The glutamine residue of the conserved GGQ motif in Saccharomyces cerevisiae release factor eRF1 is methylated by the product of the YDR140w gene.

Polypeptide release factors from eubacteria and eukaryotes, although similar in function, belong to different protein families. They share one sequence motif, a GGQ tripeptide that is vital to release factor (RF) activity in both kingdoms. In bacteria, the Gln residue of the motif in RF1 and RF2 is modified to N(5)-methyl-Gln by the S-adenosyl l-methionine-dependent methyltransferase PrmC and the absence of Gln methylation decreases the release activity of Escherichia coli RF2 in vitro severalfold. We show here that the same modification is made to the GGQ motif of Saccharomyces cerevisiae release factor eRF1, the first time that N(5)-methyl-Gln has been found outside the bacterial kingdom. The product of the YDR140w gene is required for the methylation of eRF1 in vivo and for optimal yeast cell growth. YDR140w protein has significant homology to PrmC but lacks the N-terminal domain thought to be involved in the recognition of the bacterial release factors. Overproduced in S. cerevisiae, YDR140w can methylate eRF1 from yeast or man in vitro using S-adenosyl l-methionine as methyl donor provided that eRF3 and GTP are also present, suggesting that the natural substrate of the methyltransferase YDR140w is the ternary complex eRF1.eRF3.GTP.

Stop signals in mRNA are recognized by class I release factors (RFs) 1 (1). In eubacteria, two class I RFs with different but overlapping specificity for stop codon recognition perform this task. RF1 reads UAG and UAA, and RF2 reads UAA and UGA. However, in archebacteria and eukaryotes, a single class I RF, aRF1 and eRF1, respectively, recognizes all three stop codons. aRF1 and eRF1 are homologous to each other, and although they are functionally similar to the eubacterial RFs, they belong to a different structural family (2) and display different chain foldings (3,4). One clear sequence motif alone is found in common between eubacterial RF1 and RF2, aRF1 and eRF1: a tripeptide motif Gly-Gly-Gln (GGQ) (5). In the crystal structure of human eRF1, this tripeptide is at the tip of domain 2, which is rich in basic amino acids and projects from the bulk of the molecule ( Fig. 1) (3). In a model of Escherichia coli RF2 based on both the crystal structure (4) and on cryoelectron microscopy observations of the factor bound to ribosomes (6,7), the GGQ tripeptide is positioned rather similarly (Fig. 1). This part of RFs from all of the kingdoms is thought to enter the peptidyltransferase center and trigger hydrolysis of peptidyl-tRNA (5). The GGQ motif is essential to the function of both prokaryotic and eukaryotic RFs. Replacement of either Gly residue by Ala inactivates the factors (3,5,8,9). Replacement of the Gln residue by certain amino acids in either human (9) or E. coli (8) RFs allows significant release activity to be retained in in vitro termination assays, but no substitution has been found for the Gln residue in Saccharomyces cerevisiae (3) or E. coli (8) RFs that supports cell growth, even at low levels. The three-dimensional structure of eRF1 in solution is similar if not identical to that in the crystal (10), whereas the crystal structures of both E. coli RF2 (4) and Thermotoga maritima RF1 (11), which are thought to approximate to the structures of the factors free in the cytoplasm, are much more compact.
An additional type of RF, referred to as a class II factor, participates in protein synthesis termination in eubacteria and eukaryotes. These factors, named RF3 and eRF3, respectively, are ribosome-dependent and class I RF-dependent GTPases (12). RF3 catalyzes the dissociation of class I factors from the ribosome following peptide release (13). eRF3 may perform a similar role in eukaryotes (1,14), although this has not yet been demonstrated but has also been shown to interact with other cell components and may have other roles (15). eRF1 and eRF3 form a complex in the absence of ribosomes, which involves the C-terminal domain of each protein (16). No such interaction between bacterial RF1 or RF2 and RF3 has been observed in the absence of ribosomes, but when both are bound to the ribosome, they are thought to interact in a way that involves the N-terminal domain of RF1 or RF2, which would then be functionally analogous to the C-terminal domain of eRF1 ( Fig. 1) (17,18).
In E. coli, both RF1 and RF2 are post-translationally modified by methylation of the Gln residue in the GGQ motif to give N 5 -Me-Gln (19). The methyltransferase (MTase) involved has been identified and is coded in E. coli and in most bacteria by a gene, hemK (now named prmC), located immediately downstream of prfA, encoding RF1 (20,21). In data bank annotations, PrmC has been classed among MTases methylating N 6 -adenine or N 4 -cytidine in DNA because of the presence of S-adenosyl L-methionine (AdoMet) binding motifs and of an NPPY motif thought to be a hallmark of this class of MTases. It is now understood that the real specificity of the (D/N)PPY family of MTases is for a nitrogen conjugated to a planar system, which may be an amide moiety or a nucleotide base (22). Pairs of apparent homologues to bacterial prmC, one member of which probably encodes a protein targeted to the mitochondrion, are present in the genomes of eukaryotic organisms including yeast, fly, mouse, and man (20). So far, no genetic or biochemical evidence has existed to ascribe a function to any of these supposed MTases.
The perfect conservation of the GGQ motif in RFs from all of the kingdoms and the widespread presence of genes in higher organisms potentially encoding homologues of prmC prompted us to determine whether eRF1 in S. cerevisiae is modified in a similar way to RF1 and RF2 in E. coli. Here, we present evidence that this is the case, that the YDR140w gene product is required for the modification of eRF1, and that eRF3 and GTP are also required for the modification to take place.
Strain yLM6, in which eRF1 His 6 tagged at the C terminus is expressed from the chromosome, was constructed by in vivo recombin-ation between chromosomal DNA in FYBL1-8B and a PCR product carrying the modified eRF1 sequence and a HIS3-selectable marker gene. PCR was performed on DNA from plasmid pHis3 encoding the HIS3 gene using the oligonucleotides eRFup (5Ј-CTGAGGATGAATAT-TATGACGAAGATGAAGGATCCGACTATGATTTCATTTCACACCA-CCACCACCACCACTAATGCGCTAGGAGTCACTGC-3Ј) and eRFdown (5Ј-GAATTTAATTTAAATCTGGCATCTAGTGATTAAATTCTTT-TTGATTCGATAAAGGAAAGCGCGCCTC-3Ј). eRFup contained the 50 nucleotides before the eRF1 stop codon (normal type), a sequence encoding Ser-His 6 (boldface), and 17 nucleotides upstream of the HIS3 gene (italic). eRFdown contained 50 nucleotides complementary to the region downstream of eRF1 (normal) and 17 nucleotides downstream of HIS3 (boldface). The his Ϫ S. cerevisiae strain yIBPC27n was transformed with the PCR product, and His ϩ recombinants (YLM6) were selected on minimal medium (SD), 2% glucose (23). The correspondence of the recombinants to the expected sequence was verified by PCR amplification of the region around the SUP45-HIS3 junction.
E. coli Strains and Plasmids-For expression of YDR140w in E. coli, the gene was amplified by PCR as described above with the exception that the upstream and downstream oligonucleotides (ydr140up, 5Ј-GGAAGGTGACATATGCTACCGAC-3Ј, and ydr140low, 5Ј-ATATAGG-GATCCGTCTTGAGTAA-Ј) encoded NdeI and BamHI restriction sites, respectively (in boldface), and cloned between the same sites in the expression vector pET11a (Novagen), yielding plasmid pVH371. After transformation of BL21(DE3) carrying a vector to overexpress minor tRNA isoacceptors (BL21-Codon Plus, Stratagene), the expression of YDR140w was induced with 1 mM isopropyl 1-thio-␤-D-galactopyranoside for 2 h at 37°C at an optical density of 0.5 in rich medium. The overproduced protein was present almost exclusively as inclusion bodies, which were used for purification of the protein and the preparation of rabbit polyclonal antibodies. eRF1 and Truncated eRF3 Vectors and Expression-The full-length cDNA encoding human eRF1 with C-terminal His 6 tag fusion was cloned into pET23b(ϩ) vector (Novagen) under the control of phage T7 RNA polymerase promoter as described (9,24). The truncated variant of S. cerevisiae eRF3 (Sup35) starting from the third methionine was prepared using the pET-based bacterial expression system (Novagen). The DNAs encoding this eRF3 were generated by PCR using Pwo DNA polymerase (Roche Diagnostics, Mannheim, Germany) and the Sup35 gene as a template (25) cloned in the pBCKS ϩ vector and kindly provided by M. Ter-Avanesyan. The following primers used for PCR were: forward, 5Ј-ATTGGATCCTTTTGGTGGTAAAGATCACG-3Ј, and reverse, 5Ј-AATTGGATCCTTACTCGGCAATTTTAACAAT-3Ј. The purified PCR fragments were cloned into the BamHI site of pET15b inframe with the N-terminal His 6 tag. All of the PCR-generated constructs were checked for polymerase fidelity and preservation of open reading frame by sequencing. Human eRF1 and truncated yeast eRF3 were expressed in E. coli strain BL21(DE3) and purified using nickel-nitrilotriacetic acid Superflow (Qiagen) as described (26). Rabbit polyclonal anti-eRF1 antibodies were prepared commercially.
Overproduction of YDR140w Protein in S. cerevisiae-His-tagged YDR140w protein was produced in yeast strain W303-1B transformed with plasmid pYeVH407 and grown in rich medium (YPGE: 1% bacto yeast extract, 2% bacto peptone, 0.5% glucose, 3% ethanol). Expression was induced at an optical density at 600 nm of ϳ9 by the addition of galactose to a final concentration of 2%. Control cultures of cells carrying the parent plasmid pYeDP60 without insert were grown in parallel. Growth was stopped after an additional 15 h (optical density ϳ14). Cells were collected by centrifugation, resuspended in buffer A (30 mM Tris-HCl, pH 8.0, 1 M NaCl, 5 mM ␤-mercaptoethanol, 0.1 mM PMSF, and EDTA-free anti-protease (Roche Applied Science) as recommended by the manufacturer), and broken by passage through a French press at 8000 p.s.i. After centrifugation for 30 min at 10,000 ϫ g, the supernatant was loaded on to a 0.5-ml column of nickel-nitrilotriacetic acid resin (Qiagen). The column was washed with buffer A and eluted with buffer A containing 100 mM imidazole. Fractions containing protein  (1) responsible for stop codon recognition (darker residues are those that affect termination specificity when mutated), the central domain (2), which carries the GGQ motif (black) at its tip, and the C-terminal domain 3 that interacts with eRF3 (16). The model of E. coli RF2 is that proposed by Rawat et al. (7) and is based on the crystal structure of the factor (4) (Protein Data Bank 1GQE) with conformational rearrangement of domains 1 and 3 to fit data from cryo-EM studies of 70 S ribosomal complexes with RF2 (7) (Protein Data Bank 1MI6). The model shows the N-terminal domain 1 that interacts with RF3 on the ribosome (17,18), the central 2/4 superdomain that carries the SPF tripeptide (dark gray) required for specific UGA stop codon recognition (34), and domain 3 that includes the GGQ motif (black).
were concentrated by ultrafiltration (Amicon Ultra 10 molecular weight cut-off, Millipore) and dialyzed against 10 mM Tris-HCl, pH 7.6, 50 mM KCl, 10 mM Mg acetate, 6 mM ␤-mercaptoethanol, 0.1 mM PMSF, and 50% glycerol and stored at Ϫ20°C. An analysis by SDS-PAGE and Western blotting using polyclonal anti-YDR140w antibodies showed the presence of YDR140w protein only in extracts from cells containing pYeVH407. Several protein impurities were visible in preparations of YDR140w but were present in the control extracts as well.
Purification of His-tagged eRF1-Cultures of strain yLM6 in 4l of rich medium (YPD), 2% glucose (23) were grown to an optical density at 600 nm of ϳ5. The cells were resuspended in 20 ml of 50 mM sodium phosphate buffer, pH 7.5, 5 mM ␤-mercaptoethanol, and 0.1 mM PMSF with EDTA-free anti-protease at twice the concentration recommended by the manufacturer. The cells were broken by passage through a French press, and the supernatant after centrifugation for 1 h at 30,000 ϫ g was made 8 M in urea by the addition of the solid product and readjusted to pH 7.5 with NaOH. The solution was loaded on a column (20 ml) of Mono Q fast-flow (Amersham Biosciences). The column was washed with 50 mM sodium phosphate buffer, pH 7.5, 8 M urea, and 50 M PMSF (buffer B) and eluted at 2 ml/min with a 100-ml gradient from 0 to 0.6 M NaCl in buffer A. The fractions containing eRF1 were identified by SDS-PAGE and Western blot, pooled, and concentrated by ultrafiltration (Amicon Ultra 10 molecular weight cut-off, Millipore) to 1 ml. nickel-nitrilotriacetic acid resin (0.5 ml, Qiagen), previously washed with buffer B, was added, and protein was allowed to bind to the resin for 15 h with gentle agitation at 10°C. The resin was packed as a column, washed with 100 mM NaH 2 PO 4 , 10 mM Tris buffer, and 8 M urea, pH 7.5 (adjusted with NaOH), and eluted with 100 mM NaH 2 PO 4 , 10 mM Tris buffer, and 8 M urea, pH 4.5 (adjusted with HCl). The fractions containing protein were pooled, concentrated to 0.25 ml by ultrafiltration, made at 1% in SDS and 10 mM in dithiothreitol, and dialyzed overnight against 50 mM Tris-HCl, pH 6.8, 1% SDS, and 1 mM dithiothreitol and reconcentrated by ultrafiltration to ϳ0.1 ml. The proteins were then fractionated by SDS-PAGE (10% gel), and Histagged eRF1 was located by Western blot performed on a lane containing ϳ5% of the preparation. A gel fragment containing the remaining amount of the identified protein was excised, washed, dehydrated, and reswollen in 20 l of a trypsin solution (sequencing grade, Roche Applied Science, 15 ng/l in 5 mM CaCl 2 , 25 mM NH 4 HCO 3 ) and incubated overnight at 37°C. Peptides were desalted on C18 Zip-Tips (Millipore) for analysis by mass spectrometry.
Mass Spectrometry-MALDI-TOF mass spectrometry analysis was performed with an Applied Biosystems Voyager-DE PRO spectrometer equipped with a pulsed nitrogen laser (337-nm, 3-ns pulse) at the Department of Protein Engineering and Metabolic Control (Institut Jacques Monod, Paris, France). Operating parameters for reflectron include the following: accelerating voltage (20 kV); grid voltage (75%); guidewire voltage (0.005%); and 100 laser shots/spectrum. The ions of des-Arg-bradykinin, angiotensin I, Glu-fibrinopeptide B, and neurotensin were used for external calibration. Monoisotopic masses were used with deviations for mass assignment within Ϯ0.5 Da. ␣-Cyano-4-hydroxycinnamic acid was used as the matrix, and 0.5-l volumes were applied for each analysis.
Methylation in Vitro-Methylation tests in vitro were conducted in a total volume of 50 l of 10 mM Tris-HCl, pH 7.6, 50 mM KCl, 10 mM Mg acetate, 6 mM ␤-mercaptoethanol, and 2 M [ 3 H]AdoMet (1.62 Ci/mmol) containing, where stated, 50 pmol of eRF1, 50 pmol of eRF3, 1 mM GTP, and 10 l of enzyme preparation. The source of PrmC was a 100,000-g supernatant from E. coli cells overexpressing the enzyme at a high level (19). After incubation for 1 h at 37°C, 5 ml of cold 5% trichloroacetic acid was added followed by filtration on Whatman GF/C filters and measurement of radioactivity.

Analysis of [ 3 H]Methyl-labeled Products of in Vitro
Methylation-Methylation in vitro for subsequent SDS-PAGE and analysis by fluorography was performed as described above with the exception that the specific activity of [ 3 H]AdoMet was 3.14 Ci/mmol. The proteins were separated on 9% SDS-polyacrylamide gels, treated with Amplify (Amersham Biosciences) according to the manufacturer's instructions, and exposed for 3 days at Ϫ80°C to preflashed Hyperfilm (Amersham Biosciences). Prior to Pronase (Roche Applied Science) digestion, methylation reactions were performed as described above, the reaction products were precipitated with 5% trichloroacetic acid, washed with ether, dried, and redissolved in 20 l of 100 mM Tris-HCl, pH 7.5, 5 mM CaCl 2 , and 0.5% SDS at 45°C and digested with Pronase at a final concentration of 1.5 mg/ml for 64 h at 45°C. Authentic N 5 -[ 3 H]Me-Gln was obtained from E. coli RF1 by the same procedure. The digestion products were analyzed by TLC on cellulose sheets (Polygram, Cel 400) in acetone:0.5% urea (60:40 v/v).
Western Blot Analysis-Western blot experiments were performed using rabbit anti-eRF1 or anti-YDR140w antibodies commercially prepared from pure proteins overproduced in E. coli. Proteins were separated by electrophoresis on SDS-polyacrylamide gels as described by Laemmli (27). Transfer to nitrocellulose membranes and Western blotting with antibodies (diluted ϫ 5000) were performed as described by Sambrook et al. (28) using a peroxidase-coupled secondary antibody (diluted ϫ 5000), SuperSignal chemiluminescent substrate (Pierce), and Kodak X-Omat AR.

RESULTS
Yeast eRF1 Is Methylated in Vivo-To determine whether the methylation of the Gln residue present in the GGQ motif in E. coli is conserved in yeast eRF1, the protein was isolated from yeast cells, digested with trypsin, and subjected to mass spectrometry analysis. To achieve this goal, the SUP45 gene encoding eRF1 was modified by recombination to encode a His 6 tag at the C terminus of the protein. His 6 -tagged eRF1 expressed from the modified chromosomal gene was purified under denaturing conditions by a combination of ion-exchange chromatography in 8 M urea, nickel-resin chromatography in 8 M urea, and SDS-PAGE ( Fig. 2A). As a non-methylated control protein, S. cerevisiae eRF1 was prepared by overproduction in E. coli. From the sequence of eRF1, no other tryptic peptide of the mass of the tryptic peptide containing the GGQ motif, GGQSALR, is expected if the peptide is monomethylated (702.37). On the other hand, another tryptic peptide, HNYVR, has the same mass (688.37) as the non-methylated peptide GGQSALR. An analysis of tryptic digests of His 6 -tagged eRF1 from normal yeast cells by MALDI-TOF mass spectrometry showed the presence of a peptide of mass 702.3 (Fig. 2C) that was absent from the analyses of the factor overproduced in E. coli (Fig. 2B), as expected if eRF1 was methylated on the Gln residue of the GGQ motif.
Identification of a MTase for eRF1 Methylation in Vivo-A search of the S. cerevisiae genome revealed two genes, YDR140w and YNL063w, with significant homology to prmC in E. coli, going beyond the motifs that are known to be involved essentially in the binding AdoMet. The second of these genes probably encodes a protein targeted to the mitochondrion (20). To determine whether the YDR140w gene might be required for eRF1 methylation, it was inactivated by insertion of a kanamycin-resistant cassette. The resulting strain (yVH11) was viable but grew slowly with a doubling time of 185 min in Hartwell complete synthetic medium containing 2% glucose as compared with 90 min for the parent strain (results not shown). The complementation of yVH11 by plasmid pYeVH407 carrying the YDR140w gene restored the growth rate to that of the wild type strain, yLM6. These results are consistent with the experiments of Niewmierzycka and Clarke (29) and the growth data obtained during systematic deletion studies in S. cerevisiae (30). When His-tagged eRF1 was prepared from the mutant strain yVH11, digested with trypsin, and analyzed by mass spectrometry, no fragment with a mass of 702.3 corresponding to monomethylated GGQSALR could be observed (Fig. 2D), consistent with the possibility that the YDR140w mutant strain could no longer methylate eRF1.
Methylation in Vitro of S. cerevisiae and Human eRF1 by YDR140w Protein-The mass spectrometry data described above provided evidence that the GGQ motif in eRF1 from S. cerevisiae is methylated and strongly suggested that the YDR140w gene is required for this methylation. Further evidence was sought by trying to methylate S. cerevisiae eRF1 with YDR140w protein in vitro. YDR140w was found to be efficiently overproduced in E. coli but failed to yield MTase able to methylate eRF1 under any of the conditions used (see below). Therefore, the MTase, His-tagged at the C terminus, was overproduced in wild-type yeast strain W303-1B from the high-copy plasmid pYeVH407. The level of expression of YDR140w under these conditions (see "Experimental Procedures") was low, but highly enriched preparations of the His-tagged MTase could nevertheless be obtained from the yeast cell extracts using nickel-resin chromatography. Western blots using polyclonal antibodies raised against YDR140w protein overpro-duced in E. coli showed the presence of YDR140w protein only in the extracts from yeast cells containing plasmid pYeVH407 and not in control extracts from cells transformed with the parent plasmid pYeDP60 (data not shown). Enriched MTase extracts were used in methylation assays with purified E. coli RF1, S. cerevisiae eRF1, and human eRF1 in which the incorporation of radioactivity from [methyl-3 H]AdoMet into trichloroacetic acid-precipitable material was measured. No methylation of any of the RFs was observed (Table I). However, when S. cerevisiae eRF3 and GTP were included in the methylation assay, both S. cerevisiae eRF1 and human eRF1 became methylated (Table I). Neither protein was methylated in the absence of GTP or when GTP was replaced by GDP. E. coli PrmC, although active with E. coli RF1, was unable to methylate eRF1 from yeast or man, even in the presence of eRF3 and GTP. Preparations of RF1 used in these assays contained Ͻ25% of protein active in peptide release as measured in single round release assays, which may explain that only ϳ25% RF1 can be methylated in vitro (20). Although difficult to estimate, we suspect that only a small part of the eRF1 preparations used here is active in peptide release.
The products of the in vitro methylation reactions were studied further by SDS-PAGE, fluorography, and Western blot analysis (Fig. 3). This showed that the majority of the radioactivity was incorporated into a protein that migrated similarly to His 6 -tagged eRF1 when S. cerevisiae eRF1 and eRF3 were present in the methylation assay. Some radioactivity was present in more rapidly migrating bands. However, these reacted with anti-eRF1 antibody similar to the major band of intact His 6 -tagged eRF1, suggesting that they were proteolysis products of the intact factor (Fig. 2). Overall, these results suggest that the substrate of the YDR140w MTase is a ternary complex among eRF1, eRF3, and GTP rather than eRF1 alone. This finding is consistent with the fact that the eukaryotic class I and class II factors form a complex in the absence of ribosomes, whereas the prokaryotic factors do not.
Analysis of eRF1 methylated in vitro confirmed that the product of the AdoMet-dependent reaction contained N 5 -Me-Gln. 3 H-Labeled protein was exhaustively digested with Pronase, and the resulting amino acids were separated by TLC. The radioactive material co-migrated with authentic N 5 -Me-Gln obtained from E. coli RF1 by Pronase digestion (Fig. 4). Taken together with the data obtained by mass spectrometry, these results confirm that the target of eRF1 methylation is the Gln residue of the GGQ motif. DISCUSSION Here we show that eRF1 from S. cerevisiae is methylated on the Gln residue of the conserved GGQ motif, as in bacterial RFs, despite the absence of further sequence homology between these distinct protein families. Only two instances of Gln methylation have previously been described, both in eubacteria. The first case was the methylation of Gln-150 in ribosomal protein L3 in E. coli (31), which may be important for ribosome assembly. Gln-150 in L3 is not universally conserved in bacteria, but when present, the gene encoding the L3 MTase, prmB (or yfcB), also appears to be conserved. Methylation of bacterial RF1 and RF2 is the only other documented case of Gln methylation, and the MTase responsible, PrmC (or HemK), appears to be universal in eubacteria. Inactivation of prmC is lethal for E. coli K12 strains on minimal media and almost abolishes growth on rich media. However, this phenotype is peculiar to E. coli K12 laboratory strains and is related to the nature of the amino acid residue at position 246 in E. coli K12 RF2, very close to the Gln-252 residue methylated by PrmC. This residue is Thr, whereas in every other sequenced RF1 and RF2 gene in all eubacteria including E. coli strains other than K12, the residue is Ala or Ser. In K12 strains, revertants to normal growth of prmC mutant strains are readily obtained and show the replacement of Thr-246 in RF2 by Ser or Ala. Either of the two circumstances, the presence of Thr rather than Ala/Ser or the lack of Gln methylation, reduces the termination efficiency of RF2 in vitro (19). The effect of the two factors is cumulative, and we interpret these observations to mean that the termination machinery can tolerate the effect either of the presence of Thr rather than Ala/Ser or the lack of Gln methylation but that the accumulated effect of both factors reduces termination efficiency at some stop codons to a level incompatible with cell growth. For reasons that remain to be determined, PrmC in Yersinia pseudotuberculosis is required for virulence, although not for growth in rich liquid media (32).
A previous analysis (20) of HemK-related proteins showed three clearly defined subgroups. The smallest family was typified by YfcB (or PrmB) (31), which was identified biochemically as the N 5 -Me-Gln MTase modifying ribosomal protein L3 in E. coli (20). The second subfamily included HemK, for which the gene is now renamed prmC, that was shown to be the N 5 -Me-Gln MTase specific for eubacterial RFs. Included in the HemK subfamily are proteins from S. cerevisiae (YNL063w), Schizosaccharomyces pombe (SPAC29B12.05c), man (Q9Y5R), mouse (Q921L7), and Drosophila (CG9531) that diverge significantly from the main group and are predicted to be targeted to mitochondria (20). These proteins may be MTases specific for mitochondrial RFs, but no biochemical or genetic evidence is at present available to support this supposition. These two subgroups are classed together in data banks as bearing the In-terPro signature IPR004556 (HemK family). The third subgroup comprises ϳ24 proteins from archebacterial and eukaryotic organisms and corresponds to the InterPro family IPR004557 (HemK-rel-arch family). The YDR140w protein that we characterize here is the first protein of this family to which biochemical experiments have allowed a function to be assigned. These proteins diverge in sequence from the HemK family in both the N-and C-terminal regions. A partial alignment of sequences from both the PrmC/HemK and HemK-relarch families is shown in Fig. 5. The crystal structure of PrmC/ HemK from T. maritima has been solved and shows the protein to consist of two domains, an N-terminal domain that may contribute to the binding of the RF substrate and a C-terminal domain typical of class I AdoMet-dependent MTases (22). Fig.  3 shows that the residues that constitute the N-terminal do- a MTases were the RF MTase PrmC from E. coli, partially purified His-tagged YDR140w protein overproduced in S. cerevisiae, or a control preparation made similarly except that cells were transformed with the parent plasmid pYeDP60 instead of plasmid pYeVH407.
b Substrates were E. coli RF1, S. cerevisiae eRF1 (eRF1(sc)), human eRF1 (eRF1(h)) in the presence or absence of S. cerevisiae eRF3 and GDP or GTP. H]AdoMet and S. cerevisiae eRF3 (N-terminally truncated and His 6 -tagged) and GTP in the case of yeast and human eRF1 were precipitated with trichloroacetic acid and digested with Pronase, and the amino acids were separated by TLC as described under "Experimental Procedures." Triangles, S. cerevisiae His 6 -tagged eRF1; squares, E. coli RF1; circles, His 6 -tagged human eRF1. B, amino acid standards chromatographed in parallel and visualized with ninhydrin. main in PrmC/HemK proteins are absent from the HemK-relarch family and that the main region of sequence similarity lies in the center of the molecule, spanning the Gly-rich AdoMet binding motif VL(D/E)XGXGXG (Fig. 5, m1) to the NPPY motif (Fig. 5, m2) of the catalytic center.
The striking observation that the methylation of S. cerevisiae eRF1 requires the presence of eRF3 to be consistent with the fact that the proteins readily form a heterodimer in the absence of ribosomes. The need for eRF3 also adds a new perspective to the observation that yeast cells require eRF3 for viability. Whether this requirement for eRF3 is related to the role of the factor in termination or to some other role, such as messenger circularization (33) or the regulation of mRNA deadenylation (15), is uncertain. In either case, our results suggest that the cells lacking eRF3 would be unable to methylate eRF1, which in turn might contribute to a growth defect like that observed in YDR140w mutant cells. The need for YDR140w for optimal growth of yeast cells is quite possibly due to a role in eRF1 methylation, but this has not been proven. We cannot exclude that YDR140w methylates proteins other than eRF1 and that such a role may underlie the need for the MTase.
It is known that the region of eRF1 involved in binding to eRF3 is essentially confined to the C-terminal domain of the molecule at a considerable distance from the GGQ motif (Fig.  1). This finding suggests that the role of eRF3 in the GTP form may be to induce a conformational change in eRF1 that makes the Gln available to the MTase. The strong sequence similarity between the eukaryotic and archebacterial members of the HemK-rel-arch family, together with the structural and functional resemblance between eRF1 and aRF1, suggests that aRF1are also methylated. If this is the case, it should be noted that the process may be independent of a class II RF because no homologue of eRF3 appears to be present in any completely sequenced archebacterial genome. Further studies will be required to determine how the two types of RF MTase recognize their respective substrates and to understand the role of Gln methylation in the process of translation termination.