover RMT1. Deletion of the rmt3 or rmt5 coding sequence did not perturb the association between RMT1 and Pab2 ( Fig. 2; lanes 9 and 10, respectively). These results are consistent with our tandem affinity purification data and indicate that the arginine methyltransferase RMT1 associates with Pab2 in fission yeast. An asterisk is present over the canonical aliphatic residues forming the core of the heptad repeats that constitute the coiled-coil sequence. The GenBank TM accession numbers used for the alignment are as follows: NP_004634.1 for humans; Q28165 for bovine; NP_062275.1 for mouse; AF257236 for Xenopus; AF116341 for Drosophila; CAB16904.1 for S. pombe. Alignments and shading were generated using ClustalW and Boxshade, both available via the World Wide Web.  1 and 6). Antibodies for Western blotting (WB) were rabbit polyclonal antibodies specific to RMT3, RMT1, and the Myc epitope (upper, middle, and bottom panels, respectively).
Pab2 Is Asymmetrically Dimethylated by RMT1-The C-terminal arginine-rich region of human PABP2 is dimethylated at arginines by PRMT1 and PRMT3 in vitro (21); yet, the physiological methyltransferase responsible for PABP2 methylation remains to be determined. Given the specific association between RMT1 and Pab2 (Fig. 2), we examined whether Pab2 was arginine methylated in fission yeast. Extracts of cells that were previously transformed with an empty vector or a vector expressing Myc-Pab2 were subjected to immunopurification using a Myc monoclonal antibody. Fig. 3A demonstrates that similar levels of Myc-Pab2 were immunoprecipitated from wild-type (lane 2; upper panel) and arginine methyltransferases-null cell extracts (lanes 3-5; upper panel), but not from vector control-transformed cells (lane 1; upper panel). An affinity-purified antibody specific for asymmetric dimethylarginines (aDMA) (33) was used to determine the methylation status of immunopurified Pab2. Reprobing of the membrane used for the Myc immunoblot with the aDMA-specific antibody detected aDMA-modified Pab2 in immunoprecipitates prepared from extracts of wild-type, rmt3⌬, and rmt5⌬ cells (lanes 2, 4, and 5, respectively; lower panel), but not from rmt1null cells (lane 3; lower panel). Immunoblotting with a symmetric dimethylarginine-specific antibody (37) did not detect such modification in Myc-Pab2 (data not shown). These results establish RMT1 as the physiological methyltransferase responsible for Pab2 arginine methylation. Furthermore, our data reveal that no alternate pathways exist to complement Pab2 methylation in the absence of RMT1 in S. pombe.
We next set out to identify the region in Pab2 that is arginine methylated. The arginine-rich domain has been shown to be the methylation site of several nucleic acid-binding proteins (48), including human PABP2 (21). We therefore generated a construct that expresses a C-terminal truncated version of Myc-Pab2 lacking the arginine-rich domain. As can be seen in Fig. 3B, the C-terminal-truncated form of Myc-Pab2 expressed in wild-type cells was not found to contain methylated arginine (lane 3; lower panel); yet, both full-length and truncated versions of Myc-Pab2 were successfully immunopurified (lanes 2-3, upper panel). These results define the C-terminal arginine-rich domain as the major site of arginine methylation in fission yeast Pab2.
Fission Yeast Pab2 Is a Nonessential Nuclear Protein-We constructed a diploid strain in which one of the two pab2 alleles was disrupted to address whether pab2 is an essential gene in fission yeast. Germination of the spores after meiosis and tetrad microdissection resulted in a 2:2 segregation ratio of geneticin resistance (Fig. 4A), indicating that pab2-null cells are viable in S. pombe. RT-PCR was used to confirm the absence of pab2 expression in geneticin-resistant cells. Analysis of total RNA prepared from cells derived from the dissection of tetrad 6 (Fig.  4A) demonstrated that geneticin-resistant cells from spores 6B and 6D lacked pab2 transcripts (Fig. 4B, lanes 2 and 4), whereas geneticin-sensitive cells from spores 6A and 6C expressed pab2 mRNAs (Fig. 4B, lanes 1 and 3). As a control, the rmt1 transcript was detected in all cells derived from tetrad 6 (Fig. 4B,  lanes 1-4), but not from rmt1-null cells (lane 5). Examination of cell growth revealed that pab2-null cells exhibit growth defects at lower temperatures. As can be seen in Fig. 4C, cell

PAB2 Regulation by Arginine Methylation
growth of S. pombe deleted for the pab2 gene is delayed at 18 and 25°C as compared with isogenic pab2ϩ yeast. In contrast, both wild-type and pab2⌬ cells grow comparably when incubated at 30°C. Thus, unlike for Drosophila (18), pab2 is not essential for mitotic growth in fission yeast.
We next monitored Pab2 localization in live S. pombe cells using a GFP fusion (Fig. 4D). As shown in Fig. 4E, GFP-Pab2 localized to cell nuclei and was excluded from the cytoplasm at steady state (Fig. 4E, panels a-c). In addition, GFP-Pab2 was found to localize to single nuclear foci (Fig. 4E, panel b). A systematic analysis of all fission yeast gene products by fluorescence microscopy found similar localizations for Pab2 (49).
Arginine methylation modulates the nucleocytoplasmic transport of specific proteins (33,34). We thus examined whether the absence of arginine methylation in fission yeast Pab2 affected its subcellular localization. Examination of the steady state accumulation of GFP-Pab2 expressed in rmt1-null cells demonstrated a similar nuclear localization profile as in wild-type cells (Fig. 4E,  panels d-f). These results indicate that arginine methylation does not affect Pab2 ability to be imported into the nucleus.

Pab2 Is a Poly(A)-binding Protein Required for Poly(A) Tail Length
Control-Nuclear poly(A)-binding proteins specifically bind synthetic adenylate chains in vitro (15,50). Fission yeast Pab2 and a deletion mutant lacking the C-terminal arginine-rich domain were expressed in E. coli as GST fusions and purified on glutathione-Sepharose columns to investigate their ability to bind poly(A) RNA. In poly(A) pull-down experiments, GST-Pab2 was efficiently recovered in the poly(A)-Sepharose fraction (Fig. 5A, lane 8).
In contrast, GST as well as the C-terminal-truncated form of Pab2 did not bind to the poly(A)-Sepharose resin (Fig. 5A, lanes 7 and 9). These results are consistent with studies indicating that the argininerich C-terminal domain of mammalian PABP2 contributes to RNA binding (15).
We then sought to examine the RNA binding specificity of GST-Pab2. Recombinant Pab2 was added to poly(A)-Sepharose with or without the addition of a 10-fold excess of specific polyribonucleotides. As can be seen in Fig. 5B, soluble poly(A) reduced the copurification of GST-Pab2 on the poly(A)-Sepharose resin by more than 10-fold (compare lanes 5 and 6). Conversely, competition experiments using excess soluble poly(C) had no effect on the ability of GST-Pab2 to bind the poly(A)-Sepharose resin (Fig. 5B, compare lanes 5 and 7). These results demonstrate that similarly to human and Drosophila PABP2 (16,50), fission yeast Pab2 binds specifically to poly(A) in vitro.
It was previously reported that arginine methylated mammalian PABP2 binds synthetic poly(A) chains as efficiently as unmethylated PABP2 (15). Two different sources of PABP2 were used in these binding experiments: methylated PABP2 purified from calf thymus and unmethylated PABP2 purified from E. coli. Because of the inherent differences associated from the use of proteins purified from such different sources, we decided to take advantage of the fact that fission yeast Pab2 expressed in rmt1-null cells is not arginine methylated (Fig. 3). Cell extracts prepared from wildtype and rmt1⌬ cells expressing GFP-tagged Pab2 were incubated with poly(A)-Sepharose. Following extensive washing steps, it was found that GFP-Pab2 derived from extracts of wild-type and rmt1null cells bound poly(A)-Sepharose with similar efficiencies (Fig.  5C, compare lanes 8 and 9). As a control of binding specificity, a non-poly(A)-binding protein did not copurify with poly(A) chains (lane 7). Our results are therefore consistent with that of Kuhn et al. (15) and indicate that the poly(A) binding ability of Pab2 is not modulated by arginine methylation.
Once we had established that Pab2 specifically bound to poly(A) in vitro, we examined whether Pab2 is involved in the regulation of poly(A) tail synthesis in fission yeast. Poly(A) tail distribution was therefore analyzed by 3Ј-end labeling of total RNA followed by RNase digestion, leaving the poly(A) chains intact. RNAs prepared from two independent segregants in which the genomic copy of the pab2 gene has been deleted (see Fig. 4, A and B) showed a marked increase in the level of hyperadenylated RNAs (Fig. 5D,  lanes 2 and 4) compared with the distribution of poly(A) tails from two isogenic segregants with wildtype pab2 alleles (Fig. 5D, lanes 1  and 3). Maximum poly(A) tail length increased from ϳ120 nucleotides in wild-type cells to Ͼ226 nucleotides in pab2-null cells. Poly(A) tail length of RNAs prepared from rmt1-null cells was found to be of similar size distribution as RNAs from pab2 ϩ cells (Fig. 5D, compare lanes 5 to 1 and  3). We thus conclude that fission yeast Pab2 functions in poly(A) tail length control. Furthermore, our results suggest that the role of Pab2 in the regulation of poly(A) tail length is not influenced by arginine methylation.
Arginine Methylation Modulates Pab2 Oligomerization-Oligomerization of PABP2 has been linked to the formation of nuclear aggregates and cell toxicity (51). Interestingly, in vitro  1, 4, and 7), GST-Pab2 (lanes 2, 5, and 8), and a truncated version of GST-Pab2 lacking the C-terminal arginine-rich domain (GST-Pab2⌬C28; lanes 3, 6, and 9) were incubated with poly(A)-Sepharose beads. Following a 1-h incubation, the flow-through (FT; lanes 4 -6) was collected, and the beads were washed extensively before the proteins were eluted (lanes 7-9). Detection of the proteins was by Western blotting (WB) using a polyclonal GST antibody. The position of the molecular mass markers is indicated on the right in kDa. B, equal amounts of GST-Pab2 were incubated with poly(A)-Sepharose beads in the presence of a 10-fold excess of poly(A) (lanes 3 and 6), poly(C) (lanes 4 and 7), or without any competing oligonucleotides (lanes 2 and 5). Following a 1-h incubation, the flow-through (FT, lanes 2-4) was collected, and the beads were washed extensively before the proteins were eluted (lanes 5-7). Detection of GST-Pab2 was by Western blotting (WB) using a polyclonal GST antibody. C, equal amounts of whole cell extracts (WCE, lanes 1-3) prepared from wild-type (lanes 1-2, 4-5, and 7-8) and rmt1-null cells (lanes 3, 6, and 9) that were previously transformed with vectors expressing GFP-RMT5 (lanes 1, 4, and 7) and GFP-Pab2 (lanes 2-3, 5-6, and 8-9) were incubated with poly(A)-Sepharose beads. Following a 1-h incubation, the flow-through (FT; lanes 4 -6) was collected, and the beads were washed before the elution (lanes 7-9). Detection of the proteins was by Western blotting (WB) using an affinitypurified GFP antibody. The position of GFP-RMT5 and GFP-Pab2 is indicated on the left. D, poly(A) tail length was analyzed by 3Ј end-labeling of total RNA extracted from wild-type (spores 6A and 6C; lanes 1 and 3, respectively), pab2-null (spores 6B and 6D; lanes 2 and 4, respectively), and rmt1-null (lane 5) cells. Following RNase degradation of non-poly(A) sequences, the poly(A) tails were separated by electrophoresis through a 15% polyacrylamide-8 M urea gel. The positions of 3Ј-end-labeled DNA fragments are indicated on the right.
binding studies indicate that the C-terminal domain of mammalian PABP2 promotes self-association (15,51). Because our data (Fig. 3) and the work of others (21) demonstrate that the C-terminal arginine-rich domain of nuclear poly(A)-binding proteins is arginine-methylated, we investigated the role of arginine methylation on Pab2 oligomerization. To examine the effect of Pab2 methylation on self-association, cell extracts were prepared from wild-type and rmt1-null cells that coexpressed GFP-and Myc-tagged Pab2. Myc immunoprecipitates prepared from cell extracts were followed by immunoblot analysis using Myc-and GFP-specific antibodies. Whereas GFP-Pab2 was recovered from Myc immunoprecipitates prepared from extracts of yeast coexpressing Myc-and GFP-tagged Pab2 (Fig.  6, lane 5), GFP-Pab2 was not detected in immunoprecipitates of control extracts that did not contain Myc-Pab2 (Fig. 6, lane 4). These results are consistent with previous observations indicating that nuclear poly(A)-binding proteins form oligomers (15,51,52). Analysis of Myc immunoprecipitates prepared from extracts of rmt1-null cells that were previously transformed with plasmids expressing Myc-Pab2 and GFP-Pab2 showed a 10-fold increase in the level of GFP-Pab2 that copurified with Myc-tagged Pab2 (Fig. 6, lane 6). Pab2 oligomers were RNAindependent as RNase treatment did not abolish the association between Myc-Pab2 and GFP-Pab2 (data not shown). These data indicate that arginine methylation of Pab2 modulates its self-association.
Pab2 Overexpression Causes Toxicity That Is Exacerbated in rmt1-null Cells-Mutations in the human pabp2 gene are linked to oculopharyngeal muscular dystrophy (OMPD) (19), a disease characterized by fibrous inclusions in the nuclei of skeletal muscle fibers (20). Interestingly, the extent of PAPB2 oligomerization is a key factor associated with the formation of nuclear aggregates found in cells of OPMD-afflicted individuals (20,51). Having established that Pab2 oligomerization was modulated by arginine methylation (Fig. 6), we investigated the biological relevance of this biochemical effect. For this, the gene dosage of pab2 was altered by using a vector expressing Pab2 under the control of the thiamine-regulated nmt1ϩ promoter. Full derepression of the nmt1ϩ promoter occurs between 18 -24 h following the transfer from thiamine-containing to thiamine-free medium (53). Thiamine-dependent growth defects linked to Pab2 overexpression were apparent in wildtype fission yeast (Fig. 7). The Pab2-dependent growth inhibition observed in wild-type cells required the C-terminal arginine-rich domain (Fig. 7). Importantly, rmt1-null cells were more sensitive than wild-type fission yeast to the growth inhibitory effect elicited by Pab2 overexpression (Fig. 7). These results indicate that overexpression of the nuclear poly(A)binding protein leads to cellular toxicity in S. pombe. Furthermore, our data suggest that elevated dosage of unmethylated Pab2 is more toxic to cells than arginine methylated Pab2.

DISCUSSION
Mechanisms of poly(A) tail synthesis have mainly been studied in the yeast S. cerevisiae and in mammals. Our current study establishes fission yeast as a valuable system to investigate the molecular machinery and mechanisms involved in poly(A) tail metabolism. Significantly, we believe our results provide strong evidence for the identification of a homolog of mammalian nuclear poly(A)-binding proteins (PABP2) in fission yeast. This conclusion is supported by several observations: (i) the extensive amino acid identity and similarity between S. pombe Pab2 and mammalian PABP2 (Fig. 1), (ii) the domain organization of fission yeast Pab2, including a coiled-coil region, a single RNA recognition motif, and a C-terminal arginine-rich domain (Fig.  1), (iii) the nuclear localization of Pab2 (Fig. 4), (iv) arginine methylation of the C-terminal arginine-rich domain (Fig. 3), (v) the ability of Pab2 to specifically and directly interact with poly(A) in vitro (Fig. 5), (vi) the involvement of Pab2 in poly(A) tail length control in vivo (Fig. 5D), and (vii) the oligomerization properties of Pab2 (Fig. 6).
Pab2 and Poly(A) Tail Length Control-How does the absence of Pab2 expression result in excessively elongated poly(A) tails? Most of the current knowledge on poly(A) tail length control in vivo was gained from work using S. cerevisiae. Based on this biochemical and genetic evidence, models in which the major poly(A)-binding protein of budding yeast, Pab1, regulates poly(A) tail length by (i) regulating poly(A) polymerase (PAP) processivity (54) and (ii) recruiting subunits of the poly(A) nuclease (PAN) complex to achieve mRNA-specific poly(A) length control were proposed (55,56). Yet, the role of Pab1 as the primary regulator of mRNA poly(A) tail length dur-  1 and 4) or with Myc- Pab2 (lanes 2, 3, 5, and 6). GFP-and Myc-specific antibodies were used for Western blot analyses (upper and lower panels, respectively). FIGURE 7. Pab2 overexpression induces growth inhibition that is exacerbated in rmt1-null cells. 10-fold serial dilutions of wild-type (WT) and rmt1null (rmt1D) cells that were previously transformed with a vector control, or vectors expressing wild-type and the C-terminal-truncated form of Pab2 lacking the arginine-rich domain (Pab2⌬C28) were spotted onto EMM plates with (left) or without (right) thiamine.
ing de novo synthesis has recently been debated. The poly(A)binding protein Nab2 was also found to be a key player in the regulation of poly(A) tail synthesis in S. cerevisiae. This conclusion is supported by experimental evidence indicating that Nab2-depleted cells produce hyperadenylated mRNAs and that recombinant Nab2 influences poly(A) tail length control in in vitro polyadenylation assays (11,57). However, the exact mechanism by which Nab2 regulates poly(A) length in vivo remains unclear.
Evidence also links 3Ј-end processing to mRNA nuclear export (11, 58 -62). Interestingly, hyperadenylated mRNAs are found in a diverse set of S. cerevisiae strains defective in mRNA export (11,63,64). In these mutant strains, nuclear accumulation of hyperadenylated transcripts, and thereby cell growth, are somewhat restored by inactivating components of the nuclear exosome (59,62) and the poly(A) nuclease complex (60). On the basis of these findings, a quality control system that monitors proper 3Ј-end formation before the release of mRNAs for nuclear export was suggested. Yet, the mechanism by which mRNAs become hyperadenylated upon inactivation of specific export factors remains to be determined.
It is therefore possible that the steady-state accumulation of hyperadenylated transcripts in pab2-null cells (Fig. 5) is the indirect consequence of mRNA export defects. We do not favor this interpretation, however. First, whereas all of the mRNA export factors in which hyperadenylated transcripts have been detected are essential for cell viability, pab2-null cells are viable (Fig. 4). Moreover, hyperadenylation is detected in a strain that expresses a cold-sensitive allele of nab2 in which no apparent accumulation of nuclear poly(A) ϩ RNA is seen at the permissive temperature (11), implying that hyperadenylated mRNAs are not necessarily retained in the nucleus. We rather predict a model whereby Pab2 controls poly(A) tail length by directly regulating polyadenylation. Results obtained from in vitro assays support a model in which PABP2 promotes the transition from processive to distributive poly(A) synthesis once a specific length is reached (65). Yet, the mechanisms of length determination and polyadenylation termination remain to be determined.
In vitro data also indicate that mammalian PABP2 stimulates poly(A) synthesis by recruiting the poly(A) polymerase to the RNA substrate (17). However, detection of hyperadenylated RNAs in pab2-null cells indicates that factors other than Pab2 can stimulate tethering of PAP to the 3Ј-end of nascent transcripts as well as PAP processivity in vivo. The cleavage and polyadenylation specificity factor (CPSF) complex has indeed been demonstrated as a stimulatory component of polyadenylation in vitro (66). Furthermore, it is anticipated that the C-terminal domain of RNA polymerase II modulates poly(A) synthesis, as recent chromatin immunoprecipitation studies indicate that most mRNA maturation steps occur cotranscriptionally, including 3Ј-end processing and polyadenylation (67). Therefore, although PABP2 is required for efficient polyadenylation in vitro, our results suggest that its main role in the nucleus is in the regulation of poly(A) tail length control. In apparent contrast with this conclusion are observations of shorter, yet considerable, poly(A) tails in Drosophila pabp2 mutants (18). From this study, it was concluded that PAP is incapable of producing stable polyadenylated mRNAs in the absence of PABP2. In this case, however, the data were obtained in mutant embryos in which appreciable amounts of maternal PABP2 are detected. The detection of hyperadenylated RNAs demonstrated in our work does not display such limitations, as yeast cells in which the genomic copy of pab2 has been deleted show a complete absence of Pab2 expression. These potentially contradictory results may be explained by observations suggesting two independent functions for Drosophila PABP2, in which it appears to regulate both poly(A) tail lengthening and shortening (18). Such independent functions for PABP2 may have evolved as a requirement for regulation during fly development, which is not necessary in yeast.
Pab2 and Cell Viability-Our studies reveal that pab2 is not essential for cell viability in fission yeast (Fig. 4). This is surprising given that pab1 and nab2 are essential genes in S. cerevisiae (68). The molecular basis for the inviability of these mutants is likely to be the strong inhibition of mRNA export in the absence of Pab1 and Nab2 expression (11,34,59,60). Similar to pab2, the fission yeast homolog of S. cerevisiae Pab1 is not essential (69). Furthermore, pab1-null cells do not show any detectable mRNA export defect in S. pombe (69). Several groups have reported that cytoplasmic Pab1 is a multifunctional protein that shuttles between the nucleus and the cytoplasm in yeast and humans (59,60,69,70). It is therefore possible that in pab2-null fission yeast, Pab1 complements some of the essential nuclear functions normally performed by Pab2. Such a model is consistent with: (i) pab1 being a high-copy suppressor of nab2 mutant alleles in budding yeast (11) and (ii) recent data indicating that the mammalian Pab1 homolog copurifies with poly(A) polymerase and associates with intron-containing polyadenylated transcripts (71). PABP2 is required for cell viability and embryonic development in Drosophila (18). Yet, because Drosophila PABP2 appears to play an essential role in the expression of key cell cycle regulators in the cytoplasm, it remains uncertain whether it is PABP2 function in gene expression control or mRNA 3Ј-end processing that is responsible for lethality in fruit flies.
Pab2 Arginine Methylation, Cellular Toxicity, and Oculopharyngeal Muscular Dystrophy-Our results indicate that the C-terminal arginine-rich domain of Pab2 is methylated via the protein RMT1. Mammalian PABP2 is also methylated at arginines by PRMT1 and PRMT3 in vitro (21). Similarly, we found that the S. pombe homologs of human PRMT1 and PRMT3 can methylate fission yeast Pab2 at arginines in vitro (data not shown). Yet, our gene deletion studies in yeast clearly indicate that RMT1 is the main enzyme responsible for Pab2 arginine methylation in vivo (Fig. 3A). Such a finding is consistent with the nuclear and cytoplasmic localizations of RMT1 and RMT3, respectively (26). We have also analyzed the effect of arginine methylation on the regulation of Pab2 protein interactions. Our data demonstrate that Pab2 self-association is modulated by arginine methylation (Fig. 6). This latter observation suggests a molecular basis for the Pab2-dependent increase in cellular toxicity detected in rmt1-null cells (Fig. 7). Indeed, our results support a model in which the kinetics of self-association are favored for the unmethylated form of Pab2, thereby leading to higher levels of toxic Pab2 oligomers in rmt1-null cells as compared with wild-type cells.
The extent of PABP2 oligomerization is a key factor associated with the formation of nuclear aggregates that are characteristic of OPMD. Studies using cell culture models indeed correlate PABP2 oligomerization, aggregate formation, and cellular toxicity (20,51,72). Significantly, our data extend these results and describe the importance of arginine methylation in the control of Pab2 oligomerization and cellular toxicity. This finding may offer beneficial therapeutic approaches for OPMD, as the enzymatic activity of PRMTs is likely to be more easily targeted than protein-protein interactions to reduce PABP2 oligomerization. To date, however, the arginine methylation status of PABP2 found in OPMD-related nuclear aggregates has not been investigated. Increasing the cellular concentration of Pab2 in normal cells also resulted in reduced growth rates (Fig. 7), consistent with studies in which Nab2 (73) and wildtype PABP2 (74,75) were overexpressed. Pab2 dosage may affect poly(A) RNA trafficking and gene expression (20,75), but the exact mechanism of growth defects remains to be determined.
The identification of a nuclear poly(A)-binding protein in fission yeast will provide powerful biochemical and genetic tools to study the mechanism of poly(A) tail length control as well as the various components required for this process. Furthermore, our S. pombe model should prove to be extremely useful for better understanding the biological significance of Pab2 methylation.