Analysis of the Yeast Arginine Methyltransferase Hmt1p/rmt1p and Its in Vivo Function. Cofactor Binding and Substrate Interactions

Many eukaryotic RNA-binding proteins are modified by methylation of arginine residues. The yeast Saccharomyces cerevisiae contains one major arginine methyltransferase, Hmt1p/Rmt1p, which is not essential for normal cell growth. However, cells missing HMT1 and also bearing mutations in the mRNA-binding proteins Npl3p or Cbp80p can no longer survive, providing genetic backgrounds in which to study Hmt1p function. We now demonstrate that the catalytically active form of Hmt1p is required for its activity in vivo . Amino acid changes in the putative Hmt1p S -adenosyl- L -methio- nine-binding site were generated and shown to be un-able to catalyze methylation of Npl3p in vitro and in vivo or to restore growth to strains that require HMT1 . In addition these mutations affect nucleocytoplasmic transport of Npl3p. A cold-sensitive mutant of Hmt1p was generated and showed reduced methylation of Npl3p, but not of other substrates, at 14 °C. These results define new aspects of Hmt1 and reveal the importance of its activity in vivo. Post-translational modifications are frequently used by cells to expand the repertoire of proteins to control their activity temporally or spatially or to target their degradation. The methylation of the guanidino group of arginine residues was first recognized 30 years ago (1), but recently more work has focused on this modification. Protein arginine methyltransferases have been identified and cloned from many eukaryotes AM4, AM7, N-terminal and C-terminal 0.5-kilobase Aat fragments were between plasmid pPS1305 plasmids pPS1757 pPS1758. mutations, an E18V and the introduc-ing a silent Hin dIII site to the E18 codon, were into pPS1305 QuikChange mutagenesis (Stratagene) using gel-purified oligonucleotides AM9 the manufacturer’s protocol. The resulting plasmid, contained the hmt1-E18V allele, was pPS1759.Plasmids hmt1–20 , hmt1–46 and hmt1-E18V alle- the 3.5-kb Bam HI- Hpa fragment plasmids pPS1757, pPS1758, and pPS1759 into pRS305 had been digested Bam HI and Hpa Integration integrated alleles were confirmed for G68 mutant alleles. Synthetic lethality of integrated alleles tested by mating with PSY1191 D D cbp80 PSY773 subsequent

Post-translational modifications are frequently used by cells to expand the repertoire of proteins to control their activity temporally or spatially or to target their degradation. The methylation of the guanidino group of arginine residues was first recognized 30 years ago (1), but recently more work has focused on this modification. Protein arginine methyltransferases have been identified and cloned from many eukaryotes (2)(3)(4)(5)(6). In addition, the number of potential substrates for these enzymes has grown as genome sequencing projects have revealed numerous proteins containing the RGG motif common to substrates for arginine methylation, many of which are RNA-binding proteins (7,8). Although it is not yet clear what effect methylation may have on the activity of such proteins, the yeast arginine methyltransferase Hmt1p has been shown to facilitate the export of at least two of its substrates from the nucleus, pointing to the importance of methylation in nucleocytoplasmic transport (9).
Arginine methyltransferases were initially identified biochemically in mammalian cell lysates by their ability to transfer a radiolabeled methyl group from S-adenosyl-L-methionine (SAM) 1 to histones, myelin basic protein, or the heterogeneous nuclear ribonucleoprotein A1 (10 -14). In addition to monomethylated forms of these proteins, heterogeneous nuclear ribonucleoprotein A1 undergoes asymmetric dimethylation, resulting in N G ,N G -methylarginine residues, whereas myelin basic protein is symmetrically methylated (N G ,N G Ј-methylarginine) (3,16,11,15). Sequence comparison of methyltransferases has revealed motifs common to enzymes that methylate different molecules (17); at least one of these motifs has been implicated in binding to their common co-factor, SAM (18). Novel proteins that contain these motifs have since been identified in general data base searches and in genetic screens (2)(3)(4)(5)(6)19).
The first protein arginine methyltransferase gene in the yeast Saccharomyces cerevisiae, HMT1 (heterogeneous nuclear ribonucleoprotein methyltransferase), was identified in a screen for genes that interacted with NPL3, which encodes an RNA-binding protein (4). The same methyltransferase gene, alternately called RMT1, was also found in a systematic search of the yeast genome for proteins containing methyltransferase motifs (3). HMT1 is not an essential gene in yeast, but it is required in at least two separate genetic backgrounds: in strains with the temperature-sensitive npl3-1 allele or in strains lacking the 80-kDa cap-binding protein gene CBP80 (⌬cbp80) (4,9). A search of ESTs for human homologs of HMT1 revealed two human methyltransferase cDNAs, HRMT1L1 and HRMT1L2, and HRMT1L2 was shown to substitute for HMT1 in npl3-1 strains (19). The ability of a human arginine methyltransferase to function in yeast indicates that cellular mechanisms involving protein methylation are conserved throughout eukaryotes.
The S. cerevisiae Hmt1 protein has been shown to have methyltransferase activity in vitro, methylating a range of RGG-containing proteins including yeast proteins Npl3p, Hrp1p and Hrb1p, and human heterogeneous nuclear ribonucleoprotein A1 (4,9,20). A major poly(A) RNA-binding protein in yeast, Npl3p, contains 15 RGG motifs and has been implicated in many cellular processes including nuclear transport, ribosome biogenesis, and silencing (21)(22)(23)(24)(25). Hrp1p has three RGG motifs and is a component of cleavage factor 1, which is required for mRNA cleavage and polyadenylation (26). In vitro methylation of recombinant Hrp1p had no effect on its specific binding to the polyadenylation efficiency element; however, binding of Hrp1p to RNA inhibits its methylation by Hmt1p (27). Npl3p has been shown to be methylated in vivo (4,20), and there are at least three other proteins that are substrates for Hmt1p (3,28).
Many known and potential substrates of arginine methyltransferases, including Npl3p, Hrp1p, and Hrb1p, are proteins that shuttle between the nucleus and the cytoplasm (9,21,26). The first evidence for the importance of HMT1 in cellular processes was the finding that it is required for efficient export of Npl3p and Hrp1p from the nucleus (9). The role of Hmt1p in protein export is also specific; Hrb1p export is not inhibited by the deletion of HMT1 (9).
Similar to Npl3p and Hrp1p, eukaryotic Cbp80p is involved in RNA processing and nuclear export of RNA-protein complexes (29 -32). The lack of RGG tripeptides in Cbp80p, however, suggests that this protein is not likely to be a substrate for methylation. The lethality of ⌬hmt1 ⌬cbp80 strains may reflect the importance of methylation of Npl3p or other proteins involved in these processes in the absence of Cbp80p. Alternatively, Hmt1p may have cellular functions in addition to its role as a methyltransferase that are required in strains lacking CBP80. AATACCGGTACGGCAACCGACGTCTAAA a Introduced (ϩ) or deleted (Ϫ) restriction sites are underlined, and boldface indicates nucleotide changes that result in amino acid substitutions.
We have now tested directly whether the methyltransferase activity is crucial for its biological function. In addition we have discovered a new mutation that affects the ability of Hmt1p to methylate certain substrates and have defined its effect in vivo.

MATERIALS AND METHODS
Yeast Strains and Media-The yeast strains used in this study are listed in Table I. All strains were grown and genetic manipulations performed as described previously (33). The plasmids used here are listed in Table II. Oligonucleotides used in plasmid construction and sequencing were synthesized at Integrated DNA Technologies, Inc. and are shown in Table III. All sequences that were amplified by PCR were verified by automatic sequencing at the Dana-Farber Cancer Institute Molecular Biology Core Facility.
Construction of G68 Mutant hmt1 Alleles-HMT1 sequences were amplified by PCR from pPS1307 with oligonucleotides 5ЈHMT1a and either AM2 or AM3 to introduce mutations that would result in 1) an Arg or Ala change at amino acid 68, respectively, and 2) the loss of a KpnI site. The PCR products were digested with AgeI and NcoI, and pPS1307 was partially digested with NcoI, then digested with AgeI. The 7-kilobase pair fragment of pPS1307 was ligated to each 526-base pair PCR product, resulting in plasmids pPS1750 (G68R) and pPS1751 (G68A). The 2.7-kilobase pair XbaI fragments from pPS1750 and pPS1751 were ligated into the SpeI site of either pRS315 or pRS305, resulting in plasmids pPS1752 and pPS1753 or pPS1760 and pPS1761, respectively. The integration plasmids pPS1760 and pPS1761 were linearized with NcoI, transformed into PSY865 (⌬hmt1::HIS3) or PSY1812 (⌬hmt1::HIS3 npl3-27), 2 and selected on plates lacking leucine. Integrated alleles were confirmed by Southern blot analysis or PCR, and expression and activity of mutant Hmt1 proteins were tested by immunoblotting.
Bacterial Expression of His-Hmt1 Proteins-His-Hmt1p expression plasmids were constructed as follows. HMT1 sequences were amplified by PCR from plasmids pPS1750 (G68R), pPS1751 (G68A), and pPS1759 (E18V) using primers HMT5ЈNDE and HMT3ЈBAM. Resulting fragments were digested with NdeI and BamHI and ligated into the NdeI and BamHI sites of pET-15b. Amplified sequences were verified by automatic sequencing. Purification of recombinant His-Hmt1 proteins, His-Hrp1p, and Npl3-myc was as described in Shen et al. (9). A polyclonal rabbit antiserum was raised against purified His-Hmt1p at Covance Research Products, Inc.
Immunoblot Analysis-Strains were grown overnight in the appropriate medium and then diluted and grown at the appropriate temperature. The cells were at mid-log phase (about 10 7 cells/ml) when harvested. For expression studies, yeast cell pellets were frozen before lysis in radioimmune precipitation buffer (150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris-HCl, pH 8.0) supplemented with protease inhibitors (1 mM phenylmethylsulfonyl fluoride and 2.5 g/ml each leupeptin, chymostatin, antipain, pepstatin A, and aprotinin) using a Fast-Prep instrument (Bio 101) set at 6.5 for 30 s. Insoluble material was pelleted by centrifugation at 4°C for 10 min, and the protein concentration of the supernatant was determined by Bio-Rad assay. Indicated amounts of total protein were resolved by SDS, 10% PAGE (34) and immunoblotted essentially as described previously (35). Blots were incubated overnight in phosphate-buffered saline with 0.25% Tween (PBST), 2.5% milk powder and different dilutions of primary antibody (anti-Hmt1p serum, 1:5000; anti-Npl3p serum, 1:1000 (24); 1E4 antibody 1:500, provided by M. Swanson (Gainesville, FL) (36)). After washing in PBST, blots were incubated for 1 h at room temperature with horseradish peroxidase-conjugated secondary antibody solution (1:5000 in PBST with milk; Jackson Immunoresearch Laboratories) and washed, and proteins were detected by enhanced chemiluminescence (Amersham Pharmacia Biotech).
In Vitro Methyltransferase Activity Assays-Methyltransferase activity assays were performed in a 15-l reaction containing 20 mM MOPS, pH 7.2, 400 mM KCl, 2 mM EDTA, 30 M SAM (Sigma), 3.5 Ci of [methyl-3 H]SAM (NEN Life Science Products, 80 Ci/mmol) in addition to enzyme and substrate proteins. In the case of the G68 mutants, 2 g of enzyme and 2 g of substrate were used to test methylation of His-Hrp1p. To test the methylation of Npl3-myc by the G68 mutants, 0.2 g each of enzyme and substrate were used. The reactions were incubated at 30°C for 30 min.
For the methylation of His-Hrp1p and Npl3p-myc by wild-type Hmt1p and the E18V mutant Hmt1p, reactions were performed in the reaction mix described above, but with 0.7 M substrate and with increasing amounts of enzyme ranging from 0.07 M to 7.0 M. Methylation of the peptide FGGRGGF, which was synthesized by Charles E. Dahl (Biopolymers Facility, Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School), was carried out similarly, but with lower enzyme:substrate ratios. The final concentration for the peptide was 100 M, with enzyme concentrations ranging from 0.1 M to 10 M. Reactions were incubated at 14°C or 30°C for 30 min. All reactions were terminated by the addition of SDS-PAGE sample buffer and boiling. Samples were resolved by SDS-10% PAGE (15% for peptide) followed by Coomassie staining and fluorography (Entensify; NEN Life Science Products). Autoradiographs were exposed for 4-16 h at Ϫ80°C.
Immunofluorescence Microscopy-Localization of Npl3p by immunofluorescence microscopy was performed essentially as described previously (37). Wild-type and npl3-27 strains were grown to mid-log phase at 25°C and then shifted to 37°C for 30 min. Cells were fixed in formaldehyde for 1 h at 37°C before preparation for microscopy. A 1:1000 dilution of anti-Npl3p and a 1:100 dilution of Texas red-conjugated anti-rabbit antibody were used to detect Npl3p.
Generation of Conditional hmt1 Alleles-HMT1 sequences were amplified from pPS1307 by PCR using a variety of conditions to optimize the degree of mutagenesis. All 100-l reactions contained 1 l of Taq 2 H. Krebber, unpublished data. polymerase (Amplitaq, Perkin-Elmer), 20 nmol of pPS1307, 50 pmol each of primer 5ЈHMT1a and 3ЈHMT1a, 7 mM MgCl 2 , 1ϫ Perkin-Elmer PCR buffer. Mutagenesis was increased by using 1 mM dTTP, 1 mM dCTP, 0.2 mM dATP, and 0.2 mM dGTP rather than 0.5 mM each dNTP or by also adding 0.5 mM MnCl 2 (38). PCR fragments were purified by elution from Qiaquick columns (Qiagen) and co-transformed into PSY1191 with pPS1305 that had been digested with NcoI and MscI. Strains that contained gap-repaired plasmids were selected at 25°C on plates lacking leucine and then transferred to 5-FOA plates lacking leucine to select for strains that had lost pPS1307. Candidate conditional alleles were identified by transferring cells to YEPD plates at 13°C, 25°C, and 37°C and selecting strains that grew at 25°C but not at one of the two possible restrictive temperatures. Plasmids were rescued from these strains and retransformed into PSY1191 to test plasmid linkage of the conditional phenotype.
Amplification of HMT1 with Taq polymerase in the absence of further mutagenic conditions resulted in 2-7% lethality. After screening of 26,500 colonies from these transformations, no candidates were identified. Screening of 18,000 colonies from PCR reactions containing altered dNTP concentrations (11-17% lethality) also resulted in no candidate conditional alleles. However, when PCR conditions were altered by changing dNTP concentrations and adding manganese chloride (60% lethality), cold-sensitive alleles were found after screening 20,000 colonies. The hmt1 alleles that caused conditional growth after retransformation were sequenced from plasmids pPS1757 (hmt1-20) and pPS1758 (hmt1-46) with oligonucleotides 5ЈHMT1a, AM4, AM7, and AM8.
To separate N-terminal and C-terminal mutations in the conditional hmt1 alleles, 0.5-kilobase pair NcoI-AatII fragments were exchanged between plasmid pPS1305 and plasmids pPS1757 and pPS1758. Two mutations, one leading to an E18V substitution and the other introducing a silent HindIII site 5Ј to the E18 codon, were introduced into pPS1305 by QuikChange mutagenesis (Stratagene) using gel-purified oligonucleotides AM9 and AM10, following the manufacturer's protocol. The resulting plasmid, which contained the hmt1-E18V allele, was pPS1759.

Mutations in the SAM Binding Motif Disrupt Activity of
Hmt1p-To test the importance of methyltransferase activity of Hmt1p in vivo, we wished to design point mutations in Hmt1p that would disrupt its catalytic activity (Fig. 1). Sequence alignments of methyltransferases have revealed sequence motifs shared by a diverse set of proteins (17), and the location of these motifs in Hmt1p is shown schematically (Fig.  1A). Although residues involved in substrate binding vary among methyltransferases that target nucleic acids, proteins, and small molecules, motifs I and post-I (Fig. 1B) are involved in binding to the cofactor SAM (3,18,39,40).
A mutation in the EcoP15I DNA methyltransferase, chang- 2. A, anti-Hmt1p serum recognizes only Hmt1p. A polyclonal serum was raised in rabbits against recombinant, histidine-tagged Hmt1p. The original antigen and yeast lysates were resolved by SDS-PAGE, and immunoblot analysis was used to assess the specificity of the antiserum. Lanes 1-4 contain a dilution series of His-Hmt1p (40, 8, 1.6, 0.32 ng). Lanes 5 and 6 contain wild-type yeast lysate (10 and 1 g total protein; PSY867), and lane 7 contains 10 g of total lysate from ⌬hmt1 cells (PSY865). B, in vivo methylation of Npl3p by wild-type and mutant Hmt1 proteins. Mutant hmt1 alleles with G68 changed to either an alanine (A) or an arginine (R) were integrated into a ⌬hmt1 strain (PSY865) such that only the mutant Hmt1p was expressed. Cells were grown at 25°C to mid-log phase and lysed, and total protein was resolved on SDS-PAGE gels. Samples were analyzed by immunoblotting with anti-Hmt1p and anti-Npl3p antisera and the monoclonal antibody 1E4, which specifically recognizes methylated Npl3p. All immunoblots compare protein levels in the mutant strains with those in the ⌬hmt1 strain and in a wild-type (HMT1) strain (5-fold dilution series). Lanes 1, 2, 5, 6 contain 5 g of total protein, lane 3 contains 1 g, and lane 4 contains 0.2 g of total protein for anti-Hmt1p and anti-methyl Npl3p. Half as much total lysate was used for the anti-Npl3p blot. ing the highly conserved penultimate glycine in motif I (bold, Fig. 1) to a serine or an arginine, reduced the in vitro methyltransferase activity more than 400-fold (41). We therefore chose to mutate the equivalent residue in Hmt1p, G68. This residue was mutated to either an arginine or an alanine, as a more conservative change, and the mutant proteins were tested for methyltransferase activity and biological function.
To determine expression levels of wild-type and mutant Hmt1 proteins, an antiserum was raised in rabbits immunized with recombinant, histidine-tagged Hmt1p. Fig. 2A demonstrates that the antiserum recognizes both the original antigen (lanes 1-4), which migrates as a 45-kDa protein, and endogenous yeast Hmt1p (lanes 5 and 6), which migrates as a 40-kDa protein. The absence of a band in lysate from cells lacking HMT1 (⌬hmt1) and the lack of other molecular weight bands indicates the high specificity of the antiserum. Comparison of the His-Hmt1p and wild-type lysate dilution series allowed an approximate calculation of the amount of Hmt1p in the cell, which was ϳ4 ϫ 10 5 molecules/cell.
To test the in vivo expression and activity of mutant Hmt1 proteins, mutated alleles were integrated into the genome of ⌬hmt1 cells such that only mutant forms were expressed. Both mutant alleles expressed Hmt1p to the same level as the wildtype strain (top panel, Fig. 2B). In addition, the anti-Npl3p immunoblot shows that equivalent amounts of Npl3p were found in the lysates (middle panel, Fig. 2B).
Swanson and co-workers (36) raised a monoclonal antibody, 1E4, against purified yeast Npl3p. The ability of this antibody to recognize Npl3p in yeast lysates but not recombinant bacterially expressed Npl3p led to a series of experiments showing that this antibody specifically recognizes methylated Npl3p, although its exact epitope has not been elucidated (20). We used 1E4 to determine the activity of the mutant Hmt1 proteins in vivo. The 1E4 immunoblot (bottom panel, Fig. 2B) reveals that the methylation of Npl3p in strains expressing mutant Hmt1 proteins is significantly reduced compared with that in the wild-type strain at 25°C. As expected, Hmt1-G68A, with its less disruptive amino acid change, retains some activity (down Ͼ5but Ͻ25-fold), whereas Hmt1p with arginine at position 68 has no detectable activity (Ͼ25-fold less than wild type).
The loss of in vitro methyltransferase activity of recombinant G68 mutant Hmt1 proteins correlates with their reduced in vivo activities (Fig. 3). Histidine-tagged wild-type and mutant Hmt1 proteins were expressed in bacteria, purified, and tested for methylation activity. Equal amounts of His-Hmt1p and substrate were incubated with [methyl-3 H]SAM to test if mutant proteins were able to methylate either His-Hrp1p or Npl3myc. Hmt1-G68A (lanes 2 and 5) was less active than wild-type Hmt1p (lanes 1 and 4) when either His-Hrp1p or Npl3-myc was used as a substrate (Fig. 3). Hmt1-G68R, however, failed to methylate either substrate (lanes 3 and 6).
SAM-binding Site Mutations Disrupt Hmt1p Function in Vivo-To study the in vivo effect of these mutations that inhibit methyltransferase activity, we took advantage of strain backgrounds that require HMT1. Plasmids that express the mutant proteins were transformed into ⌬hmt1 npl3-1 and ⌬hmt1 ⌬cbp80 strains bearing an HMT1 URA3 plasmid to test for their ability to suppress the synthetic lethality. The growth of strains that had lost the URA3 plasmid bearing the wild-type HMT1 was monitored on FOA plates at 25°C (Fig. 4). Neither mutant hmt1 was able to suppress the ⌬hmt1 npl3-1 synthetic lethality after 3 days of growth (Fig. 4A), and hmt1-G68A only partially suppressed the ⌬hmt1 ⌬cbp80 synthetic lethality after 4 days of growth (Fig. 4B). Thus, methyltransferase activity correlates with growth in the ⌬cbp80 background, whereas complete wild-type methyltransferase activity appears to be necessary in the npl3-1 background. Therefore, the methyltransferase activity of Hmt1p is important for its function in vivo.
To test whether methyltransferase activity of Hmt1p is important for Npl3p export from the nucleus, we used the npl3-27 allele, which encodes a mutant Npl3 protein that is imported slowly into the nucleus (37). In the presence of HMT1, wildtype Npl3p is predominantly nuclear at 37°C, whereas npl3-27p is found throughout the cell (Fig. 5 (21)). In the presence of G68 mutant hmt1 alleles, however, the steady state localization of npl3-27p becomes more nuclear. The severity of this phenotype correlates with loss of methyltransferase activity. Nuclear npl3-27p is found in Ͻ50% of hmt1-G68A cells, whereas npl3-27p is concentrated in the nucleus of 60 -80% of hmt1-G68R cells. Thus methyltransferase activity of Hmt1p   FIG. 6. Cold-sensitive hmt1 alleles generated in ⌬hmt1 ⌬cbp80 background. A, PCR mutagenesis of HMT1 and screening of mutant alleles in PSY1191 (⌬hmt1 ⌬cbp80 ϩ HMT1 URA3 CEN) revealed two cold-sensitive alleles of hmt1. PSY1191 was transformed with LEU CEN plasmids bearing wild-type HMT1, the original mutant alleles (hmt1-20 and hmt1-46; hmt1), separated N-terminal and C-terminal mutations (Nter. and C-ter.), or the single point mutant allele hmt1-E18V. Transformants were transferred to FOA plates lacking leucine and grown at 14°C (3 weeks) or 25°C (3-4 days). B, the location of mutations in the hmt1 alleles and the division between N-terminal and C-terminal mutations used in A are shown. Restriction sites and oligonucleotides (arrows) used in mutagenesis are also shown. Note that both original alleles contain one mutation in common, the E18V mutation.
facilitates export of Npl3p.
Generation of Cold-sensitive Alleles of Hmt1p-To investigate further the in vivo function of HMT1, we again took advantage of its being essential in ⌬cbp80 strains to generate conditional alleles of hmt1. Error-prone PCR was used to amplify and to introduce random mutations into HMT1 sequences. The entire HMT1-coding region and some 5Ј-and 3Ј-noncoding sequences were excised from a LEU2 HMT1 CEN plasmid. The gapped plasmid and PCR product were co-transformed into a ⌬hmt1 ⌬cpb80 strain that contained a URA3 HMT1 CEN plasmid (PSY1191). After in vivo gap repair (42), the URA3 plasmid was eliminated by transferring Leu ϩ colonies to FOA. LEU2 plasmids bearing temperature-sensitive or cold-sensitive mutant hmt1 alleles were then identified by plating on rich medium at different temperatures.
Two plasmids, pPS1762 and 1763, still showed conditional suppression of synthetic lethality when retransformed into the ⌬hmt1 ⌬cpb80 strain. The hmt1 alleles in these plasmids were named hmt1-20 and hmt1-46. ⌬hmt1 ⌬cpb80 strains bearing these alleles can grow on FOA plates at 25°C but not at 14°C, demonstrating the cold sensitivity of these alleles (compare hmt1, upper left section and wild-type HMT1, top section, Fig.  6A).
Given that the conditions used for mutagenesis were predicted to result in multiple mutations, the N-and C-terminal mutations were subcloned separately into the wild-type HMT1 plasmid pPS1305. The cold sensitivity of each allele mapped to the N terminus (Fig. 6A, compare lower left and bottom sections). Sequencing of the 5Ј-noncoding region, the open reading frame and a portion of the 3Ј-noncoding region revealed that each allele did contain multiple mutations (Fig. 6B). Remarkably, the two alleles shared one mutation in common, A to T at nucleotide 53 of the open reading frame. This mutation resulted in the substitution of a valine for glutamate 18 of Hmt1p (E18V). Introduction of this point mutation into HMT1 and expression of the resultant hmt1-E18V allele in PSY1191 demonstrated that cold sensitivity is mapped to this residue (Fig.  6A, lower right section).
The E18V Mutation in Hmt1p Reduces Methylation of Npl3p-Although E18 does not lie within predicted methyltransferase motifs of Hmt1p, mutating this residue might affect methyltransferase activity of the protein. To test this possibility in vivo, the original mutant alleles and hmt1-E18V were integrated into the genome of a ⌬hmt1 strain (PSY865), and lysates were immunoblotted for expression and activity of the mutant proteins (Fig. 7). The activity of all three mutant alleles, as detected by the relative abundance of methyl-Npl3p, is reduced at 14°C compared with 25°C, from slightly less than wild type to more than 20-fold less than wild type (Fig. 7,  bottom panel). Thus cold-sensitive growth correlates with cold- FIG. 7. In vivo methylation of Npl3p by cold-sensitive Hmt1p. Cold-sensitive hmt1 alleles were integrated into the genome as in Fig. 2C. Cells were grown and lysed, and total protein was used for immunoblot analysis of the expression and activity of the mutant proteins. Lysate from ⌬hmt1 strains and a 5-fold dilution series of wild-type HMT1 lysate (same as Fig. 2) are shown for comparison. Lanes 1, 2, 5-9, 12-14 contain 5 g of total protein, lanes 3 and 10 contain 1 g, and lanes 4 and 11 contain 0.2 g of total protein for anti-Hmt1p and anti-methyl-Npl3p. Half as much lysate was used for anti-Npl3p blots.  1, 3, 5) or E18V (lanes 2, 4, 6) His-Hmt1p (lanes 1 and 2, 0.1 M; lanes 3 and 4, 1 M;  lanes 5 and 6, 10 M enzyme). Methylation of the peptide was detected as in A.
sensitive methyltransferase activity.
Histidine-tagged E18V Hmt1p was expressed in bacteria, purified, and tested for methylation of recombinant Npl3p-myc and His-Hrp1p. Although methylation of Npl3-myc by recombinant E18V Hmt1p was equivalent to that by wild-type Hmt1p at 30°C, [methyl-3 H]Npl3-myc was significantly reduced for E18V reactions incubated at 14°C (Fig. 8A). To test whether the cold sensitivity of methyltransferase activity was substrate-specific, methylation of His-Hrp1p, a synthetic peptide substrate, and GST-Hrb1p were tested at two temperatures (Fig. 8, B and C, and data not shown). A higher concentration of the peptide substrate was used to allow detection of the tritiated peptide within a responsive range of enzyme-tosubstrate ratios. Notably, hmt1-E18V appeared to methylate these substrates as well as, if not better than, wild-type Hmt1p at both temperatures, suggesting that the hmt1-E18V defect is specific for Npl3p (Fig. 8, B and C, and data not shown).
To test whether the hmt1-E18V allele was synthetically lethal with npl3-1, a hmt1-E18V strain (PSY1686) was crossed to a npl3-1 strain (PSY773; Ref. 43). After sporulation of the diploid, 20 tetrads were dissected: 6 parental ditypes, 3 nonparental ditypes, and 11 tetratypes. Only two spores of nonparental ditypes and three spores of the tetratypes were viable, and in each case the inviable spore was predicted to be hmt1-E18V npl3-1 (Leu ϩ , temperature-sensitive phenotype). Thus the cold-sensitive hmt1 allele was synthetically lethal with npl3-1, as expected if methylation of npl3-1p is important for strain viability. DISCUSSION The primary yeast protein arginine methyltransferase, Hmt1p, has been shown to be essential for viability of strains with mutations in the RNA-binding proteins Npl3p and Cbp80 (4,9). In this report, the introduction of point mutations into the co-factor binding site of Hmt1p, one of which abolishes its enzymatic activity, has revealed that its methyltransferase activity is crucial for the survival of these strains. Random mutagenesis of HMT1 led to the identification of a cold-sensitive allele, hmt1-E18V. The N-terminal E18V substitution affected in vitro methylation of Npl3p but not of other substrates.
The cofactor-binding site of methyltransferases comprises both specific residues for SAM coordination and a loop that lies between a ␤-sheet and an ␣-helix (18,40). The methyltransferase motif I contains the loop residues, and the third amino acid in the loop, corresponding to G68 in Hmt1p, is structurally constrained (40). Given the sizes of the amino acids substituted for glycine at this residue, arginine would be expected to perturb SAM binding by Hmt1p to a greater extent than alanine would. Correspondingly, no methyltransferase activity of Hmt1-G68R was detected in vivo or in vitro, whereas the activity of Hmt1-G68A was detectable but significantly reduced. The lower activity of Hmt1-G68A protein was sufficient to partially suppress the ⌬hmt1 ⌬cbp80 synthetic lethality but not that of ⌬hmt1 npl3-1.
Upon sequencing of the original hmt1-1 allele, which was found to be synthetically lethal with npl3-1 (4), we found a mutation that would result in a glycine-to-aspartate change at residue 86 (asterisk, Fig. 1B; data not shown). The location of this mutation in the post-I methyltransferase motif, which is also thought to be involved in SAM binding (3,39), suggests that the lethality of the hmt1-1 npl3-1 strain may also be due to defective cofactor binding and a concomitant reduction in methyltransferase activity.
We were interested in determining whether there are any regions of Hmt1p other than those implicated in SAM binding that are important for its biological function. Such regions might, for example, be involved in Hmt1p-substrate interactions. The change of a glutamate to a valine at residue 18 of Hmt1p caused cold sensitivity of a ⌬cbp80 strain; in a npl3-1 strain, the hmt1-E18V allele was lethal. The E18V substitution resulted in an enzyme with lowered methyltransferase activity toward Npl3p, but not other substrates, at low temperature. This in vitro substrate specificity and the location of residue 18 N-terminal to the defined methyltransferase motifs suggest that this residue might be involved in substrate binding or positioning rather than catalysis. It is not yet clear why a mutation that affects Npl3p methylation causes cold sensitivity in the ⌬cbp80 background, but both NPL3 and CBP80 have been implicated in RNA export.
The majority of the RNA in the cell is thought to be associated with proteins throughout its lifetime. For example, capbinding proteins Cbp80p and Cbp20p bind nascent mRNAs co-transcriptionally and are joined by splicing and polyadenylation factors (e.g. Hrp1p) and other nuclear RNA-binding proteins, such as Npl3p (44) (Fig. 9). These proteins are likely to escort the RNAs through the nucleus and nuclear pores to the cytoplasm and to be exchanged for cytoplasmic proteins before FIG. 9. Model: Hmt1p and nuclear transport. Proteins involved in nuclear transport that interact with Hmt1p are shown schematically. Hrp1p and Npl3p are Hmt1p substrates, and the cap-binding protein gene CBP80 interacts genetically with HMT1. These proteins are thought to play a role in packaging mRNA for export into the cytoplasm for subsequent translation (the ribosome is indicated by 60 S and 40 S subunits).
translation. In ⌬hmt1 cells, in which Npl3p export is severely slowed, Cbp80 may expedite the export of mRNAs and allow survival (9). Conversely, in the absence of Cbp80, RNA export may require the efficient methylation of Npl3p, which is inhibited at low temperature in hmt1-E18V cells. Alternatively, the E18V mutation may disrupt methylation of another substrate that is required at low temperature in the absence of CBP80. Future studies should help to define the interactions among these genes implicated in RNA binding and nuclear transport.