PRMT3 is a ribosomal protein methyltransferase that affects the cellular levels of ribosomal subunits

François Bachand, Pamela A Silver

Author Affiliations

  1. François Bachand1 and
  2. Pamela A Silver*,1
  1. 1 Department of Systems Biology, Harvard Medical School and Dana‐Farber Cancer Institute, Boston, MA, USA
  1. *Corresponding author. Harvard Medical School, Dana‐Farber Cancer Institute, 1 Jimmy Fund Way, Boston, MA 02115, USA. Tel.: +1 617 632 5102; Fax: +1 617 632 5103; E-mail: pamela_silver{at}
View Full Text


The mammalian protein arginine methyltransferase 3 (PRMT3) catalyzes the formation of asymmetric (type I) dimethylarginine in vitro. As yet, natural substrates and cellular pathways modulated by PRMT3 remain unknown. Here, we have identified an ortholog of PRMT3 in fission yeast. Tandem affinity purification of fission yeast PRMT3 coupled with mass spectrometric protein identification revealed that PRMT3 associates with components of the translational machinery. We identified the 40S ribosomal protein S2 as the first physiological substrate of PRMT3. In addition, a fraction of yeast and human PRMT3 cosedimented with free 40S ribosomal subunits, as determined by sucrose gradient velocity centrifugation. The activity of PRMT3 is not essential since prmt3‐disrupted cells are viable. Interestingly, cells lacking PRMT3 showed an accumulation of free 60S ribosomal subunits resulting in an imbalance in the 40S:60S free subunits ratio; yet pre‐rRNA processing appeared to occur normally. Our results identify PRMT3 as the first type I ribosomal protein arginine methyltransferase and suggest that it regulates ribosome biosynthesis at a stage beyond pre‐rRNA processing.


Post‐translational modification of proteins is one mechanism by which cells expand the functional complexity of their proteome. Protein arginine methylation is characterized by the addition of monomethyl and dimethyl groups onto the ω‐guanidino nitrogen of arginines (Gary and Clarke, 1998). Using S‐adenosyl‐l‐methionine (SAM) as a methyl donor, all PRMTs catalyze the monomethylation of specific arginine residues in proteins. 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 NG‐NG‐dimethylarginine (aDMA), whereas type II PRMTs catalyze the formation of symmetric NG‐NG‐dimethylarginine (sDMA) (McBride and Silver, 2001). Genes that encode types I and II PRMTs are present in the sequenced genomes of yeast, worm, fly, plants, and mammals, but not prokaryotes. Seven different PRMTs have been identified in mammals: PRMT1 (Lin et al, 1996; Abramovich et al, 1997), PRMT2 (Scott et al, 1998), PRMT3 (Tang et al, 1998), PRMT4/CARM1 (Chen et al, 1999), PRMT5 (Pollack et al, 1999), PRMT6 (Frankel et al, 2002), and PRMT7 (Gros et al, 2003). PRMT5 is the only arginine methyltransferase known to catalyze sDMA (Meister et al, 2001; Boisvert et al, 2002).

Protein arginine methylation modulates a variety of cellular processes including protein subcellular localization (Shen et al, 1998; Cote et al, 2003), protein–protein interactions (Bedford et al, 2000; Friesen et al, 2001), RNA–protein interactions (Valentini et al, 1999), transcription and chromatin remodeling (Chen et al, 1999; Kwak et al, 2003), and DNA repair (Gros et al, 2003). The embryonic lethal phenotype of mice where both alleles of either PRMT1 or PRMT4/CARM1 are disrupted underscores the biological importance of arginine methylation by PRMTs (Pawlak et al, 2000; Yadav et al, 2003). Yet, our understanding of how arginine methylation alters the biological function of proteins is limited by the few physiological substrates identified to date. HMT1/RMT1 and its human homolog PRMT1 are the predominant arginine methyltransferases, responsible for roughly 80% of the PRMT activity of budding yeast and human, respectively (Gary et al, 1996; Pawlak et al, 2000; Tang et al, 2000). Furthermore, the majority of the arginine methylated proteins reported in yeast and mammals are substrates of HMT1 and PRMT1, respectively. In contrast, physiological substrates and biological functions for other PRMTs have not yet been identified.

Human PRMT3 is a cytosolic type I arginine methyltransferase that was initially identified from a yeast two‐hybrid screen using PRMT1 as a bait protein (Tang et al, 1998). The substrate specificity and/or the enzymatic activity of PRMT3 are likely to be regulated by a conserved C2H2‐type zinc finger located in the amino‐terminus of the protein (Frankel and Clarke, 2000). Here we report the identification of a prmt3 gene in fission yeast. Using tandem affinity purification (TAP) coupled with mass spectrometry, we identified components of the translation machinery in association with PRMT3. We found that PRMT3 methylates the 40S ribosomal protein S2 and affects the relative levels of ribosomal subunits. This work reports the first physiological substrate and initial insights into the biological function of PRMT3 and sets the basis for using fission yeast as a model organism to study protein methylation.


Identification of a prmt3 homolog in Schizosaccharomyces pombe

The fission yeast S. pombe genome has been sequenced and annotated (Wood et al, 2002). BLAST searches using the amino‐acid sequence of the rat protein arginine methyltransferase 3 (PRMT3) as a query revealed the presence of an open reading frame in the S. pombe genome that encodes a protein with significant sequence homology to rat PRMT3. Amino‐acid sequences of PRMT3 from rat, mouse, human, fly, and S. pombe were aligned for comparison (Figure 1). Fission yeast PRMT3 shares 38% identity and 55% similarity with human PRMT3. Strong amino‐acid conservation is particularly observed in the S‐adenonyl‐l‐methionine binding region and the methyltransferase active site characterized by motifs I, post‐I, II, and III (Kagan and Clarke, 1994; Figure 1). Indeed, fission yeast PRMT3 demonstrated methyltransferase activity when assayed in vitro from recombinant sources (Figure 4). PRMT3 is characterized by the presence of a single C2H2‐type zinc‐finger domain in the N‐terminal region of the protein (Frankel and Clarke, 2000). The N‐terminus of the fission yeast PRMT3 shows a perfect alignment of the canonical cysteine and histidine residues that shape the zinc‐finger motif (Figure 1). Sequence searches did not find any PRMT3 ortholog in the Saccharomyces cerevisiae genome.

Figure 1.

Amino‐acid sequence alignment of PRMT3 from rat, mouse, human, fly, and fission yeast. Identical amino acids are shown in black outline and similar amino acids are shown in gray outline. Conserved protein methyltransferase motifs I, post‐I, II, and III are boxed in gray. The putative C2H2 zinc‐finger motif of PRMT3 is boxed in white. An asterisk is present under the canonical cysteine and histidine residues of the zinc‐finger motif. Alignments and shading were generated using Clustalw and Boxshade, both available via the World Wide Web. The GenBank accession numbers used for the alignment are as follows: NP_446009 for rat; NP_598501 for mouse; XP_058460 for human; AAO24922 for Drosophila; CAA17825 for S. pombe.

The S. pombe prmt3 gene is not essential for growth

As a first step toward understanding the physiological function(s) of PRMT3, we disrupted the chromosomal copy of prmt3 by one‐step gene replacement (see Materials and methods). Both the KanMX6 and the ura4+ selective markers were used to disrupt the prmt3 locus. The sporulation of heterozygous diploids for the prmt3 disruption yielded viable geneticin‐resistant and Ura4+ haploids. Northern blot analysis confirmed that no transcript corresponding to prmt3 mRNA was expressed in the disruptants (data not shown). Growth curve analyses from cells cultured in rich media did not show major differences in the doubling time of prmt3‐disrupted cells when compared to wild‐type cells (data not shown). Thus, we conclude that prmt3 is a nonessential gene in S. pombe.

Fission yeast PRMT3 localizes to the cytoplasm

We constructed a haploid strain encoding a carboxy‐terminal green fluorescent protein (GFP) fusion of PRMT3 instead of the wild‐type protein to examine the subcellular localization of S. pombe PRMT3. Expression of the fusion protein was confirmed by immunoblotting a total cell extract from the prmt3‐gfp‐engineered strain with an affinity‐purified GFP antibody (Figure 2A, lane 1). Fluorescence microscopy on living S. pombe cells revealed that the PRMT3‐GFP fusion localized predominantly to the cytoplasm (Figure 2B, panel b). As a comparison, we also constructed a haploid strain that expresses a C‐terminal GFP fusion of the S. pombe PRMT1 homolog (Figure 2A, lane 2). In contrast to PRMT3, the PRMT1‐GFP fusion was localized primarily to the nucleus with some cytosolic fluorescence (Figure 2B, panel d), similar to that seen for its S. cerevisiae counterpart HMT1 (Henry and Silver, 1996). The observation that GFP‐tagged fission yeast PRMT1 and PRMT3 localize to the nucleus and cytoplasm, respectively, is also consistent with the localization of their mammalian homologs (Tang et al, 1998; Frankel et al, 2002).

Figure 2.

Cytoplasmic localization of fission yeast PRMT3. (A) Whole‐cell extracts were prepared from S. pombe strains expressing PRMT1‐GFP and PRMT3‐GFP, and immunoblotted with an affinity‐purified GFP antibody. (B) Cells expressing carboxy‐terminal‐tagged PRMT1 and PRMT3 were grown in YES medium. Live S. pombe cells were analyzed by Nomarski microscopy (a, c) and by fluorescence microscopy (b, d) for PRMT3‐GFP (a, b) and PRMT1‐GFP (c, d) localization.

PRMT3 associates with components of the translational machinery

A TAP approach was used to identify proteins that associate with PRMT3 in fission yeast. A haploid strain was generated in which the prmt3 gene is fused to the TAP sequence (Tasto et al, 2001). Preliminary experiments demonstrated that the PRMT3‐TAP fusion retained methyltransferase activity when incubated with a recombinant substrate rich in arginine and glycine (data not shown), suggesting that the protein is functional.

Cell extracts were prepared from 12 l of untagged and PRMT3‐TAP S. pombe and subjected to two rounds of purification over IgG‐sepharose and calmodulin‐bound resins. Following the tandem purification, the proteins from a fraction of the final eluate were resolved by SDS–PAGE and visualized by silver staining (Figure 3), whereas the remainder of the eluate was subjected to liquid chromatography‐coupled tandem mass spectrometry (LC‐MS/MS). Very few polypeptides were present in the final eluate from the untagged strain (Figure 3, lane 1) and the identified proteins were subtracted from the proteins identified in the PRMT3‐TAP purification. Several proteins were found to copurify with PRMT3. Based on silver staining, the only protein present at approximately a 1:1 stoichiometry was a 28‐kDa protein (Figure 3, lane 2); this protein was identified as the 40S ribosomal protein S2 (rpS2) by mass spectrometry analysis. Peptide sequences corresponding to two other proteins with known association to the translational machinery were identified in the PRMT3‐TAP purification (Table I): the elongation factor 1 alpha‐C (eEF‐1A) and the 40S ribosomal protein S24A (rpS24A). As a control, none of these proteins were found to copurify with PRMT1‐TAP (data not shown). These results indicate that PRMT3 associates with components of the translational machinery.

Figure 3.

TAP of PRMT3 from fission yeast. Proteins copurified with PRMT3 by TAP (lane 2) were resolved using a Bis‐Tris 4–12% gradient SDS–PAGE and analyzed by silver staining. The result for an identically treated extract from wild‐type untagged S. pombe is shown in lane 1. Molecular weight markers are shown on the right in kilodaltons (kDa). The positions of PRMT3 and the 40S ribosomal protein S2 (rpS2) identified by mass spectrometry are indicated on the left.

View this table:
Table 1. Translation‐related proteins identified by LC‐MS/MS from PRMT3‐TAP purification

The 40S ribosomal protein S2 is asymmetrically dimethylated by PRMT3

Amino‐acid sequence analysis of the S. pombe 40S ribosomal protein S2 (rpS2) revealed an N‐terminus that is rich in arginine–glycine repeats with amino acids 7–25 being N‐RGFGRGGRGGRGRGRGRRG‐C. Arginine–glycine repeats are a common signature found in protein arginine methyltransferases substrates (Gary and Clarke, 1998). To test whether rpS2 is a substrate of PRMT3 in vitro, glutathione‐S‐transferase (GST)‐tagged rpS2 and PRMT3 were expressed and purified from Escherichia coli lysates. GST‐rpS2 was incubated with [3H]SAM in the presence or absence of GST‐PRMT3 (Figure 4A). Recombinant GST‐PRMT3 catalyzed the methylation of GST‐rpS2 (lane 4), whereas no methylation was observed in the absence of PRMT3 (lane 3). We also observed the methylation of the artificial substrate GST‐(RGG)n by GST‐PRMT3 upon longer exposures, albeit to a much lower extent than for GST‐rpS2 (Figure 4A, lane 2; data not shown). These results indicate that PRMT3 methylates the 40S ribosomal protein S2 in vitro.

A construct expressing GFP‐tagged rpS2 was generated to examine whether rpS2 is arginine methylated in fission yeast. A similar rpS2‐GFP fusion protein was recently used to study the nuclear export of the 40S subunit in S. cerevisiae and shown to be functional (Milkereit et al, 2003). The rpS2‐GFP construct was used to transform wild‐type as well as prmt1‐ and prmt3‐disrupted cells (Figure 4B). The rpS2‐GFP fusion protein was localized to the cytoplasm (data not shown) as previously described (Milkereit et al, 2003). Total cell extracts were prepared from yeast cultures and proteins from the extracts were resolved by SDS–PAGE and analyzed by immunoblotting using an affinity‐purified anti‐GFP (Figure 4B, lanes 1–7). The GFP‐specific antibody recognized a 58‐kDa protein corresponding to rpS2‐GFP from extracts of yeast transformed with the plasmid construct (Figure 4B, lanes 1–6), but not from untransformed cells (Figure 4B, lane 7). An affinity‐purified peptide antibody (ASYM24) specific for asymmetrically dimethylated arginines (aDMA) (Cote et al, 2003) was used to determine the methylation status of rpS2 in fission yeast. Reprobing of the membrane used for the anti‐GFP immunoblotting with the aDMA‐specific antibody demonstrated the absence of aDMA‐modified rpS2 in prmt3‐disrupted cells (Figure 4B, lanes 12 and 13); yet, rpS2 was asymmetrically dimethylated in wild‐type and prmt1‐disrupted cells (lanes 8, 9 and 10, 11, respectively). In addition, unknown substrates of 35 and 28 kDa were not detected from extracts of prmt1‐ (lanes 10 and 11) and prmt3‐disrupted cells (lanes 12 and 13), respectively. These results establish rpS2 as the first identified physiological substrate of PRMT3. Furthermore, these data suggest that no alternate pathways exist to complement the absence of PRMT3 activity in fission yeast.

PRMT3 cosediments with free 40S ribosomal subunits

Our data indicate that PRMT3 associates with components of the translation machinery (Figure 3) and post‐translationally modifies a ribosomal protein (Figure 4). The eukaryotic 40S ribosomal protein S2 is related to the E. coli 30S ribosomal protein S5 (All‐Robyn et al, 1990). Structural studies of E. coli and yeast ribosomes suggest that rpS5 and rpS2, respectively, are present on the solvent side of the small ribosomal subunit (Marion and Marion, 1988; Wimberly et al, 2000; Spahn et al, 2001). Consequently, rpS2 could be accessible for methylation by PRMT3 on the surface of the small subunit and/or as a non‐ribosome‐associated protein. To examine if PRMT3 is physically associated with ribosomal particles, extracts of cells expressing GFP‐tagged PRMT3 were subjected to ultracentrifugation through sucrose cushions and the level of association of different proteins with the ribosome pellet was analyzed by immunoblotting (Figure 5A). Quantification of immunoblots (see Materials and methods for details) from four independent experiments revealed that 6–8% of the total PRMT3 was reproducibly associated with the ribosome pellet, whereas about 1.5% of cytosolic actin was pelleted with ribosomal particles (Figure 5B). As a control of the efficiency of ribosome pelleting, more than 95% of the total 60S ribosomal protein L7 was found in the pellet (Figure 5B). These results indicate that a fraction of PRMT3 is associated with ribosomal particles.

Figure 4.

The 40S ribosomal protein S2 (rpS2) is asymmetrically methylated by PRMT3. (A) A 1 μg portion of recombinant arginine–glycine‐rich substrate (RGG)n (lanes 1 and 2) and rpS2 (lanes 3 and 4) were incubated with (lanes 2 and 4) or without (lanes 1 and 3) recombinant PRMT3 (1 μg) in the presence of 1 μCi of [3H]AdoMet for 3 h at 32°C. Proteins were separated by SDS–PAGE and stained with Coomassie blue (upper panel). The stained gel was dried and exposed for fluorography for 72 h (lower panel). The positions of recombinant GST‐PRMT3, GST‐rpS2, and GST‐(RGG)n are shown on the left of the upper panel. (B) Total cell extracts were prepared from log‐phase yeast cultures inoculated from two independent colonies. Immunoblots of total protein lysates prepared from wild‐type (lanes 1, 2, 7, 8, 9, and 14), prmt1‐disrupted (ΔPRMT1; lanes 3, 4, 10, and 11), and prmt3‐disrupted (ΔPRMT3; lanes 5, 6, 12, and 13) yeast strains that either expressed a C‐terminal GFP‐tagged rpS2 (lanes 1–6 and 8–13) or untagged rpS2 (lanes 7 and 14). An affinity‐purified GFP antibody (anti‐GFP) was used to detect the total rpS2‐GFP protein (left panel) and an affinity‐purified antibody specific for asymmetrically dimethylated arginines (anti‐ASYM24) was used to detect the methylated form of rpS2‐GFP. The position of rpS2‐GFP is shown on the right whereas the molecular weight markers are shown on the left in kilodaltons (kDa).

Figure 5.

A fraction of fission yeast PRMT3 associates with ribosomes and cosediments with free 40S subunits. (A) S. pombe cells expressing a C‐terminal GFP‐tagged PRMT3 were treated with cycloheximide, lysed, and centrifuged through a 20% (w/v) sucrose cushion. In all, 0.125, 0.25, 0.5, and 1% of the total cell lysate (Input; lanes 1–4) and 2.5, 5, 10, and 20% of the ribosome pellet (Pellet; lanes 5–8) were subjected to SDS–PAGE and immunoblot analysis using anti‐GFP, anti‐actin, and anti‐60S ribosomal protein L7 (rpL7). The 60S rpL7 protein could be detected in the input upon longer exposures (data not shown). (B) Quantitation of ribosome‐associated rpL7, PRMT3, and actin. Values are expressed in a graphical form as percentage of input. Data were quantified for at least three independent experiments (see Materials and methods). (C) Polysome profiles of ribosomes isolated from wild‐type (WT) and PRMT3‐GFP‐expressing cells resolved using a 5–45% sucrose gradient. The positions of free small (40S) and large (60S) ribosomal subunits, monosomes (80S), and polysomes (2–6) are indicated in the WT profile (upper panel). (D) Fractions from the gradient isolated from the PRMT3‐GFP‐expressing strain shown in (A) were collected, the proteins were precipitated using trichloroacetic acid, and analyzed by immunoblotting. The positions of the 40S, 60S, 80S monosomes, and P2 polysomes are indicated at the top. (E) Extracts from PRMT3‐expressing cells were separated by sucrose gradient as in (C) but centrifuged for 12 h. The positions of free small (40S) and large (60S) ribosomal subunits are indicated. (F) Fractions from the gradient resolved in (E) were precipitated and the proteins were analyzed by immunoblotting.

To determine the type of ribosome to which PRMT3 associates, we used sucrose gradient velocity sedimentation. Wild‐type and PRMT3‐GFP‐expressing cells displayed similar polysome profiles (Figure 5C), indicating the lack of significant translation defects in the PRMT3‐tagged strain. Fractions were collected from the gradients, and the proteins were precipitated, resolved by SDS–PAGE, and analyzed by Western blotting. In agreement with the data obtained in Figure 5B, the majority of PRMT3 was present in the low‐density fractions (fractions 1–4); yet, PRMT3 clearly cosedimented with free 40S subunits as determined by the sedimentation pattern of the 40S ribosomal protein S6 (Figure 5D, fractions 7 and 8). Fractions corresponding to free 60S subunits (10 and 11), 80S monosomes (12), and polyribosomes (15–18) showed barely detectable levels of PRMT3. Similar results were obtained with the TAP‐ and myc‐tagged PRMT3 (data not shown). We also analyzed fractions from longer centrifugation runs (12 h) to better resolve the free ribosomal subunits (Figure 5E). As can be seen in Figure 5F, the levels of PRMT3‐GFP steadily decreased from fraction 3 to 14, but then increased in fraction 15. Fraction 15 corresponds to the peak of free 40S ribosomal subunits as demonstrated by the sedimentation pattern of the 40S ribosomal protein S6. As a control, cytosolic actin was not detected in the fractions containing free ribosomal subunits (Figure 5F). Similar results were obtained in four independent experiments (data not shown). The results presented in Figure 5 strongly suggest that a fraction of PRMT3 physically associates with free 40S ribosomal subunits.

Human PRMT3 and the 70‐kDa S6 kinase 1 show similar distributions along ribosome profiles

Very few ribosomal protein‐modifying enzymes have been identified and characterized. One such enzyme is S6 kinase 1 (S6K1), which phosphorylates the 40S ribosomal protein S6 (Volarevic and Thomas, 2001). To compare the sedimentation pattern of PRMT3 to another 40S ribosomal protein‐modifying enzyme, we analyzed human ribosome profiles (Figure 6A). Velocity sedimentation of HeLa cell extracts indicated that human PRMT3 had a distribution similar to yeast PRMT3: the majority of human PRMT3 was present in the low‐density fractions (1–4) but a portion also cosedimented with free 40S ribosomal subunits (Figure 6B, fraction 7). Both the total S6K1 and the phosphorylated, active form of S6K1 showed a distribution pattern analogous to PRMT3: most of the protein accumulated in the low‐density fractions (1–4), with some of the protein cosedimenting with free 40S ribosomal subunits (6 and 7). As seen for PRMT3, the S6 kinase was absent from free 60S subunits, 80S monosomes, and polysomes. These results indicate that fission yeast and human PRMT3 show sucrose gradient sedimentation profiles similar to a known 40S ribosomal protein‐modifying enzyme, the S6 kinase 1.

Figure 6.

Human PRMT3 and the 70‐kDa S6 kinase 1 show similar distributions along ribosome profiles. (A) Polysome profiles of ribosomes isolated from HeLa cells were separated using a 5–45% sucrose gradient. The positions of free small (40S) and large (60S) ribosomal subunits, monosomes (80S), and polysomes (2–6) are indicated. (B) Fractions from the gradient shown in (A) were collected and the proteins were precipitated using trichloroacetic acid. Proteins were analyzed by immunoblotting using antibodies to detect human PRMT3 (hPRMT3), the ribosomal protein S6 kinase 1 (S6K1), phospho‐Thr389‐S6K1 (P‐S6K1), human 40S ribosomal protein S6 (hrpS6), and the human 60S ribosomal protein L7 (hrpL7). The positions of the 40S, 60S, 80S monosomes, and polysomes are indicated at the top.

Perturbation of the 40S:60S ribosomal subunit ratio in prmt3 disruptants

The levels of free 40S and 60S ribosomal subunits, 80S monosomes, and polyribosomes from wild‐type, ΔPRMT1, and ΔPRMT3 cells were analyzed to determine whether ribosome biosynthesis or function is perturbed in the absence of PRMT1 or PRMT3. Cell extracts were fractionated by sucrose gradient centrifugation and representative gradient profiles are shown in Figure 7A. Ribosomal proteins, rRNAs, and thus the two ribosomal subunits normally accumulate in equimolar amounts in exponentially growing cells (Warner et al, 2001). In wild‐type cells, approximately 10–15% of ribosomal subunits are not assembled into mature ribosomes (Moy and Silver, 2002); these subunits migrate as 40S and 60S peaks (Figure 7A, upper left panel). Polyribosome profiles of extracts prepared from prmt3 disruptants generated independently in different genetic backgrounds (see Supplementary data) demonstrated a striking accumulation of free 60S ribosomal subunits (Figure 7A, bottom panels). This affected the free 40S:60S stoichiometry but not the overall content of monosomes and polyribosomes. Growth of cells in synthetic defined media instead of rich media yielded similar results (data not shown). As a control, cells lacking PRMT1 did not show this imbalance of free 40S and 60S ribosomal subunits (Figure 7A, upper right panel). Determination of the A254 ratio between free 60S and 40S ribosomal subunits from five independent polyribosomes profiles revealed average values of 1.1 and 1.0 for wild‐type and prmt1‐disrupted cells, respectively; this ratio increased to 2.8 in prmt3‐disrupted cells. Thus, a 2.5‐fold increase in the amount of free 60S ribosomal subunits relative to free 40S subunits was observed in prmt3‐disrupted cells when translation was blocked with cycloheximide (Figure 7A). This 60S ribosomal subunit enrichment was confirmed by analyzing ribosome profiles of extracts prepared in low‐Mg2+ conditions that cause ribosomes to dissociate into separate subunits. As shown in Figure 7B, the A254 60S:40S ratio for wild‐type cells was 1.9; the A254 ratio from four independent experiments increased by about 20% to 2.3 in prmt3‐disrupted cells (t‐test, P<0.05). A 20% increase in the 60S:40S ratio for total ribosomal subunits is consistent with a 150% (2.5‐fold) increase in the 60S:40S ratio for the free pool of ribosomal subunits based on the fact that free subunits represent around 10–15% of total ribosomal subunits. These results indicate that cells lacking PRMT3 exhibit an imbalance in the total 40S:60S ribosomal subunits ratio, which leads to even greater disproportion between the free pool of 40S and 60S subunits.

Figure 7.

Imbalance in the free 40S:60S ratio in prmt3‐disrupted cells. (A) Polysome profiles of ribosomes isolated from wild‐type (WT; upper left panel), prmt1‐disrupted (ΔPRMT1; upper right panel), and prmt3‐disrupted (ΔPRMT3; bottom panels) cells were separated using 5–45% sucrose gradients. The positions of free small (40S) and large (60S) ribosomal subunits and monosomes (80S) are indicated. (B) Total ribosomes were isolated from wild‐type (WT) and prmt3‐disrupted (ΔPRMT3) cells and dissociated into 40S and 60S subunits by the addition of 40 mM EDTA. The dissociated ribosomal subunits were resolved onto 5–45% sucrose gradients. (C) Normal pre‐rRNA processing in prmt3‐disrupted cells. Exponentially growing cultures of wild‐type and prmt3‐disrupted S. pombe were pulse‐labeled with 200 μCi of [methyl‐3H]methionine for 3 min, and then chased for 1.5, 3, and 6 min after the addition of unlabeled methionine. Total cellular RNA was extracted from cells at each time point and 10 000 cpm was analyzed by formaldehyde–agarose gel. The positions of the different pre‐rRNAs and mature rRNAs are indicated on the right.

prmt3‐disrupted cells do not show obvious pre‐rRNA maturation defects

Mutations in non‐ribosomal protein‐coding genes that result in imbalance of ribosomal subunit levels may suggest defects in pre‐rRNA processing and/or in the assembly of ribosomal subunits. The synthesis and processing of ribosomal RNA precursors has been extensively studied in S. cerevisiae (Fromont‐Racine et al, 2003). In the budding yeast nucleolus, rDNA repeats are transcribed by RNA polymerase I to generate a 35S rRNA precursor that is rapidly converted into 27S and 20S pre‐rRNAs. The 27S pre‐rRNA is processed into the mature 25S and 5.8S rRNAs found in the 60S ribosomal subunit, whereas the 20S pre‐rRNA is cleaved to yield the mature 18S rRNA found in the 40S subunit. Because many of the nucleotides in the 35S rRNA precursor are subjected to post‐transcriptional methylation, pre‐rRNA processing is readily followed by metabolically labeling cells with [methyl‐3H]methionine.

To examine whether the ribosomal subunits imbalance seen in prmt3‐disrupted cells was the result of defects in pre‐rRNA processing/ribosome assembly, rRNA pulse‐chase analysis was performed on wild‐type and ΔPRMT3 S. pombe. Total RNA was pulse‐labeled with [methyl‐3H]methionine and then chased for 1.5, 3, and 6 min with an excess of unlabeled methionine. As shown in Figure 7C, assembly of mature 25S and 18S rRNAs was similar in prmt3‐disrupted and wild‐type cells. Consistent with the pulse‐chase experiments, the examination of the steady‐state levels of specific pre‐rRNA intermediates by Northern analysis showed no significant accumulation of pre‐rRNA intermediates and/or aberrantly processed products (data not shown). We thus conclude that the defect in ribosome subunit levels seen in prmt3‐disrupted cells is not a direct consequence of noticeable pre‐rRNA processing deficiencies.


In the present study, we have identified PRMT3 as a ribosomal (r‐) protein methyltransferase. We showed that the 40S ribosomal protein S2 is dimethylated by PRMT3. Separation of ribosomal subunits by sucrose gradient centrifugation indicates that a fraction of PRMT3 specifically cosediments with free 40S ribosomal subunits. Consistent with these observations, we show that prmt3‐disrupted cells exhibit defects in ribosomal subunit levels, yet pre‐rRNA processing appears to occur normally. These results suggest that PRMT3 regulates ribosome biosynthesis at a stage downstream of pre‐rRNA processing.

R‐proteins are subject to a variety of post‐translational modifications such as phosphorylation, acetylation, ubiquitination, and methylation (Louie et al, 1996; Lee et al, 2002). Because few r‐protein‐modifying enzymes have been identified, little is known about the functional role of r‐protein modifications except for certain r‐protein phosphorylation (Jefferies et al, 1997; Mazumder et al, 2003) and ubiquitination (Spence et al, 2000) events. In eukaryotes, both lysine and arginine residues are methylated in many r‐proteins, with aDMA the predominant methylated amino acid in both the mammalian 40S and 60S ribosomal subunits (Chang et al, 1976). The evolutionary conservation of methyl‐lysines in r‐proteins and the appearance of methyl‐arginines in eukaryotic ribosomes suggest that the methylation of r‐proteins has a functional significance. The identification of PRMT3 as the first type I r‐protein methyltransferase will enable a better understanding of the functional significance of this important modification of r‐proteins.

The gene that encodes the 40S ribosomal protein S2 is essential for viability in budding yeast (Giaever et al, 2002). Amino‐acid sequence analysis of rpS2 from a variety of species indicates a highly conserved N‐terminal arginine–glycine (RG)‐rich region (Suzuki et al, 1991); arginine residues within RG‐rich regions are often methylated (Gary and Clarke, 1998; McBride and Silver, 2001). Given the strong conservation of the rpS2 N‐terminal RG‐rich region, our data indicating methylation of fission yeast rpS2 by PRMT3, and the recently confirmed interaction between PRMT3 and rpS2 from the genome‐wide interaction map of Drosophila proteins (Giot et al, 2003), we predict that rpS2 is also methylated by PRMT3 in metazoans. Interestingly, genetic evidence in Drosophila supports a specific developmental role for rpS2 that is different from its function as a structural component of the ribosome (Cramton and Laski, 1994). Arginine methylation of rpS2 is not essential for normal cell growth since prmt3‐disrupted cells are viable and lack aDMA‐modified rpS2 (Figure 4). Nonlethal mutations that cause defects in ribosome assembly or translation often result in yeast cells that are cold‐sensitive and/or that are sensitive or resistant to protein synthesis inhibitors (Goyer et al, 1993; Remacha et al, 1995). prmt3‐disrupted cells grew as well as wild type when incubated at 17°C or in the presence of cycloheximide, rapamycin, or geneticin (data not shown). These observations suggest that cells depleted of methylated rpS2 do not have general translational defects.

Mutations in genes that cause accumulation of free 60S ribosomal subunits have been previously described in S. cerevisiae (Moritz et al, 1990; Kressler et al, 1997; Moy and Silver, 2002; Milkereit et al, 2003). The accumulation of free large ribosomal subunits is often due to mutations that result in defective assembly of the small subunit and cause a reduction in the cellular content of free 40S subunits. rRNA pulse‐chase and Northern analyses, however, did not reveal any obvious deficiencies in pre‐rRNA processing/ribosome assembly in prmt3‐disrupted cells. Mutations that lead to defective export of 40S ribosomal subunits can also increase the 60S:40S ribosomal subunits ratio (Moy and Silver, 2002; Milkereit et al, 2003). In contrast to budding yeast, assays to monitor the export of ribosomal subunits have not yet been established in fission yeast; yet, based on the localization of rpS2‐GFP, no significant difference was seen in 40S ribosomal subunit accumulation in the cytoplasm of wild‐type and ΔPRMT3 cells (data not shown). We cannot exclude the possibility, however, that subtle delays in the kinetics of ribosome assembly and/or subunit export could lead to the ribosomal subunit imbalance observed in cells lacking PRMT3. Nevertheless, mechanisms that are not as well understood such as ribosomal subunit stability/turnover, cell cycle‐dependent regulation of ribosome biosynthesis, and the translational regulation of r‐protein mRNAs could also be responsible for the defects in ribosomal subunit levels observed in prmt3‐disrupted cells.

It is still unclear whether rpS2 is methylated before or after its association with the 40S ribosomal subunit. Similarly, it has not yet been demonstrated whether the 40S ribosomal protein S6 is phosphorylated pre‐ or post‐40S assembly. Our data revealed that a fraction of PRMT3 is present in ribosome‐enriched preparations and associates with free small ribosomal subunits (Figure 5). The presence of 6–8% of cellular PRMT3 in the ribosome‐enriched preparations is consistent with the transient nature of an enzyme–substrate interaction that would be predicted for the post‐translational modification of rpS2. These results, combined with the knowledge that rpS2 is located on the solvent side of the small subunit, may indicate that rpS2 is methylated by PRMT3 post‐40S assembly.

What could be the functional role of methylated rpS2? Ribosomal protein methylation was recently implicated in the translational control of r‐protein mRNAs in developing Dictyostelium discoideum (Mangiarotti, 2002). Upon development of D. discoideum, r‐protein mRNAs left over from vegetative growth are selectively excluded from polysomes (Steel and Jacobson, 1987). These r‐protein mRNAs associate with a class of 40S subunit that is enriched in methylated rpS24 early after activation of D. discoideum development (Mangiarotti, 2002). This presumably prevents the synthesis of ribosomal proteins during this specific stage of development by inhibiting the engagement of r‐protein mRNAs in polysomes. Previous studies have implicated the covalent modification of r‐proteins in the transcript‐specific regulation of translation (Jefferies et al, 1997; Mangiarotti, 2002; Mazumder et al, 2003). These observations support the ‘ribosome filter hypothesis’ in which ribosome alterations, including r‐protein and rRNA modifications, create structural heterogeneity among ribosomes that modulates ribosome–mRNA affinities (Mauro and Edelman, 2002). Interestingly, alterations of the oncogenic Ras and Akt signaling pathways were recently shown to have a much more immediate impact on the recruitment/disengagement of different mRNAs to polysomes than on the genome‐wide transcriptome (Rajasekhar et al, 2003). These observations suggest the existence of mechanisms where signaling pathways rapidly alter the translational efficiency of a variety of mRNAs. Similar to the phosphorylation of the 40S ribosomal protein S6, the methylation of the 40S ribosomal protein S2 could potentially regulate the translationability of specific cellular transcripts in a temporal fashion.

Arginine methylation of ribosomal proteins is common to all eukaryotes. Our understanding of the biological role of r‐protein methylation is less well understood due to the inability to isolate enzymes that catalyze these modifications. The characterization of PRMT3 as a ribosomal protein arginine methyltransferase and the identification of rpS2 as a substrate of PRMT3 will allow a better understanding of the effect of ribosomal protein methylation on the regulation of the translational machinery.

Materials and methods

Tandem affinity purification and mass spectrometry

Purification of the carboxy‐terminal TAP‐tagged S. pombe PRMT3 was performed as previously described (Tasto et al, 2001) with slight modifications. A 12 l measure of yeast culture was harvested during log‐phase growth and lysed with a bead‐beater (Biospec Inc.) in ice‐cold NP‐40 lysis buffer (6 mM Na2HPO4, 4 mM NaH2PO4, 0.5% NP‐40, 150 mM NaCl, 2 mM EDTA, 50 mM NaF, 0.1 mM Na3VO4, protease, and RNase inhibitors). Following affinity purification over an IgG‐sepharose (Amersham Pharmacia Biotech) column, the PRMT3‐bound complex was released by cleavage for 2 h with 500 U of TEV protease (Invitrogen) at 14°C. A second round of affinity purification using calmodulin‐binding resin (Stratagene) was performed followed by extensive washing of the resin. The bound proteins were eluted from the resin by two 500 μl applications of 20 mM EGTA‐containing buffer. One‐third of the eluate was subjected to trichloroacetic acid (TCA) precipitation, SDS–PAGE, and silver staining; the remaining was analyzed by LC‐MS/MS at the Southern Alberta Mass Spectrometry Center (Calgary, Canada).

Ribosome fractionation

The preparation of ribosome pellets from yeast lysates was performed essentially as described (Miyoshi et al, 2002). Log‐phase S. pombe cells were supplemented with cycloheximide to a final concentration of 0.1 mg/ml for 5 min at 32°C. Cells were washed and lysed in ice‐cold complex stabilization buffer (CSB) containing 0.3 M sorbitol, 20 mM HEPES–KOH (pH 7.5), 1 mM EGTA, 5 mM MgCl2, 10 mM KCl, 10% glycerol, 2 mM DTT, 0.1 mg/ml cycloheximide, 40 U/ml of Rnasin, and protease inhibitors. Following lysis of the cells by 6–8 pulses of 25 s vortexing in the presence of glass beads, the cells were centrifuged at 4000 rpm for 5 min and the resulting supernatant was cleared of mitochondria and debris by centrifugation at 15 000 rpm for 20 min at 4°C. A total of 20 A260 units of extracts were layered over 2‐ml sucrose cushions (CSB containing 20% w/v sucrose instead of sorbitol) and centrifuged at 45 000 rpm for 2 h at 4°C in a Beckman TLA100.4 rotor. Ribosome pellets were resuspended in SDS sample buffer and proteins from whole‐cell lysates (input) and ribosome pellets were separated using 10% SDS–PAGE. Immunoblots were quantified using a Fluor‐S MultiImager system (Bio‐Rad, CA). A standard curve of five two‐fold serial dilutions of input in the linear range was obtained for each immunoblotted protein. Four two‐fold serial dilutions of pelleted proteins were plotted against the standard curve to determine the percentage of input protein in the ribosome pellet.

Polysome profiling

Extracts from S. pombe cells were prepared for polysome analysis based on the method described by Akiyoshi et al (2001). Material from 100‐ml cultures (OD600 0.6–1.0) was used for each sucrose gradient. At the time of harvest, the translational machinery was blocked by the addition of cycloheximide to a final concentration of 0.1 mg/ml for 5 min at 32°C. Following centrifugation of the cells and washing in ice‐cold polysome lysis buffer (20 mM Tris–HCl (pH 7.5), 50 mM KCl, 10 mM MgCl2, 1 mM DTT, 0.1 mg/ml cycloheximide, and 0.2 mg/ml heparin), the cell pellets were resuspended in 800 μl of polysome lysis buffer supplemented with protease (Roche) and RNase inhibitors (Promega). Cell suspensions were transferred to prechilled 15‐ml conical tubes containing 500 μl of glass beads and subjected to 6–8 pulses of full‐speed vortexing for 25 s followed by 100 s pauses. After sedimentation of glass beads and cell debris by centrifugation at 3500 rpm for 5 min at 4°C, extracts were transferred to prechilled microcentrifuge tubes and centrifuged for an additional 15 min at 14 000 rpm at 4°C. A total of 15 A260 units of extracts were layered onto 5–45% sucrose gradients prepared in polysome lysis buffer and centrifuged for 3–12 h at 39 000 rpm and 4°C in a Beckman SW41 rotor. The gradients were then fractionated by upward displacement with 55% (w/v) sucrose using a gradient fractionator (Brandel Inc.) connected to a Bio‐Rad Econo EM‐1 UV monitor for continuous measurement of the absorbance at 254 nm. When required, 18 0.6‐ml fractions were collected, and proteins were precipitated with TCA (15% final), washed with cold acetone, and air‐dried. The relative concentration of 40S and 60S ribosomal subunits was determined using low‐Mg2+ conditions as previously described (Kressler et al, 1997).

For all other methods, see supplementary data.

Supplementary data

Supplementary data are available at The EMBO Journal Online.

Supplementary Information

Supplementary data [emboj7600265-sup-0001.pdf]


We thank Charlie Hoffmann for strains and help with S. pombe genetics. We also thank Kathy Gould for TAP constructs and Iain Hagan for GFP expression vectors. We thank G Adelmant, S Komili, A Brodsky, M Yu, and M Gama‐Carvalho for critical reading of the manuscript. FB was supported by fellowships from the Canadian Institutes of Health Research (CIHR) and the Human Frontier Science Program. This work was supported by grants from the National Institute of Health (NIH) to PAS.


View Abstract