The role of the cysteine residues of ThiI in the generation of 4-thiouridine in tRNA.

The enzyme ThiI is common to the biosynthetic pathways leading to both thiamin and 4-thiouridine in tRNA. We earlier noted the presence of a motif shared with sulfurtransferases, and we reported that the cysteine residue (Cys-456 of Escherichia coli ThiI) found in this motif is essential for activity (Palenchar, P. M., Buck, C. J., Cheng, H., Larson, T. J., and Mueller, E. G. (2000) J. Biol. Chem. 275, 8283-8286). In light of that finding and the report of the involvement of the protein IscS in the reaction (Kambampati, R., and Lauhon, C. T. (1999) Biochemistry 38, 16561-16568), we proposed two mechanisms for the sulfur transfer mediated by ThiI, and both suggested possible involvement of the thiol group of another cysteine residue in ThiI. We have now substituted each of the cysteine residues with alanine and characterized the effect on activity in vivo and in vitro. Cys-108 and Cys-202 were converted to alanine with no significant effect on ThiI activity, and C207A ThiI was only mildly impaired. Substitution of Cys-344, the only cysteine residue conserved among all sequenced ThiI, resulted in the loss of function in vivo and a 2700-fold reduction in activity measured in vitro. We also examined the possibility that ThiI contains an iron-sulfur cluster or disulfide bonds in the resting state, and we found no evidence to support the presence of either species. We propose that Cys-344 forms a disulfide bond with Cys-456 during turnover, and we present evidence that a disulfide bond can form between these two residues in native ThiI and that disulfide bonds do form in ThiI during turnover. We also discuss the relevance of these findings to the biosynthesis of thiamin and iron-sulfur clusters.

The metabolism of many sulfur-containing biomolecules remains incompletely understood. Among the metabolic pathways requiring further elucidation are those leading to ironsulfur clusters (1)(2)(3)(4)(5), biotin (6 -8), molybdopterin (9), lipoic acid (10), thiamin (8,11), and sulfur-containing bases in RNA (12). The sulfur-containing nucleosides include 4-thiouridine (s 4 U), 1 which is found at position 8 of some bacterial tRNA (Fig. 1) and serves as a photosensor for near-UV light (12). The s 4 U undergoes a photoactivated 2 ϩ 2 cycloaddition with cytidine 13 when the tRNA is exposed to light of a wavelength similar to the 334 nm absorbance maximum of s 4 U (13-15). The resulting cross-linked tRNA are poor aminoacylation substrates (16), and the accumulation of uncharged tRNA arrests growth by triggering the stringent response (17,18). Lipsett and co-workers (19,20) investigated the enzymology of s 4 U biosynthesis in Escherichia coli and reported that the overall reaction utilized cysteine as the sulfur donor and required ATP as a substrate. Lipsett and co-workers (20) concluded that two enzymes were required and that one of them also plays a role in thiamin biosynthesis and requires the cofactor PLP for activity (21,22). By using a genetic screen based on the role of s 4 U as a photosensor (18,(22)(23)(24), the genetic loci of two genes required for s 4 U biosynthesis (named nuvA and nuvC) were mapped (22)(23)(24).
By using the same genetic screen, we identified the gene thiI as essential for s 4 U formation (25) shortly after Downs and co-workers (26) identified the same gene as essential for thiamin biosynthesis. We have cloned and overexpressed thiI from E. coli (25). Kambampati and Lauhon (27) isolated another enzyme, IscS, that sufficed along with ThiI for in vitro generation of s 4 U in tRNA, and they have since confirmed (28) that iscS mutants lack s 4 U and are thiamin auxotrophs. 2 IscS is a NifS-like protein that functions in iron-sulfur cluster formation, is PLP-dependent, and proceeds through an enzymic persulfide intermediate (29,30). Based on sequence alignments that revealed similarity between the segment of ThiI around Cys-456 and other sulfurtransferases, we investigated the importance of Cys-456 for the function of ThiI and found that C456A ThiI was inactive both in vivo and in vitro (31). The report of the role of IscS in s 4 U biosynthesis fit nicely with our independent conclusion that ThiI would proceed through a persulfide intermediate on Cys-456 by providing a source of S 0 to form that persulfide group. Since our report, Kambampati and Lauhon (32) have established that sulfur flows from cysteine to IscS to ThiI to s 4 U in tRNA.
Based on all the evidence, we proposed two alternative mechanisms for the biosynthesis of s 4 U (Fig. 2), and both immediately suggest a role for another cysteine residue in ThiI (31). E. coli ThiI has four cysteine residues other than Cys-456, and sequence alignments of known ThiI proteins (all from prokaryotes) reveal that only one cysteine residue, Cys-344 in the E. coli enzyme, is completely conserved, 3 suggesting that this amino acid serves a critical role. We now report that this supposition is borne out by our investigations of the role of the cysteine residues in ThiI. * This work was supported in part by National Institutes of Health Grant GM59636. 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 1 The abbreviations used are: s 4 U, 4-thiouridine; PLP, pyridoxal 5Јphosphate; DTT, dithiothreitol; DTNB, 5,5Ј-dithiobis(2-nitrobenzoic acid) (Ellman's reagent); IscS⅐His 6 , IscS bearing the C-terminal His 6 tag encoded by pET29b; PCR, polymerase chain reaction; Ni-NTA, nickelnitrilotriacetic acid. 2 The congruence of nuvA and nuvC to thiI and iscS remains unclear due to several apparent discrepancies between the chromosomal locations of the genes and the phenotypes of nuvA and nuvC mutants relative to thiI and iscS mutants. 3 (33). A Higgins analytical CLIPEUS C18 5-m column (50 ϫ 4.6 mm) was purchased from Bodman Industries (Aston, PA). The tRNA substrate was the in vitro transcript of E. coli tRNA Phe , and we described the preparation in detail elsewhere (34). Ni-NTA superflow resin, QIAquick Gel Extraction kits, and QIAprep Spin Miniprep kits were purchased from Qiagen (Chatsworth, CA). A BioCad SPRINT perfusion chromatography system (PE Biosystems, Foster City, CA) was used for protein purification by chromatography over Poros 20 HS resin (PE Biosystems). Oligonucleotides (OPC-purified grade) were purchased from The Great American Gene Co. (Ramona, CA). QuikChange TM site-directed mutagenesis kits were purchased from Stratagene (La Jolla, CA), and a Robocycler Gradient 96 Thermocycler (Stratagene) was used for the PCR component of the site-directed mutagenesis. DNA sequencing was performed either at the University of Delaware Cell Biology Core Facility using a Long Readir 4200 DNA sequencer (Li-Cor, Inc., Lincoln, NE) or at the Delaware Biotechnology Institute and University of Delaware Center for Agricultural Biotechnology core facility using an ABI Prism model 377 DNA sequencer (PE Biosystems). The complete sequences of all plasmids described in this publication are posted at www.udel.edu/chem/mueller. Generation, Overexpression, Purification, and Physical Characterization of Altered ThiI-To substitute cysteine residues in ThiI with alanine, the QuikChange TM mutagenesis protocol was used with appropriate primers (Table I) as we described previously (34). All of the ThiI variants described in this paper contain the 20-amino acid N-terminal His 6 tag provided by pET15b, but all amino acid positions are numbered in terms of native ThiI (with no N-terminal His 6 tag). The parent plasmid was pBH113S, which has N49S thiI in pET15b; as described previously (31), this N49S mutation arose spontaneously during prop-agation of pBH113, which is wild-type thiI in pET15b (25). All of the ThiI variants described here are N-terminally His 6 -tagged and have the N49S change; for the sake of simplicity, we will refer to them as "ThiI" (specified as either wild type or by the substituted cysteine residue).
The overexpression, purification, and storage of each altered ThiI was accomplished using the methods that we have described in detail elsewhere (35). Protein expression was induced by addition of isopropyl-␤-D-thiogalactopyranoside to the growth medium, and the cells were harvested 3 h later. Chromatography of cell extracts over Ni-NTA resin yielded essentially homogeneous ThiI, which was changed into appropriate buffer and used immediately or precipitated by addition of solid ammonium sulfate (to nominal 75% saturation) for storage at 4°C. By using the methods that we have described elsewhere (35), extinction coefficients at 280 nm were determined, and far-UV CD spectra were recorded for each altered ThiI. The generation, isolation, and properties of the C456A ThiI have been described previously (31).
In Vivo Characterization of Altered ThiI-The effect of substituting the cysteine residues of ThiI on its activity in vivo was assessed in two ways. First, the thiI mutant VJS2890(DE3) was transformed with a plasmid encoding an altered ThiI, and the transformants were subjected to near-UV screening. The screening of cells producing each altered ThiI was performed twice, except for the cells expressing C202A and C207A ThiI, which were performed four times; qualitatively identical results were obtained in every case. Second, the UV spectrum of tRNA isolated from saturated cultures of the transformants was recorded and examined for the characteristic peak due to s 4 U ( max ϳ334 nm). The procedures for both of these in vivo characterizations have been described in detail elsewhere (25).
Assay for s 4 U Generation-This assay is essentially the one that we have described previously (31) except that recombinant IscS⅐His 6 was substituted for cell extracts. The generation of s 4 U was monitored by following the incorporation of 35  ]s 4 U was quantitated by the method that we have described in detail elsewhere (31). In this method, unreacted L-[ 35 S]cysteine is removed by size-exclusion chromatography, and the tRNA is digested to nucleosides, which are resolved by reverse phase high pressure liquid chromatography (36,37); the level of 35 S in the s 4 U is determined by in-line scintillation counting.
Cloning of IscS-Genomic DNA was purified from E. coli JM109 cells (Promega) using the Wizard Genomic DNA Purification protocol (Promega) according to the manufacturer's instructions. PCR amplification of iscS in the genomic DNA was achieved using appropriate primers (Table I), Taq DNA polymerase (Promega), and HotStart tubes (Molecular Bio-Products Inc., San Diego) according to the HotStart protocol. The primers (Table I) specified an NdeI site that includes the start ATG of iscS and an EcoRI site that follows the TAA that terminates iscS. A Robocyler Gradient 96 thermal cycler (Stratagene) was used for the PCR, as described elsewhere (25). The PCR product was purified by agarose gel electrophoresis and recovered using the QIAquick gel extraction protocol (Qiagen). The isolated PCR product was ligated into pGEM-T using the pGEM-T Easy Vector System II (Promega) as specified by the manufacturer, and restriction analysis confirmed the generation of the target plasmid. This plasmid was digested with NdeI and EcoRI (New England Biolabs, Beverly, MA), and the iscS fragment was isolated by agarose gel electrophoresis and ligated into pET29b (Novagen) that had been opened with the same enzymes. The success of the construction was confirmed by restriction analysis, and the plasmid was named pBH400.
Sequencing of iscS in pBH400 revealed two discrepancies from the published sequence as follows: a T 3 C transition in codon 244 that results in the substitution of proline for leucine and a deletion of the T in the penultimate codon that moves the stop codon specified by the "reverse" PCR primer out of frame and brings the C-terminal His 6 tag encoded by pET29b into frame. IscS bearing the C-terminal His 6 tag will be denoted IscS⅐His 6 . Both alterations of iscS in pBH400 were corrected using the QuikChange TM protocol (Stratagene) with appropriate primers (Table I) as we have described previously (34). The plasmid encoding L244P IscS (with no C-terminal His 6 tag) is pBH401; the plasmid encoding IscS⅐His 6 (with Leu-244 restored) is pBH402; the plasmid encoding native IscS (with Leu-244 restored and no C-terminal His 6 tag) is pBH403.

FIG. 1. 4-Thiouridine in tRNA.
A, the overall reaction, which is catalyzed by sequential action of the enzymes IscS and ThiI; cysteine, ATP, and uridine-8 of certain bacterial tRNA are the substrates. B, the location of s 4 U in bacterial tRNA; s 4 U occurs at position 8, which is shown by a filled circle in the "cloverleaf" structure of tRNA.
Overexpression and Purification of IscS⅐His 6 -The procedure for overexpression and purification of IscS⅐His 6 and L244P IscS⅐His 6 was identical to that for the overexpression and purification of ThiI (35) except that cultures of BLR(DE3) pLysS/pBH402 or BLR(DE3) pLysS/ pBH400 were grown and induced in LB medium containing kanamycin (30 g/ml) and chloramphenicol (34 g/ml). Isolated IscS⅐His 6 (ϳ40 mg/liter of culture) was dialyzed against 50 mM potassium phosphate buffer, pH 7.5, containing magnesium chloride (5 mM), potassium chloride (100 mM), and EDTA (0.1 mM). The enzyme remains stable for weeks to months when stored at 4°C in the same buffer at moderate concentration (3-4 mg/ml). By comparing A 280 nm and the concentration of IscS measured by the biuret assay (35), we calculated ⑀ 280 nm ϭ 25,400 M Ϫ1 cm Ϫ1 , which we now use to determine the concentration of IscS⅐His 6 .
Joint Overexpression of ThiI and IscS-BLR(DE3) pLysS/pBH402 (encoding IscS⅐His 6 ) and BLR(DE3) pLysS/pBH403 (encoding native IscS) was transformed with pBH113 (encoding wild-type ThiI) using the TransformAid TM protocol (MBI Fermentas, Hanover, MD). The joint overexpression of ThiI and IscS was accomplished as described for the overexpression of either protein alone. ThiI overexpressed with native IscS was cleanly separated from the latter by chromatography over Ni-NTA resin. ThiI overexpressed with IscS⅐His 6 was co-purified with the latter by chromatography over Ni-NTA resin; the two proteins were then separated by chromatography over Poros 20 HS resin, eluting with a linear gradient (over 13 column volumes) of potassium chloride (0 -1.5 M) in 50 mM potassium phosphate buffer, pH 7.5, containing DTT (1 mM). The UV-visible spectrum was recorded after exchange of isolated ThiI into 50 mM potassium phosphate buffer, pH 7.5, containing magnesium chloride (5 mM), potassium chloride (100 mM), and EDTA (0.1 mM) by size-exclusion chromatography over a spin column of Sephadex G-25 equilibrated in the same buffer.
DTNB Titrations-Samples of ThiI were subjected to DTNB titrations under both native and denaturing conditions. The enzyme samples were fresh preparations that had been isolated using buffers to which no reductant had been added. Any thiol-bearing components of the cell extracts (either small or macro-molecules) should have been removed by the chromatography over Ni-NTA and the exchange (by size-exclusion chromatography or dialysis) of ThiI-bearing column fractions into 50 mM potassium phosphate buffer, pH 7.5, containing magnesium chloride (5 mM), potassium chloride (100 mM), and EDTA (0.1 mM). The protein concentration ranged from 4.7 to 16.6 M, and DTNB (20 eq relative to ThiI) was added as a solution (10 mM) in the same buffer. Control incubations were prepared by adding the same amount of DTNB to the same buffer (with no protein). The final mixtures (400 l) were incubated 20 -30 min at room temperature, and the intensely yellow dianion of 5-thio-2-nitrobenzoic acid was quantified by subtract-FIG. 2. Proposed mechanisms for s 4 U generation. Based on the data presented here, we now consider it likely that Cys-344 of ThiI fulfills the role that we previously (31) ascribed to an unspecified thiol group. A, mechanism providing for the direct nucleophilic attack on an activated uridine species by the terminal sulfur of the persulfide on Cys-456 of ThiI. B, mechanism in which ThiI functions to generate hydrogen sulfide, which subsequently serves as the nucleophile to displace the activated oxygen of uridine 8. All events occur at an active site of ThiI, and attachment of the cysteine residues to ThiI is denoted with a squiggle. General acids are denoted as A, and general bases are denoted as B; tRNA denotes the rest of the tRNA molecule that contains the depicted uridine residue. The activation of uridine by adenylation is postulated based on the demonstrated importance of a sequence motif shared between ThiI and enzymes that adenylate carbonyl oxygens to activate them for subsequent nucleophilic attack (35) and by preliminary determinations that AMP is a product of s 4 U generation (C. Buck, unpublished observations).
ing the A 412 nm of the control incubation from the A 412 nm of each sample and using an extinction coefficient of 13,600 M Ϫ1 cm Ϫ1 (38). To denature the wild-type ThiI, solid guanidinium chloride (ϳ0.23 g, ϳ2.4 mmol) was added to the samples and dissolved to achieve a final guanidinium concentration of ϳ4.6 M; the volume and A 412 nm values were measured after 20 min at room temperature. All titrations were performed in duplicate except for the C456A ThiI under native conditions, which was performed in quadruplicate. Titrations of ThiI with 1 eq of DTNB were performed under native conditions as described for titration with 20 eq of DTNB. All replicate experiments returned values within 5% of each other.
Assay for Disulfide Bond Formation during Turnover-To test for disulfide bond formation in ThiI during turnover, the s 4 U generation assay was run without DTT, with a reduced concentration of cysteine (which can reduce disulfide bonds), and in phosphate buffer to allow separation of IscS and ThiI by cation-exchange chromatography. The assay (1 ml) consisted of 150 mM potassium phosphate buffer, pH 8. After 5 h at 37°C, ThiI was separated from IscS and other reaction components by chromatography over Poros 20 HS resin as described above except that DTT was omitted from the buffers. The ThiI eluted in two fractions, which were combined (1 ml; 2-3 M), concentrated in a Microcon-10 device, and subjected to DTNB titrations under denaturing conditions as described above. As a control for oxidation under these conditions, ThiI was also incubated in the absence of tRNA and cysteine and subjected to the same work up. To measure the extent of s 4 U formation under these conditions, a duplicate reaction containing [ 35 S]cysteine was run in parallel; 0.73 Ϯ 0.02 eq (relative to ThiI) of s 4 U was generated. Assuming that turnover results in the formation of one disulfide bond in ThiI, the expected number of free thiol groups in ThiI isolated after turnover is (0.73) (3 thiol groups) ϩ (0.27) (5 thiol groups) ϭ 3.5 thiol groups.

RESULTS
Cloning and Overexpression of IscS-The PCR-based cloning of iscS from E. coli genomic DNA resulted in the generation of an IscS overexpression plasmid based on pET29b (pBH400). Sequencing revealed a deletion in the penultimate codon of iscS during the PCR, resulting in a frameshift that fused a Cterminal His 6 tag to IscS (denoted IscS⅐His 6 ). In addition, a proline residue was encoded by codon 244 rather than the leucine specified by the sequence in the genomic data base. Overexpressed and purified L244P IscS⅐His 6 proved fully competent for in vitro assay of ThiI activity. Site-directed mutagenesis was used to generate overexpression plasmids for IscS with Leu-244 restored (pBH402) and with both Leu-244 and the penultimate codon restored (pBH403), which brings the stop codon back into frame and eliminates the C-terminal His 6 tag. IscS⅐His 6 and L244P IscS⅐His 6 behaved identically in our assays of ThiI activity; since they are easily purified, we used them in preference to native IscS (L244P IscS⅐His 6 was used in earlier experiments, and IscS⅐His 6 was used in later experiments).
Generation, Overexpression, and Physical Characterization of Altered ThiI-Site-directed mutagenesis afforded ThiI with each cysteine residue singly changed to alanine. All of the ThiI variants discussed below also contain a serine in place of the wild-type Asn-49. We discovered this mutation only after the generation of the altered ThiI, and we determined that the N49S ThiI is insignificantly (15-20%) more active than the wild-type enzyme (with Asn-49) in our current assay system. We did not restore Asn-49 in each of the variant ThiI, for the existing protein context provided an internally consistent evaluation of the effects of altering the cysteine residues. ThiI variants with one cysteine residue replaced with alanine were identical to wild-type ThiI regarding overexpression, purification, and storage. Extinction coefficients at 280 nm were determined for each altered ThiI, and they varied less than 5% from that of wild-type enzyme (63,100 M Ϫ1 cm Ϫ1 ) as follows: C108A, 64,500 M Ϫ1 cm Ϫ1 ; C202A, 64,000 M Ϫ1 cm Ϫ1 ; C207A, 66,000 M Ϫ1 cm Ϫ1 ; C344A, 63,900 M Ϫ1 cm Ϫ1 . The far-UV CD spectrum of each altered ThiI was essentially identical to the spectrum of wild-type enzyme (data not shown). Together, these data indicate that the substitutions of alanine for cysteine did not disrupt the global protein fold.
In Vivo Characterization of Altered ThiI-The thiI mutant E. coli VJS2890(DE3) (33) was transformed with plasmids expressing wild-type and altered ThiI, and the transformants were screened for near-UV sensitivity. VJS2890(DE3) is near-UV-resistant because it lacks a functional thiI and so fails to generate s 4 U in tRNA (the photosensor). VJS2890(DE3) was fully complemented by plasmids encoding wild-type, C108A, and C202A 4 ThiI, but VJS2890(DE3) harboring the plasmid encoding C344A ThiI retained the near-UV-resistant phenotype. The plasmid encoding C207A ThiI conferred an intermediate phenotype; only one-third to one-half the number of col-

Role of Cysteine Residues in ThiI
onies appeared on the near-UV-exposed side as appeared on the masked side of the plate, and all of the colonies on the exposed side were very small; after 24 h further incubation in the dark, both sides of the selection plate were indistinguishable regarding colony number and size. 5 The UV-visible spectra were recorded for tRNA isolated from VJS2890(DE3) harboring plasmids encoding wild-type and altered ThiI (Fig. 3). The levels of s 4 U in tRNA correlate well with the selection results, with little or no s 4 U indicated in the tRNA from VJS2890(DE3) expressing C344A ThiI and a reduced amount of s 4 U in the tRNA from VJS2890(DE3) expressing C207A ThiI. Activity Assays-The assay procedure was changed from our previous report (31) by inclusion of purified IscS⅐His 6 rather than cell extracts. A molar ratio of 4:1 IscS⅐His 6 :ThiI was used in the assays because it affords a linear response in rate when the concentration of ThiI is varied, and additional IscS⅐His 6 does not increase the reaction rate. No s 4 U generation was observed when ThiI was omitted from the reaction mixture, even with elevated concentrations of IscS⅐His 6 (4.8 M). The rates of s 4 U formation catalyzed by wild-type, C108A, and C202A ThiI did not differ appreciably. C207A ThiI displayed a 4-fold reduction in rate, and C344A ThiI was impaired for s 4 U generation by 2700-fold ( Fig. 4 and Table II). C456A ThiI (31) was re-examined in the assay system containing purified IscS⅐His 6 rather than cell extracts, and no s 4 U generation was detected even at elevated concentrations of C456A ThiI (1.2 M) and IscS⅐His 6 (4.8 M) and prolonged incubation times (90 min). Together, the measured activities of IscS alone and the C344A and C456A ThiI make it unlikely that the activity of C344A ThiI is due to contamination with wild-type enzyme expressed from the host cell genome.
Joint Overexpression of ThiI and IscS-BLR(DE3) pLysS cells harboring pBH113 (encoding wild-type ThiI) and either pBH402 (encoding IscS⅐His 6 ) or pBH403 (encoding native IscS) were generated. Induction of both types of cells with isopropyl-␤-D-thiogalactopyranoside resulted in overexpression of both IscS and ThiI, with IscS present at roughly 4-fold greater concentration than ThiI as determined by SDS-polyacrylamide gel electrophoresis analysis. ThiI co-overexpressed with both variants of IscS was isolated, and their UV-visible spectra were recorded. Only the absorbance centered near 280 nm expected for polypeptides was observed; no spectroscopic features indic-ative of iron-sulfur clusters (or any other cofactor) were present.
Quantitation of Thiol Groups in ThiI-Under denaturing conditions (ϳ4.6 M guanidinium chloride), DTNB titrations revealed 5.1 Ϯ 0.2 free thiol groups in wild-type ThiI. Under native conditions, DTNB titrations revealed 3.9 Ϯ 0.3 and 3.3 Ϯ 0.3 reactive thiol groups, respectively, in wild-type and C344A ThiI. The variation from integral values is within the uncertainty of the method. To determine whether or not a disulfide bond forms during turnover, ThiI was isolated after s 4 U generation in the absence of an exogenous reductant other than the substrate cysteine. DTNB titrations under denaturing conditions revealed 4.1 Ϯ 0.3 free thiol groups in ThiI (determined in triplicate). A parallel assay (performed in quadruplicate) of s 4 U generation showed that 73% of ThiI turned over under these reaction conditions, so 3.5 free thiol groups were expected. Although the observed 4.1 free thiol group is significantly higher than the expected value, this discrepancy likely arises from the obligatory presence of cysteine, which can reduce disulfide bonds. As a control for oxidation unrelated to catalytic activity, ThiI was incubated under identical conditions except that cysteine and tRNA were omitted. DTNB titrations under denaturing conditions of the isolated ThiI revealed 4.9 Ϯ 0.2 thiol groups (determined in triplicate), demonstrating that s 4 U generation significantly decreased the number of free thiol groups in ThiI, consistent with the formation of a disulfide bond during turnover.
Evidence of a Disulfide Bond between Cys-344 and Cys-456 -To test whether or not Cys-344 and Cys-456 can form a disulfide bond, titrations of ThiI with 1 eq of DTNB were performed; if a disulfide bond forms, then one expects 2 eq of the chromophore 5-thio-2-nitrobenzoate to be released. DTNB will first react with the free thiol group of one cysteine residue to release 1 eq of chromophore and form a mixed disulfide between the cysteine residue and "half" of DTNB. A second cysteine residue can then displace another equivalent of chromophore to   4. Generation of s 4 U in the activity assay for ThiI. One representative run is shown for each ThiI variant; the data shown here was used along with duplicate runs to generate the data in Table II. The ThiI variants are denoted: Ⅺ, wild-type; E, C108A; ‚, C202A; ᭛, C207A; ϫ, C344A. All of the variants were present at a concentration of 1 nM in the assay except for C344A ThiI, which was present at 1.2 M.
form the purely enzymic disulfide. When wild-type ThiI was treated with 1 eq of DTNB, 1.8 Ϯ 0.2 eq of chromophore were produced. With C456A ThiI, 1.3 Ϯ 0.2 eq of chromophore were generated, whereas 0.9 Ϯ 0.1 eq of chromophore resulted from treatment of C344A ThiI with 1 eq of DTNB. ThiI, then, can readily form a disulfide bond unless either Cys-344 or Cys-456 is replaced with alanine, which is most simply explained by a disulfide bond between Cys-344 and Cys-456. DISCUSSION Based on sequence alignments with sulfurtransferases, we previously investigated (31) the role of Cys-456 of ThiI in the biosynthesis of s 4 U and found that residue to be essential. The catalytically essential Cys-456 is found on a C-terminal extension of ϳ100 amino acids that was then found only in ThiI from E. coli and its close relatives Salmonella typhimurium and Hemophilus influenzae among the organisms for which a ThiI sequence was known (31). Currently, sequences of ThiI from 23 organisms (all prokaryotes) are known, and 8 have the Cterminal extension. 3 Some of the bacteria with the shorter ThiI have open reading frames that are highly similar to the Cterminal extension found in the E. coli ThiI 3 , which suggests that this C-terminal extension may function as an autonomous sulfurtransferase domain. Consistent with our proposed domain structure, Kambampati and Lauhon (32) report that light trypsinolysis cleaves ThiI into two major fragments, the larger of which is inactive for sulfur transfer and of a size (M r ϳ45, 000) consistent with our postulated N-terminal domain.
Our findings concerning Cys-456 coupled with the report of Kambampati and Lauhon (27) that IscS was involved in s 4 U biosynthesis led us to offer two mechanisms for s 4 U biosynthesis (Fig. 2). The key features of these two mechanisms are formation of a persulfide on Cys-456 by trans-persulfidation at the expense of the persulfide on IscS and the liberation of the terminal sulfur of the Cys-456 persulfide as a formal equivalent of S 2rather than S 0 . Subsequently, Kambampati and Lauhon (32) confirmed the transfer of sulfur from free cysteine to IscS to ThiI to tRNA, which led those authors to propose (independently of our results) a sulfur transfer scheme that is similar to our mechanisms although less detailed. Our mechanisms (Fig. 2) both postulate the participation of a thiol group to liberate the formal S 2equivalent (in s 4 U itself or as hydrogen sulfide), which immediately suggests the possible participation by another enzymic cysteine residue. As observed by us 3 and Kambampati and Lauhon (32), sequence alignments reveal that only Cys-344 is absolutely conserved among all ThiI for which sequences are known. To test the role of the conserved Cys-344 and the other three cysteine residues, we separately substituted each one with alanine in E. coli ThiI and measured the effect on ThiI function both in vivo and in vitro.
All of the altered ThiI appear to maintain the global fold of wild-type enzyme as judged by overexpression and storage behavior, extinction coefficients, and far-UV CD spectra that do not vary significantly from those of wild-type ThiI. The substitution of Cys-108 and Cys-202 had negligible effects on the measured activity of ThiI both in vivo (Fig. 3) and in vitro (Table II). In agreement with the prediction from sequence alignments, Cys-344 proved critical for ThiI activity both in vivo (Fig. 3) and in vitro (Table II), although C344A ThiI retains some activity under our assay conditions. C207A ThiI is an intermediate case, impaired but functional in vivo (Fig. 3) and in vitro (Table II). This mild effect on activity could arise from a local conformational change upon substitution of Cys-207 that does not significantly perturb the solution behavior or CD spectrum. These findings, however, are also consistent with a more direct role for Cys-207 in s 4 U generation that is either nonessential or may be assumed by some other component in the cell or the assay mixture. In either case, detailed kinetic studies, a structure, or both will be required to determine the role of Cys-207.
We are left, then, with a need to accommodate a critical but not absolute role for Cys-344. One explanation for the roles of IscS and the cysteine residues in ThiI would be formation of an ironsulfur cluster in ThiI (IscS (29) was named for its role in ironsulfur cluster formation). The hypothetical ThiI iron-sulfur cluster might serve as an intermediate in the transfer of sulfur from free cysteine to tRNA, similar to the proposals for biotin (6,7) and lipoic acid biosynthesis (10). In this case, a small amount of iron-sulfur cluster would be generated in the assay mixture, producing highly active ThiI. No evidence has been presented to suggest that ThiI bears such a cluster, but only overexpressed ThiI has been examined, leaving open the possibility that the high level of ThiI in the overexpression cells swamps the capacity to generate iron-sulfur clusters. Although we did not consider the requirement of an iron-sulfur cluster likely, the experiment was straightforward: we overexpressed IscS and ThiI in the same cells. Even though IscS was present at roughly 4-fold higher concentration than ThiI, isolated ThiI displays no spectroscopic evidence of an iron-sulfur cluster (or any other cofactor).
The main premise that led us to undertake the substitution of each cysteine with alanine was that an enzymic disulfide bond might form during catalysis (Fig. 2). Because the resting state of ThiI could be the form with the disulfide bond, we titrated the wild-type enzyme with DTNB in the presence of guanidinium chloride, and the denatured wild-type ThiI showed 5 thiol groups. Since ThiI has five cysteine residues, we conclude that ThiI is fully reduced in its resting state (at least in cells that overexpress it).
By having ruled out a disulfide bond in resting ThiI, we sought evidence that a disulfide bond forms during turnover. If true, omitting reductant would limit the enzyme to a single turnover, for the ThiI would be left in the inactive disulfidebonded state. Quantitation of the s 4 U formed and titration of free thiol groups in the resultant ThiI would allow assessment of the proposed disulfide bond. In practice, the substrate cysteine is also a reductant for disulfide bonds, so we are hampered in attaining the desired single turnover conditions. We proceeded by omitting DTT from the assay mixture and lowering the concentration of cysteine. Under these conditions, less than 1 eq of s 4 U was formed relative to ThiI, supporting the need for exogenous reductant for multiple turnovers. Titration of ThiI isolated from the reaction mixture revealed fewer thiol groups than before the reaction. Both results are in agreement with disulfide bond formation during turnover, but does the disulfide bond form between Cys-456 and Cys-344? Titration of native, wild-type ThiI with 1 eq of DTNB produced 2 eq of the thiolate product 5-thio-2-nitrobenzoate, which requires formation of a disulfide bond on ThiI. When C456A and C344A ThiI were subjected to titration with 1 eq of DTNB, only 1 eq of the thiolate product formed (the other "half" of DTNB was left as a mixed disulfide with ThiI). The simplest explanation of these results is formation of a disulfide bond between Cys-456 and Cys-344.
The profoundly reduced activity of C344A ThiI (2700-fold, Table II) is also consistent with formation of a disulfide bond between Cys-456 and Cys-344 during turnover. But if this catalytically critical disulfide bond exists, why does C344A ThiI display residual activity? If the side chain of Cys-344 were surface-accessible, a thiol group from an exogenous molecule (such as DTT in the assay mixture) might substitute, albeit inefficiently, for the missing thiol group in C344A ThiI. To probe this possibility, we compared the reactivity of wild-type and C344A ThiI toward DTNB in the absence of denaturant.
Wild-type ThiI registered 4 thiol groups, and C344A ThiI registered 3. These results indicate that the side chains of Cys-344 is among the four (of five) cysteine side chains in ThiI that appear to be surface-accessible, which supports the residual in vitro activity of C344A ThiI arising from rescue by thiol groups of DTT. All of our results, then, are consistent with our hypothesis that Cys-344 forms a disulfide bond with Cys-456 during turnover of ThiI (Fig. 2).
We now consider the extrapolation of our results and proposals to the biosynthesis of other sulfur-containing molecules. Our mechanistic possibilities accommodate the dual function of ThiI in s 4 U and thiamin biosynthesis, especially considering the evidence that the C terminus of E. coli ThiI constitutes a sulfurtransferase domain of ϳ100 amino acids that may behave nearly independently from the N-terminal portion of ThiI (31). The persulfide on Cys-456, then, would serve as a nucleophile at either another active site on ThiI to make the tRNAdisulfide intermediate ( Fig. 2A) or at an active site in the complex of the proteins ThiF and ThiS to form a disulfide intermediate on the way to the thiocarboxylate of ThiS, which is the ultimate sulfur donor for thiamin biosynthesis (33). If the C-terminal domain of ThiI constitutes a hydrogen sulfide-generating domain (Fig. 2B), the dual role of ThiI in s 4 U and thiamin biosynthesis is also readily accommodated; the hydrogen sulfide is shunted either toward the activated uridine in tRNA in another ThiI-active site or toward the adenylate of the terminal carboxylate of ThiS in an active site in the ThiF-ThiS complex. This latter possibility is attractive in light of the reports by both the groups of Begley et al. (11) and Lauhon and Kambampati (28) that ThiI stimulates but is not required for the production of the ThiS thiocarboxylate in vitro; hydrogen sulfide is generated slowly by IscS (27,29), which would allow thiocarboxylate formation in the absence of ThiI. Furthermore, Lauhon and Kambampati (28) report that inorganic sulfide (at 5 mM) can substitute for IscS and cysteine in the in vitro generation of the thiocarboxylate of ThiS, and those authors also raise hydrogen sulfide generation as a likely role for ThiI. However, the absence of detectable activity by C456A ThiI must be accommodated because IscS should suffice if the postulated C-terminal domain of ThiI serves only to generate hydrogen sulfide. Scenarios such as an activating conformational change triggered by the formation of the Cys-456 persulfide can account for the observations, but these issues clearly require further investigation.
A mechanism of hydrogen sulfide generation similar to the one we propose (Fig. 2B) may also operate in the generation of iron-sulfur clusters. IscS would again serve as an S 0 donor to make a persulfide group on another protein, which could be a scaffold protein such as NifU (4) or IscU (3,5) or a protein that bears a permanent iron-sulfur cluster. The enzymic persulfide then would undergo nucleophilic attack by another cysteine residue to make an enzymic disulfide bond and liberate sulfide for the assembling cluster. The cysteine side chains would be regenerated as thiol groups when the disulfide is reduced, likely by a thioredoxin/thioredoxin reductase or glutathione/ glutaredoxin pair. Two cysteine residues in the iron-sulfur cluster proteins could be dedicated to sulfide generation, or cysteine residues that serve as ligands for iron could undergo this chemistry. If the cluster ligands are used, the process could repeat until all but one cysteine residue was bound to iron in the incipient cluster. It follows naturally from this scenario that the last sulfide in the cluster is generated from the last cysteine persulfide, and the final step of cluster formation would therefore be the reductive cleavage of the persulfide group to generate the last sulfide and the final cysteine ligand. Intriguingly, when clusters are assembled in vitro using cys-teine and IscS or NifS to provide sulfide, the initial cluster is oxidized by two electrons relative to the iron(II) used in assembly (3)(4)(5). This cluster oxidation state is expected from reductive cleavage of the persulfide group. Our proposal for sulfide delivery to growing iron-sulfur clusters is chemically reasonable and consistent with current reports of cluster generation, and we offer it to encourage further experiments.
The studies presented here along with our previous report concerning the essential role of Cys-456 (31) provide a solid framework for understanding the nature of sulfur transfer from cysteine through IscS and ThiI to tRNA. ThiI contains neither an iron-sulfur cluster nor a disulfide bond in its resting state. In the absence of reductants, ThiI supports s 4 U generation but cannot achieve more than one turnover and suffers a loss of thiol groups. All of the evidence supports the formation of a disulfide bond between Cys-456 and Cys-344 during turnover, and indirect evidence strongly suggests that this disulfide bond forms readily in native ThiI. Since the disulfide bond between Cys-456 and Cys-344 is a feature of both of our proposed mechanisms (Fig. 2), we cannot yet exclude nor favor one or the other, but we have already begun studies that should allow us to do so.
Note Added in Proof-Begley and co-workers (40) have recently reported an acyldisulfide linkage between the C terminus of ThiS and Cys-184 of ThiF and propose that this species rather than the thiocarboxylate of ThiS is the true intermediate in thiamin biosynthesis. While this finding clouds the involvement of ThiI in thiamin biosynthesis, it does provide a gratifying example of a nucleophilic attack by an enzymic persulfide group, which was unprecedented when we offered the first mechanism proposing such an event (31).