Conserved Bases in the TΨC Loop of tRNA Are Determinants for Thermophile-specific 2-Thiouridylation at Position 54*

2-Thioribothymidine (s2T) is a post-transcriptionally modified nucleoside of U54 specifically found in thermophilic bacterial tRNAs. The 2-thiocarbonyl group of s2T54 is known to be responsible for the thermostability of tRNA. The s2T54 content in tRNA varies depending on the cultivation temperature, a feature that confers thermal adaptation of protein synthesis in Thermus thermophilus. Little is known about the biosynthesis of s2T, including the sulfur donor, modification enzyme, and the tRNA structural requirements. To characterize 2-thiolation at position 54 in tRNA, we constructed an in vivo expression system using tRNAAsp with an altered sequence and a host-vector for T. thermophilus. We were able to detectin vivo activity of s2T54 thiolase using phenyl mercuric gel electrophoresis followed by Northern hybridization. 2-Thiolation at position 54 was identified in the precursor form of the tRNA, indicating that 2-thiolation precedes tRNA processing. To ascertain the elements that determine 2-thiolation in tRNA, systematic site-directed mutagenesis was carried out using the tRNAAspgene. Conserved residues C56 and A58 were identified as major determinants of 2-thiolation, whereas tertiary interaction between the T and D loops and non-conserved nucleosides in the T loop were revealed not to be important for the reaction.

A characteristic structural feature of tRNA is post-transcriptional modification. The roles of modified nucleosides in tRNA function are important and wide-ranging. They are known to include codon recognition, reading-frame maintenance, stabilization of the tertiary structure, and serving as identity determinants for amino acid specificity (1).
The melting temperatures of tRNAs from the extreme thermophile Thermus thermophilus sp. are 3-10°C higher than those of corresponding tRNA species from the mesophilic bacterium Escherichia coli, a feature that cannot be explained solely by the higher G-C pair content in Thermus tRNAs (2). Analyses of modified nucleosides in tRNAs from T. thermophilus revealed a thermophile-specific sulfur-containing modified nucleoside that was identified as 2-thioribothymidine (s 2 T) 1 (2, 3), a 2-thiolated derivative of 5-methyluridine (ribothymidine (T)) located at position 54 in the T loop of almost all tRNAs (4). Because s 2 T54 is also present in tRNA from hyperthermophilic Archaea such as Pyrococcus furiosus, which contains about 0.77 mol % of s 2 T when cultured at 100°C (5), 2-thiolation of T54 is postulated to be a common modification responsible for the thermostabilization mechanism of tRNA in both thermophilic eubacteria and Archaea. The 2-thiolation of T54 increases along with elevation of the cultivation temperature without any changes in other modifications such as 1-methyladenosine at position 58 (m 1 A58) or 2Ј-O-methylguanosine at position 18 (Gm18) (6); more than half of the tRNAs in T. thermophilus HB8 cells grown at more than 80°C were found to contain s 2 T54 instead of T54, whereas at 50°C only a small proportion had s 2 T. In addition, the tRNA melting temperature increased concomitantly with increases in the s 2 T content (6). These findings indicate that 2-thiolation of T54 is responsible for the thermostability of T. thermophilus tRNA under diverse cultivation temperatures, thereby ensuring the thermal adaptation of protein synthesis.
The temperatures of the inflection point in the specific CD signal (7) and in the characteristic chemical shift in the NMR spectra of s 2 T in T. thermophilus tRNA (8) show a good match with the melting temperature monitored by UV absorbance, suggesting a close correlation between the local conformation of s 2 T54 and the structural stability of tRNA. The mechanism of tRNA structure stabilization conferred by s 2 T has been elucidated by proton NMR analysis (9); the ribose puckering of s 2 T preferentially takes the C3Ј-endo-gg-anti conformation as do all residues in A-form RNA because of the steric effect of the bulky 2-thiocarbonyl group toward the 2Ј-hydroxyl group. This inherent rigidity of s 2 T54 gives stability to the elbow region formed by D loop-T loop interaction, resulting in the thermostability of the overall tRNA tertiary structure (10).
Although s 2 T54 is clearly a key modification for tRNA stability and function at elevated temperatures, information on its biosynthesis is very limited. Features that remain to be elucidated include the sulfur donor, modification enzymes or related genes, and the tRNA structural elements necessary for 2-thiolation. A genetic approach is, thus, indispensable for the characterization of this modification. Using the leuB gene as a selective marker, some of the present authors recently developed a host-vector system for T. thermophilus (11) that facilitates studies on the thermostability of proteins or RNAs from the bacterium. Here, we describe the detection of 2-thiolation activity in vivo in a reporter tRNA expressed using this T. thermophilus host-vector system. A systematic mutation analysis enabled us to characterize 2-thiolation in the maturation process of the tRNA and to identify the structural requirements for 2-thiolation at position 54.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Media-T. thermophilus TTY1 (a ⌬leuB⌬pyrE strain derived from T. thermophilus HB27) (11) was used as the host strain throughout. E. coli JM109, DH5␣, and MC1061 were used as hosts for the genetic manipulation of plasmids. Rich and minimal media for T. thermophilus were used according to the literature (12). Uracil (20 g/ml) was supplied to the minimal medium. T. thermophilus was cultured at 70°C in all the experiments.
Site-directed Mutagenesis-Plasmids harboring 24 tRNA variants were prepared using a QuikChange TM site-directed mutagenesis kit (Stratagene). Target mutations were introduced into the plasmids using 30-mer forward primers and the complementary reverse primers, both of which possessed the target mutation in the middle section. pCR-XL-tRNA Asp * and pCR-XL-tRNA Asp *(U8A) were exploited as template plasmids for mutagenesis. After the sequences of these tRNA variants were confirmed by a DNA sequencer (as above), the tRNA operons were inserted into pT8leuB to construct the respective tRNA expression vectors.
Purification of tRNA Asp* from T. thermophilus-About 1000 A 260 units of total RNA were obtained by the acid guanidinium thiocyanatephenol-chloroform extraction method (14) from a 3L late log-phase culture of the TTY1 transformant expressing tRNA Asp *. Total RNA was fractionated on a DEAE-Sepharose fast flow column (1 ϫ 40 cm) with a linear gradient of NaCl and MgCl 2 consisting of 500 ml of elution buffer A (20 mM Tris-HCl (pH 7.5), 200 mM NaCl, and 8 mM MgCl 2 ) and elution buffer B (20 mM Tris-HCl (pH 7.5), 450 mM NaCl, and 16 mM MgCl 2 ) with gravitational flow. Fractions containing tRNA Asp * were determined by dot-hybridization using the DNA probe AS-1, combined, precipitated with ethanol, and dissolved in a binding buffer (1.2 M NaCl, 30 mM Tris-HCl (pH 7.6) and 15 mM EDTA). To procure tRNA Asp * on its own, the solid-phase DNA probe method (15) was employed using a 3Ј-biotinylated oligonucleotide with a sequence identical to that of the DNA probe AS-1. The tRNA Asp * obtained was further purified by denaturing polyacrylamide gel electrophoresis.
RNA Sequencing-Purified tRNA Asp * was sequenced by the method of Donis-Keller (16). Partial enzymatic digestion was carried out with the following base-specific RNases: RNase T 1 (Amersham Biosciences), RNase U 2 (Seikagaku Kogyo), RNase PhyM (Amersham Biosciences), and RNase CL 3 (Roche Molecular Biochemicals). Digested fragments were electrophoresed separately in lanes on a 15% denaturing polyacrylamide gel along with the undigested control and alkaline-hydrolyzed tRNAs. Modified nucleotides were identified by the post-labeling method (17).
Mass Spectrometry-An LCQ ion trap mass spectrometer (Thermo-Finnigan) equipped with an electrospray ionization source and a MAGIC 2002 liquid chromatography system (Michrom BioResources) was used to analyze RNA fragments digested with RNase T 1 . Purified tRNA Asp * (0.4 g) was digested with RNase T 1 (2.5 units) in 25 mM ammonium acetate (pH 5.3) at 37°C for 1 h, and the digest was subjected to mass spectrometric analysis. Oligonucleotides produced by RNase T 1 digestion were detected by LC/MS in the negative ion mode as described by Qiu and McCloskey (18)  Detection of 2-Thio Modification in tRNA Asp* and Its Precursor-The 2-thiolation at position 54 in tRNA Asp * and its precursor were detected by retardation in an electrophoretic system consisting of a 10% polyacrylamide gel (10 ϫ 12 ϫ 0.1 cm 3 ) containing 7 M urea polymerized in the presence of 25 or 100 M [(N-acryloylamino)phenyl]mercuric chloride (APM), which was synthesized as described by Igloi (19). Total RNA (1.6 -3.2 g) was resolved on the APM gel. After electrophoresis, the gel was soaked in 0.2 M ␤-mercaptoethanol for 1 h to reduce and/or break the sulfur linkages formed between the thiolated-RNA and the APM in the gel. The RNA was transferred onto a nylon membrane (Hybond-N ϩ , Amersham Biosciences) by blotting using TBE Buffer, which was hybridized with the 5Ј-32 P-labeled DNA probes for tRNA Asp * or its precursor. The retarded band resulting from the presence of thiocarbonylated nucleotides in RNA was visualized by a Fuji BAS1000 bioimaging analyzer.
Primer Extension Analysis of 5Ј End of Precursor tRNA-The 5Ј end of the precursor tRNA Asp * was determined by the primer extension technique. Total RNA from TTY1/pT8leuB or TTY1/pEx-Asp* (U8A/ G19C) was heat-denatured and annealed with the 5Ј-32 P-labeled primer RT-1 (15-mer), 5Ј-CAACTACACC TACAC-3Ј. Reverse transcription was performed with Moloney murine leukemia virus reverse transcriptase (Toyobo) according to the manufacturer's instructions. To determine the initiation position of the transcription, the primer extension product was electrophoresed together with the dideoxy sequencing products of the template plasmid pCR-XL-tRNA Asp *(U8A) using the same primer (RT-1) on a 15% denaturing polyacrylamide sequencing gel.

Expression of T. thermophilus tRNA Asp Gene with an Altered
Sequence in E. coli and T. thermophilus Cells-To detect activity of the s 2 T (54)-thiolase in T. thermophilus cells and to investigate its recognition elements in tRNA in vivo, we first expressed an artificial tRNA gene encoded on a plasmid. We selected T. thermophilus HB8 tRNA Asp as our model tRNA species because the presence of s 2 T at position 54 has been verified in this tRNA (20). We introduced mutations at 11 base pairs in the acceptor, anticodon, and T stems of tRNA Asp (Fig.  1B) so that the resultant tRNA (named tRNA Asp *) could be discriminated from the native tRNA Asp by Northern hybridization using the DNA probes for tRNA Asp *. The tRNA Asp * gene was integrated into the tRNA Ser operon from T. thermophilus strain HB8 by replacing the tRNA Ser gene, as depicted in Fig.  1, A and B). The reason for employing this strategy is that the tRNA Ser operon is the only T. thermophilus tRNA gene reported so far that includes both 5Ј-and 3Ј-adjacent apparent promoter and terminator sequences. The tRNA Asp * operon was synthesized by PCR using six DNA oligos and ligated into the EcoRI and EcoRV sites of an E. coli-T. thermophilus shuttle vector, pT8leuB (11), resulting in an expression vector for tRNA Asp * (pEx-Asp*).
Total RNA from cells of E. coli JM109 harboring pEx-Asp* was separated by 10% denaturing PAGE followed by Northern hybridization with the probe AS-2. Two distinct RNA bands associated with tRNA Asp * could be observed ( Fig. 2A). The longer band (asterisk), which appeared in close proximity to 5 S rRNA, was considered to be a precursor of tRNA Asp *. The shorter band, located in the E. coli class I tRNA cluster, was presumed to be the mature form of tRNA Asp *. The result demonstrated that the tRNA operon from T. thermophilus could function even in E. coli cells, which can be explained by the fact that the promoter sequence of this gene (TTGACG (Ϫ35)/TA-CACT (Ϫ10) (Fig. 1A) is similar to the E. coli tRNA promoter consensus sequence (TTGACA (Ϫ35)/TATAAT (Ϫ10) (21). However, the heterologous expression of Thermus tRNA in E. coli cells resulted in strong accumulation of the tRNA precursor (ϳ70% of the total expression), indicating that the slight disparity with the consensus sequence and the high copy number of the shuttle vector derived from pUC118 in E. coli cells (22,23) may have led to the tRNA precursor being produced too abundantly to be processed sufficiently by tRNA maturases such as RNase P (24) or 3Ј ribonucleases (25).
When the expression of tRNA Asp * in T. thermophilus was examined by Northern hybridization with the probe AS-1 (Fig.  2B), tRNA Asp * originating from the vector was detected mainly in the mature form, although as much as 30% of the total product was observed as the putative precursor (asterisk). Judging from the intensity of the Northern blotting, the expression of total RNA in T. thermophilus cells was lower than that in E. coli cells. This can be accounted for by the likely low copy number of pEx-Asp* in T. thermophilus, because this expression vector has a replication origin derived from pTT8, whose copy number is about eight (26).
Expressed tRNA Asp* with the s 2 T (54) Modification-Our next task was to determine whether tRNA Asp * expressed in T. thermophilus cells was processed normally and modified in the same manner as the native tRNA and, in particular, whether or not it contained s 2 T. To isolate tRNA Asp *, total RNA from T. thermophilus cells harboring pEx-Asp* was fractionated by anion exchange column chromatography so as to enrich the fractions that included tRNA Asp *. In this step, we were able to successfully concentrate the tRNA Asp * fractions and exclude those containing the precursor. tRNA Asp * was then purified to homogeneity by solid-phase DNA probing (Fig. 3A). End-labeling with 32 P and sequencing by Donis-Keller's enzymatic digestion method (Fig. 3B) showed that the purified tRNA Asp * had the expected sequence, including both the 5Ј and 3Ј ends. Abnormalities in the bands suggested the presence of some modified residues (the expected modifications are parenthesized in Fig. 3B). The primary sequence of tRNA Asp * was further determined by post-labeling and LC/MS analysis, which enabled us to identify 7 post-transcriptional modifications at 8 positions  Fig. 3C). The post-labeling method clearly identified both T and s 2 T at position 54 (data not shown), which is consistent with a previous report that s 2 T is derived from a partial modification of T induced by the cultivation temperature (6). The presence of s 2 T at position 54 was further examined by LC/MS with RNase T 1 -digested fragments of tRNA Asp * (Fig. 4). Although the frag-  (11). The sequences of the aminoacyl stem, T stem, and anticodon stem were altered from those of the native tRNA Asp gene to allow the expressed tRNA to be discriminated from native tRNAs in T. thermophilus cells. AS-1, AS-5Ј, and AS-3Ј indicate the positions of these DNA probes used to detect mature tRNA Asp * and its precursors.
FIG. 2. Detection of expressed tRNA Asp* by PAGE followed by Northern hybridization. A, total RNA (0.8 g) from cells of MC1061/ pT8leuB and JM109/pEx-Asp* was separated by 10% denaturing gel electrophoresis, and the gel was stained by ethidium bromide (lanes 1  and 2, respectively). Northern hybridization patterns of the same gel with the probe AS-2 are shown in lanes 3 and 4, respectively. tRNA Asp * and its precursors are indicated by the arrow and asterisk, respectively. B, electrophoresis on 10% denaturing PAGE of total RNA (1.6 g) from TTY1/pT8leuB (lane 1), TTY1/pEx-Asp* (lane 2), and their Northern blots with the probe AS-1 (lanes 3 and 4, respectively). The arrow and asterisk show tRNA Asp * and its precursors, respectively. ment s 2 T⌿CGp was clearly detected at retention time 20.1 min as singly and doubly charged ions, the fragment T⌿CGp was hardly detected even though the LC/MS was highly sensitive, suggesting that tRNA Asp * preferentially possesses s 2 T at position 54.
Our experimental results demonstrated that tRNA Asp * was expressed in T. thermophilus cells in the mature form and that it carried 8 modifications including s 2 T at position 54. We thus successfully detected s 2 T synthesis activity in vivo.
Detection of 2-Thiouridylation at Position 54 in tRNA Asp* by APM Gel Electrophoresis-Because the procedures we employed to detect s 2 T54 are unsuitable for routine assays of multiple samples, we searched for a simpler method that does not require purification of the expressed tRNA. Igloi (19) reports an affinity electrophoresis system in which a polyacrylamide gel is co-polymerized with APM. In this system, which was reported to be successful in detecting tRNA with the s 4 U or 5-methylaminomethyl-2-thiouridine modification (19), the electrophoretic mobility of thiolated tRNAs is retarded compared with that of non-thiolated tRNAs due to the specific interaction between the thiocarbonyl group and the mercuric compound. By combining this technique with Northern hybridization using a probe specific for the relevant tRNA, the thiolated nucleosides in a particular tRNA species can be detected (27). We therefore used this approach to identify the 2-thiouridylation at position 54 in tRNA Asp * expressed in T. thermophilus cells.
When APM gel electrophoresis was performed, specific retardation of tRNA Asp * was observed (data not shown). However, because tRNA Asp * has 2 thiouridines, s 4 U8 and s 2 T54, this retardation would have been mediated by both of them. To differentiate gel retardation due to 2-thiolation at position 54 from that caused by s 4 U8, we first eliminated s 4 U8 from the expressed tRNA Asp * by introducing a U to A point mutation at position 8 in the expression vector pEx-Asp*. As shown in Fig.  5A (lanes 1 and 4), the mutant tRNA Asp *(U8A) without the s 4 U8 modification was expressed in T. thermophilus cells in its mature form together with a few possible precursors from the resultant plasmid (pEx-Asp*(U8A)). Total RNA from this mutant strain was subjected to APM gel electrophoresis using two APM concentrations, 25 and 100 M (ϩ and ϩϩ, respectively, in Fig. 5). In the gels stained by ethidium bromide (lanes 2 and 3), most of the intrinsic tRNAs exhibited strong retardation due to s 4 U8. Furthermore, when compared with the 5 S rRNA band, which contains no thiolated nucleotides, the tRNA cluster was retarded in accordance with the concentration of APM. Next, the band corresponding to the expressed tRNA Asp *(U8A) was detected by Northern hybridization using the probe AS-1 (Fig.  5A, lanes 4 -6). Slight but significant retardation concomitant with the increment in the APM concentration could be observed. This was thought to be almost certainly caused by the presence of the 2-thio group at position 54 in tRNA Asp *(U8A) given that band retardation due to 2-thiolation at position 54 is likely to be much less than the strong retardation resulting from 4-thiolation of U8. This interpretation is supported by the reported successful detection of the other 2-thiouridine derivative, 5-methylaminomethyl-2-thiouridine, by the APM method (19).
To confirm that the observed band retardation did actually derive from 2-thiolation at position 54, we constructed another mutant vector, tRNA Asp *(U8A/U54A), in which a U to A point mutation was introduced at position 54 in addition to that at position 8. As shown in Fig. 5B (lanes 7 and 10), tRNA Asp *(U8A/U54A) was expressed in its mature form together with a large accumulation in the precursor form, indicating that the U54A mutation had an inhibitory effect on tRNA maturation. In the presence of APM, the bands corresponding to the mature form of tRNA Asp * (U8A/U54A) (lanes 11 and 12) appeared at the same position as the bands for tRNA Asp *(U8A) and tRNA Asp *(U8A/U54A) on the gels without APM (lanes 4 and 10, respectively). Moreover, no difference in mobility was observed even when the concentration of APM was increased (lanes 11 and 12).
The above results clearly demonstrated that the retardation of tRNA Asp *(U8A) in APM gel electrophoresis actually resulted from the presence of the 2-thio group at position 54 in the tRNA and confirmed the feasibility of using this simple assay system to investigate the sequence requirements of tRNA for in vivo 2-thiolation at position 54.
Evidence That 2-Thiolation Precedes tRNA Processing-In the APM gel electrophoresis, specific retardation of the tRNA Asp * (U8A) precursor was observed (asterisks in lanes 4 -6 of Fig. 5A) and verified by the fact that the tRNA Asp *(U8A/ U54A) precursor showed no such retardation (Fig. 5B, lanes  10 -12). These observations suggested that 2-thiolation at po-sition 54 occurs in the precursor form and precedes tRNA processing. To investigate this hypothesis experimentally, we constructed the vector tRNA Asp *(U8A/G19C). As shown in Fig.  6 (lanes 3 and 4), precursors of tRNA Asp *(U8A/G19C) (asterisks) were shifted in the APM gel in the same manner as the mature tRNA (indicated by arrows), which is a clear indication that the 2-thiolation occurred in precursor tRNAs.
To differentiate the type of precursor tRNA that is 2-thiolated, we performed Northern blot analysis. Two oligo DNA probes, AS-5Ј and AS-3Ј, were designed to detect tRNA precursors with 5Ј-leader and 3Ј-trailer sequences, respectively (Fig.  1B). The fact that the mature tRNA was not detected in lanes 1 and 2 of Fig. 6 shows that these probes were able to specifically detect precursor tRNAs. Because the longer precursors (** in Fig. 6) were hybridized by both probes, this type of precursor was considered to have the 5Ј-leader and 3Ј-trailer sequences. The shorter precursors (*) were only detected by AS-3Ј, suggesting that they have the 3Ј-trailer sequence but no 5Ј-leader. Primer extension analysis revealed that the 5Ј end of the precursors was at position Ϫ7 from the 5Ј end of mature tRNA (ϩ1) (Fig. 6B). Judging from the presence of heterologous Northern bands (Fig. 6A), the longer and shorter precursors contained several species of different sizes. Because the primer extension analysis indicated that these precursors have a uniform 5Ј end, it is presumed that their unequal lengths arise from 3Ј end heterogeneity of transcription termination (28) and/or 3Ј-trimming (25). All the precursors were shifted in the APM gel (Fig. 6, lane 4), from which we conclude that 2-thiolation occurs in the precursor form with both 5Ј-leader and 3Ј-trailer sequences. Hence, thermophile-specific 2-thiolation most probably precedes tRNA processing.
Identification of Recognition Sites in tRNA for 2-Thiolation at Position 54 -Having established the validity of the relatively simple APM gel electrophoresis system in detecting s 2 Tthiolase activity in T. thermophilus cells, we utilized the same technique to investigate recognition elements in tRNA Asp * (U8A) for s 2 T-thiolase, which is responsible for 2-thiolation at position 54. Although the interaction between U8 and A14 is crucial for the formation of the proper tRNA three-dimensional structure (29,30), we have shown that the mutant tRNA Asp * (U8A) was efficiently 2-thiolated, which indicates that the en-  lanes 1-3) and TTY1/pEx-Asp*(U8A/U54A) (lanes 7-9) was applied onto a 10% polyacrylamide gel containing 7 M urea with different concentrations (Ϫ, 0 M; ϩ, 25 M; ϩϩ, 100 M) of APM, and the gel was stained by ethidium bromide. RNA in the gel was transferred to a nylon membrane and subjected to Northern hybridization with probe AS -1 (lanes 4, 5, 6, 10, 11, and 12). tire tRNA structure is not required for 2-thiolation at position 54. To identify which sites are required, we first examined the effect of D loop-T loop interaction. We did so because earlier work revealed that s 2 T plays a critical role in stabilizing the interaction between the D and T loops by extending the A-form double-stranded helix into the T loop (10), thereby giving stability to the tRNA structure as a whole. To disrupt the D-T loop interaction, we constructed three mutants bearing point mutations at G18 (G18A, G18U, and G18C) and one with a mutation at G19 (G19C), because both G18 and G19 are involved in the tertiary interaction between the two loops (29,30). When these mutated tRNAs Asp *(U8A) were subjected to APM gel electrophoresis, they were all retarded in the APM gel (Fig. 7A), from which we concluded that disruption of D-T loop interaction has no effect on the 2-thiolation at position 54. Therefore, the T loop itself appeared to be the most likely recognition site for s 2 T-thiolase.
To verify this, we investigated 18 point mutations at all the T loop positions of tRNA Asp *(U8A) except for U54 (i.e. at positions 55-60), in which each base was systematically substituted by each of the other three bases (Table I; Figs. 7B and  8A). The mutants bearing U55A, U55G, C56G, C56U, A58C, A58G, and A58U gave no mature tRNA detectable by Northern hybridization but only strong accumulations of longer precursors (data not shown). The most likely reason is that these mutations had a marked influence on the tRNA processing activity similar to the effect of the U54A mutation noted earlier (Fig. 5). In these cases, we evaluated the 2-thiolation activity from the shifts of the precursors, because we demonstrated above that 2-thiolation occurs in the precursor form when both the 5Ј-leader and 3Ј-trailer sequences are present (Fig. 6). Seven of the mutations, G57A, G57C, G57U, G59A, G59C, G59U, and U60G, had almost no influence on the 2-thiolation activity (Table I and Fig. 7B).
A slight reduction in 2-thiolation activity was observed in the case of mutant U55C (Fig. 7B), whereas U55A and U55G showed no activity (Table I). Although conserved U55 (normally modified to ⌿55 in the cell) is involved in the tertiary basepairing of G18 and U55 (29,30), the slight reduction in the thiolation of the U55C mutant is considered to result from sequence-specific recognition by s 2 T-thiolase and not from disruption of D-T loop interaction, because there was no effect on the 2-thiolation of three G18 mutants (Fig. 7A).
Mutation at G59 to any other nucleoside (G59A, G59C, and G59U) did not affect the 2-thiolation activity (Fig. 7B), which is consistent with the fact that any of the four nucleosides can occur at position 59 in the T. thermophilus tRNA consensus sequence (Fig. 8B).
In the case of position 60, the U60G and U60C mutants were both 2-thiolated with efficiencies similar to and slightly lower than that of the wild type, respectively. In contrast, mutant U60A was not 2-thiolated (Fig. 7B). Because a pyrimidine is located at position 60 in T. thermophilus tRNAs (Fig.  8B), it is reasonable that the U60C mutation resulted in 2-thiolation. A possible explanation for non-thiolation of the U60A mutant is that A60 may be base-paired with U54 (the target site for s 2 T-thiolase) instead of the conserved U54-A58 base-pairing. This would result in the formation of a T stem with 6 base pairs, destroying the canonical 7-membered T loop structure, which may not be recognized by s 2 T-thiolase.
A strong reduction in 2-thiolation activity was observed in the cases of mutations at positions C56 and A58 (which is normally modified to m 1 A58 (1-methyladenosine at position 58) in cells) (Fig. 7B, Table I). These bases are highly conserved residues in Thermus tRNAs (Fig. 8B), and A58 is directly involved in tertiary base pairing with the target site U54 (which is normally modified to either T or s 2 T) (10). Taking these facts into consideration together with our results, it is supposed that 2-thiolation activity at position 54 requires the canonical T loop structure formed by the conserved U54-A58 interaction together with the conserved residue C56.
ϩ ϩ a Mutants of tRNA Asp* (U8A) are shown. b ϩ represents expression of mature tRNA; Ϫ represents no expression of mature tRNA (longer precursors only were expressed).
c In the cases of mutants expressing mature tRNA, the extent of 2-thiolation modification was determined as the ratio of the band strength at the shift position in the presence of APM to that of the whole amount of total expressed RNA and classified as fully modified (ϩ, 1-0.65), partially modified (Ϯ, 0.65-0.25), or not modified (Ϫ, 0.25-0). In the cases of mutants not expressing mature tRNA, the band-shift tendency of the precursor is indicated as ϩ or Ϫ (although only Ϫ is presented in the table).

DISCUSSION
Although almost 30 years have passed since we first found s 2 T54 in tRNAs from T. thermophilus (3), there has been no report during that time on the biosynthetic pathway of the s 2 T modification or on the identification of a putative s 2 T-thiolase, the key enzyme presumed to be responsible for the thermal adaptation of protein synthesis. In the present study, we first attempted to detect s 2 T thiolation activity in T. thermophilus cells by utilizing an E. coli-T. thermophilus shuttle vector (11), into which the T. thermophilus tRNA Asp gene with an altered sequence (tRNA Asp *) was introduced. This strategy enabled us to successfully express the non-native tRNA Asp * in T. thermophilus cells and identify s 2 T54, thereby demonstrating that s 2 T-thiolase activity could in fact be detected in T. thermophilus cells.
Several approaches have been employed to elucidate recognition mechanisms of tRNA modification enzymes toward tRNA. If the relevant enzyme(s) can be purified or a recombinant enzyme is available, it may be possible to reconstruct the modification reaction in vitro using unmodified tRNA as a substrate. There has been much work in this area, especially with methyltransferases (31). Almost all known methyltransferases utilize S-adenosylmethionine as a common methyl donor, and methylation can be detected from the radioactivity of the 14 C-or 3 H-labeled methyl group. If an enzyme(s) cannot be purified, its characterization in vivo may be possible, as was done, for example, by microinjecting labeled-tRNA into the cytoplasm of the Xenopus oocyte (32). However, this approach does not always reflect the in vivo situation, particularly when heterologous tRNA species are examined.
Because the putative s 2 T-thiolase in T. thermophilus cells has never been identified or isolated, we developed our in vivo characterization method involving expression of the non-native tRNA Asp *. Although artificial, the tRNA Asp * gene was normally expressed to give a mature tRNA product with the full sequence and nucleoside residue modifications expected (Fig.  3C). Although both s 2 T54 and T54 were detected by postlabeling, only s 2 T54 was detected in the LC/MS analysis (Fig.  4), indicating the preponderance of the 2-thioribothymidine modification in tRNA Asp *. The native tRNA Asp from T. thermophilus HB8 was reported to possess 50% s 2 T54 when cultured at 70°C (20). The apparent difference in the degree of 2-thiolation at position 54 in the native and non-native tRNAs Asp may arise from dissimilarities in the amounts of tRNA expressed, the altered sequence of the constructed tRNA, and/or the fact that the host strains were different (tRNA from strain TTY1, a derivative of strain HB27, has a higher molar content of s 2 T than strain HB8 (33)).
The combination of APM gel electrophoresis and Northern hybridization proved to be a simple and sensitive method for the systematic analysis of 2-thiolation at position 54 in the series of tRNA mutants that we constructed. Using this method, we found that 2-thiolation precedes the processing of the tRNA precursor. Although we do not know whether 5-methylation of U54 occurs in the precursor form, it has been shown that m 5 U is synthesized in the tRNA su3 Tyr precursor with a 5Јleader sequence when it is expressed in E. coli (34) and that m 5 U occurs in intron-containing tRNA Phe and tRNA Tyr precursors from Saccharomyces cerevisiae (35). As Grosjean et al. have speculated (32), this 5Ј-methyl modification might stabilize tRNA precursors to promote the subsequent processing and other modification steps. We have shown here that T. thermophilus precursor tRNAs are also thiolated at residue 54, which probably occurs together with methylation at position 5 of U54 in the early stages of tRNA processing. Our finding is consistent with the evidence for mesophilic tRNAs. In T. thermophilus, the 2-thiolation apparently contributes particularly to the structural stabilization of precursor tRNAs at higher temperatures.
Our experiments with a series of structural mutants enabled us to identify the s 2 T (54)-thiolase recognition elements in tRNA. The determinants for 2-thiolation were ascertained to be conserved T loop residues, and the conformation of the T loop formed with these residues. The U54-A58 reverse Hoogsteen base pair is especially important for 2-thiolation. The residues necessary for recognition by the s 2 T-thiolase (Fig. 8A) were revealed to be broader than those making up the T. thermophilus tRNA T loop consensus sequence (Fig. 8B). This raises the possibility that almost all tRNA species can be recognized and modified to s 2 T54 and is consistent with the fact that 2-thiolation of T54 occurs in 68% of unfractionated tRNAs in T. thermophilus HB27 cultured at 80°C (33).
Two distinct modification steps are involved in s 2 T synthesis, 5-methylation and 2-thiolation. Because we have never detected 2-thiouridine at position 54 either in the present work or in many previous studies (2, 36), 5-methylation of U54 most probably occurs before its 2-thiolation. This raises the question whether 2-thiolation of U54 requires 5-methylation. The modification enzyme for 5-methylation of U54, known as tRNA(m 5 U54)-methyltransferase (RUMT), is encoded in trmA in E. coli (37) and in TRM2 in S. cerevisiae (38). The recognition pattern of the E. coli RUMT has been investigated (39,40). It is most likely that the T. thermophilus counterpart of RUMT has similar characteristics to the E. coli enzyme because the two species are both eubacteria. If this is the case, most of the s 2 T-thiolase recognition sites would be shared with RUMT, because it is known that RUMT strictly recognizes the conserved T loop structure and that D loop-T loop interaction is not necessary for this recognition (39,41), which is exactly what we observed in this study. On the other hand, a distinct difference in the recognition mechanisms of these two modification enzymes is evident with respect to residue 57. We found that in T. thermophilus the G57C mutation caused no reduction in 2-thiolation (Figs. 7B and 8), whereas the same mutation resulted in a drastic decrease in E. coli RUMT recognition activity to about only 4% of the k cat /K m of the wild-type tRNA (39). This difference in sensitivity of the mutation at residue 57 may mean that 2-thiolation and 5-methylation of U54 are independent modifications. It has been reported that in E. coli tRNAs, C-5 modification and 2-thiolation in the biosynthesis of 5-methylaminomethyl-2-thiouridine at position 34 are independent processes (42)(43)(44). The same may be true for 5-methylation and 2-thio-  Table I). Only data for mutants that could be thiolated are included. The bases shown as gray circles are strictly required for 2-thiolation; those shown as white circles are not strictly required. The mutant U55C is partially thiolated. B, T loop consensus sequence derived from sequences of 11 species of T. thermophilus tRNAs contained in the tRNA Compilation Data base (4). Positions 54, 55, 56, and 58 (gray circles) are completely conserved, whereas positions 57 and 60 are conserved as a purine and a pyrimidine, respectively. Only position 59 is not conserved. The dashed line between U54 and A58 depicts the reverse Hoogsteen base pair. lation of U54. Another possibility is that the apparent difference in the recognition mechanisms of s 2 T-thiolase and RUMT may arise from variances between in vivo and in vitro experiments. In any event, further investigations, including the use of a deletion strain of the trmA gene in T. thermophilus cells, are needed to elucidate this.
A clearer picture of the s 2 T-thiolase recognition mechanism and temperature-dependent control of the T to s 2 T ratio in tRNA is likely to emerge when s 2 T-thiolase is isolated and the 2-thiolation reaction is reconstituted in vitro, which are the goals of work currently in progress.