Temperature-dependent Biosynthesis of 2-Thioribothymidine of Thermus thermophilus tRNA*

2-Thioribothymidine (s2T) is a modified nucleoside of U, specifically found at position 54 of tRNAs from extreme thermophilic microorganisms. The function of the 2-thiocarbonyl group of s2T54 is thermostabilization of the three-dimensional structure of tRNA; however, its biosynthesis has not been clarified until now. Using an in vivo tRNA labeling experiment, we demonstrate that the sulfur atom of s2T in tRNA is derived from cysteine or sulfate. We attempted to reconstitute 2-thiolation of s2T in vitro, using a cell extract of Thermus thermophilus. Specific 2-thiolation of ribothymidine, at position 54, was observed in vitro, in the presence of ATP. Using this assay, we found a strong temperature dependence of the 2-thiolation reaction in vitro as well as expression of 2-thiolation enzymes in vivo. These results suggest that the variable content of s2T in vivo at different temperatures may be explained by the above characteristics of the enzymes responsible for the 2-thiolation reaction. Furthermore, we found that another posttranscriptionally modified nucleoside, 1-methyladenosine at position 58, is required for the efficient 2-thiolation of ribothymidine 54 both in vivo and in vitro.

Post-transcriptional modification is a characteristic feature of RNA molecules. In transfer RNA, post-transcriptional modification plays various roles that are required for the translation process, including fidelity control of codon recognition, reading frame maintenance, and stabilization of the tertiary tRNA structure (1).
2-thioribothymidine (s 2 T) 2 is a 2-thiolated derivative of 5-methyluridine (ribothymidine (rT)), located at position 54 of Thermus thermophilus tRNA (2), and has been shown to stabilize tRNA structure in a high temperature environment (3). In the extreme thermophilic eubacteria, Thermus thermophilus (4), and the archaea, Pyrococcus furiosus (5), the 2-thiolation level of rT54 in tRNA increases with cultivation temperature; the melting temperature (T m ) of the tRNA increases concomitantly with incremental increases in s 2 T content. These findings indicate that 2-thiolation of rT54 is responsible for the thermostability of thermophile tRNA at a variety of cultivation temperatures, thereby ensuring the adaptation of the protein synthesis machinery to specific thermal environments.
However, the body of knowledge regarding s 2 T biosynthesis is limited. First, the enzymes responsible for the tRNA modification and the genes involved in the 2-thiolation reaction have not been identified to date. Previously, we demonstrated that the tRNA structural elements required for 2-thiolation are the conserved bases in the TC loop and the structure created by those bases (6). In this report, we investigated the sulfur donor for s 2 T biosynthesis in vivo. We also examined the 2-thiolation reaction in vitro, in order to characterize the temperature dependence of this reaction, and considered the contribution of neighboring nucleoside modifications to the efficiency of 2-thiolation of rT54, both in vivo and in vitro.

EXPERIMENTAL PROCEDURES
Strains and Media-T. thermophilus HB8 was used as the wild-type strain throughout this study. Escherichia coli JM109 was used as the bacterial host for the genetic manipulation of plasmids. Other strains used in this study are listed in Table 1. Rich medium and minimal medium (MM) for T. thermophilus were used as described previously (9). Both wild-type and mutant cells were cultivated in the rich medium without or with 30 g/ml kanamycin, respectively, unless otherwise stated. For in vivo labeling of RNA, minimal medium without sulfur compounds (MMϪS) was prepared. The salts used had chloride ions (Cl Ϫ ) substituted for sulfate ions ( 35 S-Labeled compounds were purchased from American Radiolabeled Chemicals. Total RNA was extracted from the cells using ISOGEN (Wako). For alkaline treatment of tRNA, total RNA was incubated at 37°C for 1 h in 100 mM HEPES-KOH (pH 9.0). Then total RNA was separated by 10% PAGE containing 7 M urea, and the gel was dried and exposed to an imaging plate, followed by analysis using a BAS5000 bioimaging analyzer (Fuji Photo Systems). 35 S-Labeled Nucleoside Analysis by High Performance Liquid Chromatography (HPLC)-35 S-Labeled RNA was digested by Nuclease P 1 (Yamasa) and bacterial alkaline phosphatase (Takara) in 20 mM HEPES-KOH (pH 7.5) for 10 h at 37°C. The hydrolysate of RNA was analyzed by an L-7100 liquid chromatography system (Hitachi) with an ODS reversed-phased column (Inertsil ODS-3, 2.1 ϫ 250 mm; GL Science), using the solvent system described previously (10). Eluted fractions were collected every 30 s, UV absorbance at 254 nm was measured, and the radioactivity of each fraction was measured using an LSC6100 liquid scintillation counter (Aloka). Authentic 4-thiouridine (s 4 U) was purchased from Sigma. Authentic s 2 T was purified from the hydrolysate of total tRNA from T. thermophilus as described above, using a preparative column (Inertsil ODS-3, 10 ϫ 250 mm; GL Science), and the molecular weight and UV spectra were confirmed using a mass spectrometer, LCQ-DUO (ThermoFinnigan), and a spectrometer, Gene Spec III (Hitachi), respectively.
Construction of thiI, iscS, sufS, and trmI Mutant Strains-The target genes of T. thermophilus HB8 were disrupted by the insertion or the replacement of the highly thermostable kanamycin nucleotidyltransferase (htk) gene (11). To identify the candidate open reading frames, we made use of genome sequence data provided by the genome sequencing project of T. thermophilus HB8, which was in progress in Japan (now available on NCBI as an accession number of NC_006461).
To obtain the thiI::km insertion strain (NS0801), we cloned an ϳ3.2-kb region containing the thiI gene (TTHA1796) and flanking regions into a pT7Blue vector (Novagene), and the htk gene cassette was ligated into the NcoI site, located at the 16th amino acid of the thiI open reading frame. The htk gene cassette was amplified from pUC18-pJHK3 (11) using primers containing NcoI sites. Pairs of oligonucleotide DNAs used are as follows: thiI-F/thiI-R and htk-NcoI-F/htk-NcoI-R. The sequences of oligonucleotide DNAs used in gene disruptions are listed in Table 2. For the iscS::km insertion strain (NS0821), we cloned a 1.6-kb region that included the iscS gene (TTHA0456) and flanking regions. The htk gene cassette was ligated into the BlnI site, located at the 122nd amino acid. Pairs of oligonucleotide DNAs used are as follows: iscS-F/ iscS-R and htk-BlnI-F/htk-BlnI-R. To construct the sufS::km insertion strain (NS0822), we used a linear DNA fragment constructed using the PCR ligation. The 5Ј-and 3Ј-regions (0.5 kb) of sufS (TTHA1735) were amplified from the genome by PCR, and the htk cassette was amplified from pUC18-pJHK3 (11). The fragments had complementary sequence enabling the 5Ј-region, htk, and the 3Ј-region to be ligated by PCR. Pairs of oligonucleotide DNAs used are as follows: sufS-5Ј-F/sufS-5Ј-R, sufS-3Ј-F/sufS-3Ј-R, and sufS-htk-F/sufS-htk-R. To construct the trmI (TTHA0609) deletion strain (NS0802), we used a linear DNA fragment constructed by PCR ligation as above. For deletion of trmI, the 5Ј-and 3Ј-flanking regions of trmI were amplified, respectively. Pairs of oligonucleotide DNAs used are as follows: trmI-5Ј-F/trmI-5Ј-R, trmI-3Ј-F/ trmI-3Ј-R, and trmI-htk-F/trmI-htk-R.
Using the plasmids and fragments prepared as described above, T. thermophilus HB8 was transformed according to the literature (12) and then selected on a rich plate containing 300 g/ml kanamycin at 70°C. Site-specific homologous recombination of transformants was confirmed by PCR amplification, followed by restriction enzyme digestion or Southern hybridization.
Modified Base Analysis of Total tRNA by LC/MS-Cells were cultivated at 70°C up to late log phase. Total RNAs were extracted from cells using Isogen (Wako), and the tRNA fraction was further purified by 10% PAGE containing 7 M urea. Nucleoside analysis of total tRNA was performed by LC/MS as described previously (10). The s 2 T content was calculated from the area under the UV peak and normalized by the peak of pseudouridine () in each data set.
In Vitro Thiolation Assay-s 4 U-deficient tRNA fMet (Kenjo Miyauchi, University of Tokyo) was prepared from the E. coli JD20605 strain (Dr. Miki Takeyoshi, Fukuoka Dental College). tRNA Phe from Saccharomyces cerevisiae was purchased from Sigma. tRNA fMet with 1-methyladenosine (m 1 A) 58 was prepared by recombinant T. thermophilus TrmI protein, which was expressed in the E. coli strain Rosetta (DE3) (Novagen), with plasmid pML24, and purified as described previously (13). tRNA fMet from the E. coli thiI mutant (8 g) was methylated at 60°C for 1 h in a reaction mixture containing 50 mM Tris-HCl (pH 8.0), 10 mM MgCl 2 , 0.5 mM dithiothreitol, 50 M S-adenosylmethionine (Sigma), and 2 g of TrmI. The reaction was stopped by phenol extraction, and the tRNA was ethanol-precipitated.
Unmodified tRNA fMet (with a C1G mutation) and tRNA fMet (with C1G/U54A mutations) were prepared by in vitro transcription using T7 RNA polymerase (14). Template for in vitro transcription was constructed by PCR using synthetic oligonucleotide DNAs carrying the tRNA gene under the T7 promoter sequence (14). Oligonucleotide DNAs used for the construction of a plasmid-bearing gene for tRNA fMet are fMet-C1G-F (ggccGAATTC taatacgact cactataggc ggggtggagc agcctggtag ctcgtcgggc tcataacccg aagg) and fMet-C1G-R (cgcgcAAGCT Tggatggaag acctggttgc gggggccgga tttgaaccgg cgaccttcgg gttatgag). The PCR products were cloned into the EcoRI and HindIII sites of pUC19. For in vitro transcription, the pair of primers fMet-F (taatacgact cactataggc)/fMet-R (ttgcgggggc cggatttgaa c) or fMet-F/fMet-U54A-R (ttgcgggggc cggatttgaT ccggcgacct tc) was employed to PCR-amplify the template plasmid pUC19-tRNA fMet C1G, respectively. Transcripts of tRNA genes were prepared at 37°C for 3 h in a reaction mixture containing 40 mM HEPES-KOH (pH 7.8), 5 mM dithiothreitol, 1 mM spermidine, 8 mM MgCl 2 , 1 mM each NTPs, 5 mM GMP, 50 g/ml bovine serum albumin, 2 g/ml template DNA, and T7 RNA polymerase (14). Then the reaction for CCA addition was performed as described (15). tRNAs were dephosphorylated using bacterial alkaline phosphatase (Takara) and labeled at their 5Ј termini with [␥-32 P]ATP (110 TBq/ Cell extract used for the in vitro assay was prepared from early log phase cells that had been cultivated at either 80 or 50°C. Cells were resuspended with standard buffer containing 50 mM HEPES-KOH (pH 7.6), 100 mM KCl, 10 mM MgCl 2 , and 0.2 mM phenylmethylsulfonyl fluoride. Resuspended cells were sonicated at 4°C and then centrifuged at 20,000 ϫ g for 15 min at 4°C. The supernatant (S20) was used as "lysate." Additionally, we performed a 100,000 ϫ g centrifugation (S100) followed by gel filtration of the supernatant with a Nap5 column (Amersham Biosciences), to prepare the lysate for the m 1 A dependence assay (Fig. 5C).
RNA Sequencing with APM-RNA sequencing (17) was performed as described previously (6). RNA sequencing with APM was performed using standard methods, except for gels containing 100 M APM.
Temperature Sensitivity Assay-Cells were cultivated overnight at 70°C. Cultures diluted to 10 Ϫ2 , 10 Ϫ3 , or 10 Ϫ4 were spotted onto rich plates and incubated for 32 h at 70 or 80°C, respectively.
Transfer RNA Melting Analysis-Melting curves for tRNAs were measured in 50 mM sodium cacodylate buffer, pH 7.5, in the presence of 10 mM MgCl 2 and 200 mM NaCl. tRNA hyperchromicities at 260 nm were recorded using a spectrophotometer, RESPONSE II (Gilford).

Cysteine and Sulfate Ion Are the Sulfur Donors for s 2 T Biosynthesis in
Vivo-To identify the sulfur donor for s 2 T modification, we performed an in vivo labeling experiment. It is known that cysteine is a sulfur donor for tRNA modifications, such as 4-thiouridine (s 4 U), 5-methylaminomethyl-2-thiouridine (mnm 5 s 2 U), 6-isopentenyl-2-methylthioadenosine (ms 2 i 6 A), and 2-thiocytidine (s 2 C), in E. coli (20). Also in Salmonella typhimurium, cysteine is a sulfur donor for tRNA modi-fications as above, and instead of ms 2 i 6 A, tRNA contains 6-(4-hydroxyisopentenyl)-2-methylthioadenosine (ms 2 io 6 A) (21). In light of this knowledge, we tested the possibility that cysteine and other sulfur compounds are incorporated into the s 2 T of T. thermophilus tRNA.
T. thermophilus were cultivated for 1 h in minimal medium containing 35 S-labeled methionine, cysteine, or sodium sulfate. Total RNA was extracted from the cells and analyzed by denaturing PAGE. A [ 35 S]methionine-labeled band appeared in one tRNA species, and the band disappeared with mild alkaline treatment of the sample (Fig. 1A, lanes 1  and 2). This indicates that the band was probably derived from [ 35 S]methionylated tRNA Met . When grown in the presence of labeled cysteine or sulfate ions, bands corresponding to all of the tRNA species were uniformly labeled with 35 S and were unaffected by mild alkaline treatment (Fig. 1A, lanes 3-6). These bands are believed to be derived from tRNAs modified with 35 S.
T. thermophilus tRNAs possess s 4 U at position 8, one of the major sulfur-modified nucleosides together with s 2 T54 (4). Therefore, labeled RNA was digested into nucleosides and analyzed by an ODS reversedphase column, in order to confirm that the sulfur atom of cysteine and 5Ј-ggacctcgta ccgcccccaa gcgtaaccaa catgattaac-3Ј htk s 2 T Biosynthesis of Thermophile tRNA the sulfate ion were actually incorporated into s 2 T. Two major 35 Slabeled peaks were detected in the RNA hydrolysate (Fig. 1B). These were identified as s 4 U and s 2 T by comparing their elution times with those of authentic nucleosides. The chromatogram pattern was essentially the same for both [ 35 S]sulfate and [ 35 S]cysteine labeling (data not shown). These results clearly demonstrate that the sulfur atoms from cysteine and the sulfate ion were incorporated into the s 2 T of tRNA in vivo. iscS and sufS Mutations Did Not Affect s 2 T Biosynthesis in T. thermophilus-IscS (iron sulfur cluster) is the cysteine desulfurase that is responsible for all of the thio-modifications in E. coli (20) and S. typhimurium (21). This enzyme catalyzes the initial sulfur transfer reaction from cysteine. It is believed that, in T. thermophilus (see above), the sulfur atom from cysteine is also incorporated into s 2 T, so that an IscS homologue in T. thermophilus may be a candidate enzyme for s 2 T biosynthesis. A genomic data base search of T. thermophilus HB8 revealed that there are two putative cysteine desulfurase-encoding genes. Alignment of the deduced sequence of these enzymes with the sequence of E. coli IscS and SufS (named after the mobilization of sulfur), strongly suggests that the two cysteine desulfurase-like proteins of T. thermophilus are IscS and SufS homologues (data not shown). The difference between the two enzymes is indicated by the motif around a conserved catalytic cysteine; a consensus sequence (SSGSACTS) around Cys 328 of E. coli IscS can effectively distinguish this enzyme from SufS, which has a consensus sequence of RXGHHCA (22). Both IscS and SufS are known to be involved in Fe-S cluster assembly, together with other components of the ISC and SUF machinery (23,24). However, only IscS (not SufS) is involved in tRNA thiolation in E. coli (20).
We constructed two mutant strains of iscS or sufS by homologous recombination and confirmed the genomic organization around the target genes by Southern hybridization or PCR (data not shown). Formation of s 2 T in tRNA was not affected in these mutant strains, as determined by nucleoside analyses of total tRNA with reversed-phase HPLC, coupled with a mass spectrometer. s 2 T was detected as the protonated form (m/z ϭ 275) and protonated free-base form (m/z ϭ 143). The s 2 T contents of iscS and sufS mutants were 1.2 and 0.9, when compared with that of the wild type (set at 1), respectively. These results suggest redundant roles for IscS and SufS in the synthesis of s 2 T. However, it is possible that another cysteine-utilizing enzyme, distinct from IscS and SufS, is present in T. thermophilus.
Detection of 2-Thiolation Activity in a T. thermophilus Cell Extract-To characterize the 2-thiouridylation reaction in detail, we tried to detect its activity in vitro. We used the affinity electrophoresis method (APM gel electrophoresis) developed originally by Igloi (16), since we have succeeded previously in the detection of the 2-thiolation of U54 using this method (6). First, we used tRNA fMet from the E. coli thiI mutant strain as the substrate, since the nucleotide sequence of the tRNA is almost identical to that of tRNA fMet from T. thermophilus. At the unmodified nucleoside level, the differences are that the U50 of E. coli tRNA is C50 in T. thermophilus (25) (Fig. 2A (a and b)) and that the tRNA from the thiI mutant has no s 4 U (26). s 4 U is known to be retarded greatly in APM gels (16).
In addition, we constructed an s 4 U-deficient strain of T. thermophilus (thiI mutant). The gene thiI encodes a thiotransferase that is required for s 4 U8 synthesis in E. coli (26). We constructed a thiI mutant of T. thermophilus (NS0801), and tRNA recovered from the mutant cells was analyzed by LC/MS. The tRNA modification of NS0801 was the same as wild type, with the exception of the complete loss of s 4 U (data not shown). This indicates that thiI of T. thermophilus encodes a genuine thiotransferase for s 4 U8 synthesis.
Using the s 4 U-deficient, native tRNA fMet from E. coli as the substrate, ATP-dependent 2-thiouridylation was detected, suggesting the formation of s 2 T (Fig. 2B (a)). When the wild-type lysate was used, migrating bands corresponding to both 4-thiouridylated tRNA (greatly retarded) and 2-thiouridylated tRNA (slightly retarded) were detected (lane 3), but only a single retarded band, corresponding to 2-thiouridylated tRNA, was detected when lysate from the mutant NS0801 was used (lane 5).
To confirm that the 2-thiouridylation occurred at position 54 in the tRNA, we performed the same experiment using tRNA transcripts ( Fig.  2A (c)), in which C1 was substituted by G1, to obtain more efficient transcription in vitro (14). The tRNAs recovered from the reaction mixture were analyzed using APM gel electrophoresis, by which 2-and 4-thiouridylations of tRNA fMet transcript of the U54A mutant were examined as compared with those of the wild type ( Fig. 2B (b)). The wild-type tRNA transcript was thiolated at a slightly lower level than the native tRNA fMet (compare lane 5 in Fig. 2B, a and b), and this may be caused by the presence of other modified nucleosides in the native tRNA. The U54A mutant was effectively 4-thiouridylated ( Fig. 2B (b), lane 8) but not 2-thiouridylated ( Fig. 2B (b), lane 10), suggesting that 2-thiouridylation occurred at position 54.
In order to analyze the modified nucleosides, we performed RNA sequencing of the tRNA fMet from the E. coli thiI mutant, reacted in vitro with the lysate from NS0801 (thiI::km). First, we separated the reacted tRNA into the upper (retarded) and lower (not retarded) bands using a preparative scale APM gel electrophoresis, like the electrophoretic pattern shown in lane 5 of Fig. 2B (a). The bands were then excised, recovered from the gel, and subjected to RNA sequencing. The ladder pat- terns of the lower and upper band samples were almost identical, except for the A in position 58 (see below) of a normal sequencing gel (Fig. 3A). When analyzed using the APM-containing sequencing gel (Fig. 3B), the ladder pattern of the lower band sample was essentially the same as that found using a normal gel (Fig. 3A). Although the ladders below G53 of the upper band sample electrophoresed in a similar manner for both the normal and APM gels, the ladders above position 54 disappeared through upper shifting in the APM gel (shown by the arrow in Fig. 3B), suggesting the presence of a thiocarbonyl nucleoside at position 54, probably attaching at position 2 of the uridine base.
At position 58 (Fig. 3A, asterisk), the tRNA of the upper band was not digested by RNase U 2 (compare this with the position of the tRNA in the lower band), suggesting that the upper band possesses m 1 A58. Thus, the tRNA of the upper band possesses both s 2 T54 and m 1 A58, but the tRNA of the lower band possesses rT54 and unmodified A58. During the reaction, the methyl moiety is thought to be incorporated into A58 by a specific methyltransferase included in the lysate. This result suggests that methylation of A58 is required for s 2 T biosynthesis (see below).
Temperature Dependence of the 2-Thiouridylation Reaction-The s 2 T content of thermophile tRNA is known to vary depending on the cultivation temperature (4,5), but the mechanism underlying this regulation is still unknown. In particular, we investigated the question of whether or not either the abundance or activity of the 2-thiolation enzyme(s) is temperature-regulated. We analyzed the temperature  2 and 3), and lysate from NS0801 (thiI::km) (lanes 4 and 5). ATP was included in lanes 1, 3, and 5. b, transcripts tRNA fMet (with C1G mutation) and tRNA fMet (with C1G/U54A mutations) were separated in lanes 1-5 and 6 -10, respectively. Samples were as follows: no lysate (lanes 1 and 6), lysate from HB8 (lanes 2, 3, 7, and 8), and lysate from NS0801 (thiI::km) (lanes 4, 5, 9, and 10). ATP was included in lanes 1, 3, 5, 6, 8, and 10. The positions of 4-thiouridylated, 2-thiouridylated, and substrate tRNAs are indicated on the right. dependence of 2-thiouridylation activity in vitro. Since s 4 U8 content is known to be both constant and independent of cultivation temperature (4), we analyzed 4-thiouridylation activity as a control. To estimate the 2-thiolation of rT54, we used lysate from the mutant NS0801 (thiI::km) and tRNA Phe from S. cerevisiae, which contains m 1 A58 (25). Thus, any effects caused by methylation of A58 could be disregarded. We used lysate from HB8 and a transcript of tRNA fMet possessing C1G and U54A mutations for estimation of 4-thiolation of U8.
Using these reactions, the thiolation activity for tRNA was measured by APM gel electrophoresis (Fig. 4A (a and b)), and the band intensities were quantified (Fig. 4A (c)). The substrate tRNAs were greatly degraded above 70°C, and this precluded the measurement of thiolation activity above that temperature. The 2-thiolation activity of rT54 was very low at 50°C but increased with temperature, rising 7-fold at 70°C. However, the 4-thiolation activity increased by only 2-fold throughout the same temperature range.
We measured the enzymatic activities of the lysate obtained from the cells cultured at 50 and 80°C (Fig. 4B). The 4-thiolation activity of the lysate from the cells grown at 80°C was almost the same as that of the 50°C lysate, whereas the 2-thiolation activity of the 80°C lysate was 13-fold higher than that of the 50°C lysate. These results suggest that expression of 2-thiolation enzyme activity increases at higher temperatures. Thus, the 2-thiolation enzyme increases in both activity and abundance at higher temperatures, and these elevated levels could explain why the s 2 T content of tRNA increases with the cultivation temperature.
Effect of Methylation of A58 for s 2 T Biosynthesis-Analysis of the products of the 2-thiolation reaction in vitro suggested that methylation of A58 is required for s 2 T biosynthesis (see above). To clarify the role of A58 methylation in 2-thiolation of rT54, we analyzed an m 1 A-deficient strain of T. thermophilus HB8. Since trmI is known to encode the m 1 A58 methyltransferase in T. thermophilus HB27 (13), we constructed a trmI knock-out strain of T. thermophilus HB8 and analyzed the modification of the tRNA (Fig. 5A). In this mutant, m 1 A58 was completely lost, and the s 2 T content decreased to 15% of that of the wild-type strain, as determined using a UV chromatogram (Fig. 5A). The mass spectrum of these modified nucleosides was also confirmed (data not shown). Since rT was eluted at the same retention time, with abundant unmodified G, it could not be detected by the UV chromatogram. In the trmI mutant, rT was detected with the m/z of 259 and 127, corresponding to the protonated molecule and protonated free base of rT, respectively, at the same retention time as G (data not shown). 2-thiouridine, which is known to be eluted near G, was not detected (data not shown). Thus, we conclude that in the trmI mutant, the thiolation reaction of rT54 to s 2 T54 is severely inhibited.

s 2 T Biosynthesis of Thermophile tRNA
We then analyzed the thiolation level of a single tRNA species using the APM-PAGE method. tRNA Ile was 3Ј-labeled using [␣-32 P]dCTP, and in order to exclude the effect of s 4 U8, the anticodon region was hydrolyzed by limited digestion with RNase T 1 . Intact tRNA Ile contains both s 4 U8 and s 2 T54 (Fig. 5B, lanes 5 and 6). In the trmI strain, the s 2 T54 content in the 3Ј-region of the tRNA Ile was about 20% of the wild type (Fig. 5B, lanes 7 and 8).
We further examined the effect of m 1 A58 on 2-thiolation of tRNA in vitro. As substrate, we used the tRNA fMet from the E. coli thiI mutant, with or without modification of m 1 A58. This was incorporated into tRNA by recombinant TrmI protein, and the modification was confirmed by RNA sequencing (data not shown). The lysate was prepared by centrifugation (100,000 ϫ g) of the cell extract from the NS0801 (thiI::km), followed by gel filtration of the supernatant, in order to minimize methylation during incubation. In the in vitro assay system, tRNA with the m 1 A58 modification was more effectively 2-thiolated than that without (Fig. 5C). These results suggest that 1-methylation of A58 is required for effective biosynthesis of s 2 T. At 5 min of incubation, the difference in 2-thiolation between the modified and unmodified substrate tRNAs was about 2-fold, but after the longer incubation (Ͼ10 min), the extent of 2-thiolation of the m 1 A-deficient substrate tRNA increased considerably when compared with the methylated tRNA. This is caused by the methylation of the m 1 A-deficient tRNA by the lysate during the longer incubation period (confirmed by RNA sequencing; data not shown).
Finally, we examined the phenotype of the trmI mutant (NS0802), which has a particularly low s 2 T54 content (about 15% of wild type; see Fig. 5A). This mutant exhibits a temperature-sensitive phenotype. Although the mutant and wild type demonstrate similar growth at 70°C, the mutant can no longer grow at 80°C, a temperature at which the wild type can still grow (Fig. 6A). Thus, this phenotype resembles the trmI strain of HB27 (13). Since the temperature sensitivity is thought to be caused by lowered thermostability of the tRNA, we compared the melting temperatures of total tRNA between the wild type (85.6°C) and the trmI mutant (NS0802) (83.7°C) and found a 2°C difference (Fig.  6B). A58 methylation does not contribute to T m (27); thus, it is evident that the lowered T m of the total tRNA from the trmI mutant is the result of the decrease of the 2-thiouridylation (i.e. s 2 T is only 15% of the wildtype tRNA level). These results imply that s 2 T modification is responsible for both the thermal stability of the tRNA and the adaptation of T. thermophilus cells to higher temperatures. However, characterization of a mutant deficient in s 2 T only is necessary for a better understanding of the precise roles of s 2 T and m 1 A.

DISCUSSION
To date, there has been little reported on thermophile-specific tRNA s 2 T54 biosynthesis, a process that confers thermal stability on tRNAs. In this paper, we characterized some aspects of the modification of rT to s 2 T, including the relevant enzymes and sulfur donors.
Considering the biosynthetic pathways for sulfur-containing modified nucleosides that have been recently characterized, it appears that activation of both the tRNA and the sulfur atom is necessary for the enzymatic reaction converting rT to s 2 T. It has been proposed that there are two distinct routes for the biosynthesis of thiolated nucleosides (e.g. s 4 U, mnm 5 s 2 U, ms 2 i 6 A, ms 2 io 6 A, and s 2 C) in tRNA (20,21,28,29). First, the sulfur atom is transferred to a cysteine residue at the active site of IscS cysteine desulfurase (30,31). In one pathway, sulfur is directly transferred from IscS to the tRNA-modifying enzyme in the persulfide form and then is finally transferred to the tRNA (32). In the other pathway, the sulfur on IscS is incorporated into the Fe-S clusters by the ISC machinery (IscSUA-HscBA-Fdx in E. coli) and is transferred to other protein(s) that have iron-sulfur clusters (e.g. the enzyme responsible for ms 2 i 6 A synthesis, MiaB, is known as an iron-sulfur protein (33)(34)(35)).
In this work, we first identified cysteine as one of the sulfur donors for biosynthesis of s 2 T. However, neither of the mutations iscS or sufS (a iscS paralogue) affected the s 2 T content in T. thermophilus. In E. coli and   JANUARY 27, 2006 • VOLUME 281 • NUMBER 4 S. typhimurium, the sulfur atom for all thio-modifications of tRNAs is derived from cysteine, using the sulfur delivery protein IscS (20). Other cysteine desulfurases, SufS and CsdA (cysteine sulfinite desulfinase), are not involved in thio-modification (20). Strikingly, mutation of iscS did not alter s 2 T biosynthesis, giving rise to speculation that there may be a supplementary role for SufS in s 2 T synthesis in T. thermophilus, which could be investigated using a double mutant strain of iscS/sufS. However, in E. coli, the double deletion is lethal (24), and we suspect that a similar effect would occur in T. thermophilus, due to the lack of Fe-S cluster formation. An as yet unknown pathway that uses the sulfate ion may also exist in T. thermophilus, since the in vivo labeling experiment demonstrated that the sulfur atom from the sulfate ion is actually incorporated into the modified nucleosides of tRNA.
By constructing the in vitro assay system, we have successfully detected and characterized s 2 T synthesis in vitro. The ATP requirement of the s 2 T reaction was clearly observed in this system, and this is consistent with the fact that some tRNA modification enzymes are derived from P-loop ATPases (e.g. ThiI (s 4 U8 synthesis (32,36)), MnmA (mnm 5 s 2 U34 (31)), TilS (lysidine 34 (37)), TtcA (s 2 C32 (38)), and YbbB (5-methylaminomethyl-2-selenouridine 34 (39)). It has been demonstrated that some of these enzymes bind the substrate tRNA (31,32,37,39) and activate the target nucleoside with ATP, forming an adenylate intermediate (40). Thus, a tRNA binding ATPase might also be involved in s 2 T54 synthesis, and/or ATPase(s) may be involved also in sulfuractivating steps.
Additionally, as seen with s 4 U8 synthetase ThiI, which has a rhodanese domain that serves as a sulfurtransferase (41,42), a thiotransferase protein (or domain), together with the ATPase domain described above, may also be required for s 2 T54 synthesis.
In our in vitro assay, cysteine dependence was neither observed with the S20 extract nor with the gel-filtrated S100 fraction (data not shown). Thus, it is likely that the sulfur transfer reaction we observed came from an activated sulfur intermediate (possibly a persulfide or an Fe-S cluster) transferred by the modifying enzyme to the substrate tRNA.
Very little is known about the mechanisms controlling the amount of modified nucleosides, with the exception of the increase of methylation of G18 to 2Ј-O-methylguanosine 18 that has been observed in the tRNA of T. thermophilus HB27 in a high temperature environment. This increase may not necessarily be due to an increase in the abundance of enzyme but to an increase in methyltransferase activity (43).
The s 2 T content of tRNA in T. thermophilus HB8 increases with elevation of cultivation temperature in vivo (4). This phenomenon is an interesting strategy for bacterial adaptation to high temperature environments. Our in vitro results indicated that the molecular mechanism for this adaptation may be an increase in both the expression and activation of 2-thiolating enzyme(s) at higher temperatures. Since there may be various enzymes involved in this modification process (see above), these characteristics may be derived from properties of individual enzymes involved in this pathway. Identification of all of these enzymes and investigation of these properties will clarify the precise mechanism of the temperature dependence of s 2 T biosynthesis.
Differences between the s 2 T-thiolase from HB8 (this paper) and the Gm-methylase of HB27 (43) may indicate alternative mechanisms required for the control of the abundance of each post-transcriptional modification. m 1 A58 is required for effective 2-thiolation of rT54 both in vitro and in vivo (Figs. 3, 5, and 6), and this suggests that methylation of A58 is responsible for the recognition of the substrate tRNA in s 2 T biosynthesis. In a recent paper (6), we identified the residues required for effective 2-thiolation of U54 and demonstrated that A58 shows a strict require-ment for 2-thiolation. It is possible that 2-thiolation enzyme(s) recognize the positive charge resulting from methylation at the 1 position of A58. It is also possible that the TrmI protein may enable the 2-thiolase to recognize substrate tRNAs and that it is the TrmI protein itself, and not the methyl moiety of m 1 A58, that is required for effective 2-thiolation of rT54. However, the addition of TrmI to the reaction mixture did not enhance the levels of 2-thiolation (data not shown).
During our investigation of the 2-thiolation of tRNA, characterization of the trmI mutant strain of T. thermophilus HB27 was published (13), and the temperature-sensitive phenotype of that strain was the same as for our HB8 mutant (NS0802) (Fig. 6). Droogmans' group (13) proposes that m 1 A58 is responsible for the thermal adaptation of T. thermophilus. However, aside from an m 1 A deficiency, the trmI mutant has quite a low level of s 2 T. We propose that the major reason for a lower tRNA T m and for the temperature-sensitive phenotype is the reduction in the abundance of s 2 T. Although the observed difference in the T m of the tRNA between the wild-type and m 1 A/s 2 T-deficient strains is not great (about 2°C; Fig. 6B), it may contribute significantly to the ability to grow in thermal environments over 80°C. Characterization of a mutant that is deficient in s 2 T only is necessary for clarification of the relative function of s 2 T and m 1 A residues in the thermal stabilization of tRNA.
Our future studies will involve the purification of the s 2 T synthesis enzymes using the in vitro assay system in order to develop a clearer understanding of the mechanisms controlling the relative contents of rT and s 2 T and their roles in temperature adaptation.