The Sulfur Carrier Protein TusA Has a Pleiotropic Role in Escherichia coli That Also Affects Molybdenum Cofactor Biosynthesis*

Background: The sulfur carrier protein TusA is involved in sulfur transfer to different biomolecules in the cell. Results: TusA is involved in sulfur transfer for molybdopterin. Conclusion: Direction of sulfur transfer to biomolecules is mediated by the interaction partners of IscS. Significance: This study furthers the understanding of how sulfur transfer is regulated in bacteria. The Escherichia coli l-cysteine desulfurase IscS mobilizes sulfur from l-cysteine for the synthesis of several biomolecules such as iron-sulfur (FeS) clusters, molybdopterin, thiamin, lipoic acid, biotin, and the thiolation of tRNAs. The sulfur transfer from IscS to various biomolecules is mediated by different interaction partners (e.g. TusA for thiomodification of tRNAs, IscU for FeS cluster biogenesis, and ThiI for thiamine biosynthesis/tRNA thiolation), which bind at different sites of IscS. Transcriptomic and proteomic studies of a ΔtusA strain showed that the expression of genes of the moaABCDE operon coding for proteins involved in molybdenum cofactor biosynthesis is increased under aerobic and anaerobic conditions. Additionally, under anaerobic conditions the expression of genes encoding hydrogenase 3 and several molybdoenzymes such as nitrate reductase were also increased. On the contrary, the activity of all molydoenzymes analyzed was significantly reduced in the ΔtusA mutant. Characterization of the ΔtusA strain under aerobic conditions showed an overall low molybdopterin content and an accumulation of cyclic pyranopterin monophosphate. Under anaerobic conditions the activity of nitrate reductase was reduced by only 50%, showing that TusA is not essential for molybdenum cofactor biosynthesis. We present a model in which we propose that the direction of sulfur transfer for each sulfur-containing biomolecule is regulated by the availability of the interaction partner of IscS. We propose that in the absence of TusA, more IscS is available for FeS cluster biosynthesis and that the overproduction of FeS clusters leads to a modified expression of several genes.

The Escherichia coli L-cysteine desulfurase IscS mobilizes sulfur from L-cysteine for the synthesis of several biomolecules such as iron-sulfur (FeS) clusters, molybdopterin, thiamin, lipoic acid, biotin, and the thiolation of tRNAs. The sulfur transfer from IscS to various biomolecules is mediated by different interaction partners (e.g. TusA for thiomodification of tRNAs, IscU for FeS cluster biogenesis, and ThiI for thiamine biosynthesis/tRNA thiolation), which bind at different sites of IscS. Transcriptomic and proteomic studies of a ⌬tusA strain showed that the expression of genes of the moaABCDE operon coding for proteins involved in molybdenum cofactor biosynthesis is increased under aerobic and anaerobic conditions. Additionally, under anaerobic conditions the expression of genes encoding hydrogenase 3 and several molybdoenzymes such as nitrate reductase were also increased. On the contrary, the activity of all molydoenzymes analyzed was significantly reduced in the ⌬tusA mutant. Characterization of the ⌬tusA strain under aerobic conditions showed an overall low molybdopterin content and an accumulation of cyclic pyranopterin monophosphate. Under anaerobic conditions the activity of nitrate reductase was reduced by only 50%, showing that TusA is not essential for molybdenum cofactor biosynthesis. We present a model in which we propose that the direction of sulfur transfer for each sulfur-containing biomolecule is regulated by the availability of the interaction partner of IscS. We propose that in the absence of TusA, more IscS is available for FeS cluster biosynthesis and that the overproduction of FeS clusters leads to a modified expression of several genes.
Sulfur, an important element in all living cells, is incorporated into proteins not only in the form of cysteine and methionine, but also as iron-sulfur (FeS) clusters and sulfur-containing cofactors and vitamins, and into RNA through a variety of modifications (1,2). Sulfur is delivered to these various biosynthetic pathways by complex processes involving the successive transfer of sulfur as a persulfide between multiple proteins. In Escherichia coli, the iron-sulfur cluster (ISC) 2 system fulfills a housekeeping function (3). The initial sulfur mobilization step is catalyzed by IscS forming a protein-bound persulfide after the conversion of L-cysteine to L-alanine (4). IscS delivers the sulfur to several sulfur-accepting proteins such as IscU, TusA, and ThiI and, as shown recently, to rhodanese-like proteins such as YnjE (5,6), thereby providing sulfur for different metabolic pathways like the assembly of FeS cluster, thiamine, biotin, and lipoic acid, tRNA modification, or molybdenum cofactor (Moco) biosynthesis. Recent studies by Shi et al. (7) map the different binding sites for proteins that interact with IscS. Although the ISC operon encodes the major FeS assembly pathway, a second system exists in E. coli, which is encoded by the sufABCDSE operon and contains components of a secondary pathway of iron-sulfur cluster assembly; this system is required mainly for the synthesis of FeS clusters under oxidative stress and iron starvation (8).
The TusA protein functions as a sulfur mediator for the synthesis of 2-thiouridine of the modified wobble base 5-methylaminomethyl-2-thiouridine (mnm) 5 s 2 U in tRNA (9). It interacts with IscS and stimulates its L-cysteine desulfurase activity 3-fold (9). The persulfide sulfur on TusA-Cys-19 is then transferred to TusD in the TusBCD complex. TusE likely accepts the sulfur from TusD and then interacts with the MnmA-tRNA complex, in which MnmA catalyzes 2-thiouridine formation at position 34 of the tRNA in an ATP-dependent manner by using the persulfide sulfur of TusE. In total, little is known about the function of the modification of 2-thiouridine. It is proposed that in E. coli thiomodified tRNA Lys confers efficient ribosome binding (10,11) and 2-thio-modified tRNA Glu is required for specific recognition by glutaminyl-tRNA synthetase (12). Thus, the 2-thio-modification of uridine 34 plays a critical role in the decoding mechanism. Recently, Maynard et al. (13) identified 2-thiouridine modification of tRNA Lys as additionally responsible for enhanced susceptibility to viral infection by inhibition of programmed ribosomal frameshifting.
Despite its role as a sulfur transfer protein for 2-thiouridine formation, a fundamental role for TusA in the general physiology of E. coli was discovered by analyzing a strain deficient in tusA (14). High salt concentrations were determined to be crucial for the normal growth of a tusA-deficient strain. Later, Ishii et al. (15) showed that the FtsZ-ring formation appears to be severely impaired in tusA-deficient E. coli cells, resulting in the formation of a non-divided filamentous cell.
Because of its pleiotropic effect, we were interested in studying the role of tusA in Moco biosynthesis (16). In E. coli Moco biosynthesis is divided into four steps: 1) conversion of GTP into cyclic pyranopterin monophosphate (cPMP); 2) insertion of two sulfur atoms into cPMP, forming molybdopterin (MPT); 3) insertion of molybdenum to form Moco; and 4) additional modification of Moco by the covalent addition of GMP or CMP to the C4Ј-phosphate of MPT, forming either the MPT guanine dinucleotide (MGD) cofactor or the MPT cytosine dinucleotide (MCD) cofactor (17,18). Two MGD moieties are ligated to the molybdenum atom and form the bis-MGD cofactor (19), which is characteristic for the majority of molybdoenzymes in E. coli, such as nitrate reductase. Enzymes of the xanthine oxidase family in E. coli, like the aldehyde oxidoreductase PaoABC, contain the MCD cofactor.
In the formation of cPMP from GTP, MoaA and MoaC are involved, with MoaA belonging to the class of S-adenosylmethionine-dependent enzymes, which contain a conserved 4Fe4S cluster binding site at the N terminus. MoaA, in addition, contains a C-terminal 4Fe4S cluster, which is unique to MoaA proteins. Further, for the formation of MPT from cPMP, two sulfur atoms are incorporated into the C1Ј and C2Ј positions of cPMP, a reaction catalyzed by MPT synthase (20). The E. coli MPT synthase is a heterotetrameric enzyme consisting of two MoaE and two MoaD subunits (20,21). In its active form, MoaD contains a C-terminal thiocarboxylate group that acts as a direct sulfur donor for the synthesis of the dithiolene group of Moco (20,22,23). For the synthase to act catalytically, it is necessary to regenerate its transferable sulfur, a reaction for which the MoeB protein and ATP is required (20). It has been shown that MoeB solely activates the C terminus of MoaD by the formation of an acyl-adenylate, and the L-cysteine desulfurase IscS in its persulfide-bound form has been identified as the primary physiological sulfur-donating enzyme for the generation of thiocarboxylate on MPT synthase (24,25). Recently, it has been shown that the rhodanese-like protein YnjE works in conjunction with IscS to make the sulfur transfer to MoaD more specific (5).
Here, we report that a disruption in the tusA gene has a pleiotropic effect on the transcription of genes in E. coli. Microarray analyses showed that the transcription of several genes involved in Moco biosynthesis and FeS cluster biosynthesis, and additionally of genes encoding for hydrogenase and several molybdoenzymes, was increased. Surprisingly, the activity of several molybdoenzymes in the ⌬tusA strain was drastically decreased. Further analyses showed that TusA has an effect on the direction of sulfur transfer of IscS to the various sulfur-accepting biomolecules. In the absence of TusA, sulfur transfer to FeS clusters is increased, thereby effecting the expression of several genes in E. coli. Here we present a model explaining the different effects of TusA under various growth conditions in E. coli with a focus on Moco biosynthesis.

Bacterial Strains, Plasmids, Media, and Growth Conditions-
The strains and plasmids used in this study are listed in Table 1. All isogenic BW25113 mutant strains used for various assays were obtained from the Keio collection (26). Verification of the mutant strains was performed either by PCR amplification or by functional complementation studies of the respective genes. If necessary, DE3 lysogens of E. coli strains were generated by using the DE3 lysogenization kit (Novagen) following the manufacturer's instructions. Bacterial cultures were generally grown in LB medium, and when required, 150 g/ml ampicillin, 25 g/ml kanamycin, or 50 g/ml chloramphenicol was added to the medium.
RNA Isolation for Microarray Analysis-Experiments were performed with E. coli strains BW25113 and JW3435 (⌬tusA) grown both aerobically and anaerobically and supplemented with 20 mM nitrate as electron acceptor to mid-log phase. Total RNA was isolated with the High Pure RNA Isolation Kit (Roche Applied Biosystems) with two DNase I treatments. RNA concentration was determined by its absorbance at 260 nm. DNA contaminations were checked by PCR on each RNA sample. The quality of RNA was analyzed by using a Bioanalyzer (Agilent Technologies).
Global Transcriptional Analysis-The ChipShot TM labeling clean-up system (Promega) was used to generate fluorescent labeled cDNA via direct incorporation of Cy3-and Cy5-labeled nucleotides (GE Healthcare). The reverse transcription reaction was performed in the presence of 10 g of total RNA and random hexamers according to the manufacturer's recommendations. Two independent cDNA preparations were labeled once with each dye (reverse dye labeling) to account for sampling differences, biases in dye coupling or emission efficiency of Cy dyes. Labeled cDNA was purified from contaminating fluorescent dNTPs and degraded RNA using the Chip-Shot labeling clean-up system (Promega). Dye incorporation efficiency was determined by absorbance readings at 260, 550, and 650 nm, and the frequency of incorporation (pmol of dye incorporated/ng of cDNA) was calculated according to Promega's instructions. Optimally labeled samples were combined, vacuum-dried, and resuspended to a final volume of 50 l in GE hybridization buffer (HI-RPM). The combined target cDNA samples were hybridized to the spotted slides for 17 h at 65°C in an Agilent hybridization chamber. Following hybridization, slides were washed in Agilent wash buffers as recommended by the manufacturer and dried.
Agilent microarrays (E. coli gene expression microarray, 8 ϫ 15 K (clustering), G4813A-020097, Agilent) were then scanned with an Axon Genepix 4400A (MDS Analytical Technologies) at a resolution of 5 m/pixel. Scans stored as 16-bit TIFF (tagged information file format) image files and data analysis were performed with GenePix Pro 7.2. Low quality spots (spots smaller than 40 m in diameter and/or exhibiting a signal-noise ratio higher than 2) were flagged and filtered out. Median values for each spot were log transformed (log 2 ) and normalized by the Lowess method to account for any difference in total intensity between the scanned images using Acuity microarray analysis software (version 4.0; Axon Instrument Inc.). Two independent microarray experiments were performed. A 1.0-fold difference from the 1:1 hybridization ratio was taken as indicative of differential gene expression.
Two-dimensional Gel Electrophoresis-50-ml cultures of strains BW25113 and JW3435 (⌬tusA) were grown aerobically at 37°C for 8 h. Cells were harvested, resuspended in 1 ml of 10 mM Tris-HCl (pH 7.5), and sonified. Cell lysates were centrifuged at 18,000 ϫ g for 30 min. The proteins of the supernatants were purified with the 2D Clean-Up Kit (GE Healthcare) from nucleic acids, lipids, and salt and resuspended in a rehydration solution (10% glycerol, 8 M urea, 2% (w/v) CHAPS, 2% (v/v) IPG buffer (pH 4 -7 L), 19 mM DTT, and 4 mM AEBSF). Linear immobilized pH 4 -7 L gradient strips (13 cm, GE Healthcare) were rehydrated overnight in rehydration solution containing 600 g of total protein. First dimension isoelectric focusing was performed by using a PROTEAN IEF Cell (Bio-Rad) as follows: 15 min at 150 V, 15 min at 300 V, 10 min at 500 V, 10 min at 1,000 V, 10 min at 1500 V, 10 min at 2000 V, 10 min at 2500 V, and 10 min 3000 V. The last step was performed at 3500 V until 20,000 volt hours were reached. For all steps a linear voltage step was used and the current was limited to 50 A/gel. After the isoelectric focusing, the IPG strips were re-equilibrated for 15 min in 30% (v/v) glycerol, 50 mM Tris-HCl (pH 8.8), 6 M urea, 2% (w/v) SDS, and 1% (w/v) DTT, washed with water, and additionally incubated for 15 min in 30% (v/v) glycerol, 50 mM Tris-HCl (pH 8.8), 6 M urea, 2% (w/v) SDS, and 2.5% (w/v) iodacetamide. For the second dimension, IPG strips were placed on 12% SDS-polyacrylamide gels (Bio-Rad system, 1-mm spacer) and fixed with agarose. The gels were run for 30 min at 200 V (10 mA/gel) and for 4.5 h at 400 V (20 mA/gel). Gels were stained with colloidal Coomassie Brilliant Blue and destained with water. MALDI-TOF MS-Peptide mapping was performed on a Bruker ULTRAFLEX time-of-flight (TOF/TOF) instrument in the positive mode using the reflectron for enhanced resolution and a matrix of ␣-cyano-4-hydroxycinnamic acid. For MS/MS analyses, selected parent ions were subjected to laser-induced dissociation, and the resulting fragment ions were separated by the second TOF stage of the instrument. 1-l samples at an approximate concentration of 1-10 pmol/l were mixed with equal amounts of matrix. This mixture was spotted onto a stainless steel target and dried at room temperature before analysis.
Enzyme Assays-Strains BW25113, JW3435 (⌬tusA), JW3307 (⌬tusD), JW1119 (⌬mnmA), JW0764 (⌬moaA), JW1994 (⌬yeeD), JW1915 (⌬yedF), and JW3435/1994/1915 (⌬tusA/⌬yeeD/⌬yedF) (26) were grown aerobically and anaerobically for 10 h in 15 ml of LB medium at 37°C in the presence of 20 mM KNO 3 or TMAO. 15 ml of culture was harvested, resuspended in 50 mM Tris-HCl (pH 7.5) for nitrate reductase (NR) activity measurements or in 100 mM potassium-phosphate buffer (pH 6.5) for TMAO reductase (TMAOR) activity measurements, and sonified. Aliquots of 50 -200 l of each extract were analyzed for NR activity in a total volume of 4 ml of an assay mixture containing 0.3 mM benzyl viologen, 10 mM KNO 3 , and 20 mM Tris (pH 7.5). For TMAOR activity, the same volumes of cell extract were analyzed in a total volume of 4 ml of an assay mixture containing 0.4 mM benzyl viologen, 5 mM TMAO, and 100 mM potassium-phosphate buffer (pH 6.5). The reaction was initiated by the injection of sodium dithionite into the anaerobic assay mixture. The oxidation of reduced benzyl viologen at 600 nm was measured for 1 min, and the units of activity were calculated as described elsewhere (27). The protein concentration of each extract was determined using the Bradford assay (Pierce). Hydrogenase activity was estimated spectrophotometrically by measuring oxidized benzyl viologen reduction at 600 nm in the presence of H 2 -saturated buffer (28).
For the determination of overall L-cysteine desulfurase activity, plasmids pJD35 (encoding TusA), pJD54 (encoding IscU), and pACYCDuet were expressed in BW25113(DE3) in the presence of 20 M IPTG for 8 h. Cells were harvested, resuspended in 50 mM Tris-HCl, 200 mM NaCl, 10 M pyridoxal-5phosphate (PLP), and 2 mM dithiothreitol (pH 8.0), sonified, and centrifuged for 20 min at 18,000 ϫ g at 4°C. The L-cysteine desulfurase activity of the supernatants was measured by determining the total sulfide concentration as described previously (29). Supernatants were incubated with 500 M L-cysteine in a total volume of 800 l for 10 min at 30°C. The reactions were stopped by the addition of 100 l of 20 mM N,N-dimethyl-pphenylenediamine in 7.2 M HCl and 100 l of 30 mM FeCl 3 in 1.2 M HCl. After an incubation time of 20 min, the precipitated protein was removed by centrifugation, and methylene blue was measured at 670 nm. A standard curve was generated using known amounts of sodium sulfide as the sulfur source.
Detection of the Total MPT Content-50 ml of LB cultures of strains BW25113, JW3435 (⌬tusA), JW1119 (⌬mnmA), JW0764 (⌬moaA), and JW0767 (⌬moaD) were grown aerobically and anaerobically for 8 h in the presence of 20 mM nitrate, harvested by centrifugation, and resuspended in 3 ml of 100 mM Tris-HCl (pH 7.2). The lysate was obtained by sonication. The total MPT content of the strains was determined by form A detection as described previously (5).
Detection of cPMP-cPMP was converted to compound Z purified from E. coli cell extracts of the strains BW25113, JW3435 (⌬tusA), JW1119 (⌬mnmA), JW0764 (⌬moaA), and JW0767 (⌬moaD). 50 ml of LB cultures of each strain containing 20 mM nitrate were grown aerobically or anaerobically for 8 h, harvested by centrifugation, and resuspended in 3 ml of H 2 O. The lysate was obtained by sonication. cPMP was converted to its fluorescent derivative, compound Z, by adjusting the pH of the supernatant to 2.5 with HCl and adding 200 l of 1% (w/v) I 2 , 2% (w/v) KI for 14 h. Excess iodine was removed by the addition of 110 l of 1% (w/v) ascorbic acid, and the sample was adjusted with 1 M Tris to pH 8.3. Compound Z was further purified on a QAE ion exchange column (Sigma) with a 500-l bed volume, which was equilibrated in H 2 O. Compound Z was eluted from the matrix in a volume of 5 ml (5 ϫ 1 ml fractions). The reactions were analyzed by subsequent injection (100 l) onto a C18 reversed phase HPLC column (4.6 ϫ 250-mm ODS Hypersil; particle size, 5 m) equilibrated with 10 mM potassium phosphate (pH 3.0) with 1% methanol at an isocratic flow rate of 1 ml/min. In-line fluorescence was monitored by an Agilent 1100 series detector with excitation at 370 nm and emission at 450 nm. Plasmids pJD34 (encoding TusA) and pJD49 (encoding TusA-C19S) were introduced into ⌬tusA strain. As a control, pET11b containing ⌬moaD was used. Cells were cultivated in the presence or absence of 50 M IPTG. cPMP accumulated in these cells was determined as described above.
Analysis of the Sulfuration Level of MPT Synthase Purified from Different E. coli Mutant Strains-For expression of MPT synthase, the plasmid encoding active MPT synthase (expressing MoaD, MoaE, and MoeB) was introduced into E. coli BW25113(DE3), JW3435 (⌬tusA) (DE3), and JW1119 (⌬mnmA) (DE3) cells, respectively. Expression and purification of E. coli MPT synthase were performed following published procedures (23). Protein concentration was determined using the calculated extinction coefficient at 280 nm. MPT synthase reactions were performed at room temperature in a total volume of 400 l of 100 mM Tris (pH 7.2). The reaction mixtures contained 15 M overexpressed MPT synthase. The reaction was initiated by the addition of an excess amount of cPMP, which was purified according to published procedures (30). The MPT produced was oxidized to form A and quantified following published procedures (27,31).
Expression and Purification of EcPaoABC, EcTorAD, hSO, and RcXDH-Plasmids pMN100 (coding for PaoABC), pSL207 (coding for RcXDH), pTorAD (coding for TMAO reductase), and pTG718 (coding for hSO) were expressed in E. coli BW25113 and JW3435 (⌬tusA) cells. The enzymes were purified by nickel-nitrilotriacetic acid chromatography following published procedures (32)(33)(34)(35). The enzyme activities of the four enzymes were measured spectrophotometrically as described elsewhere (33, 36 -38). Moco/MPT was quantified by conversion to form A as described previously (31). Oxidation was performed either at room temperature for 14 h (for hSO and Rhodobacter capsulatus XDH) or for 30 min at 95°C (for E. coli TorAD and E. coli PaoABC). The molybdenum and iron contents of the purified proteins were quantified by inductively coupled plasma optical emission spectrometry (ICP-OES) analysis with a PerkinElmer Life Sciences Optima 2100 DV ICP-OES. 500 l of a 10 M solution was mixed with 500 l of 65% nitric acid and incubated overnight at 100°C before the addition of 4 ml of water.
Metal Analysis-For metal analysis of total crude extract (A) and membrane fraction (B), strains BW25113, JW3435 (⌬tusA), JW1119 (⌬mnmA), and JW0764 (⌬moaA) were grown aerobically for 10 h in 100 ml of LB at 37°C. Cultures were harvested, washed three times with 10 ml of 50 mM Tris-HCl (pH 7.5), resuspended in 3 ml of the same buffer, and sonified. 2 ml of disrupted cells were centrifuged at 18,000 ϫ g for 30 min. The membrane fraction was then washed three times with 5 ml of 50 mM Tris-HCl (pH 7.5) and resuspended in 1 ml of the same buffer. The molybdenum, iron, nickel, and copper contents of the E. coli strains were quantified in total crude extract and membrane fraction. As a reference, the multi-element standard solution XVI (Merck) was used.
␤-Galactosidase Activity-E. coli strains BW25113, ⌬tusA, ⌬iscR, MC4100, and ⌬fnr carrying tusA-lacZ reporter plasmid pJD27, tusBCD-lacZ reporter plasmid pJD28, ydjXYZyn-jABCD-lacZ reporter plasmid pJD8, ynjE-lacZ reporter plasmid pJD14, iscRSUA-lacZ reporter plasmid pJD66, pepT-lacZ reporter plasmid pJD83, or pGE593 were grown in LB under aerobic or anaerobic conditions. Following growth to A 600 nm 0.6 -1.0, ␤-galactosidase activities were measured by the SDSchloroform method, and the units of activity were calculated as Miller units for each sample using the equation MU ϭ 1000 ϫ where t equals the time of reaction and V equals the volume (ml) of cells added to the assay tubes.
Expression of the Genes from the ISC Operon, tusA, and iscU-Plasmids pDB1282 (coding for Azotobacter vinelandii IscS, IscU, IscA, HscB, HscA, Fdx, and IscX), pJD34 (coding for TusA), pJD54 (coding for IscU), pET15b, and pEB327 were expressed in the BW25113(DE3) strain. Expression of TusA, IscU, and pET15b was induced by the addition of 20 M IPTG for 8 h. Expression of both the ISC genes and pEB327 was induced by the addition of 0.1% arabinose. Cells were harvested, and NR and TMAOR activity was measured in the same manner as described above for the mutant strains.
Surface Plasmon Resonance (SPR) Measurements-All binding experiments were performed with the SPR-based instrument Biacore TM T200 on CM5 sensor chips at a temperature of 25°C and a flow rate of 10 l/min using Biacore control T200 software and evaluation T200 software (GE Healthcare) as described previously (39). The autosampler racks containing the sample vials were cooled to 4°C. Immobilization of proteins yielded the following resonance units (RU)/flow cell: BSA, 1204 RU; IscS, 1118 RU; TusA, 538 RU; IscU, 1049 RU. For immobilization, 10 mM acetate buffer with variable pH (pH 4.0 for TusA and IscU and pH 5.0 for IscS) was used.
As a running buffer, 20 mM phosphate, 150 mM NaCl, and 0.005% Tween 20 (pH 7.4) was used. IscS, IscU, and TusA with concentrations of 0.8, 1.6, 3.1, 6.3, 12.5, and 25 M were injected for 4.5 min at a flow rate of 30 l/min followed by 15 min of dissociation using the KINJECT command and regeneration of the sensor surface with 50 mM HCl for 1 min. As a control, BSA was used as ligand. Binding curves were corrected by subtracting buffer injection curves for all four flow cells.

RESULTS
Microarray Analysis-In previous studies, TusA was shown to be involved in the modification of tRNA, in cell morphology, in cell division, and in receptivity for viral infection (9,(13)(14)(15). Because of its pleiotropic effects we were interested in analyzing the role of TusA further. To reveal an effect of the absence of TusA on gene expression, we chose a microarray approach and compared different growth conditions in E. coli. One major effect of TusA influences growth in the exponential ohase as reported previously. Additionally, we wanted to analyze the differences in gene expression in relation to oxygen. E. coli BW25113 and ⌬tusA strains were grown in LB until the stationary phase or mid-exponential phase was reached under aerobic or anaerobic conditions at 37°C. For expression profiling, samples were taken, total RNA was extracted, converted to Cy3 and Cy5-labeled cDNA, and hybridized to E. coli DNA microarrays. Microarray analysis showed that deletion of tusA led to changes in the regulation of a set of genes of which there is a complete list in supplemental Tables S1-S3. Negative values indicate increased expression in the ⌬tusA strain, and positive values show higher expression in BW25113 wild type. Among those, we selected some genes of interest encoding for proteins of Moco biosynthesis, molybdoenzymes, hydrogenases, and FeS cluster biosynthesis (Table 2) for further description.
The most pronounced difference in the ⌬tusA strain grown aerobically to mid-exponential phase was the higher expression of the moaABCDE operon, encoding enzymes involved in cPMP/MPT biosynthesis. This effect was the same under the different growth conditions analyzed. The expression of the molybdate transport system (modABC) was decreased under aerobic conditions in the stationary phase in the absence of tusA. The role of modF is thus far unclear. For most molybdoenzymes the expression was increased in the ⌬tusA strain under anaerobic conditions and under aerobic conditions in the stationary phase. However, in the mid-exponential phase and under aerobic conditions, the expression of genes coding for the molybdoenzymes DMSO reductase, YdhV, YnfE, and NarG was decreased in the ⌬tusA strain. Additionally, the log 2 ratio of the operon coding for hydrogenase 3 (hycABCDEFG) and its maturation factors (hypA, hypN, and hypF) indicated an elevated expression level in the ⌬tusA strain ( Table 2). The expression of the sufABCDSE operon was also increased under anaerobic conditions when tusA was deleted, whereas it was decreased under aerobic conditions. The expression of the iscA and hcsB genes, involved in housekeeping FeS cluster biosynthesis, was also decreased under aerobic conditions in the stationary phase. Genes for tRNA thiolation were also altered, with an increased expression of mnmA and a decreased expression of tusB.
Two-dimensional Gel Electrophoresis-To confirm the results obtained by microarray analysis of the ⌬tusA strain, we wanted additionally to check the changes at the protein level of this mutant strain compared with the corresponding wild type. Therefore two-dimensional gel electrophoresis was used as described under "Experimental Procedures." The soluble pro- Gene expression values (log 2 median of ratio) for all genes/operons are differently regulated in ⌬tusA strain cultivated under different growth conditions. Genes with a log 2 median of ratio larger than ͗1.0͘ (corresponding to genes induced more than 2.3-fold) are considered differentially expressed. Negative values indicate higher expression in ⌬tusA strain; positive values show higher expression in BW25113 wild type. Stat., stationary phase. tein fractions obtained from strains BW25113 and ⌬tusA grown aerobically to the beginning of the stationary phase were separated by two-dimensional gel electrophoresis. Fig. 1 shows the colloidal Coomassie-stained two-dimensional gels for BW25113 and the ⌬tusA strains. The intensity of 16 spots within the gels was identified as different, indicating their different levels of expression. The spots were excised from the gels and analyzed by MALDI peptide mapping. The identified proteins are listed in Table 3, where five proteins were identified in five independent experiments to be expressed differentially in either BW25113 or ⌬tusA. One of the proteins identified to be present in higher amounts in ⌬tusA is the MoaE protein (marked with the number 1 in Fig. 1). Expression of peroxyredoxin OsmC and periplasmic dipeptide transport protein DppA was decreased in the ⌬tusA mutant. In contrast, expression of the chaperone DnaK as well as flagellin FliC was increased in the ⌬tusA strains ( Fig. 1 and Table 3). In comparison with the microarray data, mainly the MoaE protein was identified to be present in higher amounts. Limitations to this two-dimensional gel electrophoresis might explain why the other proteins were not identified in higher amounts. These disadvantages and the small size of, for example, MoaD and MoaC might contribute to the fact that only MoaE was detectable in our proteomic approach.

Transcriptional Analysis of tusA-lacZ and ynjE-lacZ Promotor Fusions in E. coli BW25113 under Aerobic and Anaerobic
Conditions-The microarray analysis identified differences in the expression of genes in the ⌬tusA mutant under aerobic and anaerobic conditions. To study the expression of tusA in relation to oxygen, we constructed a tusA-lacZ fusion. We analyzed the expression level of tusA-lacZ in comparison with a tusD-lacZ fusion, the promoter of which controls the genes of the tusBCD operon. Additionally, we analyzed the expression of ynjE, as YnjE was recently identified to act as a sulfur donor for Moco biosynthesis in E. coli. Because it was not clear whether ynjE is expressed as a single transcriptional unit or together with the ydjXYZynjABCD operon, we compared the expression of both lacZ fusions. The plasmid-born lacZ fusions were stud-ied under aerobic and anaerobic conditions in BW25113, and ␤-galactosidase activities were determined as Miller units. As shown in Table 4, the tusA-lacZ transcriptional fusion showed no changes in response to the oxygen concentration, with a high expression level of about 11,000 Miller units. In comparison, the expression of genes coding for proteins involved in tRNA thiolation is 3-fold higher under anaerobic conditions in comparison with the aerobic conditions, as revealed by the tus-BCD-lacZ promotor fusion. The results in Table 4 additionally show that ynjE is expressed mainly under anaerobic conditions (7.5-fold induction) and represents its own transcriptional unit. Although the ydjXYZynjABCD-lacZ fusion showed a generally low expression level, the expression was induced 10-fold under anaerobic conditions. However, the role of this operon remains unclear. The results show that the differences in expression of the genes identified by transcriptomic and proteomic analyses in response to the oxygen concentration are not based on differences in the expression of tusA, because the expression of tusA is not altered by the oxygen concentration in the cell. In contrast, the transcription of the ynjE gene, which is involved in Moco biosynthesis, is elevated under anaerobic conditions. This may imply that YnjE has a main role under anaerobiosis, which might be related to a higher expression of selected molybdoenzymes under these conditions.
Analysis of the Activities of Hydrogenases, NR, and TMAOR in a ⌬tusA Strain-The microarray analysis showed that the gene expression of hydrogenase 3 and nitrate reductase in addition to their maturation factors are increased up to 4-fold in the ⌬tusA mutant under anaerobic or aerobic conditions. We wanted to test whether the increased expression level correlates with an increase in enzyme activity. Thus, we tested the activity of hydrogenase, NR, and, as additional molybdoenzyme, TMAOR, in BW25113 cells and in the ⌬tusA mutant after aerobic and anaerobic growth on nitrate or TMAO, respectively. For comparison regarding the effect of the absence of thiolated tRNA, we used the ⌬mnmA strain as a control.
Hydrogenase activity was determined by detecting the reduced benzyl viologen-based hydrogen utilization. As shown in Table 5, the hydrogen oxidation was increased almost 4-fold in the ⌬tusA strain, whereas it remained the same in the ⌬mnmA strain in comparison with the wild type strain. The increase in activity is consistent with the 4-fold increased expression shown in Table 2.
Further, we analyzed the activity of TMAOR and NR. The respective E. coli strains were cultivated in the presence of 20 mM nitrate or 20 mM TMAO under aerobic conditions until the stationary phase was reached. NR activities and TMAOR activities in crude cell extracts were determined by analyzing the oxidation of benzyl viologen. As shown in Table 5, the activities of both NR and TMAOR were largely reduced down to 4 -6% when purified from the ⌬tusA strain in comparison with the wild type strain. In all cases, the effect on enzyme activity was not based on an altered level of thiolated tRNA, as the ⌬mnmA strain showed enzyme activities comparable to BW25113. Thus, the absence of TusA influenced the activity of the two tested molybdoenzymes drastically.

Analysis of the Activities of Molybdoenzymes with Different
Cofactor Modifications in the E. coli ⌬tusA Strain-Molybdoenzymes with three different cofactor modifications were identified in E. coli belonging to the three classified families: DMSO reductase, sulfite oxidase, and xanthine dehydrogenase. To further characterize the effect of the tusA deletion on the molybdoenzymes of each family, we tested one characteristic enzyme from each family in regard to its activity in the ⌬tusA strain. For this purpose we selected E. coli TMAOR containing bis-MGD, E. coli PaoABC containing MCD, R. capsulatus xanthine dehydrogenase (RcXDH) containing sulfurated Mo-MPT, and human sulfite oxidase containing unmodified Mo-MPT. The respective molybdoenzymes were overexpressed in E. coli ⌬tusA and BW25113 cells under aerobic conditions, purified by nickel-nitrilotriacetic acid chromatography, and analyzed for their MPT content and their saturation with molybdenum and iron (when appropriate). Consistent with the data shown above, all four molybdoenzymes (EcPaoABC, human sulfite oxidase, RcXDH, and EcTMAOR) showed a loss of enzyme activity of 84.2-94% when purified from the ⌬tusA strain ( Table 6). Analysis of the metal content showed that whereas the purified enzymes were completely saturated with iron (except for EcTMAOR, which contains Moco as sole cofactor), the molybdenum contents were down to a level of 3.7 to 15.3%, consistent with their enzyme activities (Table 6). Additionally, the MPT content (detected as form A) was reduced to 81.5-100% in a similar manner. Thus, the loss of enzyme activity correlated with a reduced Moco content in these enzymes, and the loss of activity was the same for molybdoenzymes from each family. This result shows that the effect of the absence of TusA has to be a general effect on Moco biosynthesis independent of further specific modifications of Moco.

Analysis of the Effect of Oxygen on the Activity of NR and TMAOR in Different E. coli Mutant Strains-The microarray
analysis showed the effect of oxygen concentration on the expression of different genes in the ⌬tusA mutant strain. As the results above showed that the activity of NR and TMAOR was largely reduced under aerobic conditions in the ⌬tusA strain, we additionally wanted to test the activity under anaerobic conditions. To exclude an effect of tRNA thiolation on enzyme activity, we additionally tested NR and TMAOR activity in the ⌬tusD and ⌬mnmA strains. Additionally, two homologues of TusA were identified in E. coli, YeeD and YedF (14,40), which share an amino acid sequence homology of 39% for YedF and 33% for YeeD. Thus far, YeeD and YedF had not been characterized and nothing is known about the role of these proteins in E. coli. To test whether YeeD or YedF could replace TusA in its role, at least partially, we additionally analyzed NR and

Detection of differently regulated proteins in E. coli JW3435 (⌬tusA) strain by two-dimensional gel electrophoresis
Strains BW25113 and ⌬tusA were grown aerobically at 37°C for 8 h. Two-dimensional gel electrophoresis was performed as described under "Experimental Procedures" and in the legend for Fig. 1. All proteins were identified by MALDI peptide mapping. Spot numbers refer to the numbering on the two-dimensional gel electrophoresis pictures (Fig. 1). Protein names, sizes, and functions are indicated. Protein scores greater than 70 are significant (p Ͻ 0.05). In total, five independent two-dimensional gel electrophoreses were performed. In respect to their increase/decrease of expression in ⌬tusA in comparison with BW25113 wild type; the numbers indicate the quantity of identification in these five independent measurements.   TMAOR activities in ⌬yeeD and ⌬yedF mutants in addition to a ⌬tusA⌬yedF⌬yeeD triple mutant. The strains were cultivated in the presence of either 20 mM nitrate or 20 mM TMAO under aerobic and anaerobic conditions, respectively. As shown in Fig. 2, the deletion of tusD or mnmA did not affect the activity of NR and TMAOR in comparison with the corresponding wild type strain under the different growth conditions. This shows that an impairment of 2-thiouridine formation does not affect molybdoenzyme activities. As described above, NR and TMAOR activity were decreased up to 96% in the ⌬tusA strain under aerobic conditions. However, under anaerobic conditions, the effect of the ⌬tusA mutant was less pronounced, showing a decrease in NR activity of 52% and of TMAOR activity of 74% in comparison with the wild type strain.

Spot
To determine whether the remaining activities of NR and TMAOR in the ⌬tusA mutant strain were based on the partial replacement of TusA function by its two homologues, YeeD and YedF, we analyzed the enzyme activities in single, double, and triple mutant strains of these genes. The results in Fig. 2 show that neither YedF nor YeeD are able to replace TusA in its function, because single yedF and yeeD deletions did not affect molybdoenzyme activities and the triple ⌬tusA⌬yedF⌬yeeD mutant showed the same effect on enzyme activity as the single ⌬tusA strain. The results show conclusively that whereas TusA is almost essential for the activity of NR and TMAOR under aerobic conditions, the absence of TusA only partially affects molybdoenzyme activities under anaerobic conditions. Thus, under anaerobic conditions other proteins different from YedF or YeeD can take over the role of TusA for Moco biosynthesis.
Analysis of the Effect of TusA on the MPT and cPMP Content in Addition to the Sulfuration Level of MPT Synthase-The results above show that TusA has a major effect on Moco biosynthesis under aerobic conditions, which is independent of the modification of the cofactor. Because TusA is involved in sulfur transfer to tRNA, it might additionally be involved in sulfur transfer for the generation of the dithiolene moiety of Moco. So far, it has been shown that the direct donor for the sulfur in Moco is the MoaD protein, inserting sequentially the sulfur of its C-terminal thiocarboxylate group into cPMP, thus forming MPT (22). As a consequence, cPMP is accumulated in a moaD mutant. The direct sulfur donor for MoaD is the IscS protein, which generates the thiocarboxylate group on MoaD in conjunction with MoeB (25). YnjE is also involved in this step, making the sulfur transfer reaction from IscS more specific (5). To test whether TusA is additionally involved in the sulfur transfer reaction from IscS to MoaD, we analyzed the sulfuration level of MoaD isolated from a ⌬tusA mutant. The sulfuration level of MoaD was analyzed by testing its ability to convert cPMP to MPT in conjunction with its partner protein, MoaE (forming active MPT synthase). MoaD, MoaE, and MoeB were expressed under aerobic conditions, purified from the BW25113(DE3), ⌬tusA(DE3), and ⌬mnmA(DE3) strains, and incubated with an excess amount of cPMP. The MPT produced was quantified by its conversion to the fluorescent derivate, form A. The activity of MPT synthase isolated from the wild type strain was set at 100%. As shown in Fig. 3A, MPT synthase purified from the ⌬tusA strain showed a residual activity of 10% in comparison with the MPT synthase isolated from BW25113(DE3). The ⌬mnmA strain did not affect MPT synthase activity, showing that this effect is independent of tRNA thiolation.
Because the results showed that MPT synthase activity is highly reduced in the ⌬tusA mutant strain, we determined the total cPMP and MPT concentrations. Strains were grown aerobically and anaerobically in the presence of 20 mM nitrate. For comparison, a ⌬moaA strain was chosen that does not contain cPMP or MPT.
In crude extracts of different mutant strains, cPMP and MPT were oxidized to their stable fluorescence derivatives, compound Z and form A, respectively (Fig. 3, B and C). The amount of cPMP determined in the ⌬moaD strain was set individually to 100% under aerobic and anaerobic conditions (Fig. 3B). Although in the wild type, the ⌬mnmA mutant, and the ⌬moaA mutant, no cPMP was identified, the ⌬tusA mutant accumulated 75% cPMP under aerobic conditions and 40% cPMP under anaerobic conditions when compared with the ⌬moaD strain.
For the overall MPT levels determined in these strains, the MPT content of the wild type under the two growth conditions was set individually to 100% (Fig. 3C). Whereas the ⌬mnmA strain showed comparable levels as the wild type strain, the MPT level in the ⌬tusA strain was reduced to 80% under aerobic growth conditions and to 70% under anaerobic growth conditions. In comparison, no MPT was detected in the ⌬moaA and ⌬moaD strains.
The results are consistent with the activity of molybdoenzymes analyzed in the ⌬tusA strain, showing that under aerobic conditions, less Moco is produced. Additionally, the results show that TusA has a role in sulfur transfer for the conversion of cPMP to MPT, because in a ⌬tusA mutant, cPMP is accumulated. However, TusA is not essential in this reaction, because under anaerobic conditions the absence of TusA only leads to a reduction of about 50% of MPT.
Additionally, we analyzed the metal contents of the crude extract (Fig. 4A) and the membrane fraction (Fig. 4B) of these mutant strains. The results show that whereas the molybdenum content in the ⌬tusA strain correlated well with the activity of NR in this strain, the concentration of other metals (iron, cop- per, or nickel) was not influenced by the absence of TusA. The decrease in molybdenum content in the membrane fraction is based on the reduced synthesis of membrane-bound molybdoenzymes like NR. The overall reduced molybdenum content shows a feedback regulation of impaired Moco biosynthesis and molybdate uptake (as revealed by the decreased expression of the mod operon in the ⌬tusA strain).
Analysis of the Effect of Overexpression of the ISC Genes on Molybdoenzyme Activity-The above results show that TusA is involved in the sulfur transfer step for Moco biosynthesis. However, because TusA has a more pronounced effect under aerobic conditions than under anaerobic conditions, where the absence of TusA leads to only a 50% reduction of molybdoenzyme activities, we suspected that TusA might not be involved directly in Moco biosynthesis and that the effect is more general in the distribution of sulfur in the cell. Because the expression of a lot of genes for molybdoenzymes that are regulated by FNR was increased under anaerobic conditions in a ⌬tusA strain, we suspected that the availability of TusA might influence the level of FeS clusters in the cell, as suggested previously by Maynard et al. (13). To analyze the effect of an increased level of FeS cluster biosynthesis on molybdoenzyme activity, we overexpressed the A. vinelandii iscSUAhscABfdxiscX operon in BW25113 under aerobic and anaerobic conditions. The activity of NR and TMAOR was determined and related to the activity determined in BW25113. The results in Fig. 5A show that NR activity was reduced to 10% under aerobic conditions and 41% under anaerobic conditions, whereas TMAOR activity was reduced to 30% under aerobic conditions and 53% under anaerobic conditions in comparison with the strain in which the ISC operon was not expressed. Thus, the reduced activities of NR and TMAOR in the presence of a higher level of FeS clusters in the cell are FIGURE 2. Analysis of NR and TMAOR activity in different E. coli strains. NR activity (A) and TMAOR activity (B) were determined by using cell lysates of the strains BW25113, JW3435 (⌬tusA), JW3305 (⌬tusD), JW1119 (⌬mnmA), JW1994 (⌬yeeD), JW1915 (⌬yedF), and JW3435/1994/1915 (⌬tusA⌬yeeD⌬yedF), which were grown aerobically (white bars) and anaerobically (black bars) in the presence of 20 mM nitrate and TMAO, respectively. Reduced benzyl viologen was used as an electron donor. The activity of the wild type strain under aerobic and anaerobic growth conditions was set to 100% each and compared with the mutant strains under the same growth conditions. The activities were related to total protein concentration in the crude extract. Error bars are derived from three independent measurements within a standard deviation of 10% at maximum. consistent with the reduced activities of different molybdoenzymes when TusA is absent.
Further, the opposite should occur when TusA is overexpressed in the cell. Then, sulfur transfer from IscS is directed away from FeS cluster biosynthesis. To test this hypothesis, we overexpressed TusA and IscU for comparison in BW25113 cells. The results in Fig. 5B show that the effect of an overexpression of IscU on NR and TMAOR activity is similar to the expression of the whole iscSUAhscABfdxiscX operon. Expression of IscU resulted in a reduced NR and TMAOR activity of 25-39% in comparison with cells in which IscU was not expressed. Surprisingly, when TusA was overexpressed, the activity of NR and TMAOR was not increased (Fig. 5B). Both NR and TMAOR activities were decreased to a level of 25-32%, in a manner similar to the IscU mutant. Here, we suspect that again the FeS level has an effect on Moco biosynthesis for NR and TMAOR. Because MoaA contains two 4Fe4S clusters, the synthesis of cPMP is affected when TusA is overexpressed, as and purified after ammonium precipitation and size exclusion chromatography. To analyze the sulfuration level of MPT synthase, 15 M purified MPT synthase was incubated with an excess of cPMP. After 45 min, the reactions were stopped by the addition of acidic iodine, and MPT formed during the reaction was quantified as form A using the peak area. Fluorescence was monitored with excitation at 383 nm and emission at 450 nm. The activity of the wild type strain was set to 100%. Results were derived from at least three measurements, and standard deviations are within a range of 10%. B, quantification of cPMP concentration in E. coli strains JW3435 (⌬tusA), JW1119 (⌬mnmA), JW0764 (⌬moaA), and JW0767 (⌬moaD) after oxidation to its stable fluorescent derivative, compound Z. Fluorescence was monitored with excitation at 370 nm and emission at 450 nm. The amount of compound Z of the ⌬moaD strain was set to 100%. The concentrations were related to cell density determined at A 600 nm . C, total MPT content of strains JW3435 (⌬tusA), JW1119 (⌬mnmA), JW0764 (⌬moaA), and JW0767 (⌬moaD). MPT was quantified after conversion to its fluorescence derivative, form A, and is shown as relative form A fluorescence quantified from the peak areas. The fluorescence obtained from the wild type strain was set to 100%. Fluorescence was monitored with excitation at 383 nm and emission at 450 nm. Black bars, anaerobic growth conditions; white bars, aerobic growth conditions. Total crude extracts and membrane fractions were prepared from E. coli strains BW25113, JW3435 (⌬tusA), JW1119 (⌬mnmA), and JW0764 (⌬moaA). ICP-OES was used to analyze levels of molybdenum (black), iron (light gray bars), nickel (dark gray bars), and copper (gray bars) in total crude extract (A) and in the membrane fraction of each strain (B). The metal content of the wild type strain was set to 100%. All metal contents were related to the total protein concentration. Results were derived from at least three independent measurements, and standard deviations are within a range of 20%.
fewer FeS clusters are available for the insertion into MoaA. Thus, overexpression of TusA directs IscS away from FeS cluster biosynthesis.
Analysis of the L-Cysteine Desulfurase Activity in Vivo-TusA was shown to enhance the L-cysteine desulfurase activity of IscS in vitro, whereas IscU has no such an effect on IscS, as reported recently (9,41). IscS has different binding partners, namely IscU, CyaY, TusA, ThiI, and IscX (7). The binding sites for these proteins on IscS were mapped. Some proteins were shown to have overlapping binding sites. Thus, the binding of the different proteins to IscS might be regulated by their availability, suggesting that one protein outcompetes the other when available in higher concentrations. To test whether the direction of sulfur transfer from IscS to TusA is regulated by the concentration of TusA or IscU in the cell, we performed coexpression experiments in vivo. TusA, IscU, or both TusA and IscU were overexpressed in BW25113 cells, and the effect on L-cysteine desulfurase activity was analyzed in cell lysates after aerobic growth. The results in Fig. 5C show that the L-cysteine desulfurase activity increased 1.9-fold in the lysates where TusA was expressed. The expression of IscU did not affect the L-cysteine desulfurase. In contrast, the L-cysteine desulfurase activity containing both IscU and TusA showed a further enhancement (a factor of 2.1). This shows that in the presence of IscU, TusA is still able to bind to IscS, resulting in an increase in its activity.
Analysis of Protein-Protein Interactions by Surface Plasmon Resonance Measurements-To determine the dissociation constants of TusA and IscU for interaction with IscS, SPR measurements were employed for real-time detection of specific interactions using the purified proteins. TusA, IscS, and IscU were immobilized on the CM5 chip via amine coupling.
The results obtained by SPR measurement for the protein pairs listed in Table 7 showed the lowest K D values for the interaction of immobilized TusA with IscS (K D ϭ 0.304 M). IscS and immobilized IscU showed a K D value of 0.574 M, whereas no dissociation constant for the interaction of IscU with TusA was obtained. A comparable value was obtained for the interaction of immobilized IscS with TusA (K D ϭ 0.966 M). The interaction between immobilized IscS and IscU (K D of 14.6 M) was lowered, showing that IscS was immobilized in a conformation that inhibited this interaction but not the interaction with TusA. In general, the results obtained from SPR FIGURE 5. Analysis of NR and TMAOR activities after overexpression of the ISC operon, TusA, and IscU. A, plasmid encoding the A. vinelandii ISC operon was introduced into BW25113 (DE3), and the cells were grown for 8 h at 37°C under aerobic (white bars) and anaerobic (black bars) conditions in the presence of either 20 mM nitrate or 20 mM TMAO, respectively. NR and TMAOR activities were determined in cell lysates. Reduced benzyl viologen was used as the electron donor. The activity of the wild type strain containing an arabinose-induced ampicillin-resistant plasmid was set to 100% for each condition. The activities were related to total protein concentration in the crude extract. Error bars are derived from three independent measurements within a standard deviation of 10% at maximum. B, plasmids encoding IscU (pJD54), TusA (pJD34), and pET15b (Ϫ) were introduced into BW25113 (DE3), and the cells were grown for 8 h at 37°C in the presence of either 20 mM nitrate or 20 mM TMAO under aerobic conditions. TMAOR activities (white bars) and NR activities (black bars) were determined in cell lysates as described for Table 5. C, plasmids encoding IscU (pJD54), TusA (pJD35), and pACYCDuet (Ϫ) were introduced into BW25113 (DE3), and the cells were grown for 8 h at 37°C under aerobic conditions. Cleared cell lysates were incubated with 500 M L-cysteine for 10 min at 30°C, and L-cysteine desulfurase activity was measured by determining total sulfide produced using methylene blue as reported previously (29). D, plasmids encoding TusA (pJD34), TusA-C19S (pJD49), and pET15b (Ϫ) were introduced into ⌬tusA(DE3) and ⌬moaD, respectively, and the cells were grown for 8 h at 37°C in 20 mM nitrate containing LB under aerobic conditions with (black bars) and without (white bars) the addition of IPTG. Quantification of cPMP concentration was performed after oxidation to its stable fluorescent derivative, compound Z. Fluorescence was monitored with excitation at 370 nm and emission at 450 nm. The amount of compound Z of the ⌬moaD strain was set to 100%. The concentrations were related to protein concentration in crude extract. n.d., not determined.
show that TusA and IscU have dissociation constants similar to those of IscS, so that a direct competition in dependence on their cellular concentration is possible.
Transcriptional Analysis of Different lacZ Fusions in E. coli ⌬tusA and ⌬fnr Strains-From the results in the above studies, we suspected that the availability of IscS for certain pathways is regulated by its interaction partners. One way to test whether FeS levels are increased in the absence of TusA is the analysis of the activity of FNR. In its active form, FNR contains a 4Fe4S cluster, which enables FNR to bind to DNA and regulate the expression of genes. For this purpose, we analyzed the activity of FNR-regulated genes in different E. coli mutant strains. We constructed a pepT-lacZ fusion, which encodes peptidase T, as the expression of pepT has been shown to be activated by FNR (46). Additionally, we analyzed the expression of ynjE and ydjXYZynjABCD, genes that have been shown also to be regulated by FNR (46). Table 8 shows that the pepT-lacZ fusion has a 2-3-fold higher expression in the absence of tusA, implying a higher activity of FNR. Additionally, the expression of ynjE and ydjxyZynjABCD was enhanced 3-fold in a ⌬tusA strain. In contrast, the expression of these genes/operons was decreased 2-fold in the ⌬fnr strain. For comparison, we analyzed the expression of the iscRSUA operon in a tusA-and fnr-deficient strain. Although the expression of the operon was increased 13% in the ⌬tusA strain, the expression was decreased 10% in the ⌬fnr strain. Because the expression of the iscRSUA operon is regulated by the amount of FeS clusters in the cell via IscR, the result might be explained thus, that in the ⌬tusA strain, FeScontaining molybdoenzymes are more highly expressed, simulating a higher demand for FeS clusters. In contrast, in the fnr mutant, the expression of genes for FeS-containing proteins is not activated, and thus the iscRSUA operon is down-regulated. In total, our transcriptional data imply a higher activity for FNR in the tusA-deficient strain, which would explain the different regulation of genes shown in our transcriptomic data (Table 2  and Supplemental Tables S1-S3).
Increased Amounts of TusA-C19S Lead to Decreased cPMP Accumulation in the ⌬tusA Mutant-To further dissect the influence of TusA on FeS cluster biosynthesis, we analyzed the influence of different expression levels of an inactive TusA variant on cPMP biosynthesis. For the synthesis of cPMP from GTP, the activity of the 4Fe4S cluster containing protein MoaA is required. Thus, a limited availability of FeS clusters should result in a decreased synthesis of cPMP due to the inactivity of MoaA.
To analyze the effects of TusA binding to IscS, we constructed the inactive TusA-C19S mutant and expressed the protein in the ⌬tusA strain. TusA-C19S is still able to bind to IscS (data not shown); however, cPMP is accumulated under aerobic conditions because of the impairment in sulfur transfer for MPT biosynthesis. By the addition of IPTG, the expression of TusA-C19S was increased. Cells were grown under aerobic conditions, and the amount of cPMP was quantified and related to a ⌬moaD mutant strain (set to 100%), which is completely impaired in the synthesis of MPT. As shown in Fig. 5D, the ⌬tusA strain expressing TusA-C19S accumulated fewer amounts of cPMP in comparison with the ⌬moaD strain. The amount of accumulated cPMP was decreased to 42% when IPTG was added. In comparison, the ⌬tusA strain expressing TusA wild type did not accumulate cPMP, as sulfur is available for the synthesis of MPT. Thus, we propose that under the conditions tested, the higher amounts of expressed TusA-C19S compete with IscU for binding to IscS. Consequently, FeS cluster biosynthesis is decreased, resulting in a reduced level of active MoaA, which then results in the accumulation of cPMP.

DISCUSSION
The studies presented in this article show that TusA has a role in addition to tRNA thiolation in the cell. We suggest that TusA is involved in sulfur transfer for the synthesis of MPT and, in addition, is involved in the balanced regulation of the availability of IscS to various biomolecules in E. coli. In this report we studied the effects of a deletion of tusA in the cell in detail. Our studies show a pleiotropic effect of a ⌬tusA mutant on the regulation of numerous genes in the cell (Table 2 and supplemental Tables S1-S3). Microarray analysis showed an increased expression of the moaABCDE operon in the absence of tusA independent of the growth conditions analyzed. Several molybdoenzymes, like NarGHI, NarZYV, XdhD, and FdhF, were found to be expressed up to 4-fold higher in the absence of tusA under anaerobic conditions or in the stationary phase. However, the determined activity of several molybdoenzymes from different families was largely decreased, up to 95%, under aerobic conditions. Further, we ascertained that cPMP is accumulated in the ⌬tusA mutant and that this effect is stronger under aerobic conditions than in anaerobiosis. Because tusA is expressed in equal amounts under aerobic and anaerobic conditions, it could be excluded that this effect is based on different levels of TusA under various growth conditions. However, the results also showed that TusA is not essential for Moco biosynthesis, as 50% of the enzyme activities remained under anaerobic conditions. Because the expression of genes regulated by FNR (hypABCDE, narGHJI, moaABCDE, soxS, cyoABCDE, sdhAB, tdcABCDEFG, and feoB) was increased in the ⌬tusA mutant, this let us to suspect that due to the absence of TusA, the level of IscS available for FeS cluster biosynthesis is increased. A higher level of FeS clusters in the cell might stabi- lize holoFNR, thus stimulating the transcription of FNR-regulated genes (like the genes for NR). However, a higher level of FeS clusters in the cell might also reduce the sulfur transfer of IscS to other biosynthetic pathways. To test this idea, we expressed the A. vinelandii ISC operon in E. coli cells. The results showed that an overexpression of the ISC operon has the same effect as IscU alone, or in the absence of TusA, on the activity of molybdoenzymes, decreasing their activity. Thus, a shift of the IscS pool in the cell has drastic effects on various pathways in the cell. Our conclusions are consistent with the results reported recently by Maynard et al. (13); these authors identify 2-thiouridine modification of tRNA Lys as responsible for the enhanced susceptibility of viral infection by inhibition of programmed ribosomal frameshifting. They show that the ISC FeS cluster biogenesis pathway exerts a protective effect and that a higher rate of binding of IscU to IscS is responsible for determining the rate of translational frameshifting (13). Here, computational modeling suggested a competitive binding of IscU and TusA to IscS (13). As a consequence of an increased FeS cluster pathway in the ⌬tusA mutant, a protective effect during phage infection was observed, whereas an increased tRNA thiolation via TusA enhanced viral infection. Our results are consistent with their computational modeling results and the final model. Our results show that IscU and TusA have comparable dissociation constants with IscS; thus, they directly compete over binding to IscS. We suggest conclusively that one part of IscS in the cell is mostly in complex with IscU for FeS cluster biosynthesis and the other part may be available for other interaction partners such as TusA. Consistent with this hypothesis are studies by Marinoni et al. (42), who solved the crystal structure of the IscS-IscU complex from Archaeoglobus fulgidus. In the crystal structure the participation of the catalytic IscS-Cys321 as a transient ligand in the formation of the 2Fe2S cluster on IscU was resolved (42). Thus, during FeS cluster formation, a portion of IscS would not be available for acceptor proteins like TusA or ThiI. When the concentration of one of the interaction partner is changed (tested here for IscU and TusA), the IscS pool is shifted to one or the other direction, with a drastic effect on gene regulation.
With these findings, we propose a model (Fig. 6) that additionally tries to explain the differences in gene regulation observed in the micoarray data, which are very complex. We suggest that TusA is involved in sulfur transfer for the biosynthesis of Moco but, additionally, influences the FeS cluster biosynthesis in the cell. Fig. 6A shows the situation of an E. coli wild type strain under anaerobic conditions. Here, the ISC operon is induced when FeS clusters are required, and IscS forms a complex with IscU for FeS cluster biogenesis. Among others, one regulator for the expression of genes under anaerobiosis is FNR, which contains an oxygen labile 4Fe4S cluster (43). FNR activates the transcription of the moaABCDE operon and the genes for nitrate reductase in E. coli (44). The moaABCDE operon codes for proteins involved in cPMP and MPT biosynthesis, with MoaA itself harboring two 4Fe4S clusters. The other IscS pool interacts with proteins like TusA, ThiI, CyaY, or IscX, for sulfur transfer to tRNA, thiamine, or MPT. Because TusA is not essential for the formation of MPT, its role might be replaced by other sulfur transfer proteins under anaerobic conditions. One sulfur carrier is the rhodanese-like protein YnjE, which is preferentially sulfurated by IscS (5). Additionally, our results showed that the expression of ynjE is increased under anaerobic conditions and in the absence of TusA, thus making it more available for sulfur transfer.
In the ⌬tusA strain studied in this report, we propose that the pool of IscS is shifted, resulting in the misregulation of several pathways. Under aerobic conditions, generally fewer amounts of IscS are present, as the ISC operon is not induced (Fig. 6B). Also, FNR is mainly inactive, because the 4Fe4S cluster is converted to a 2Fe2S cluster, thereby inactivating FNR as a transcriptional regulator (45). In the absence of TusA, more IscS is available for the interaction with IscU, increasing the amount of FeS clusters in the cell. This results in an increased expression of the moaABCDE operon. Additionally, the amount of YnjE is decreased under aerobic conditions, and in the ⌬tusA mutant the SUF system is less expressed. This leads to an impairment of the sulfur ability of all possible sulfur donors for MPT synthase required for the conversion of cPMP to MPT, thus resulting in an inactivity of almost all molybdoenzymes.
Under anaerobic conditions, the amount of IscS is generally increased. In the absence of TusA, the IscS pool is also shifted in the direction of IscU for FeS cluster biosynthesis (Fig. 6C). Increasing amounts of FeS clusters then switch FNR to its active state, inducing the transcription of genes like soxS, moaAB-CDE, fdnGHI, dppA, and sodA (44). Our studies showed that the genes for hydrogenase 3 were induced in the ⌬tusA mutant, leading to 4-fold higher hydrogen utilization. Increased levels of FeS clusters into hydrogenase might result additionally in higher amounts of active enzyme.
The expression of different molybdoenzymes were differently affected in the ⌬tusA mutant. Because the expression profile was recorded in strains grown on nitrate, the expression was additionally regulated by the activity of NarL. The availability of nitrate leads to activation of the transcriptional regulator NarL of many anaerobic electron transport-and fermentation-related genes (46,47). The activity of NarL is dependent on the presence of nitrate, leading to its phosphorylation by NarX. NarL activates the expression of genes like narGHI, which are  FEBRUARY 22, 2013 • VOLUME 288 • NUMBER 8 required for nitrate respiration. These genes were also identified to be induced in the ⌬tusA strain under aerobic conditions. Also identified as induced in the ⌬tusA strain under anaerobic conditions (Fig. 6C). Additionally, NarL represses genes like dmsA, ynfE, and ydhV, which are involved in alternative respiratory pathways. The effect of the expression of molybdoenzymes taking together the regulation of FNR and NarL, which was determined by microarray analysis, is boxed in Fig. 6C. Further, the activity of IscR might also be affected. IscR is negatively autoregulated and contains an FeS cluster that acts as a sensor of FeS cluster assembly (48). IscR activates the expression of the SUF system leading to increased expression of SufS under anaerobic conditions (49). In our microarray experiments, the expression of the SUF system was increased in the absence of tusA under anaerobic conditions, confirming this hypothesis. Additionally, under anaerobiosis the expression of ynjE is induced. Thus, the activity of various molybdoenzymes under anaerobic conditions in the ⌬tusA strain was only reduced to a level of 50% (in the case of NR), as YnjE or SufS are available as sulfur donor for MPT biosynthesis and can replace IscS under these conditions. IscR usually shuts down the expression of the ISC operon when FeS clusters are present in sufficient amounts. However, we see an increased expression of the ISC operon in the ⌬tusA mutant. This might be due to a higher expression of several molybdoenzymes, under these conditions, which also contain FeS clusters.

Role of TusA in Synthesis of Sulfur-containing Cofactors
Conclusively, our studies show that the pleiotropic effect of a tusA deletion might be caused by changes in the FeS cluster concentration in the cell, leading to major differences in gene regulation. We propose that TusA is involved in regulating the IscS pool and shifting it away from IscU, thereby making IscS available for sulfur transfer for the biosynthesis of MPT. Whether TusA acts as a direct sulfur donor to provide the sulfur to MoeB/MoaD for Moco biosynthesis or other proteins like YnjE are involved as mediators in this reaction has to be investigated further in future studies.