Thio-modification of Yeast Cytosolic tRNA Requires a Ubiquitin-related System That Resembles Bacterial Sulfur Transfer Systems*

The wobble uridine in yeast cytosolic tRNALys2UUU and tRNAGlu3UUC undergoes a thio-modification at the second position (s2 modification) and a methoxycarbonylmethyl modification at the fifth position (mcm5 modification). We previously demonstrated that the cytosolic and mitochondrial iron-sulfur (Fe/S) cluster assembly machineries termed CIA and ISC, including a cysteine desulfurase called Nfs1, were essential for the s2 modification. However, the cytosolic component that directly participates in this process remains unclear. We found that ubiquitin-like protein Urm1 and ubiquitin-activating enzyme-like protein Uba4, as well as Tuc1 and Tuc2, were strictly required for the s2 modification. The carboxyl-terminal glycine residue of Urm1 was critical for the s2 modification, indicating direct involvement of the unique ubiquitin-related system in this process. We also demonstrated that the s2 and mcm5 modifications in cytosolic tRNAs influence each other's efficiency. Taken together, our data indicate that the s2 modification of cytosolic tRNAs is a more complex process that requires additional unidentified components.

Many modified nucleotides are found in tRNAs of various organisms, and the post-translational modification of tRNA molecules is thought to be necessary to maintain their structure and thereby to exert their proper function in translation (1). In yeast, uridine of the first position of anticodon, in cytosolic tRNA (cy-tRNA), 2 for lysine (cy-tRNA Lys2 UUU ) and glutamate (cy-tRNA Glu3 UUC ) contains sulfur instead of oxygen at the second position (the s 2 modification) and 5-methoxylcarbonylm-ethyl at the fifth position (the mcm 5 modification). These cy-tRNAs read split codon boxes; they decode the general NAAtype codon and can wobble into the NAG codon (2). Thus, the s 2 and mcm 5 modifications in U 34 are thought to be important in maintaining stable codon-anticodon pairing during decoding of these cy-tRNAs on the ribosome.
For the s 2 modification of cy-tRNA Lys2 UUU and cy-tRNA Glu3 UUC , the cysteine desulfurase Nfs1 located in the mitochondria was essential, indicating that a sulfur atom used for the s 2 modification should originate from the cysteine sulfur atom located inside the mitochondria (3). Nfs1 is also known to provide sulfur for the iron-sulfur (Fe/S) cluster biosynthesis, which involves the mitochondrial ISC and cytosolic CIA machineries (4 -6). We previously demonstrated that the s 2 modification of cy-tRNAs was dependent not only on Nfs1 but also on other ISC and CIA proteins such as Cfd1 (7). Because the cytosolic Fe/S cluster assembly mediated by CIA must precede Fe/S cluster biosynthesis in the mitochondria, which is mediated by ISC (8), our previous observation suggests that at least one cytosolic Fe/S cluster-containing protein plays an indispensable role in the s 2 modification of cy-tRNAs (7). It may also be possible that the sulfur atom forming an Fe/S cluster may itself be directly used for the s 2 modification. Besides the mitochondrial Nfs1 and Fe/S cluster assembly machineries, ISC and CIA, other cytosolic components directly involved in the s 2 modification remain to be elucidated. However, for the mcm 5 modification of U 34 in cy-tRNA Lys2 UUU and cy-tRNA Glu3 UUC , several cytosolic proteins, including the Elongator proteins (Elp1-6) and Kti11-13 proteins, have been shown to be involved (9). Among them, Elp3 is an Fe/S clustercontaining protein (10), but its knock-out mutant was reported to retain the s 2 -modified uridine derivatives in their nucleoside pool (9).
Here we identify Tuc1 (previously named Ncs6) as an indispensable factor for the s 2 modification. In addition, we found that Tuc2 (previously named Ncs2), Urm1, and Uba4 were also strictly required for the s 2 modification of cy-tRNAs. All of these factors are somehow genetically linked to Cla4, whose depletion causes a synthetic lethality with the tuc1 null mutation (11). Although Urm1 and Uba4 were initially identified as a ubiquitination-like protein modifier system (12), the exact function of this system remains unclear. We propose here that a cytosolic sulfur transfer system mediated by Urm1 and Uba4 plays a key role in the s 2 modification of cy-tRNAs. * This work was supported by Grant-in-aid for Scientific Research 19570140 from the Japan Society for the Promotion of Science (to Y. N.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence may be addressed. Fax: 81-726-84-6516; E-mail: med004@art.osaka-med.ac.jp. 2 The abbreviations used are: cy-tRNA, cytosolic tRNA; cy-tRNA Lys2 UUU , cytosolic tRNA for lysine; cy-tRNA Glu3 UUC , cytosolic tRNA for glutamate; s 2 modification, thio-modification at the second position; s 2 U, 2-thiouridine; mcm 5 modification, 5-methoxylcarbonylmethyl modification at the fifth position; mcm 5 U, 5-methoxycarbonyl methyluridine; cm 5 U, 5-carboxymethyl uridine; Fe/S, iron-sulfur; ISC machinery, iron-sulfur cluster biogenesis machinery; CIA machinery, cytosolic iron-sulfur cluster assembly machinery; TAP, tandem affinity purification tag; APM, [(N-acryloylamino)phenyl] mercuric chloride; U 34 , uridine at the first anticodon sequence in tRNA; E1, ubiquitin-activating enzyme; HPLC, high pressure liquid chromatography.
The TUC1 gene was amplified by PCR and subcloned into the plasmid pESC-LEU (Stratagene, La Jolla, CA), generating the plasmid pESC-Tuc1-myc, which expressed Tuc1 followed by a c-Myc epitope under the GAL1 promoter. The plasmid was introduced into ⌬tuc1 cells to generate the recombinant strain ⌬tuc1/Tuc1-myc. Cells grown on SD-Leu plates were cultivated with SGGly-Leu medium to express the recombinant protein, and recovery of the s 2 modification was confirmed with the APM-Northern analysis. The TUC2 gene was also amplified by PCR and subcloned into the plasmid pESC-HIS to generate pESC-Tuc2-FLAG, generating Tuc2 followed by a FLAG epitope sequence under the GAL10 promoter. The plasmid was introduced into ⌬tuc2 cells to generate ⌬tuc2/Tuc2-FLAG cells. Cells grown on SD-His plates were cultivated with SGGly-His to express the recombinant protein, and recovery of the s 2 modification was confirmed. The Tuc2-TAP strain was also purchased (Open Biosystems, the yeast TAP strains).
To examine temperature-sensitive and/or caffeine-sensitive growth, each strain grown to the mid-log phase in appropriate liquid media was harvested and transferred to SD medium. After a 4-h incubation at 30°C, each cell suspension that measured an A 600 ϭ 0.1 unit was serially diluted, spotted onto agar plates, and incubated at 30°C or 37°C.
Gel Retardation Analysis Using APM Coupling with Northern Hybridization (APM-Northern Analysis)-The presence of thiouridine in the prepared tRNA can be verified by the retardation of electrophoretic mobility on polyacrylamide gels containing [(N-acryloylamino)phenyl] mercuric chloride (APM). Equal amounts (2 g) of total tRNA were separated with a denaturing gel containing 7 M urea and 24 M of APM and analyzed with a specific DNA probe against cy-tRNA Lys2 UUU , cy-tRNA Glu3 UUC , and cy-tRNA Arg UCU , as shown previously (3). The relative proportion of the s 2 modification in each tRNA was determined with three independent experiments as described previously (7).
Detection of the Adenylated Intermediate of Tuc1-The TUC1 gene was inserted into pET28a (Novagen) and used to express a His 6 -Tuc1 protein in Escherichia coli BL21(DE3)Star (Invitrogen). E. coli soluble cell extracts containing His 6 -Tuc1 were incubated with a mixture of 4 g of tRNA prepared from ⌬tuc1 cells at 37°C for 30 min in a 10-l reaction mixture with 50 mM Tris-HCl buffer (pH 7.5) containing 50 mM KCl, 12 mM Mg(OAc) 2 , 1 mM dithiothreitol, and 5 M of [␣-32 P]ATP or [␥-32 P]ATP (ϳ300 Ci/mmol, BD Biosciences). After quenching the reaction by adding 90 l of 5% formic acid, tRNA was extracted and precipitated and then subjected to PAGE under denaturing conditions. The gel was stained with ethidium bromide and also exposed on an imaging plate to visualize the radioactivity using the BAS2500 system (FujiFilm).
Sulfur Incorporation into the Recombinant Uba4 Protein in Vitro-URM1 and UBA4 genes were cloned into pACYC-Duet1 (Novagen) to express Urm1 and Uba4 with a carboxylterminal S-tag (Uba4-S-tag either separately or simultaneously). Expression of the recombinant proteins in BL21(DE3)Star (Stratagene) cells was induced by adding 0.5 mM isopropyl galactopyranoside for 3 h at 37°C. When grown to A 600 ϭ 0.5, E. coli cells were collected and resuspended in Luria Bertani medium containing half-concentration of salts, followed by further incubation for 30 min at 37°C in the presence of 50 g/ml blasticidin-S HCl (Invitrogen). The cells were then collected and suspended in 20 mM Tris-HCl (pH 7.5) with 0.1% Triton X-100. After disruption by freeze-thawing, cells were incubated with L-[ 35 S]cysteine (1075 Ci/mmol, Tokyo Medical CRO, Japan) for 30 min at room temperature. The reaction mixtures were spun down, and then the supernatants were subjected to acrylamide gel electrophoresis under nonreducing conditions. 35 S-Labeled protein bands were detected by contacting the gel to the imaging plate and analyzed with BAS2500 (FujiFilm).
HPLC Analysis of the Total tRNA Nucleoside-Total tRNAs prepared were digested with 4 units of ribonuclease P1 (Yuasa Shou Co., Ltd.) in 100 l of 10 mM sodium acetate (pH 5.2) for 1 h. After addition of Tris-HCl (pH 7.5) to a final concentration of 100 mM, the resulting nucleosides were then precipitated with 2-propanol. The total nucleosides suspended in 50 mM Tris-HCl were loaded onto a chromatography column (COS-MOSIL 5C 18 column, 4.5 mm ϫ 25 cm, Nakalai Tesque, Japan) and developed with a gradient of 25 mM sodium acetate and 50% acetonitrile as described previously (1). The eluted nucleosides were monitored at 256 nm.
Gel Filtration Chromatography with Yeast Cytosol Coupling Immunological Detection of Proteins-Cytosolic fractions were prepared from TAP-Tuc1 and TAP-Tuc2 cells that expressed Tuc1 and Tuc2 proteins with TAP tags at their carboxyl termini, as described previously. They were fractionated using Superdex-200 gel filtration column chromatography using SMART System (GE Healthcare) with a running buffer of 150 mM NaCl and 50 mM HEPES-NaOH (pH 7.5). Aliquots of each fraction were separated using electrophoresis on a 12.5% SDSpolyacrylamide gel followed by Western blotting with an anti-TAP antibody (Open Biosystems) and an anti-Urm1 antibody (Invitrogen).

Tuc1 and Tuc2 as Well as Urm1 and Uba4 Are Strictly
Required for the s 2 Modification of cy-tRNAs-During a search for possible cytosolic proteins that are directly involved in the s 2 modification of cy-tRNAs, we noticed that Ncs6 (YGL211W gene product, now termed Tuc1), which is likely to be a cytosolic protein, contained a PP-loop-type ATPase domain as well as some conserved cysteine residue-containing domains. Because proteins containing this domain arrangement have been shown to be involved in some kinds of tRNA modifications (13), and because Cfd1 and Nbp35, which participate in both the s 2 modification of cy-tRNAs and the cytosolic Fe/S cluster assembly, also possess the P-loop motif and can transfer certain sulfur-containing compounds, we regarded Tuc1 (Ncs6) as a likely candidate that could be directly involved in the s 2 modification of cy-tRNAs. The s 2 modification of total tRNAs prepared from ⌬tuc1 cells was examined to observe the retardation of tRNA during migration in the APM-containing gel (see "Experimental Procedures" for details). We analyzed three cytosolic tRNAs as follows: cy-tRNA Lys2 UUU and cy-tRNA Glu3 UUC undergo s 2 modification in addition to mcm 5 modification, whereas cy-tRNA Arg UCU undergoes only mcm 5 modification. As shown in Fig. 1, the migration of cy-tRNA Arg UCU was not altered in either wild-type BY4742 or ⌬tuc1 cells. In contrast, the migration of cy-tRNA Lys2 UUU and cy-tRNA Glu3 UUC prepared from wild-type cells was significantly retarded as compared with those prepared from ⌬tuc1 cells, the latter of which corresponded well to the migration of cy-tRNA Arg UCU . These data indicate that the s 2 modification of both cy-tRNA Lys2 UUU and cy-tRNA Glu3 UUC was severely impaired in ⌬tuc1 cells.
Tuc1 was originally designated Ncs6 (need Cla4 to survive 6) because depletion of Tuc1 causes a synthetic lethality in a ⌬cla4 strain (11). Cla4 is known as a Ste20-like protein kinase involved in signal transduction and whose deletion exhibits multiple defects (14). NCS2 (now termed TUC2), URM1, and UBA4 genes were also demonstrated to be essential in a ⌬cla4 strain, but their exact functions still remain unclear (15). Therefore, we examined whether deletion of TUC2, URM1, or UBA4 causes any defect in the s 2 modification of cy-tRNAs. As shown in Fig. 1, we observed that the s 2 modification of both cy-tRNA Lys2 UUU and cy-tRNA Glu3 UUC was almost completely impaired in ⌬tuc2, ⌬urm1, or ⌬uba4 cells. However, in contrast to these stains, ⌬cla4 did not exhibit any such defects in the s 2 modification of both cy-tRNA Lys2 UUU and cy-tRNA Glu3 UUC , indicating that Tuc2, Urm1, and Uba4, but not Cla4, are necessary for the s 2 modification of cy-tRNA Lys2 UUU and cy-tRNA Glu3 UUC . To know whether these novel factors involved in the s 2 modification of cy-tRNAs form a complex in the cytosol, we performed gel filtration chromatography of cytosolic fractions prepared from yeast cells expressing Tuc1-TAP and Tuc2-TAP proteins followed by Western analysis using an anti-TAP tag and anti-Urm1 antibodies ( Fig. 2A). Note that expression of these tagged proteins in the respective deletion strains was able to restore the cytosolic s 2 modification (data not shown), indicating that they are functional. Although the calculated molecular mass of Tuc1-TAP protein is 60,523 Da, the protein was eluted as a large molecular weight complex (Ͼ250 kDa) (fraction 4 in Fig.  2A). However, Tuc2-TAP, whose calculated molecular mass is 77,008 Da, eluted at ϳ140 kDa, which most likely corresponds to a homodimer (fraction 6 in Fig. 2A). Urm1 protein was eluted at ϳ24 kDa, which was also most likely a homodimer. Taken together, Urm1, Tuc1, and Tuc2 exist as distinct entities in the yeast cytosol.
Tuc1 Belongs to the Family of PP-loop Type ATPases and Can Catalyze Adenylation of tRNAs-MnmA, a PP-loop ATPase required for the s 2 modification of U 34 in E. coli tRNAs, exhibits pyrophosphatase activity in vitro, and it can activate U 34 by adenylation just prior to sulfur incorporation (16). Tuc1 is also characterized as a PP-loop ATPase, whereas the unique cysteine motif that is lacking in MnmA protein exists at the aminoterminal region of Tuc1 (2). We therefore investigated whether Tuc1 possesses similar adenylation activity. When the non-s 2 modified tRNAs prepared from ⌬tuc1 cells were incubated with E. coli extracts containing His 6 -Tuc1 protein in the presence of [␣-32 P]ATP, radiolabeled tRNAs were seen. However, when [␥-32 P]ATP was used, no such radiolabeled species were detected (Fig. 2B). Therefore, like MnmA in E. coli, yeast Tuc1 can form an adenylated tRNA intermediate that is likely required for the s 2 modification of yeast cy-tRNAs.
Carboxyl-terminal Glycine of Urm1 Is Important for the s 2 Modification of cy-tRNAs-Urm1 and Uba4 were initially identified as a novel protein conjugation system, where Urm1 acts as a ubiquitin-related modifier, and Uba4 functions as a ubiquitin-activating enzyme (E1)-like protein (12). Later it was proposed that this protein conjugation system be termed urmylation, although the target proteins that undergo urmylation remain unclear (15). In parallel with the ubiquitin system, Urm1 forms a thioester with Uba4 between the carboxyl-terminal glycine of Urm1 and the active cysteine of Uba4 (12). To examine whether the carboxylterminal glycine of Urm1 is also required for the s 2 modification of cy-tRNAs, a mutant form of Urm1 whose carboxyl-terminal glycine was masked by the Myc tag (Urm1-myc) of 14 amino acids was expressed in ⌬urm1 cells under the control of the GAL10 promoter. Although s 2 modification of cy-tRNA Lys2 UUU and cy-tRNA Glu3 UUC in ⌬urm1 cells was fully recovered by expressing the wild-type Urm1, no such recovery was observed when Urm1myc was expressed (Fig. 3A). Western analysis confirmed that expression levels of Urm1 and Urm1-myc in ⌬urm1 cells were not different from each other (Fig. 3B). From these data, we concluded that the carboxyl-terminal glycine of Urm1 is essential for the s 2

FIGURE 3. Complementation analysis of the s 2 modification in ⌬urm1 cells by expressing Urm1 with or without carboxyl-terminal Myc tag.
A, total tRNA prepared either from the ⌬urm1/Urm1 or ⌬urm1/Urm1-myc cells grown under the GAL1/10 promoter-induced (ϩ) or -repressed (Ϫ) conditions were examined for the s 2 modification of cy-tRNA Lys2 UUU (cy-K) and cy-tRNA Glu3 UUC (cy-E). Parenthesis and an arrowhead indicate corresponding tRNAs with and without s 2 modification. Proportions of the s 2 -modified tRNAs are shown as bar graphs presenting the average values obtained from three independent experiments, with error bars indicating the ranges of the values (lower panel). B, expression of Urm1 and Urm1-myc in ⌬urm1 cells used in A was confirmed by Western analysis using anti-Urm1 and anti-Myc antibodies. Whole cell extracts were prepared from transformed yeast cells grown under the GAL1/10 promoter-induced (ϩ) or -repressed (Ϫ) conditions. Note that the URM1 and URM1-MYC genes were placed under the GAL1/10 promoter and transformed into in ⌬urm1 cells so that no band was detected with either antibody in the extracts prepared from cells grown under the GAL1/10 promoter-repressed conditions. modification of cy-tRNAs and that this unique ubiquitin-related system involving Urm1 and Uba4 plays a critical role in the s 2 modification.
Uba4 Can Bind the Cysteine-derived Sulfur with the Aid of Urm1-In bacteria, sulfur that is incorporated into thiamine is derived from L-cysteine, and this sulfur incorporation reaction is known to be catalyzed, at least in part, by ThiS and ThiF, both of which were demonstrated to bind cysteine-derived sulfur (17). Interestingly, Urm1 and Uba4 show significant sequence similarities to ThiS and ThiF, respectively. We therefore asked if Urm1 and/or Uba4 can bind cysteine-derived sulfur. To achieve this, Urm1 and Uba4 (with S-tag) proteins were expressed in E. coli cells either separately or simultaneously (Fig. 4A). After de novo protein synthesis was blocked by blasticidin-S, cell extracts were incubated with [ 35 S]cysteine. As shown in Fig. 4B, the 57-kDa Uba4-S-tag protein was found to bind the cysteine-derived 35 S label only in the presence of coexpressed Urm1. These data suggest that Uba4 might function as a sulfur carrier protein required for the s 2 modification of cy-tRNAs. Although Furukawa et al. (12) previously identified a significant amount of Urm1-Uba4 conjugate when tagged versions of both proteins were overexpressed in yeast cells, no such Urm1-Uba4 conjugate was detected in E. coli cell extracts when both proteins were expressed at high levels (Fig. 4A).
mcm 5 Modification of U 34 Influences the Efficiency of s 2 Modification of cy-tRNAs-As described above, U 34 of cy-tRNA Lys2 UUU and cy-tRNA Glu3 UUC undergoes both mcm 5 and s 2 modifications. Trm9 is a methyltransferase for the mcm 5 modification (18), and both Kti11 and Elp3 are involved in the early steps of this modification (9,19). In ⌬kti11, ⌬elp3, and ⌬trm9 cells, the mcm 5 modification of cy-tRNA was almost completely inhibited (9,19). In those mutants, although the s 2 modification of cy-tRNAs was reported to occur, the impact on the efficiency of s 2 modification was not investigated. As shown in Fig. 1, the s 2 modification was significantly, but not totally, inhibited in ⌬kti11, ⌬elp3, and ⌬trm9 mutants, suggesting that impairment of mcm 5 modification has some inhibitory effects on the s 2 modification in cy-tRNAs. In contrast, to examine the impact on efficiency of mcm 5 modification by eliminating the s 2 modification in cy-tRNAs, we compared HPLC elution profiles of nucleosides of bulk tRNA prepared from the wild-type, ⌬elp3, ⌬tuc1, and ⌬tuc2 cells. As shown in Fig. 5A, we observed a common major nucleoside peak in ⌬elp3, ⌬tuc1, and ⌬tuc2 samples but not with the wild-type sample. The nucleoside was purified, and its molecular mass was determined to be 302.03, which shows good agreement with that of 5-carboxymethyluridine (cm 5 U) (302.24), a premature form of mcm 5 U (Fig. 5B). This may indicate that impairment of s 2 modification also affects the efficiency of methylation of cm 5 U 34 .  ⌬elp3 and ⌬kti11 Strains Are More Sensitive to Stress Than ⌬urm1, ⌬uba4, ⌬tuc1, and ⌬tuc2 Strains-Previously, ⌬elp3 and ⌬kti11 strains were demonstrated to exhibit severe growth defects when they were grown at elevated temperature and/or with caffeine, a purine analogue (20). Therefore, we compared the growth of ⌬urm1, ⌬uba4, ⌬tuc1, and ⌬tuc2 strains with those of ⌬elp3 and ⌬kti11 strains as well as that of wild-type cells under such stress conditions. At normal temperature (30°C), all mutants exhibited equal growth, but only slightly slower than the wild type (Fig. 6). However, when grown at 37°C and/or with 10 mM caffeine, severe growth defects were obvious for ⌬elp3 and ⌬kti11 strains but not for ⌬urm1, ⌬uba4, ⌬tuc1, and ⌬tuc2 strains. Thus, the mcm 5 modification may be more important for tolerance to these stresses than the s 2 modification at the same U 34 .

DISCUSSION
This study identifies the following four additional yeast cytosolic components that are strictly required for the s 2 modifica-tion of U 34 of cy-tRNAs: ubiquitin-like protein Urm1, ubiquitin-activating enzyme E1-like protein Uba4, cytosolic PP-loop ATPase Tuc1, and unknown function protein Tuc2 (Fig. 7). Therefore, the s 2 modification of cy-tRNAs requires more protein factors than initially anticipated. While preparing this manuscript, studies were published suggesting that Tuc1 and Tuc2 (previously called Ncs6 and Ncs2, respectively) are involved in the s 2 modification of cy-tRNAs because their mutants exhibited no detectable amounts of 2-thiouridine nucleoside derivatives (2,19). Our experimental data are in agreement with this previous observation.
Both Yeast Tuc1 and bacterial MnmA are classified as part of the PP-loop ATPase family (2). We showed here that Tuc1 is truly required for the s 2 modification of cy-tRNAs (Fig. 1) and also that it can catalyze the formation of the adenylated form of yeast tRNAs (Fig. 2B). MnmA was previously demonstrated to recognize the tRNA nucleoside U 34 to form an adenylated intermediate that is thought to be a prerequisite for sulfur incorporation at that position (16). A similar reaction was also found in another N-type ATP pyrophosphatase, TilS and MesJ, which facilitates adenylation of C 34 of isoleucine tRNA to incorporate L-lysine to form lysidine (21). Therefore, by analogy with MnmA and TilS/MesJ, Tuc1 is likely to function in the adenylation of U 34 of cy-tRNA Lys2 UUU and cy-tRNA Glu3 UUC that should be necessary for sulfur incorporation (Fig. 7). Another MnmA-like protein, Mtu1, is localized in yeast mitochondria and functions in s 2 modification of U 34 of mitochondrial tRNAs (22). We previously demonstrated that the s 2 modification of cy-tRNAs is Fe/S protein-dependent but that the s 2 modification of mt-tRNAs is Fe/S cluster-independent, although both require the mitochondrial cysteine desulfurase Nfs1 (7). Thus, yeast cytosolic and mitochondrial s 2 modifications of tRNAs are mechanically distinct from each other, at least in part, but appear to require two homologous PP-loop ATPases localized separately in each compartment. Interestingly, Tuc1 more closely resembles bacterial TtcA, which is required for the s 2 modification of C 32 (2), than MnmA or Mtu1, whereas no s 2 C modification has been found in yeast so far. This may indicate different evolutionary histories for cytosolic and mitochondrial s 2 modification of U 34 of tRNAs.
Tuc1 can activate U 34 to be s 2 -modified by preceding adenylation, but subsequent s 2 -modification itself requires additional cytosolic components. Tuc2 might be one of these cytosolic components, although the exact function of Tuc2 remains unclear at this time. Very recently, Tuc1 and Tuc2 orthologues (termed Ctu1 and Ctu2, respectively) identified in the nematode and fission yeast were found to form a complex (23). However, our data demonstrated in Fig. 2A indicate that, in budding yeast, the two proteins exist as distinct entities in the cytosol. In this context, it is noteworthy that Tuc1 appears to form a complex larger than 250 kDa. Identification and characterization of other constituents of the complex may help to elucidate a detailed molecular mechanism for the final s 2 modification step.
The strict requirement of Urm1 and Uba4 for the s 2 modification of cy-tRNAs is surprising because the two proteins were originally identified as part of a ubiquitin-related protein modifier system (12,15). Urm1 is conserved among eukaryotes and shows a weak homology to ubiquitin (24). More intriguingly, Urm1 resem- bles bacterial sulfur carrier proteins, ThiS and MoaD, both in their sequence and structure (25,26). Correspondingly, Uba4 shows significant sequence similarity to E1 as well as ThiF and MoeB, which are involved in thiamine and molybdenum cofactor biosyntheses in cooperation with ThiS and MoaD, respectively (12). The carboxyl-terminal glycine of ThiS is activated by ThiF to form an acyl-adenylated intermediate (27). Such a reaction is mechanistically similar to that seen between ubiquitin and the E1 enzyme (Fig. 7). Different from the ubiquitin system that converts the acyladenylate to acyl-thioester linkage with the active site cysteine residue of E1, ThiS undergoes an acyl-disulfide linkage with an active site cysteine of ThiF (28), where the additional sulfur donated by the bacterial cysteine desulfurase IscS is supplied as a form of cysteinyl persulfide sulfur by ThiI (29). In the case of acyl-thioesterlinked ubiquitin-E1 conjugate, ubiquitin is finally transferred to a lysine amino group of a specific target protein with the aid of ubiquitin-conjugating enzyme and ubiquitin-protein isopeptide ligase enzymes, whereas the acyl-disulfide bound on ThiS is used as a sulfur donor for thiamine biosynthesis (Fig. 7).
By analogy, the critical roles of Urm1 and Uba4 in the s 2 modification of cy-tRNAs can be explained as follows (Fig. 7). The carboxyl-terminal glycine residue of Urm1, whose importance is shown in this study (Fig. 3), is activated by Uba4 to form acyl-adenylate using ATP, followed by the formation of an acyldisulfide linkage between Urm1 and Uba4 (Fig. 7). Although we could not detect a complex between Urm1 and Uba4 when expressed in E. coli, such a complex was previously identified in yeast (12). Thus, the s 2 modification of cy-tRNAs involving Urm1 and Uba4 may resemble the sulfur-transfer system found in bacterial thiamine biosynthesis more than the eukaryotic ubiquitination system (Fig. 7). That recombinant Uba4 can bind sulfur derived from cysteine in E. coli when Urm1 was co-expressed, as shown in Fig. 4B, may suggest the formation of such an acyl-thioester-linked Urm1-Uba4 conjugate, although this conjugate might be unstable during our assay system so that only Uba4 carrying a cysteinyl persulfide could be observed.
Recently, Schmitz et al. (30) reported that yeast Uba4 had the sulfur transferase activity in vitro. They also demonstrated that persulfurated Uba4 was able to form a thiocarboxylate group of Urm1 in vitro. Therefore, their experimental data are in good agreement with our observations and our working model.
In this study, we also demonstrated that the s 2 and mcm 5 modifications in cy-tRNAs influence each other. As shown in Fig. 1, the defect in the mcm 5 modification of cy-tRNA could not totally inhibit the s 2 modification (only about 50% reduction was seen). Actually, total tRNA nucleoside pool purified from ⌬elp3 cells was reported to contain at least s 2 U (9). However, we successfully observed the accumulation of cm 5 U, a possible direct precursor of mcm 5 U group in ⌬elp3 or in ⌬tuc1 or ⌬tuc2 cells (Fig.  5). Because ELP3 is a key protein that mediates the mcm 5 U formation in yeast cells, the observation that the degree of cm 5 U accumulation in ⌬tuc1 or ⌬tuc2 seems to be comparable with that found in ⌬elp3 cells strongly suggests that loss of an s 2 modification does affect the mcm 5 modification. Although yeast mutants that abolish either modification exhibit slight growth defects, the complete absence of both modifications is lethal because ⌬elp3⌬tuc1 or ⌬elp3⌬tuc2 double knock-out mutants are not viable (2). Therefore, the s 2 and mcm 5 modifications in cy-tRNAs are probably important for the stability of these tRNAs and/or to maintain translation efficiency, although they may have interrelated but complementary effects. Indeed, a significant number of smaller tRNA fragments are accumulated when either modification was inhibited (Fig. 1).
Despite the identification of the novel cytosolic components shown in this study, it remains unclear how a sulfur atom is provided in yeast cytosol for the process of the s 2 modification of cy-tRNAs (Fig. 7). One sulfur atom should presumably be supplied as a protein-bound persulfide that originated from cysteine desulfuration activity of mitochondrial Nfs1, and such incorporated sulfur is somehow exported to the cytosol and finally used for the s 2 FIGURE 7. A plausible model of ubiquitin-related sulfur transfer system required for the s 2 modification of cy-tRNAs in yeast. Protein modifier systems of ubiquitination found in eukaryotes (A) and bacterial sulfur transfer system in thiamine biosynthesis (B) are compared with a plausible model for the s 2 modification of yeast cy-tRNAs that involves the Urm1-Uba4 system and a PP-loop ATPase Tuc1 and Tuc2. Cysteine desulfurase Nfs1 primarily supplies a sulfur atom necessary for the s 2 modification (3). The sulfur atom is then transferred somehow via the sulfur-transfer system involving Urm1 and Uba4 and also with Tuc2 and as-yet-unidentified cytosolic Fe/S protein to the activated tRNA by adenylation where Tuc1 functions in this process (C). Dark gray arrows shown in B and C indicate proposed sulfur transfer pathways. s 2 Modification of Yeast Cytosolic tRNAs OCTOBER 10, 2008 • VOLUME 283 • NUMBER 41 modification of cy-tRNAs. We found that both mitochondrial ISC and cytosolic CIA proteins for the Fe/S protein maturation system in yeast are involved in the s 2 modification of cy-tRNAs (7). Therefore, currently unidentified cytosolic Fe/S proteins may be involved somewhere in this process. At present, it remains unclear whether the newly identified Tuc1, Tuc2, or Uba4 proteins are Fe/S proteins. Further experiments are needed to elucidate molecular details on the involvement of this unique ubiquitin-related system in the s 2 modification of cy-tRNAs.