Identification and characterization of the tRNA:Psi 31-synthase (Pus6p) of Saccharomyces cerevisiae.

To characterize the substrate specificity of the putative RNA:pseudouridine (Psi)-synthase encoded by the Saccharomyces cerevisiae open reading frame (ORF) YGR169c, the corresponding gene was deleted in yeast, and the consequences of the deletion on tRNA and small nuclear RNA modification were tested. The resulting DeltaYGR169c strain showed no detectable growth phenotype, and the only difference in Psi formation in stable cellular RNAs was the absence of Psi at position 31 in cytoplasmic and mitochondrial tRNAs. Complementation of the DeltaYGR169c strain by a plasmid bearing the wild-type YGR169c ORF restored Psi(31) formation in tRNA, whereas a point mutation of the enzyme active site (Asp(168)-->Ala) abolished tRNA:Psi(31)-synthase activity. Moreover, recombinant His(6)-tagged Ygr169 protein produced in Escherichia coli was capable of forming Psi(31) in vitro using tRNAs extracted from the DeltaYGR169c yeast cells as substrates. These results demonstrate that the protein encoded by the S. cerevisiae ORF YGR169c is the Psi-synthase responsible for modification of cytoplasmic and mitochondrial tRNAs at position 31. Because this is the sixth RNA:Psi-synthase characterized thus far in yeast, we propose to rename the corresponding gene PUS6 and the expressed protein Pus6p. Finally, the cellular localization of the green fluorescent protein-tagged Pus6p was studied by functional tests and direct fluorescence microscopy.

To characterize the substrate specificity of the putative RNA:pseudouridine (⌿)-synthase encoded by the Saccharomyces cerevisiae open reading frame (ORF) YGR169c, the corresponding gene was deleted in yeast, and the consequences of the deletion on tRNA and small nuclear RNA modification were tested. The resulting ⌬YGR169c strain showed no detectable growth phenotype, and the only difference in ⌿ formation in stable cellular RNAs was the absence of ⌿ at position 31 in cytoplasmic and mitochondrial tRNAs. Complementation of the ⌬YGR169c strain by a plasmid bearing the wild-type YGR169c ORF restored ⌿ 31 formation in tRNA, whereas a point mutation of the enzyme active site (Asp 168 3 Ala) abolished tRNA:⌿ 31 -synthase activity. Moreover, recombinant His 6 -tagged Ygr169 protein produced in Escherichia coli was capable of forming ⌿ 31 in vitro using tRNAs extracted from the ⌬YGR169c yeast cells as substrates. These results demonstrate that the protein encoded by the S. cerevisiae ORF YGR169c is the ⌿-synthase responsible for modification of cytoplasmic and mitochondrial tRNAs at position 31. Because this is the sixth RNA:⌿-synthase characterized thus far in yeast, we propose to rename the corresponding gene PUS6 and the expressed protein Pus6p. Finally, the cellular localization of the green fluorescent proteintagged Pus6p was studied by functional tests and direct fluorescence microscopy.
Pseudouridine (⌿) 1 and 2Ј-O-methylated nucleotides are the most abundant modified residues in RNAs. Indeed, numerous ⌿ residues are present in rRNAs and tRNAs from Archaea, Bacteria, and Eukarya and in UsnRNAs from Eukarya (for reviews see Refs. [1][2][3][4][5][6]. In rRNAs and UsnRNAs, modified nucleotides are clustered frequently in domains playing an important functional role (7)(8)(9)(10)(11). The strong phylogenetic conservation of some ⌿ residues suggests that they are implicated in stabilizing the RNA three-dimensional structure or RNA-RNA intermolecular interactions (11)(12)(13)(14) or in RNA recognition by protein factors (discussed in Refs. 3 and 15). A structural importance has already been demonstrated for residues ⌿ 32 and ⌿ 38/39 in tRNAs (16 -20). Also, the crystal structure of Escherichia coli tRNA Gln 2 complexed with its cognate aminoacyl-tRNA synthase revealed the existence of a water molecule that bridges the N 1 atom of each ⌿ 38 and ⌿ 39 with the phosphate oxygens of the RNA backbone (21). Similarly, water bridge was proposed for the tRNA Asp anticodon loop (15,22). However, despite this progress in understanding the structural significance of ⌿ residues in RNAs, their importance for cellular mechanisms was understood only in a few cases (23)(24)(25). One way to get information in this field is to identify the enzyme(s) responsible for their synthesis. The corresponding gene(s) can then be deleted and the resulting phenotype tested. Pseudouridine residues in RNAs are formed post-transcriptionally by a group of enzymes called RNA:⌿-synthases (26,27). For a long time, the E. coli RNA:⌿-synthases were the only ones to be identified. A thorough search in DNA data banks for homologs of E. coli RNA:⌿-synthases led to the identification of four families (TruA, TruB, RluA, and RsuA; Refs. 27 and 28). Representatives of each family have been fully characterized in E. coli. Among them, E. coli proteins TruAp and TruBp modify tRNAs at positions 38 -40 (26) and 55 (29), respectively. The E. coli protein RluAp acts both at position 32 in tRNAs and position 746 in 23S rRNA (30,31), whereas the two other members of the RluA family in E. coli, RluCp and RluDp, act at six distinct sites in 23S rRNA (25,32,33). The remaining member of this family (YqcB ϭ TruCp) is solely implicated in the formation of ⌿ 65 in tRNAs. 2 E. coli RsuAp, the prototype of the RsuA family, is specific for the formation of ⌿ 516 in E. coli 16S rRNA (34), and another member of this family in E. coli, RluEp, catalyzes the formation of ⌿ 2457 in 23S rRNA (35). A comparison of the amino acid sequence of the E. coli RNA:⌿synthases and orthologs in other organisms reveals the presence of an invariant aspartate residue within a relatively well conserved motif (27,32). Its substitution by asparagine or glutamic acid inactivates the enzyme, indicating that the conserved aspartate residue is essential for activity (31, 36 -39).
Whereas in E. coli nearly all the enzymatic activities required for formation of the ⌿ residues found in rRNAs and tRNAs were characterized, identification is not as complete in yeast. The situation for rRNAs is understood. Formation of the numerous ⌿ residues in yeast cytoplasmic rRNA is guided by H/ACA small nucleolar RNAs and catalyzed by a unique enzyme of the TruB family, Cbf5p (40,41), whereas the unique ⌿ residue present in yeast mitochondrial rRNA is formed by an RNA:⌿-synthase of the RluA family, Pus5p (39). However, the situation is more complex for yeast tRNAs. The enzymes responsible for U-to-⌿ conversions at 11 of the 15 sites in cytoplasmic tRNAs and three (possibly five) of the eight sites in mitochondrial tRNAs ( Fig. 1; see Ref. 42) have been identified. Two of these enzymes, Pus1p and Pus3p are TruAp-related proteins (43,44). Pus1p is a multisite-specific RNA:⌿-synthase acting at eight different positions in various cytoplasmic tRNAs (positions 26, 27, 28, 34, 35, 36, 65, and 67) as well as at position 44 in U2 snRNA (11,45). Pus1p was also proposed to catalyze ⌿ formation at positions 27 and 28 in mitochondrial tRNAs, but this is still not demonstrated (45). Pus3p catalyzes ⌿ formation of at positions 38 and 39 in the anticodon loops of several cytoplasmic and mitochondrial tRNAs (44). Pus4p, similar to TruB in E. coli, is responsible for the formation of the highly conserved ⌿ residue in tRNA at position 55 in both cytoplasm and mitochondria (46). The enzymes catalyzing the formation of ⌿ residues at positions 1, 13, 31, and 32 in cytoplasmic tRNAs and at positions 31, 32, and 72 in mitochondrial tRNAs remained to be identified ( Fig. 1).
Six ⌿ residues were identified in the Saccharomyces cerevisiae UsnRNAs (11). Thus far, only the enzyme responsible for the formation of residue ⌿ 44 in U2 snRNA was identified (Pus1p) (11). Enzymes responsible for ⌿ formation at positions 5 and 6 in U1 snRNA, 35 and 42 in U2 snRNA, and 99 in U5 snRNA remained to be characterized.
Using the sequence homology approach, nine genes encoding putative RNA:⌿-synthases were identified in the S. cerevisiae genome. Five of them correspond to the already characterized RNA:⌿-synthases (Pus1p, Pus3p, Pus4p, Pus5p, and Cbf5p; see above). Despite its partial characterization, Pus2p has not been assigned to an RNA substrate yet, and the substrate specificity of the three other putative RNA:⌿-synthases, which are encoded by the YGR169c, YDL036c, and YOL066c (RIB2) ORFs that are related to the E. coli RluA family, have not been studied.
To complete RNA:⌿-synthase identification in S. cerevisiae, we replaced the YGR169c ORF, which shows strong sequence homology with E. coli tRNA:⌿ 32 -synthase RluAp, by the kanamycin-resistant gene. We tested the effect of this gene deletion on cytoplasmic and mitochondrial tRNA pseudouridylation. Taking into account the existence of a dual substrate specificity for yeast Pus1p (11), we also tested whether the YGR169c ORF protein product may be involved in UsnRNA pseudouridylation. The results demonstrate that despite a pronounced sequence homology with E. coli tRNA:⌿ 32 -synthase RluAp, the yeast YGR169c-encoded protein acts at position 31 in both cytoplasmic and mitochondrial tRNAs.

EXPERIMENTAL PROCEDURES
Deletion of YGR169c ORF in S. cerevisiae-The YGR169c ORF (deposited in the Swiss Protein Database under Swiss-Prot no. P53294, also referred to as YG3X_YEAST) was deleted in the S. cerevisiae strains BMA64 (Mat a/␣, ura 3-1, ade 2-1, leu 2-3, 112 his3-11, 15 trp1⌬, can 1-100) and BMA64A (Mat a, ura 3-1, ade 2-1, leu 2-3, 112 his3-11, 15 trp1⌬, can 1-100) by the replacement of ϳ87% of the coding region with the kan r resistance marker. Replacement was done by the PCR-targeting method using the kan r gene flanked by short regions homologous to the targeted locus of the S. cerevisiae DNA (47,48). The kan r gene was amplified from plasmid pFA6a-kanMX2. Both haploid and diploid strains were transformed with 3 or 6 g of PCR-generated DNA fragments using the lithium acetate method (49,50). Positive transformants (⌬YGR169c) were selected on yeast extract-peptone-dextrose plates bearing Geneticin G418 (200 mg/liter). The disruption was confirmed by PCR amplification on yeast colonies. The sequences of oligonucleotide primers used in this study can be obtained on request.
In Vitro Tests for RNA:⌿-Synthase Activity-Cell-free S-100 extracts from the wild-type and ⌬YGR169c yeast strains were prepared as described earlier (51). The experimental procedure for testing enzymatic formation of ⌿ at positions 13, 32, and 35 in tRNAs, using as substrates T7 transcripts of yeast tRNA Asp (anticodon GUC) and Arabidopsis thaliana pre-tRNA Tyr (anticodon GUA, including the intron) were described previously (39).
CMCT/RT Mapping of ⌿ Residues in RNA-Total RNA from the wild-type and ⌬YGR169c yeast strains was prepared as described previously (11). The presence of pseudouridine residues in tRNAs (positions 31 and 32) and UsnRNAs from the ⌬YGR169c and wild-type S. cerevisiae strains was tested by the CMCT/RT approach (52) with the modifications described previously (11). N-cyclohexyl-NЈ-[␤-(N-methylmorpholino)ethyl]guanidine substituents on ⌿ residues were identified by primer extension analysis. The oligonucleotide primers used for U1, U2, and U5 snRNAs were described previously (11).
The oligonucleotide primers used for the analysis of pseudouridylation at position 31 was complementary to nucleotides 40 -57 in yeast cytoplasmic tRNA Met (anticodon CAU) and nucleotides 44 -57 in yeast mitochondrial tRNA Met (anticodon CAU). Similarly, the presence of a ⌿ residue at position 32 in cytoplasmic tRNA Gly (anticodon GCC) and in mitochondrial tRNA Ser (anticodon GCU) was tested with oligonucleotides complementary to residues 44 -57 in each tRNA (39).
Complementation of the ⌬YGR169c S. cerevisiae Strain by Plasmid p416GalS-YGR169c-To generate an S. cerevisiae plasmid with an active YGR169c gene, the YGR169c ORF was amplified by PCR from the genomic DNA of strain BMA64A using oligonucleotides that generated an NheI and HindIII restriction site, respectively. The amplified DNA fragment was subcloned at the SmaI site of plasmid pUC18. The resulting construct (pUC18-YGR169) was cleaved by the NheI and HindIII nucleases, and the fragment containing ORF YGR169c was inserted downstream from the GalS promoter between the XbaI and HindIII sites of plasmid p416GalS (53). A point mutation in the YGR169c ORF (Asp 168 3 Ala, GAC3 GGC) was introduced by PCRmediated site-directed mutagenesis using the Quick Change kit (Stratagene). The haploid BMA64A ⌬YGR169c strain was transformed with plasmids containing the wild-type or mutated YGR169c ORFs using the standard lithium acetate procedure (49,50).
Overexpression, Purification, and Test of the Activity of the Recombinant His 6 -Pus6p-Plasmid pET28-Pus6p was built for overexpression of the recombinant Ygr169 protein (Pus6p) in E. coli. To this end, the NheI-HindIII fragment of plasmid pUC18-YGR169 was inserted into plasmid pET28b (Novagen) cut by the same two nucleases. The resulting construct was used to transform E. coli BL21CodonPlus(DE3) RIL cells (Stratagene). Recombinant His 6 -Pus6p was overproduced at 37°C by isopropyl-1-thio-␤-D-galactopyranoside induction (at a 1 mM final concentration) of cells grown in LB medium containing 50 g/ml of kanamycin. The protein was purified from the E. coli cells by adsorption chromatography on nickel-nitrilotriacetic acid agarose (Qiagen) as described earlier for yeast Pus1p tRNA:⌿-synthase (45) and Trm4p tRNA: m 5 C-methyltransferase (54). The protein fraction eluted by 250 mM imidazol was dialyzed against 50 mM Tris-HCl buffer, pH 7.5, containing 50% glycerol and stored at Ϫ20°C. Analysis of the purified protein by SDS-polyacrylamide gel electrophoresis revealed the presence of a component with the expected molecular mass of the His 6 -Pus6 protein (47 kDa).
tRNA:⌿-Synthase activity of the purified recombinant His 6 -Pus6 protein was tested on total RNA extracted from the ⌬YGR169c strain in the following conditions: total yeast RNA (10 g) was incubated for 2 h with ϳ0.2 g of the recombinant protein in 10 l of reaction mixture containing 100 mM Tris-HCl, pH 8.0, 100 mM ammonium acetate, 5 mM MgCl 2 , 2 mM dithiothreitol, and 0.1 mM EDTA. After incubation, the modified RNA was phenol-extracted, ethanol-precipitated, and used for CMCT/RT mapping of ⌿ residues as described above.
Cellular Localization of GFP-tagged Pus6p-For production of Pus6 protein N-terminally tagged with GFP, the NheI-HindIII fragment from pUC18-YGR169 was inserted into plasmid pPS808 (55) cut by the XbaI and HindIII nucleases. The resulting construct was used to transform the wild-type BMA64A and ⌬YGR169c BMA64A yeast strains. Chromosomal DNA was detected by DAPI staining performed according to the standard procedure (50). GFP-Pus6p was observed in living cells by direct fluorescent microscopy.

RESULTS
Deletion of the YGR169c ORF Encoding a Putative RNA:⌿-Synthase Does Not Affect Growth and UsnRNA Pseudouridylation-The yeast S. cerevisiae ORF YGR169c was disrupted in the S. cerevisiae BMA64A haploid strain through the replacement by the kanamycin resistance gene as described under "Experimental Procedures." The haploid yeast strain bearing the deleted YGR169c ORF (⌬YGR169c) was viable and did not show any detectable growth phenotype on solid or liquid media, rich medium (yeast extract-peptone-dextrose), and minimum synthetic medium (data not shown). Altogether, this demonstrated that YGR169c is not an essential gene. Thus, we could compare the RNA:⌿-synthase activities in the WT and ⌬YGR169c strains. Activities were tested either in vitro using cell extracts and tRNA transcripts or in vivoby analysis of the pseudouridylation pattern of various cellular RNAs. Because the S. cerevisiae RNA:⌿-synthases that remained to be identified were tRNA-or UsnRNA-specific, we focused our study on the effect of the deletion on ⌿ formation in UsnRNAs and tRNAs.
The UsnRNA pseudouridylation in the WT and ⌬YGR169c strains was analyzed by the CMCT/RT approach. Only U1, U2, and U5 snRNAs, which contain ⌿ residues, were examined. We found that deletion of the YGR169c ORF had no effect on ⌿ formation in U1, U2, and U5 snRNAs (data not shown). Thus, the Ygr169 protein is not involved in UsnRNA pseudouridylation. We then tested for tRNA:⌿-synthase activity.
The YGR169c-encoded Enzyme Is Not Involved in Formation of ⌿ Residues at Positions 13,32,and 35 in tRNAs-Earlier data showed that the activity of the tRNA:⌿-synthases acting at positions 13 and 32 can be tested using in vitro transcribed yeast tRNA Asp as a substrate and a yeast extract (45). Similarly, the tRNA:⌿ 35 -synthase activity distinct from the Pus1p activity can be tested with a yeast extract using an in vitro produced cytoplasmic A. thaliana pre-tRNA Tyr as a substrate (45,56). With each of these in vitro produced tRNA transcripts, we tested the corresponding tRNA:⌿-synthase activities in the S-100 extracts from the WT and ⌬YGR169c strains. Formation of ⌿ residues at the expected positions was examined by the nearest-neighbor approach using two-dimensional thin layer chromatography (TLC) of tRNAs digested by RNase T2 (45). As shown in Table I, deletion of the YGR169c ORF had no effect on the capacity of yeast extract to form ⌿ 13 and ⌿ 32 in cytoplasmic tRNA Asp transcript or ⌿ 35 in A. thaliana pre-tRNA Tyr , indicating that the Ygr169 protein is not implicated in modification of cytoplasmic tRNAs at position 13, 32, or 35.
To complete the analysis, the presence of ⌿ 32 in the mitochondrial tRNAs from the WT and ⌬YGR169c strains was tested with the CMCT/RT approach. The mitochondrial tRNA Ser , which contains a ⌿ residue at position 32, was used for the assay. As a control, the presence of ⌿ 32 was also tested in the cytoplasmic tRNA Gly . As evidenced by a strong RT stop detected in the CMCT/RT analysis (data not shown), residue ⌿ 32 was present in both the mitochondrial tRNA Ser and cytoplasmic tRNA Gly from the ⌬YGR169c strain. Thus, despite the pronounced sequence homology of the YGR169c-encoded protein with the E. coli tRNA:⌿ 32 -synthase RluA, the Ygr169 protein does not fulfill the same function in yeast.
Cytoplasmic and Mitochondrial tRNAs Are Unmodified at Position 31 in the ⌬YGR169c Strain-Up to now, ⌿ at position 31 in yeast tRNAs was only detected in one cytoplasmic tRNA (tRNA Met ) and seven mitochondrial tRNAs ( Fig. 1; see Ref. 42). The presence of ⌿ 31 in cytoplasmic and mitochondrial tRNAs from the ⌬YGR169c strain was inspected by the CMCT/RT technique. As shown in Fig. 2, the reverse transcription stop corresponding to ⌿ 31 , which is detected in both the cytoplasmic and mitochondrial tRNAs Met of the WT strain (A and B, lanes 3 and 4, respectively), was missing completely in the ⌬YGR169c strain (lanes 5 and 6). The disappearance of the pseudouridylation at position 31 in the ⌬YGR169c strain was also observed for mitochondrial tRNA Lys and tRNA Ser (not shown). Altogether these results strongly suggested that protein Ygr169 is involved in the formation of ⌿ residue at position 31 in both cytoplasmic and mitochondrial tRNAs.
Protein Ygr169 Is Directly Implicated in ⌿ Formation at Position 31 in tRNAs-The modification defect observed in the ⌬YGR169c strain and the presence of the conserved motif characteristic for RNA:⌿-synthases with an invariant aspartate residue in the Ygr169p sequence strongly suggested that protein Ygr169 directly catalyzes the U-to-⌿ conversion at position 31 in tRNAs. As a control, we complemented the ⌬YGR169c strain with plasmid p416GalS bearing a WT or a mutated copy of the YGR169c ORF (Asp 168 3 Ala substitution). As shown in Fig. 2, A and B, transformation of the deleted strain by the plasmid bearing a WT YGR169c but not a mutant YGR169c led to restoration of ⌿ 31 formation in both cytoplasmic and mitochondrial tRNAs (A and B, lanes 7 and 8, respectively). Hence, substitution of the putative catalytic aspartate residue (Asp 168 ) by an alanine in the characteristic conserved GRTD sequence of RNA:⌿-synthases abrogate the complementation capacity of the YGR169c ORF.
Finally we tested the capacity of the His 6 -tagged Ygr169 protein produced in E. coli to form residue ⌿ 31 in both cytoplasmic and mitochondrial tRNAs extracted from the  ⌬YGR169c strain. As shown in Fig. 3, A and B, after incubation with the recombinant protein a strong stop in the CMCT/RT pattern was detected for both cytoplasmic and mitochondrial tRNAs Met (lanes 7 and 8). Taken together, our results demonstrate that ORF YGR169c encodes the tRNA:⌿ 31 -synthase of S. cerevisiae. Because this is the sixth RNA:⌿-synthase characterized thus far in yeast, we propose to rename the correspond-ing gene PUS6 and the expressed protein sixth pseudouridine synthase Pus6p.
Cellular Localization of Pus6p-Based on the disappearance of ⌿ 31 in both cytoplasmic and mitochondrial tRNAs upon PUS6 deletion and taking into account the fact that mitochondrial tRNAs are transcribed in mitochondria and remain in this cellular compartment, the Pus6p enzyme was expected to have a dual localization (cytoplasmic and mitochondrial). Indeed, the PSORT (57) (Web site address psort.nibb.ac.jp) and NNPLS (58) (Web site address predict.sanger.ac.uk/nnpsl/) software both predict that Pus6p is mostly mitochondrial with a possible presence of a cytoplasmic fraction.
To get more information on the subcellular localization of Pus6p we produced a Pus6 protein N-terminally tagged with the GFP. Expression of the GFP-Pus6p fusion in the ⌬YGR169c strain restored U-to-⌿ conversion at position 31 in both cytoplasmic (data not shown) and mitochondrial tRNAs Met (Fig.  3C), indicating that the fusion protein conserved the ⌿-synthase activity in both the cytoplasmic and mitochondrial compartments. The results presented in Fig. 4 are in agreement with the predicted dual localization of the GFP-Pus6p in both the cytoplasm and mitochondria, because a uniform fluorescence of the cytoplasmic compartment was observed (GFP column). In contrast, the nucleus, visualized by DAPI staining, had only a low fluorescence level (Fig. 4, DAPI column), indicating that Pus6p is likely absent in this compartment.

A Single Gene Product Is Responsible for ⌿ 31 Formation in Both Cytoplasmic and Mitochondrial tRNAs-In this work we
showed that the S. cerevisiae ORF YGR169c, now renamed gene PUS6, encodes the site-specific tRNA:⌿-synthase, which catalyzes U-to-⌿ conversion at position 31 of tRNAs. PUS6 is indeed likely to be the only gene in yeast that is responsible for ⌿ 31 formation in both the cytoplasmic and mitochondrial tRNAs, because deletion of the PUS6 gene abolished formation of ⌿ 31 both in the cytoplasmic tRNA Met (the only cytoplasmic tRNA harboring ⌿ 31 ) and in several mitochondrial tRNAs. The existence of a single gene in yeast S. cerevisiae that is responsible for a given modification of both cytoplasmic and mitochondrial tRNAs was described already and seems to be rather general. Indeed, yeast tRNA:⌿-synthases Pus4p (46), Pus3p (44), and also the tRNA:m 2 2 G-methyltransferase Trm1p (59) and ⌬ 2 -isopentenyl-transferase Mod5p (60) acting both on cytoplasmic and mitochondrial tRNAs are produced from a single mRNA, and part of the translation product is targeted to mitochondria. In the cases of Trm1p and Mod5p, two distinct proteins are produced from the same mRNA by an appropriate choice between alternative initiation codons (for review see Ref. 60).
Based on its activity in the modification of cytoplasmic and mitochondrial tRNAs, Pus6p should be located in both compartments. As found for other mitochondrial RNA:⌿-synthases (Pus3p, Pus4p, and Pus5p; Refs. 39, 44, and 46), no characteristic N-terminal addressing signal (61,62) is present in Pus6p. However, the N-terminal 30-amino acid sequence of Pus6p can adopt an ␣-helical conformation with scattered arginine and lysine residues, two characteristic features of peptides able to target proteins in mitochondria (61,62). This particular mitochondrial addressing signal may be required for a partition of Pus6p in both mitochondria and the cytoplasm. Indeed, by use of the GFP-Pus6p fusion protein we showed that an important part of Pus6p remains in the cytoplasm. The observed absence of tagged Pus6p in the nucleus is in agreement with the absence of detectable nuclear localization signal. This implies that the site of action of Pus6p on cytoplasmic tRNA Met is the cytoplasmic compartment, and this is in contrast with the  5 and 6) indicates the presence of unmodified U residue at position 31. To verify the direct involvement of protein Ygr169 in the U-to-⌿ conversion at position 31 of tRNAs, the ⌬YGR169c strain was transformed with plasmid p416GalS bearing the wild-type (Asp 168 ) or mutated (Asp 168 3 Ala) YGR169c ORF. Total RNA was extracted from the two transformed strains, and the cytoplasmic and mitochondrial tRNA Met were analyzed by the CMCT/RT approach (shown in lanes 7, 8, 9, and 10 in A and B).
Pus1p tRNA:⌿-synthase that acts in the nucleus on the precursors of cytoplasmic tRNAs (43,45). This difference is likely explained by the fact that tRNA Met (CUA), the unique cytoplasmic tRNA substrate of Pus6p, is transcribed without intron, whereas several of the Pus1p tRNA substrates are produced with introns, and some of the U-to-⌿ conversions are intron-dependent.
Pus6p Is Specific for Position 31 in tRNAs-The 15 positions at which ⌿ residues are found in yeast cytoplasmic tRNAs are not uniformly distributed along the molecules; some are clustered (for example, positions 26 -28, 31-32, 34 -36, and 38 -39).
Until the present characterization of Pus6p, it was generally considered that consecutive ⌿ residues in tRNAs are formed by the same enzyme. Indeed, a single enzyme (TruA in E. coli or Pus3 in yeast) catalyzes the U-to-⌿ conversion at positions 38, 39, and 40 (in E. coli) (26,44). Similarly, the modification at positions 26,27,28,34,35, and 36 is ensured by the same enzyme Pus1p in yeast (43,45). By extrapolation one would expect that the two consecutive ⌿ residues at positions 31 and 32 are formed by the same enzyme. The results of the present study clearly demonstrate that at least two distinct enzymatic activities are required for U 31 and U 32 modification.
Sequence comparison of yeast tRNAs shows that when a uridine residue is present at position 31 in a tRNA, it is always converted into a ⌿ residue. The comparison of eight tRNA sequences around the modification site does not reveal any clear consensus motif that may explain such specific recognition. Although the recognition properties of Pus6p deserve further experimental studies, this apparent lack of identity elements suggests the sequence-independent recognition of a structural motif, as found for Pus1p or Pus3p RNA:⌿-synthases (44,45). It is noteworthy that another potential pseudouridylation site (⌿ 13 ) in tRNAs is present in a similar structural context (the penultimate base pair in the stem), but it is not a Pus6p substrate. This is likely because of the different three-dimensional contexts of positions 13 and 31 in tRNAs. Position 31 is well exposed in solution and readily accessible for interactions, whereas position 13 is buried in the tRNA core structure, and its 3Ј-adjacent nucleotide is engaged in a tertiary interaction. These structural differences likely justify the existence of another specific enzymatic system acting at position 13.
Cell Growth Is Unaffected by Deletion or Mutational Inactivation of the PUS6 Gene-Despite its action on both cytoplas-  A and B) and GFP-tagged Pus6 protein (C). His 6 -Pus6p was partially purified as described under "Experimental Procedures." Total RNA extracted from the ⌬YGR169c strain was used as the substrate. After incubation for 2 h at 37°C with His 6 -Pus6p, the RNA was phenol-extracted and modified by CMCT. The presence of ⌿ 31 in the cytoplasmic (cyt, A) and mitochondrial (mit) tRNAs Met (B) was tested by primer extension with the oligonucleotides 1679 and 1680 as described in the legend to Fig. 2. RT stops at position 31 are indicated. C, the ⌬YGR169c strain was transformed with plasmid pPS808 expressing the GFP-Pus6 protein under the control of the Gal1 promoter. Total RNA from the transformed strain was analyzed by the CMCT/RT approach to test for the presence of ⌿ 31 in the mitochondrial tRNA Met (as described in the legend for Fig. 2).
FIG. 4. Cellular localization of GFP-Pus6p observed by direct fluorescence microscopy. The nuclear compartment was stained by DAPI. The same cells are also shown in visible light. The first and second rows correspond to the ⌬YGR169c yeast strain transformed by plasmid pPS808-Pus6 expressing the GFP-tagged Pus6p. The third row corresponds to the same yeast strain transformed by a control plasmid pPS808 (expressing GFP only). The arrows shown in the DAPI staining column indicate the position of the nucleus. mic tRNA Met and seven mitochondrial tRNAs, Pus6p was not found to be an essential protein in yeast in the conditions tested, and no growth phenotype related to PUS6 gene deletion was observed. This situation is rather common for the characterized yeast tRNA:⌿-synthases. Indeed, independent deletions of PUS1 or PUS2 (43), PUS4 (46), or PUS5 (39) genes in S. cerevisiae did not result in a detectable growth phenotype. Even the absence of Pus1p, which modifies tRNAs at eight different positions and U2 snRNA at position 44 (11), had no dramatic influence on cell growth. After deletion of the PUS3 gene, growth was reduced, but the reasons for that are not yet clearly established (44,63). The only case in which gene deletion led to impaired growth is the rRNA-specific ⌿-synthase Cbf5p (64). However, the growth defect in the ⌬CBF5 strain seems to be related to a defect in rRNA maturation and not to the absence of ⌿ residues in rRNA (65).
The absence of a marked phenotype upon the deletion of genes coding for RNA:⌿-synthases is not restricted to yeast. Similar observations were made also for both tRNA or rRNAspecific bacterial RNA:⌿-synthases (31,38,66). However, a slow growth phenotype related to RluD deletion in E. coli was described recently (25). Yet here again it may be linked to a function of RluDp in ribosome biogenesis and not to RNA:⌿synthase activity per se (67).
However, despite the absence of a detectable growth phenotype of the ⌬PUS6 strain, our results do not exclude the possibility that its activity confers subtle selective advantages under special conditions, similar to the ones reported recently for E. coli truB-deleted strains in competition with wild-type cells (66).
Remaining Pseudouridylation Sites with Unidentified RNA: ⌿-Synthases in S. cerevisiae-Of nine yeast proteins bearing RNA:⌿-synthase sequence motifs, only three proteins (Pus2p, Rib2p, and Ydl036p) remain to be characterized experimentally in yeast, whereas at least four major pseudouridylation sites in tRNAs (positions 13, 32, and 35 in cytoplasm and position 32 in mitochondria) and five sites in UsnRNAs still have no assigned RNA:⌿-synthases. It is rather unlikely that these remaining enzymes may account for RNA pseudouridylation at all unassigned sites in tRNAs and UsnRNAs.
Concerning UsnRNAs, the existence of small nucleolar RNAguided 2Ј-O-methylation and pseudouridylation was demonstrated in vertebrates (68 -70). However the existence of a similar system for UsnRNA pseudouridylation in yeast seems unlikely, because in the absence of Cbf5p expression, UsnRNAs remain modified (11), unless another RNA:⌿-synthase can associate with pseudouridylation RNA guides.
Two hypotheses can be proposed to explain this apparent lack of genes corresponding to RNA:⌿-synthases in S. cerevisiae. The first one implies the existence of another RNA:⌿synthase family, the members of which are not homologous to the already known E. coli and S. cerevisiae enzymes. Because all the characterized yeast RNA:⌿-synthases were identified first by sequence homology with E. coli proteins, no systematic biochemical study of yeast RNA:⌿-synthases was done. Only the presence of three chromatographically distinct enzymes specific for ⌿ 13 , ⌿ 32 , and ⌿ 55 in tRNAs was noticed (71). Hence, the existence of another uncharacterized family of RNA:⌿synthases cannot be excluded. Biochemical studies will be necessary to address this question.
An alternative possibility is the existence of RNA:⌿-synthases with overlapping substrate specificities, such that the deletion of a single gene would not affect RNA modification in vivo. In other words, it may well be that the effect of deletion or inactivation of one of the characterized enzymes has been masked by the presence of another enzyme with overlapping specificity. This situation was reported already for ⌿ 35 formation in tRNA. Indeed, although yeast Pus1p is capable of catalyzing ⌿ 35 formation in vitro, deletion of the PUS1 gene does not affect ⌿ 35 formation in vivo, and extracts from the ⌬PUS1 strain are still active in ⌿ 35 formation (45). A similar situation may occur for other modification sites in tRNAs. The search for such a possibility will require the generation of strains mutated in two or three RNA:⌿-synthase genes and a detailed characterization of the specificity of recombinant RNA:⌿-synthases in vitro. 3