Evidence for the Transfer of Sulfane Sulfur from IscS to ThiI during the in Vitro Biosynthesis of 4-Thiouridine in Escherichia coli tRNA*

IscS from Escherichia coli is a cysteine desulfurase that has been shown to be involved in Fe-S cluster formation. The enzyme converts L -cysteine to L -alanine and sulfane sulfur (S 0 ) in the form of a cysteine persulfide in its active site. Recently, we reported that IscS can do-nate sulfur for the in vitro biosynthesis of 4-thiouridine (s 4 U), a modified nucleotide in tRNA. In addition to IscS, s 4 U synthesis in E. coli also requires the thiamin biosyn-thetic enzyme ThiI, Mg-ATP, and L -cysteine as the sulfur donor. We now report evidence that the sulfane sulfur generated by IscS is transferred sequentially to ThiI and then to tRNA during the in vitro synthesis of s 4 U. Treat-ment of ThiI with 5-((2-iodoacetamido)ethyl)-1-aminon-apthalene sulfonic acid (I-AEDANS) results in irreversible inhibition, suggesting the presence of a reactive cysteine that is required for binding and/or catalysis. Both ATP and tRNA can protect ThiI from I-AEDANS inhibition. Finally, using gel shift and protease protection assays, we show that ThiI binds to unmodified E. coli tRNA Phe . Together, these results suggest that ThiI is a recipient of S 0 from IscS and catalyzes the ultimate sulfur transfer step in the biosynthesis of s 4 U. polyacrylamide gels, electrophoresis performed boric

IscS from Escherichia coli is a cysteine desulfurase that has been shown to be involved in Fe-S cluster formation. The enzyme converts L-cysteine to L-alanine and sulfane sulfur (S 0 ) in the form of a cysteine persulfide in its active site. Recently, we reported that IscS can donate sulfur for the in vitro biosynthesis of 4-thiouridine (s 4 U), a modified nucleotide in tRNA. In addition to IscS, s 4 U synthesis in E. coli also requires the thiamin biosynthetic enzyme ThiI, Mg-ATP, and L-cysteine as the sulfur donor. We now report evidence that the sulfane sulfur generated by IscS is transferred sequentially to ThiI and then to tRNA during the in vitro synthesis of s 4

U. Treatment of ThiI with 5-((2-iodoacetamido)ethyl)-1-aminonapthalene sulfonic acid (I-AEDANS) results in irreversible inhibition, suggesting the presence of a reactive cysteine that is required for binding and/or catalysis.
Both ATP and tRNA can protect ThiI from I-AEDANS inhibition. Finally, using gel shift and protease protection assays, we show that ThiI binds to unmodified E. coli tRNA Phe . Together, these results suggest that ThiI is a recipient of S 0 from IscS and catalyzes the ultimate sulfur transfer step in the biosynthesis of s 4 U.
Sulfur trafficking within the cell remains a poorly understood area. The biochemical steps for sulfur incorporation into biotin, lipoic acid, and thiamin, as well as macromolecules such as protein Fe-S clusters and thionucleosides in tRNA, have not been fully elucidated. The biosynthesis of thiamin and 4-thiouridine (s 4 U) 1 in tRNA has been linked by the observation that some mutants lacking s 4 U in their tRNA were also auxotrophic for thiamin (1). In two independent studies, the thiI gene was isolated and found to complement both s 4 U (2) and thiamin (3) deficient mutants in Escherichia coli and Salmonella typhi-murium, respectively. We recently reported that IscS, a NifS homolog from E. coli, can mobilize sulfur from cysteine for the in vitro synthesis of s 4 U in E. coli tRNA (4). The formation of s 4 U requires IscS and ThiI as well as Mg-ATP and L-cysteine as the sulfur donor. Using UV spectroscopy and HPLC analysis, we have confirmed that s 4 U is the product of the in vitro reaction (4). All tRNA substrates lacking a uridine in position 8, the natural site of s 4 U modification, are completely inactive in our system, suggesting that the modification is restricted to this position. 2 We have also found that IscS, among multiple enzymes in E. coli with cysteine desulfurase activity, is specifically involved in sulfur transfer to tRNA in vitro (4).
Early work by Lipsett and co-workers (5) showed that two factors, A and C, mediated s 4 U synthesis. Two genes, nuvA and nuvC, which were later identified in E. coli near UV-resistant mutants, were reported to lack s 4 U and were complemented by Factors A and C, respectively (1,6). IscS requires PLP for activity analogous to Factor C; however, the location of iscS in the E. coli genome (57.3 min) is different from that reported for nuvC (42-46 min). The thiI gene was previously reported to complement a nuvC mutation (7), however, ThiI does not bind PLP (4,7). Thus, the assignment of iscS and thiI as nuvA or nuvC remains unclear.
IscS was previously characterized as a cysteine desulfurase that catalyzes the formation of alanine and sulfane sulfur (S 0 ) from cysteine (8). During this reaction, a cysteine persulfide is formed in the IscS active site. In the presence of excess DTT or other thiols, the products of the reaction are alanine and sulfide. Flint (8) has proposed a scheme in which IscS functions as a S 0 donor for Fe-S cluster synthesis. We have found that ThiI stimulates the cysteine desulfurase activity of IscS 2-fold (4). Gel mobility shift studies (reported herein) show that ThiI and not IscS binds to unmodified E. coli tRNA Phe . We therefore hypothesized that the persulfide sulfur formed at the active site of IscS may be transferred to ThiI for reaction with an activated form of uridine 8 in the tRNA to form s 4 U. To test this hypothesis, we sought to determine the nature of sulfur transfer in the biosynthesis of s 4 U in E. coli tRNA. In this paper, we report evidence that ThiI binds tRNA, accepts sulfur from IscS, and functions as the ultimate sulfur donor for s 4 U formation.

MATERIALS AND METHODS
General-L-[ 35 S]Cysteine and [␣-32 P]ATP were purchased from NEN Life Science Products. TPCK-treated trypsin, ATP, cysteine, and all other biochemicals were from Sigma. Overproduced and purified ThiI and IscS were prepared as described previously (4). Unmodified tRNA Phe was prepared by linearization of pCF23 (a gift from Dr. Rachel Green, Johns Hopkins Medical School, Baltimore, MD) with BstNI and runoff transcription essentially as described (9). E. coli tRNA Phe transcript was labeled with 32 P by substituting ATP with [␣-32 P]ATP in the in vitro transcription reaction mixture.
Gel Mobility Shift Assay-32 P-Labeled tRNA (3.7 M) was incubated with varying amounts (0.9 -7.2 M) of ThiI or IscS in 20 l of buffer A (50 mM Tris-HCl, pH 7.5, 50 mM KCl, 10 mM Mg(OAc) 2 , 1 mM DTT, 1 mM PMSF, 3% glycerol) for 5 min at 37°C. Samples were then applied to 8% native polyacrylamide gels, and electrophoresis was performed using 90 mM Tris/90 mM boric acid and 2.5 mM EDTA at 4°C. The gels were then dried using gel drying frames and exposed to x-ray film.
Tryptic Digestion-ThiI (0.13 mg/ml) in buffer A minus PMSF was digested at 37°C with TPCK-trypsin at a ThiI/trypsin ratio of 83:1 (w/w). In the protection experiments, ThiI was preincubated with 1 mM ATP or 1 mM cysteine or tRNA (at a molar ratio of 1:13) at 37°C for 5 * This work was supported by National Institutes of Health Grant GM57002 (to C. T. L.) and by a grant from the University of Wisconsin Graduate School. 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.
min in buffer A minus PMSF, and then reaction mixtures were digested as above. Aliquots were removed at various time intervals and analyzed by SDS-PAGE. Gels were quantitated using a densitometer and Image-Quant software (Molecular Dynamics). The intensity of the bands was quantitatively analyzed by the volume quantitation method, and the object average option was used for background correction.

RESULTS AND DISCUSSION
tRNA Binding Studies-We first looked for tRNA binding by each of the protein factors involved in the biosynthesis of s 4 U using a gel mobility shift assay. Fig. 1 shows the effect of increasing concentrations of either ThiI (lanes 2-5) or IscS (lanes 8 -11) on the mobility of 32 P-labeled unmodified E. coli tRNA Phe . Analysis by native PAGE indicates that only ThiI is able to shift the mobility of the tRNA substrate. The addition of an excess of unlabeled tRNA (25-50-fold) could block the observed mobility shift by ThiI (lanes 6 -7). To corroborate the gel shift results, we performed protease protection experiments on ThiI (Fig. 2). TPCK-trypsin degrades ThiI into a major frag-ment with an approximate molecular mass of 45 kDa (designated L in Fig. 2); tRNA protects ThiI from proteolysis (lower left panel). Neither cysteine nor ATP protects ThiI from tryptic degradation at this site. However, ATP lowers the rate of proteolysis of both ThiI and the 45-kDa fragment (Fig. 2, lower  right panel). Quantitative analysis of the protein bands in Fig.  2 revealed that ThiI proteolysis was slowed by 2-fold in the presence of ATP. ThiI tryptic digests containing primarily the 45-kDa fragment are devoid of tRNA sulfurtransferase activity (data not shown). Both the gel shift analysis and protease protection experiments clearly show that ThiI binds to tRNA.
Observation of Transfer of 35 S from IscS to ThiI during s 4 U Formation-We used 35 S-labeled cysteine in an attempt to follow the transfer of sulfur during s 4 U synthesis in vitro. Reactions containing an increasing amount of ThiI, with or without ATP and tRNA, were incubated for 10 min and then quenched with SDS buffer and subjected to electrophoresis (Fig. 3). Each lane in Fig. 3 corresponds to a separate reaction mixture. 35 S-Labeled bands were visualized by PhosphorImager analysis (Fig. 3, B and C). As expected, only IscS can be labeled with [ 35 S]cysteine alone. In reactions with increasing concentrations of ThiI (Fig. 3B, lanes 3-6), we observed a 35 S-labeled band of increasing intensity with mobility identical to ThiI. The proportion of 35 S label comigrating with IscS was observed to decrease with an increasing ThiI concentration (Fig. 3B, lanes 4 -6). However, 35 S label on IscS increased 47% in the presence of small amounts of ThiI (Fig. 3B, lane 3). Interestingly, this increase was not observed in the absence of ATP (Fig. 3C, lane 3).
When the tRNA substrate was included in the reaction mixture (Fig. 3B, lane 7), the label corresponding to ThiI and IscS was greatly diminished, and a high intensity band that correlates with the position of the tRNA was observed. Quantitation of the labeled bands shown in Fig. 3B reveals that the amount of label on ThiI and IscS was decreased by 81 and 98%, respectively, in the presence of tRNA (Fig. 3D). Together, these labeling patterns suggest that ThiI is covalently modified with sulfur mobilized by IscS and catalyzes the ultimate sulfur transfer in the synthesis of s 4 U. Under previously reported sulfurtransferase assay conditions (4), ThiI and IscS catalyze the transfer of 0.2 mol of 35 S to 1 mol of tRNA. HPLC analysis of 35 S-labeled tRNA nuclease digests confirms that s 4 U accounts for Ͼ85% of the 35 S label. In the present experiment Role of ThiI in the Biosynthesis of s 4 U 10728 (Fig. 3), the radioactivity observed in ThiI is 1% that of the label on tRNA. The higher intensity of the label in the tRNA results from the greater chemical stability of s 4 U relative to the putative protein persulfides. An additional unidentified labeled band with electrophoretic mobility in between IscS and tRNA was also observed (Fig. 3 B, lane 7). Quantitation of this labeled band gave a value 0.1% of the label in tRNA. Because we do not see this band in the Coomassie-stained gel (Fig. 3A, lane 7), and it is visible by staining with ethidium bromide (not shown), we conclude that it is a nucleic acid contaminant present in the tRNA substrate.
We observed no labeling of the negative control protein BSA (Fig. 3B, lanes 8 and 9), which is known to contain reactive cysteine residues, or of carbonic anhydrase (Fig. 3B, lane 10) when each is substituted for ThiI in the reaction mixture. As expected, IscS, in the absence of ThiI, does not catalyze sulfur transfer to tRNA (Fig. 3B, lanes 9 and 10). In the absence of ATP, the amount of label transferred to tRNA was reduced by more than 98%, and the intensity of the labeled band corresponding to ThiI remained unchanged (Fig. 3C, lane 7, and 3E). As expected, ATP was not required for the transfer of 35 S from L-cysteine to IscS and then to ThiI (Fig. 3C, lanes 2-6).
The persulfides were expected to be labile to reducing agents, and therefore we removed mercaptoethanol from the electrophoresis sample buffer. However, the gel samples received 0.5 mM DTT from the reaction mixtures, which is required for ThiI activity. It is perhaps unexpected that we observed 35 S-labeled ThiI and IscS in the gel under denaturing conditions. Because the DTT from the reaction mixture is rapidly diluted as the proteins enter the stacking gel, it is possible that a fraction of the persulfides survives long enough to be separated in the gel matrix. Flint (8) has found that the reduction of the IscS persulfide in the absence of DTT is slow, despite the presence of 2.5 mM cysteine as substrate. Attempts to exclude DTT completely in our experiments resulted in the appearance of additional high molecular weight protein bands in the gel, presumably because of oligomerization or aggregation. We also found that the presence of 5 mM DTT in the reactions abolished label on either IscS or ThiI in our gel analysis. Thus, the 35 S-label is sensitive to higher DTT concentrations, which is consistent with the proposed persulfide functionality. We removed [ 35 S]cysteine from the reaction mixture after incubation using size exclusion chromatography and analyzed the separated large molecule fraction by SDS-PAGE followed by PhosphorImager analysis. This shows that the separated protein fraction contains 35 S-labeled ThiI, and the addition of this fraction to tRNA in the presence of Mg-ATP results in the transfer of label to tRNA (data not shown). This result suggests that 35 S-modified ThiI lies on the direct pathway to s 4 U formation in tRNA.

Inhibition of ThiI Catalyzed Formation of s 4 U by I-AE-DANS-
The reaction of ThiI with the fluorescent thiol-specific alkylating agent I-AEDANS leads to irreversible inactivation of the enzyme. Fig. 4 shows that ThiI can be protected from I-AEDANS inhibition by adding ATP and, to a lesser extent, tRNA into the reaction mixture. As expected, GTP, which is inactive as a substrate for s 4 U formation, does not protect ThiI from I-AEDANS inhibition. Because tRNA and ATP have a protective effect, we are confident that inactivation of IscS, which is present in saturating amounts, is not the cause for the observed inhibition. Exposure of SDS gels to UV light after reaction with I-AEDANS shows a fluorescent band comigrating only with ThiI. An alignment of amino acid sequences of ThiI from E. coli and homologs from a range of other organisms shows that Cys 344 is present in all of the analyzed sequences (data not shown). These results support the possibility that ThiI possesses a reactive cysteine that can accept S 0 for the synthesis of s 4 U. We are currently attempting to identify the modified cysteine residue(s), as well as prepare site-directed mutants for analysis both in vitro and in vivo.
General Scheme for Sulfur Trafficking in the Synthesis of s 4 U-Based on the model proposed by Flint (8) and the evidence presented in our current work, we propose a scheme for sulfur mobilization and transfer in the biosynthesis of s 4 U in E. coli (Fig. 5). The initial mobilization of sulfur from cysteine results in the formation of S 0 covalently bound to a cysteine in the active site of IscS as a persulfide (IscS-SSH). This persulfide sulfur is then transferred, presumably to a cysteine residue on ThiI, to give the ThiI persulfide (ThiI-SSH). Finally, in the presence of ATP, the persulfide sulfur from ThiI-SSH is mobilized and transferred to tRNA. The exact mechanism of the mobilization in vivo is not yet clear, but it could involve oxidation of the protein via internal reduction of the persulfide by a cysteine residue to give a disulfide that is reduced later. Alternatively, an external reducing agent could be required to directly mobilize the sulfur from the persulfide. Because native ThiI has not yet been isolated, it is not possible to ascertain the oxidation state of the naturally isolated enzyme. It is interesting to note that in Flint's (8) original isolation of IscS in the absence of reducing agents, the enzyme was found to be part of a mixed disulfide with the phosphopantetheine moiety of acyl carrier protein.
The role of ATP in s 4 U synthesis has not yet been elucidated. Lipsett and colleagues (5) proposed that Factor A utilized Mg-ATP to synthesize a tRNA-derived unstable activated intermediate, which is then converted to s 4 U by Factor C in the presence of cysteine and PLP. Based on sequence homology searches with putative ThiI homologs from various organisms, we have proposed that ThiI uses Mg-ATP to activate the oxygen at the 4-position of uridine for attack by a sulfur nucleophile (4). Our observation that ATP protects ThiI from inhibition by I-AEDANS and lowers the rate of ThiI trypsinolysis is consistent with this suggestion.
The observed in vitro shuttling of S 0 from IscS to ThiI may represent a general strategy in vivo for efficient sulfur transfer in a number of metabolic pathways (10). It is possible that IscS or other NifS homologs in E. coli transfer S 0 to a variety of other acceptors, including Fe-S cluster-containing enzymes and other sulfurtransferases. We have recently found that E. coli iscS Ϫ mutants are viable but are unable to synthesize s 4 U or thiamin. 3 It is thus likely that IscS initiates sulfur mobilization for the synthesis of both s 4 U and thiazole in vivo. Current efforts are focused on a full characterization of the iscS Ϫ phenotype in E. coli and further elucidation of the scope of IscSinitiated sulfur transfer in thionucleotide biosynthesis.
Addendum-While this paper was in review, a paper was published by Mueller and Palenchar (11) showing that ThiI contains a unique P-loop motif ( 185 SGGXDS 190 ) that is common to the pyrophosphate synthetase family of enzymes that catalyze the adenylation and substitution of carbonyl oxygens. We note that ThiI contains significant ATPase activity independent of tRNA, 3 as was earlier reported for Factor A (5). Thus far, we have only found evidence of ADP formation. However, it is possible that the tRNA-independent ATP hydrolysis is obscuring our ability to observe AMP formation.