Identification of Two tRNA Thiolation Genes Required for Cell Growth at Extremely High Temperatures*

Thermostability of tRNA in thermophilic bacteria is effected by post-transcriptional modifications, such as 2-thioribothymidine (s2T) at position 54. Using a proteomics approach, we identified two genes (ttuA and ttuB; tRNA-two-thiouridine) that are essential for the synthesis of s2T in Thermus thermophilus. Mutation of either gene completely abolishes thio-modification of s2T, and these mutants exhibit a temperature-sensitive phenotype. These results suggest that bacterial growth at higher temperatures is achieved through the thermal stabilization of tRNA by a 2-thiolation modification. TtuA (TTC0106) is possibly an ATPase possessing a P-loop motif. TtuB (TTC0105) is a putative thio-carrier protein that exhibits significant sequence homology with ThiS of the thiamine synthesis pathway. Both TtuA and TtuB are required for in vitro s2T formation in the presence of cysteine and ATP. The addition of cysteine desulfurases such as IscS (TTC0087) or SufS (TTC1373) enhances the sulfur transfer reaction in vitro.

The cellular components of thermophilic organisms have evolved to function efficiently in high temperature environments. tRNA is also thermostabilized through post-transcriptional modifications, such as 2-thioribothymidine (s 2 T) 2 at position 54. s 2 T is a 2-thiolated derivative of 5-methyluridine (ribothymidine (rT)), which is located at position 54 in the T loop of almost all tRNAs in the eubacterium Thermus thermophilus (1) and the archaea Pyrococcus furiosus (2). The 2-thiolation content of rT54 increases with cultivation temperature; in T. thermophilus cells grown at Ն80°C, more than half of the tRNAs contain s 2 T54, whereas at 50°C only a small proportion were thiolated at this position. Additionally, the melting temperature of tRNA increases concomitantly with s 2 T incorporation (1). In an in vitro translation assay, 2-thiolated tRNA functions more efficiently at higher temperatures, whereas unthiolated tRNA functions optimally at lower temperatures (3). These results suggest that the proportion of tRNA containing s 2 T54 is a key factor in the adaptation of the thermophilic translation system to varying environmental temperatures.
NMR analysis has elucidated the mechanism by which 2-thiolation of rT54 effects the thermal stabilization of tRNA (4). s 2 T adopts the C3Јendo-gg-anti conformation, as do all residues in A-form RNA, due to steric effects between the bulky 2-thiocarbonyl group and the 2Ј-hydroxyl group. The inherent rigidity of s 2 T54 stabilizes the interaction between the D and T loops of tRNA, resulting in thermostability of the tertiary structure (3).
Although the function of s 2 T has been studied extensively, information on the biosynthesis of the 2-thiocarbonyl group of s 2 T is limited. In previous work, we undertook a preliminary characterization of the 2-thiolation reaction in a cell-free extract of T. thermophilus and found that a sulfur atom from cysteine or sulfate was incorporated into s 2 T (5). The aims of this study were to isolate novel tRNA-binding proteins involved in the s 2 T modification of tRNA in T. thermophilus, to confirm their activity through mutation and in vitro analyses, and to determine whether or not the s 2 T modification of tRNA is required for growth at elevated temperatures.

EXPERIMENTAL PROCEDURES
Strains-The strains used for this study were wild-type T. thermophilus HB27 and T. thermophilus NS0801, which lacks the gene for 4-thiouridine (s 4 U) biosynthesis (thiI) (5). Both wild-type and mutant strains were cultivated in rich medium (6) at 70°C without or with 30 g/ml kanamycin, respectively, unless otherwise stated. Escherichia coli JM109 (7) and TOP10 (Invitrogen) were used as hosts for the genetic manipulation of plasmids.
In order to prepare tRNA-bound resin, the 3Ј-ribose of the tRNA was oxidized to form a dialdehyde at room temperature, for 1 h in the dark, in a buffer containing 100 mM sodium acetate (pH 5.2), 10 mM sodium periodate, and 10 mM MgCl 2 . The oxidized tRNA was purified using a NAP 5 gel filtration column (Amersham Biosciences) and recovered by ethanol precipitation. Oxidized tRNA was biotinylated using Biotin (Long Arm) hydrazide (Vector), incubated in 100 mM sodium acetate (pH 5.2) and 10 mM MgCl 2 , at room temperature overnight in the dark. Unreacted biotin was removed by gel filtration using a NAP 5 column, followed by ethanol precipitation of the biotinylated tRNA. The precip-itate was resuspended in H buffer (50 mM HEPES-KOH buffer (pH 7.6), 100 mM KCl, 10 mM MgCl 2 , 1 mM dithiothreitol, and 0.2 mM phenylmethylsulfonyl fluoride) and mixed with Streptavidin-Sepharose HP (Amersham Biosciences) for 1 h at room temperature. The excess tRNA (unbound fraction) was removed from the resin by washing three times with H buffer. About 35 g of tRNA was immobilized on 10 l of Streptavidin-Sepharose HP.
Purification and Identification of tRNA-binding Proteins-A cell extract was prepared from an early log phase T. thermophilus HB27 culture. Cells were resuspended in H buffer with 10 mM KCl and then sonicated at 4°C and centrifuged at 7,000 ϫ g for 15 min. The supernatant was collected and centrifuged at 100,000 ϫ g for 2 h, after which the supernatant was again collected and subjected to desalting using a Nap 25 gel filtration column (Amersham Biosciences). About 30 mg of gelfiltrated S100 was obtained from a 1-liter culture of T. thermophilus HB27 (ϳ1.4 g of cells).
In order to remove nonspecific binding proteins, the desalted S100 fraction (2.5 mg) was mixed with Streptavidin-Sepharose HP and incubated for 10 min at room temperature, after which the resin was removed. The resultant fraction was then applied to tRNA-Sepharose (ϳ50 g of bound tRNA Ile ) and incubated at room temperature for 20 min, followed by washing three times with H buffer.
The resin-bound protein was mixed with an equal volume of sample buffer (125 mM Tris-HCl (pH 6.8), 2% SDS, 0.7 M 2-mercapthoethanol, and 0.02% bromphenol blue) and incubated at 95°C for 5 min, prior to separation using SDS-PAGE. Proteins were visualized using Coomassie Brilliant Blue R-250, and bands of interest were excised and subjected to in-gel tryptic digestion as described previously (9). Tryptic digests were analyzed using nanoflow-HPLC-electrospray ionization tandem mass spectrometry. This consisted of a DiNa splitless nanoflow HPLC system (KYA Technologies) and a 50 ϫ 0.15-mm inner diameter packed ODS (3-m particle size) capillary column (KYA Technologies), which are used for efficient separation of small amounts of peptide. Tryptic digest fragments were separated in 0.1% formic acid in water, using a linear gradient from 0 to 100% of 70% acetonitrile for 40 min at a flow rate of 500 nl/min. Ionization of the eluted peptides was performed using LCQ ion trap mass spectrometry (Thermo Electron), through electrospray ionization (metal nanosprayer S; GL Sciences). Proteins were identified using the MASCOT data base search engine (Matrix Science) and compared with the T. thermophilus HB27 genome data base (NC_005835 and NC_005838 on NCBI).
Using the plasmids prepared as described above, T. thermophilus HB27 was transformed as previously described (11), and transformants were selected on rich medium containing 300 g/ml kanamycin. Homologous recombination was confirmed using PCR amplification, followed by restriction digestion analysis.
Analysis of tRNA Modification-Cells were grown at 70°C to late log phase, and total RNA was extracted using Isogen (Wako). The tRNA fraction was further purified using 10% PAGE containing 7 M urea. Total tRNA was digested by nuclease P 1 (Yamasa) and bacterial alkaline phosphatase (Takara) in 20 mM HEPES-KOH (pH 7.5), at 37°C for 10 h. The hydrolysate of RNA was analyzed using liquid chromatography/mass spectroscopy (LC/MS), as described previously (12).
Temperature-dependent Growth Phenotypes of ttuA and ttuB Strains-Cells were cultivated in the rich medium (30 g/ml kanamycin was added to the mutant strains) at 70°C overnight. Diluted culture (A 600 ϭ 0.1) and serial dilutions (10 Ϫ1 , 10 Ϫ2 , 10 Ϫ3 , and 10 Ϫ4 ) were spotted onto rich medium plates and incubated for 31 h at 60°C, 17 h at 70°C, 15 h at 75°C, 24 h at 80°C, and 32 h at 82°C.
E. coli Rosetta (DE3) (Novagene) was transformed with the expression plasmids described above. Cultures were grown to A 600 ϭ 0.6, induced with isopropyl 1-thio-␤-D-galactopyranoside (0.5 mM for TtuA, 1 mM for TtuB, 0.01 mM for co-expression of TtuA/B, and 0.1 mM for IscS and SufS), and expressed for a further 4 h at 37°C. Cells expressing recombinant His-tagged TtuB, IscS, and SufS were harvested, suspended in K100 buffer (50 mM HEPES-KOH buffer (pH 7.6), 100 mM KCl, 10 mM MgCl 2 , 5% glycerol, and 7 mM 2-mercaptoethanol) with 0.2 mM phenylmethylsulfonyl fluoride, and gently sonicated. Lysates were separated using centrifugation at 6,000 ϫ g, and the supernatants were collected and heat-treated at 70°C for 15 min. A second centrifugation at 6,000 ϫ g was performed, and the supernatant was applied to a Ni 2ϩnitrilotriacetic acid-agarose (Qiagen) column, washed with K100 buffer containing 1 M NH 4 Cl and 10 mM imidazole, and eluted with K100 buffer with 250 mM imidazole. Purified proteins were desalted using a NAP 25 gel filtration column and K100 buffer. Protein-containing fractions were concentrated with Centricon columns (YM3 for TtuB and YM30 for IscS and SufS; Millipore Corp.). The His-tagged fusion pro-tein TtuA and the co-expressed TtuA/TtuB were purified as above, except for the addition of 500 mM KCl to all buffers and 200 mM imidazole to the buffer for gel filtration of TtuA in order to prevent precipitation. Centricon columns YM10 and YM3 were used for concentration of TtuA and TtuA/TtuB, respectively. Protein concentrations were determined using a Bio-Rad protein assay kit with a bovine serum albumin standard. Glycerol was added to the purified protein solutions to a final concentration of 30%, and each sample was stored at Ϫ30°C.
The 2-thiolation reaction with [ 35 S]cysteine was performed as described above with the following modifications. Four g of S. cerevisiae tRNA Phe and 20 M [ 35 S]cysteine (20 Ci) were added to 20 l of reaction mixture. [ 35 S]cysteine was purchased from American Radiolabeled Chemicals. RNAs were recovered using Isogen reagent and ethanol-precipitated. For alkaline treatment, the RNA sample was incubated at 37°C for 1 h in 100 mM HEPES-KOH (pH 9.0). The sample was then subjected to 10% PAGE using gels containing 7 M urea, and the gels were then stained with EtBr. The gel was dried, exposed to an imaging plate, and analyzed using a BAS 1000 bioimaging analyzer (Fuji Photo Systems). For nucleoside analysis, labeled tRNA Phe was purified by PAGE, and 35 S-labeled nucleosides were analyzed by an HPLC system as essentially described in the literature (5). Fractions, which were collected every 37.8 s, were measured for absorbance at 280 nm and radioactivity.

Purification of tRNA-binding Proteins from T. thermophilus Cell
Extract-In order to identify the gene encoding the protein responsible for tRNA thiolation, we isolated the tRNA-binding proteins from a cell extract of T. thermophilus and identified the corresponding genes using proteomics. Transcript tRNA Ile was biotinylated with ϳ90% efficiency (data not shown) and immobilized on a streptavidin-Sepharose resin. The immobilized tRNA Ile was used for the affinity purification of tRNAbinding proteins from a cell extract of T. thermophilus (Fig. 1). Samples were separated using a gradient SDS-PAGE. When compared, the sample that was affinity-purified with tRNA-Sepharose (lane 3) exhibited a different pattern from that of the cell extract prior to purification (lane 1), and little nonspecific binding to the resin alone was observed (lane 2). This suggests that tRNA-binding proteins were purified successfully.
The major protein bands in lane 3 were excised and analyzed using mass spectrometry. A total of 108 proteins were identified from 2,204 ORFs of T. thermophilus HB27. Of these, ϳ40% were predicted to interact with RNA (tRNA, rRNA, and mRNA); the remainder comprised DNA-binding proteins, biotin-binding proteins, metabolic pathway proteins, and proteins of unknown function.
Among the RNA-binding proteins identified, we found 14 tRNAbinding proteins; tRNA nucleotidyltransferase (CCA-adding enzyme), elongation factor Tu, Ile-tRNA-synthetase, peptidyl-tRNA hydrolase, and 10 tRNA-modifying enzymes (Fig. 1B). We used the tRNA Ile species for the purification experiment, since native tRNA Ile has eight posttranscriptional modifications (15) (22)). Additionally, enzymes that are not involved in modification of tRNA Ile were identified, such as TruD for ⌿13 (23), MiaA for 6-isopentenyladenosine at position 37 (i 6 A37) (24), and TruA for ⌿38 -40 (25). Considering these results, enrichment of modification enzymes from the cell extract turned out to have been very efficiently achieved by our strategy.
Identification of the Genes Essential for the 2-Thiouridylation Reaction-Among the proteins in the affinity-purified sample, we identified one belonging to the TtcA family (TTC0106) (Figs. 1A, 2A, and 3A). This pro- tein family includes TtcA, an enzyme that is responsible for the biosynthesis of 2-thiocytidine (s 2 C) at position 32 of tRNA (26). The amino acid sequence of TTC0106 indicates that it is more closely related to the Group II TtcA family of proteins. These proteins possess a P-loop motif (Ser-Gly-Gly-Xaa-Asp-(Ser/Thr)) and four or five Cys-Xaa-Xaa-Cys (CXXC) motifs. In contrast, TtcA from Escherichia coli is included in Group I, which possess a P-loop motif and two CXXC motifs. The P-loop motif specifically binds ATP to form an adenylated intermediate; other tRNA-modifying enzymes, such as MnmA, ThiI, and TilS, also possess this motif (27)(28)(29). By the analysis of tRNA modification using liquid chromatography/mass spectroscopy (LC/MS), we were unable to detect s 2 C in the tRNA mixture of T. thermophilus under similar conditions to those required for detection of s 2 C from E. coli (data not shown). Thus, we believe that TTC0106 is not involved in s 2 C biosynthesis but is a candidate for s 2 T biosynthesis.
TTC0105 was also affinity-purified using tRNA-Sepharose (Figs. 1A and 2B). It is a paralog of ThiS, a sulfur carrier protein essential for biosynthesis of the sulfur-containing cofactor, thiamine (30,31).
Multiple alignment of TTC0105 and related sequences indicates that these proteins possess a conserved C-terminal Gly-Gly motif (Fig.  3B). The C terminus of ThiS is thiocarboxylated with a sulfur atom from cysteine, and then the sulfur atom is incorporated into the thiamine precursor (31,32). TTC0316 appears to be an ortholog of ThiS, and not only does the sequence closely resemble ThiS from E. coli (Fig. 3B), but in the T. thermophilus genome, it is located in a thiamine biosynthesis operon (ThiESGOC, unknown ORF, and ThiD; TTC0315-0321).
The genes encoding TTC0105 and TTC0106 form a single operon (Fig. 4A) and are probably transcribed from a transcriptional start site immediately upstream of the TTC0105 coding region. To examine whether or not TTC0105 and TTC0106 are involved in s 2 T biosynthesis, we constructed insertion mutants of ttc0106 or ttc0105 and analyzed the modified nucleosides of unfractionated tRNA using LC/MS. In order to avoid polar effects in the ttc0105 insertion strain, ttc0106 was constructed so as to be transcribed together with a drug resistance gene. In both mutants, s 2 T (27.7 min) was completely absent, although its precursor rT (21.4 min) was detected (Fig. 4B); other nucleoside modifications remained unchanged. These results suggest that both TTC0105 and TTC0106 are involved in s 2 T biosynthesis. A small peak (27.5 min) from dinucleotide GmpG was observed immediately before the s 2 T peak and was derived from incomplete digestion of tRNA. This assignment was confirmed by the mass spectrum (m/z ϭ 643) and by the disappearance of the peak in trmH, a Gm18-methylase mutant (data not shown). Thus, we have renamed the proteins TTC0106 and TTC0105 to tRNA two-thiouridine-synthesizing protein A and B (TtuA and TtuB), respectively, and the corresponding genes are now desig-nated ttuA and ttuB. The precursor of s 2 T was identified as rT, suggesting that the 5-methyltransferase that modifies U at position 54 (TrmFO) (20) does not recognize the 2-thiocarbonyl group of s 2 U.
Temperature-sensitive Phenotypes of s 2 T-deficient Strains-Although the 2-thiocarbonyl group of s 2 T54 is known to be responsible for the thermostabilization of tRNA (1, 3) and tRNA thus modified functions effectively at elevated cultivation temperatures (3), a direct requirement for the thermal stabilization of tRNA in high temperature environments has not been demonstrated.
The growth phenotypes of two s 2 T-deficient strains, NS2710 (ttuA::km) and NS2720 (ttuB::km), were examined. These insertion mutants were cultivated at various temperatures on rich medium plates. Although neither ttu mutant exhibited any obvious growth defect when cultivated below 75°C, both mutants were unable to grow above 80°C (Fig. 5). This is the first direct evidence that thermal stabilization of tRNA, accomplished by the s 2 T modification, is required for cell growth at high temperatures.  MAY 19, 2006 • VOLUME 281 • NUMBER 20

Identification of Two s 2 T Biosynthesis Genes
In Vitro Formation of s 2 T Using Recombinant TtuA and TtuB-We cloned the genes encoding TtuA and TtuB into His-tagged expression vectors and overexpressed the proteins in E. coli. When TtuA and N-terminal hexahistidine (His 6 )-tagged TtuB were co-expressed in E. coli, the untagged TtuA was co-purified with the His 6 -tagged TtuB during Ni 2ϩ affinity column chromatography (Fig. 6A, lane 1). This suggests that these two proteins are able to form a heterocomplex. However, the ratio of the two proteins varied in each preparation (data not shown), indicating that the complex is not stable under the conditions used in this study. Therefore, we used individually expressed proteins for the experiments. TtuA and TtuB were expressed and purified to ϳ90% homogeneity (Fig. 6A, lanes 2 and 3). Both proteins were colorless, and the UV-visible adsorption spectra did not indicate the presence of an iron sulfur (Fe-S) cluster (data not shown).
In vitro s 2 T formation was assayed with tRNA Phe from Saccharomyces cerevisiae using thiocarbonyl-affinity electrophoresis (5). The extent of thiolation was determined by retardation of the thiolated tRNA (13,14). Initially, we attempted to reconstitute the thiolation reaction using only the TtuA and TtuB proteins, but no thiolation was observed (data not shown). We then performed the reaction in the presence of the gelfiltered cell extract (S100) from the s 4 U-deficient mutant (NS0801, thiI mutant strain of T. thermophilus (5)) in order to avoid the formation of s 4 U. As shown in Fig. 6B, lane 1, a small amount of upper-shifted band was observed that may have been 2-thiolated by S100 only. Although neither the addition of TtuA nor TtuB significantly increased 2-thiolation levels (Fig. 6B, lanes 1-3), there was significant enhancement when both were added to the reaction mixture (lane 4). Although this result clearly indicates that both TtuA and TtuB are involved in the biosynthesis of s 2 T, other component(s) present in the S100 extract are also required for the reaction.
Since TtuA and TtuB are predicted to be an ATPase and a sulfur carrier, respectively (see above), ATP and cysteine were omitted from FIGURE 5. Growth phenotypes of s 2 T-deficient strains. Cultures of wild type (HB27), NS2710 (ttuA::km), and NS2720 (ttuB::km), were serially diluted, spotted onto plates containing rich medium, and incubated at the temperatures indicated. Assays were repeated in triplicate. The gel-filtered extract (S100) was present in all reactions. The recombinant proteins (30 pmol of each) were added to the reaction mixture: TtuA (lanes 3-6), TtuB (lanes 2 and 4 -6), IscS (lanes 5 and 7), and SufS (lanes 6 and 8). C, ATP and cysteine requirement for s 2 T formation. S100 fraction, TtuA, TtuB, ATP, and cysteine were present in the reaction mixture (lane 1). ATP was omitted (lane 2), and cysteine was omitted (lane 3). In lane 2, the degraded substrate is indicated by an asterisk (see "Results"). the reaction mixture (Fig. 6C). Elimination of ATP resulted in a complete loss of s 2 T synthesis (lane 2), whereas elimination of cysteine merely resulted in a moderate decrease in the reaction (lane 3), and it is possible that a sulfur intermediate from the cell extract was used for the residual s 2 T synthesis (5). These results demonstrate that s 2 T synthesis requires both ATP and cysteine. When ATP was removed from the reaction mixture, the substrate tRNA was slightly degraded (shown by an asterisk in Fig. 6C). ATP may protect tRNA from degradation by an unknown mechanism.
IscS and SufS Enhance s 2 T Formation-IscS (iron sulfur cluster) is a cysteine desulfurase that is responsible for biosynthesis of thio-modifications in E. coli (33) and Salmonella typhimurium (34). This enzyme catalyzes the initial sulfur transfer reaction from cysteine. Also in the case of T. thermophilus, it is supposed that the sulfur atom of cysteine is finally incorporated into s 2 T in vivo (5). Thus, IscS is one of the candidate enzymes for s 2 T biosynthesis.
Sequences encoding two cysteine desulfurases, IscS and SufS, were identified from the T. thermophilus genome. These genes were cloned, overexpressed as N-terminal His 6 -tagged IscS (TTC0087) and SufS (TTC1373) fusion proteins in E. coli, and purified to ϳ90% homogeneity (Fig. 6A, lanes 4 and 5). These proteins were yellow, a feature characteristic of pyridoxal 5Ј-phosphate-binding proteins (35), and their UV-visible spectrum indicated the presence of pyridoxal 5Ј-phosphate (data not shown). Although neither the addition of IscS nor SufS significantly increased 2-thiolation levels (Fig. 6B, lanes 7 and 8), both cysteine desulfurases increased 2-thiolation in vitro in the presence of TtuA/B (Fig. 6B, lanes 5 and 6); ϳ40% of the substrate tRNA became 2-thiolated in the presence of SufS (lane 6). These results suggest the involvement of these cysteine desulfurases in s 2 T synthesis.

The Sulfur Atom of [ 35 S]Cysteine Is Incorporated into s 2 T in Vitro-
Next, we performed in vitro tRNA thiolation reactions with [ 35 S]cysteine and unlabeled S. cerevisiae tRNA Phe . After the reaction, the RNAs were separated by denaturing PAGE, and the incorporation of [ 35 S]sulfur into the RNAs was monitored by applying an imaging plate (Fig. 7A). When no recombinant proteins were added, faint bands corresponding to 5 S rRNA and the tRNA mixture derived from T. thermophilus S100 (lane 1) were detected in the EtBr-stained gel. tRNA Phe was clearly visible by EtBr staining when added to the reactions (lane 3). One major 35 S-labeled species was detected (lanes 1 and 3) by measuring the radioactivity in the gel, and the bands disappeared after mild alkaline treatment of the samples (lanes 2 and  4). Thus, these bands could be derived from [ 35 S]cysteinylated tRNA Cys species that were originally present in the T. thermophilus S100 fraction and that were cysteinylated during incubation.
When recombinant TtuA and TtuB and IscS or SufS were added (lanes 5-8 or 9 -12, respectively), substrate tRNA Phe was 35 S-labeled together with tRNA Cys from the S100 fraction (lanes 7 and 11). The 35 S labels of tRNA Phe were tolerant of mild alkaline treatment (lanes 8 and 12), suggesting that these bands are derived from tRNA Phe modified by 35 S.
In order to confirm that the sulfur atom of cysteine was actually incorporated into s 2 T, we analyzed the modifications of these labeled tRNA Phe (lanes 8 and 12). Labeled tRNA Phe was first purified by PAGE and then digested into nucleosides before being analyzed by ODS reversed phase column chromatography. One major 35 S peak (31.2 min) was detected in the tRNA hydrolysate (Fig. 7B). The elution position of this peak was the same as that of authentic s 2 T that was co-injected with the sample (shown by an arrow). The chromatogram pattern was essentially the same for tRNAs reacted with IscS (Fig. 7A, lane 8) and SufS  (IP, lower panel). The in vitro reaction was carried out with S100 alone (lanes 1-4); S100, TtuA, TtuB, and IscS (lanes 5-8); or S100, TtuA, TtuB, and SufS (lanes 9 -12). S. cerevisiae tRNA Phe was added to lanes 3, 4, 7, 8, 11, and 12. After the reaction, the samples were treated with alkaline (lanes 2, 4, 6, 8, 10, and 12). The positions of 5 S rRNA from S100 (5S rRNA), S. cerevisiae tRNA Phe (S. c tRNA Phe ), tRNA mixture from S100 (tRNAs), and tRNA Cys from S100 (T. t tRNA Cys ) are indicated on the right. B, nucleoside analysis of 35 S-labeled tRNA Phe (reacted with S100, TtuA, TtuB, and IscS; from lane 8 of A) by reverse phase chromatography. The squares indicate UV absorbance at 280 nm. The circles represent 35 S radioactivity (cpm). The elution position of co-injected authentic s 2 T is indicated by an arrow. (Fig. 7A, lane 12) (data not shown). These results clearly demonstrate that the sulfur atom of cysteine was incorporated into s 2 T in vitro by TtuA, TtuB, and cysteine desulfurases (IscS or SufS).

DISCUSSION
In order to identify tRNA modification enzymes, we used immobilized tRNA Ile to affinity-purify proteins that bound to tRNA. All of the known tRNA Ile -modifying enzymes were purified from the T. thermophilus cell extract. Therefore, this approach proved to be an effective method for the identification of RNA-modifying enzymes and other proteins that bind RNA.
We identified two candidate genes and identified their roles in s 2 T biosynthesis using insertional mutagenesis. Given what is known about the tRNA modification mechanism and sulfur transfer reaction in thiamine biosynthesis, we suggest that TtuA may interact with and activate tRNA and that TtuB may function as an intermediate sulfur donor. The in vitro assay indicated that the cysteine desulfurases, IscS and SufS, were also involved in s 2 T biosynthesis. A putative s 2 T biosynthesis pathway is presented in Fig. 8, and the potential roles played by TtuA, TtuB, IscS, and SufS are discussed below. Although the first activation of the sulfur atom from cysteine is achieved by the cysteine desulfurase in all of the known thio-modification pathways (33,34), the involvement of TtuB-like sulfur carrier protein is a novel feature specific for the s 2 T biosynthesis pathway.
TtuA has a number of Cys-Xaa-Xaa-Cys (CXXC) motifs, which are found at the redox centers of thioredoxins, and disulfide bond isomerases (40,41). This motif is also known to act as a metal binding site, either for divalent cation such as Zn 2ϩ (42), or for iron sulfur (Fe-S) clusters (43). In this study, no evidence was found for an Fe-S cluster in the UV-visible adsorption spectrum of the recombinant TtuA. However, many Fe-S clusters are known to be unstable under oxidative conditions, and further experimentation is required to confirm the absence of an Fe-S cluster in TtuA. TtcA protein, a group I member of the TtuA/TtcA family, may have an Fe-S cluster, because the synthesis of s 2 C requires IscU (44), which is involved in Fe-S cluster biogenesis. The presence and absence of Fe-S clusters may be specific features of groups I and II members of the TtuA/TtcA family, respectively. The cysteines in the central CXXC motif of TtcA are known to be essential for s 2 C formation in S. typhimurium (26) and may play a similar role in TtuA for the synthesis of s 2 T.
TtuB is a close paralog of ThiS, a sulfur carrier protein of the thiamine biosynthesis pathway (32). Thiamine is a sulfur-containing cofactor essential for enzymes involved in carbohydrate and branched-chain amino acid metabolism and is synthesized from thiazole and pyrimidine moieties (32). A sulfur atom that is derived from cysteine is eventually incorporated into the thiazole. The sulfur transfer pathway, from cysteine to thiazole, uses a C-terminal thiocarboxylate of ThiS (ThiS-COSH) as an intermediate sulfur donor. The thiocarboxylate is formed in two stages. Another ATPase, ThiF, acyl-adenylates the C-terminal of ThiS to ThiS-COAMP (30). ThiS-COAMP is then converted to the thiocarboxylate form (ThiS-COSH) by ThiF and IscS, using cysteine as a sulfur donor (31). Similar activation pathways are found in the biosynthesis of molybdenum cofactor (45) and in the eukaryotic ubiquitin conjugation system (46).
Given that TtuB shares significant sequence homology with ThiS ( Fig. 3B), it also may be thiocarboxylated at the C-terminal glycine (TtuB-COSH) by IscS or SufS using cysteine as a sulfur donor, and this putative thiocarboxylate may be the immediate sulfur donor for s 2 T formation. In this scenario, the sulfur atom of TtuB-COSH would be incorporated into s 2 T, either by direct transfer to the tRNA or via TtuA (possibly to a conserved cysteine residue). If this scenario is correct, then we would expect a C-terminal acyl-adenylated form of TtuB (TtuB-COAMP) to be synthesized by an ATPase (corresponding to the activity of ThiF in ThiS-COAMP synthesis). The results of our in vitro s 2 T reaction demonstrated that ATP was an absolute requirement for s 2 T synthesis (Fig. 6), suggesting the involvement of a TtuB-activating ATPase, in addition to the tRNA-activating ATPase (TtuA). Although it is possible that TtuA is responsible for the C-terminal activation of TtuB, a TtuB-activating ATPase is a major candidate for one of the unidentified but strictly required components of the in vitro s 2 T reaction, present in the T. thermophilus cell extract. Both IscS and SufS enhanced formation of s 2 T in the in vitro reaction (Fig. 6). This result also strongly suggests the involvement of a C-terminal thiocarboxylated form of TtuB (TtuB-COSH). In Bacillus subtilis, all four cysteine desulfurases (CSD, NifZ, YrvO, and NifS) can serve as sulfur donors for the formation of ThiS-COSH in vitro (47). IscS is a primary sulfur donor for the thio-modifications of tRNA in E. coli (33). However, in T. thermophilus, mutants of IscS and SufS remain capable of synthesizing s 2 T when the two genes are disrupted independently (5). This suggests that both cysteine desulfurases can serve as sulfur donors in T. thermophilus.
SufE and CsdE are known activation partners for the cysteine desulfurase reactions of SufS and CsdA, respectively (48 -50). Thus, these proteins may be involved in s 2 T synthesis. However, no homologs for SufE or CsdE were found to be encoded in the T. thermophilus genome (data not shown).
Untagged TtuA was co-purified with His 6 -tagged TtuB using nickel affinity column chromatography (Fig. 6A, lane 1), suggesting that TtuA and TtuB form a heterocomplex, which may function as the s 2 T synthetase in vivo. A detailed investigation of the interactions of this complex, using crystallography, may clarify the mechanism of this tRNA modification system. s 2 T is found at position 54 in the tRNA of T. thermophilus (51) and P. furiosus (2), and this modification is also found in the tRNA of Thermotoga maritima and Thermococcus sp. MSB4, although its position remains unknown (52). In the genome sequence of P. furiosus, we identified two ORFs (PF0273 (see Fig. 3B) and PF1758) that were closely related to TtuA of T. thermophilus. The proteins encoded by these ORFs may be involved in the synthesis of s 2 T and s 2 C, respectively, since P. furiosus tRNA has s 2 C, in addition to s 2 T (2). The genome of Thermococcus kodakarensis KOD1 was found to contain two ttuA-like ORFs, which may correspond to proteins involved in the biosynthesis of s 2 T and s 2 C. Although the tRNA of T. maritima possesses both s 2 T and s 2 C, only one hypothetical protein, TM0197 (Fig. 3B), exhibited homology with TtuA, suggesting a functional overlap in this organism. Both P. furiosus and T. kodakarensis have closely related TtuB homologs (Fig.  3B). However, no obvious homolog for TtuB could be identified in the T. maritima genome (data not shown), suggesting an alternative sulfur transfer system for s 2 T synthesis. Additionally, in these organisms, TtuA and TtuB are not encoded in a single operon, and this genomic organization may be specific to T. thermophilus (Fig. 4A).
The distribution of TtuA and TtuB varies among species. As noted previously, TtuA/TtcA homologs are distributed widely (26) and may participate in thio-modification reactions other than s 2 T and s 2 C. In contrast, TtuB homologs are found in only a limited number of species, and these sulfur transfer components may have evolved from the thiamine biosynthesis pathway in a progenitor.
The mutant strains ttuA and ttuB demonstrated a temperature-sensitive phenotype, when compared with the wild-type strain. This is the first result that suggests that s 2 T in tRNA is critical for bacterial growth at high temperatures. It is known that s 2 T-containing tRNA works well at higher temperatures (3) and that its abundance increases with cultivation temperature (1). However in in vitro reactions, the efficiency of translation is not dependent upon the temperature at which the cells were cultivated prior to preparation of ribosomes and translation factors (3,53). Taken together, these results indicate that control of the s 2 T content in tRNA enables the translation machinery to function over the temperature range encountered by the bacterium in nature. The observation that the mutants could grow at 75°C but not at 80°C is somewhat surprising, since tRNA structure would not be expected to undergo dramatic denaturation within this narrow temperature range. There exists the possibility that other unknown factor(s) are involved in the temperature-sensitive phenotype of these mutants.
Further clarification of the expression and activities of TtuA, TtuB, IscS, SufS, and the unidentified protein(s) that were present in the gel filtration extract will lead to a better understanding of the s 2 T biosynthesis pathway and the mechanisms underlying the control of the s 2 T content of tRNA.