Thiosulfate Transfer Mediated by DsrE/TusA Homologs from Acidothermophilic Sulfur-oxidizing Archaeon Metallosphaera cuprina

Background: The dsrE3A-dsrE2B-tusA cluster from Metallosphaera cuprina encodes putative sulfur-trafficking proteins and is flanked by genes encoding a heterodisulfide reductase-like complex and dihydrolipoamide dehydrogenase. Results: DsrE3A and TusA react with tetrathionate, yielding protein Cys-S-thiosulfonates, triggering irreversible thiosulfate transfer from DsrE3A to TusA. Conclusion: DsrE3A and TusA are important players during dissimilatory sulfur and tetrathionate oxidation. Significance: Protein-mediated transfer of thiosulfonate is unprecedented in sulfur oxidizers.

Elemental sulfur (S 0 ) and reduced inorganic sulfur compounds serve as energy sources and electron donors for a number of chemo-and photolithotrophic bacteria such as Acidithiobacillus species (1)(2)(3) and Allochromatium vinosum (4). Dissimilatory sulfur oxidation also occurs in the archaeal domain of prokaryotes and is well known for chemolithotrophic acidophiles such as Sulfolobus, Acidianus, and Metallosphaera. Species of the genus Metallosphaera typically grow by aerobic respiration on CO 2 with S 0 , pyrite, and tetrathionate (S 4 O 6 2Ϫ ) as electron donors (5,6). The best characterized archaeal enzyme involved in sulfur oxidation is probably sulfur oxygenase reductase, identified in Acidianus and present also in some Sulfolobus species. In vitro the enzyme catalyzes disproportionation of S 0 into sulfide, sulfite, and thiosulfate (7)(8)(9). Sulfur oxygenase reductase is not present in Metallosphaera (10,11).
Thiosulfate and tetrathionate are important intermediates that play key roles during sulfur oxidation by bacteria and archaea. Although the periplasmic Sox multienzyme for thiosulfate degradation is widespread in bacterial sulfur oxidizers, it is not found in acidophilic sulfur-oxidizing archaea (12,13). Instead, in organisms such as Acidianus ambivalens two thiosulfate molecules are oxidatively condensed to tetrathionate in a reaction catalyzed by the membrane-bound cytoplasmically oriented thiosulfate:quinone oxidoreductase (TQO) 5 (14). Although TQO is also present in a few bacteria, the main catalyst of tetrathionate formation in the Bacteria domain appears to be the soluble, periplasmic c-type cytochrome TsdA (15,16). Sulfide:Quinone oxidoreductase is a widespread sulfide-oxidizing enzyme not only in bacteria but also in archaeal sulfur oxidizers like Metallosphaera cuprina (11). In the genera Acidianus and Metallosphaera, electrons from sulfide as well as from thiosulfate are thus fed into the quinone pool and coupled to ATP generation via oxidative phosphorylation (17).
Many sulfur-oxidizing bacteria form conspicuous sulfur globules as intermediates during the oxidation of sulfide, polysulfides, or thiosulfate. The sulfur globules are deposited either extracellularly or intracellularly in the periplasm (e.g. in Allochromatium (18) or Beggiatoa (19) species). In A. vinosum, the degradation of the sulfur globules involves essential steps in the cytoplasm and is catalyzed by soluble and membrane-bound proteins of the Dsr system (18, 20 -22). It is well established that the Dsr mechanism involves transport of sulfur into the cytoplasm and an extensive sulfur trafficking network. DsrC is the final sulfur-accepting protein, and in its persulfurated form it serves as a direct substrate for dissimilatory sulfite reductase (DsrAB), the enzyme that catalyzes the formation of sulfite. DsrC receives sulfur from DsrEFH, which in turn is sulfurated by TusA. Sulfane sulfur is mobilized from low molecular weight persulfides and transferred to TusA by a rhodanese-like protein. Furthermore, the whole process possibly involves a DsrElike protein, termed DsrE2, encoded in the same gene cluster (rhd-tusA-dsrE2) (23,24). Notably, an rhd-tusA-dsrE2 or at least a tusA-dsrE2 arrangement also occurs in many photo-and chemotrophic sulfur oxidizers that do not contain DsrC and the Dsr pathway (25,46). Those sulfur oxidizers include archaeal sulfur oxidizers such as Acidianus hospitalis (26), Sulfolobus tokodaii (27), Metallosphaera sedula (28), and M. cuprina (11), as well as bacterial sulfur oxidizers such as members of the family Aquificaceae (29,30) and the genera Acidithiobacillus (31) and Thioalkalivibrio (32) (Fig. 1). Inevitably, in this group the putative tusA-dsrE2 genes are linked with the gene cluster hdrC1B1AhyphdrC2B2 that encodes a possible heterodisulfide-reductase complex. This complex has been predicted to be responsible for the oxidation of organic persulfides to sulfite in Acidithiobacillus ferrooxidans based on the observation that the tusA-dsrE2-hdr genes were transcriptionally up-regulated when elemental sulfur was utilized as energy source (25). Additionally, transcription of dsrE2-, tusA-, or hdr-like gene was also up-regulated in M. sedula when S 0 or tetrathionate was provided as an electron donor (28).
This study aimed at gaining information about the function and biochemical properties of DsrE-and TusA-like proteins in the acidothermophilic archaeon M. cuprina (6). To this end, a bioinformatics approach was combined with in vitro studies that demonstrated not only tight interaction but also transfer of thiosulfate between archaeal TusA and one of the DsrE homologs.

EXPERIMENTAL PROCEDURES
Strains, Plasmids, Media, and Growth Conditions-Bacterial strains and plasmids used in this study are listed in Table 1. M. cuprina Ar-4 was cultivated at 65°C and pH 3.0 in modified Allen medium with 1 g/liter yeast extract (6,33). Escherichia coli DH5␣ was used for molecular cloning of genes (Table 1). E. coli BL21(DE3)/pLysRARE was used for overproduction of recombinant proteins. All E. coli strains were grown in Luria-Bertani medium at 37°C. The antibiotics used were at 100 g/ml for ampicillin and 25 g/ml for chloramphenicol.
Genetic Cloning and Site-directed Mutagenesis-The targeted genes were PCR-amplified with Pfu DNA polymerase using genomic DNA of M. cuprina Ar-4 as template. Primers are listed in Table 1. The PCR product was purified after digestion and was cloned into vectors pET15b (Novagen, Darmstadt, Germany) and pPRIBA1 (IBA GmbH, Göttingen, Germany). Positive recombinant plasmids were selected and verified by nucleotide sequencing. Plasmids used in this study are listed in Table 1.
Replacements of cysteine residue by serine residue (site-directed mutations) were performed with gene splicing by overlap extension (34). Plasmids carrying targeted genes were used as PCR template. Forward and reverse primers (Table 1) complementary to the plasmid sequence (pET15b or pPRIBA1) were 700 -1000 bp upstream or downstream from the target gene. The double mutant dsrE3A-C 93 S/C 101 S was constructed by introducing the C 101 S mutation to pET15b-dsrE3A-C 93 S. All of the genetic constructions were sequenced to exclude any PCR amplification errors.
Molecular Evolutionary Analysis-A molecular evolutionary analysis was performed using MEGA6 software (35). The phylogeny test was executed with UPGMA (unweighted pair group method with arithmetic mean) (36), a statistical method that applies a bootstrap test with 10,000 replicates (37). The evolutionary distances were computed using the Poisson correction method (38).
Overproduction and Purification of the Recombinant Proteins-Recombinant DsrE2B, DsrE3A, and TusA were produced in E. coli BL21(DE3)/pLysRARE. Overnight precultures were used to inoculate fresh LB medium with a ratio of 1:60 (v/v). Synthesis of recombinant proteins in cells was induced by the addition of 0.1 mM IPTG when the culture reached an OD 600 of 0.6 -0.8 and was further incubated for 2.5 h at 37°C before harvesting. Cells were pelleted at 3000 ϫ g for 10 min, and were resuspended in buffer containing 50 mM Tris-HCl and 150 mM NaCl (pH 7.5). Cells were broken by sonification. Cel-lular lysate was centrifuged at 17,000 ϫ g for 20 min at 4°C. The supernatant was incubated in a water bath at 65°C for 10 min, and the denatured proteins were removed by centrifugation at 17,000 ϫ g for 20 min. The supernatant containing the recombinant protein was filtered with a 0.45-m filter membrane (Millipore, Darmstadt, Germany) before being loaded to the gravity flow column for purification. His-tagged and Streptagged proteins were purified with TALON metal affinity resin (Clontech) and Strep-Tactin Superflow (IBA), respectively, according to protocols provided by the suppliers.

TABLE 1 Archaeal and bacterial strains, plasmids, and primers used in this study
Sites that recognize restriction enzyme are underlined. The italic nucleotides are codons for serine.

Strain, plasmid, or primer
Description Reference or source Visualization of cysteine modification of cystein residues by 1,5-I-AEDANS proceeded as follows. 5-10 l of the concentrated sample was treated with 3 l of 2 mM 1,5-I-AEDANS at 4°C for at least 1 h before adding 2 l of 8 mM L-cysteine at room temperature for 30 min to react with the excessive 1,5-I-AEDANS. 1 l of 100 mM DTT was added to the reaction mixture as described (41). Our results indicated that treatment with DTT did not affect reaction of proteins with 1,5-I-AEDANS. Native loading buffer was applied to the sample followed by electrophoresis on 15% Tris-glycine SDS-polyacrylamide gels in the dark. The fluorescence of 1,5-AEDANS was detected under UV light, and proteins were stained with Coomassie Brilliant Blue R-250.
Determination of Thiosulfate Transfer-In a typical thiosulfate transfer assay, 1.5 nmol of thiosulfate donor proteins (DsrE3A or TusA after reaction with substrate) was incubated with equal amounts of thiosulfate acceptor proteins (TusA or DsrE3A) at 65°C for 30 min. Thiosulfate transfer was evaluated by determination of cysteine modification, and visualization of cysteine modification was carried out as described above.
MALDI-TOF Mass Spectrometry-For MALDI-TOF MS, the matrix was sinapinic acid in 50% acetonitrile and 0.1% trifluoroacetic acid solution. The buffer of protein samples was exchanged for 0.1% trifluoroacetic acid by using a PD MiniTrap G-25 column. About 10-pmol samples were detected in the positive linear mode with a Biflex III (Bruker Daltonics GmbH, Leipzig, Germany) or AB SCIEX TOF/TOF 5800 (AB SCIEX, Framingham, MA).
Strep-tag Pulldown Assay-During the Strep-tag pulldown assay, 10 nmol of Strep-tagged proteins and 30 nmol of non-Strep-tagged proteins were incubated together with 0.75 ml of Strep-Tactin Superflow on ice for 1 h. The mixture was then loaded to the gravity flow column to continue the pulldown assay. Reference tests were run in parallel under identical conditions, but only the non-Strep-tagged protein was incubated with Strep-Tactin Superflow.
Surface Plasmon Resonance-A Biacore 3000 instrument (GE Healthcare) was equipped with a CM5 sensor chip (GE Healthcare) at 25°C. DsrE3A (26 g/ml) proteins (ligand) in 10 mM acetic acid (pH 5.5) were covalently immobilized to the chip according to the protocol provided by the supplier, and the resonance units reached about 1500. PBST buffer (PBS containing 0.005% Tween 20 (pH 7.4)) was used as the running buffer with a flow rate of 30 l/min. TusA proteins (analyte) in PBST buffer was injected for 2 min at a flow rate of 30 l/min. Dissociation of protein-protein complexes on the sensor surface proceeded for 8 min by flowing PBST buffer over the sensor surface. A blank injection with PBST buffer was used as the control. The surface was regenerated with 5-60 l of 10 or 20 mM NaOH. A control surface without ligand was used as a reference.
Gel Filtration-DsrE3A and TusA were incubated on ice for at least 1 h before being injected with a 2-ml sample loop to AKTA TM purifier equipped with a HiLoad 16/60 Superdex 200 prep grade column (GE Healthcare). A solution of 50 mM Tris-HCl (pH 7.5) and 150 mM NaCl was used as the running buffer, and the flow rate was kept constant at 1.0 ml/min. Proteins were incubated with or without 25 M Tris(2-carboxyethyl)phosphine (TCEP) at room temperature for 1 h before injection to the AKTA TM purifier with a 0.5-ml sample loop for oligomerization analysis. The column used in these cases was a HiLoad   (Mcup_0678). Notably, a lipoamide-binding protein resembling protein H of the glycine cleavage system (Mcup_0662) and several proteins responsible for the biosynthesis of lipoamide-containing proteins (Mcup_0671-0673) are also encoded in the vicinity of these genes (Fig. 2B). In fact, related genes are also part of or reside in the immediate vicinity of all the tusA-dsrE-hdr For clarity, only those parts of the alignments are shown that contain conserved cysteine residues. The complete alignment served as the basis for the tree shown in C. In C, the evolutionary history was inferred using the UPGMA method. The optimal tree with the sum of branch length ϭ 7.16951712 is shown. The percentages of replicate trees in which the associated taxa clustered together in the bootstrap test (10,000 replicates) are shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the Poisson correction method and are in the units of the number of amino acid substitutions per site. The analysis involved the 19 amino acid sequences aligned in A. genetic clusters in other genome-sequenced archaeal and bacterial sulfur oxidizers. As an example, the organization of the respective genes in Acidithiobacillus caldus is compared with that in M. cuprina in Fig. 2B.

Occurrence, Genetic Environment, and Grouping of Archaeal
Sequence alignments and phylogenetic analyses provided us with the basis for grouping DsrE homologs into the following five categories (Fig. 2C). 1) The DsrE group consists of subunits of DsrEFH (prototype Alvin_1253). 2) The DsrE2 group con-sists of members that are genetically or functionally linked with Dsr or Hdr systems. Subgroup DsrE2A (prototype Alvin_2601) contains two strictly conserved cysteines. The prototype of subgroup DsrE2B is Mcup_0682. 3) The genes of the DsrE3 group are either immediately linked with genes for dihydrolipoamide dehydrogenase (DsrE3A, prototype Mcup_0681) or with genes for lipoamide-binding proteins (DsrE3B, prototype Atc_2345). The proteins of group DsrE3A contain two conserved cysteine residues in a Cys-X 7 -Cys motif with the first cysteine corresponding to the established DsrE active site cysteine ( Fig. 2A). 4) The members of the DsrE4 group are encoded downstream of sulfide:quinone oxidoreductase (prototype Mcup_1724). 5) Group DsrE5 is represented by Mcup_1706.
Similar genetic organizations (i.e. dsrE-tusA) were observed in sulfur-oxidizing species of the families Sulfolobaceae, Aquificaceae, Acidithiobacillaceae, Chromatiaceae, and Chlorobiaceae (Fig. 1). It appears that the dsrE gene might have been duplicated (as dsrE3A and dsrE2B) in members of the Sulfolobaceae and Aquificaceae. Genes encoding proteins or enzymes that are involved in reversible reduction of heterodisulfide bonds coupled with energy conservation (Hdr complex) or in sulfur oxidation (rhodanese (Rdh)-like protein) were found in conjunction with the dsrE-tusA cluster as shown in Fig. 1     vitro. DsrE3A and TusA were not modified by NaSH, GSSH, polysulfide, thiosulfate, or sulfite, but they reacted with tetrathionate. As shown in Fig. 3, fluorescence was not seen for tetrathionate-incubated DsrE3A and was substantially lower than after treatment with the other tested sulfur group donors for TusA. The residual fluorescence of tetrathionate-treated TusA might be either due to partial reactivity of TusA or a greater susceptibility of the TusA-Cys 18 -S-thiosulfate to hydrolysis. Control experiments showed that the tetrathionate-treated proteins were a priori unable to react with 1,5-I-AEDANS, pointing to a modification of cysteines with sulfonate or S-thiosulfonate rather than sulfane groups. This conclusion was verified by an independent experimental approach. Mass changes arising after incubation with the different tested sulfur compounds were analyzed by MALDI-TOF mass spectrometry (Fig. 4). Upon incubation with tetrathionate, DsrE3A gained a mass of 226 Da, which corresponded to a tetrathionate group or two thiosulfate groups (Fig. 4, A and B). The protein harbors two conserved cysteine residues, Cys 93 and Cys 101 . Mutants DsrE3A-C 93 S (DsrE3A-Cys 101 ) and DsrE3A-C 101 S (DsrE3A-Cys 93 ) retained the ability to interact with tetrathionate, and each mutant protein covalently attached a group of 112 Ϯ 4 Da, matching the molecular mass of thiosulfate ( Table 2). The DsrE3A-C 93 S/C 101 S derivative lacking Cys 93 as well as Cys 101 stayed unmodified after incubation with tetrathionate (Table  2). Thus, both cysteine residues, Cys 93 and Cys 101 , are individually and independently modified by attachment of a thiosulfate group upon incubation with tetrathionate as shown in Scheme 1. TusA from M. cuprina was also proven by mass spectrometry to form a Cys-S-thiosulfonate derivative upon incubation with tetrathionate (Fig. 4, C and D). A mass increase of 111 Da suggested that TusA also covalently attached a thiosulfate group. Because the mutant protein TusA-C 18 S lost the ability to react with the tetrathionate, it is deduced that the conserved cysteine residue of TusA played a key role in the reaction with tetrathionate ( Table 2).
DsrE3A-Cys-S-Thiosulfonate Transfers a Thiosulfate Group to TusA-Our previous experiments described above established that DsrE3A and TusA were capable of mobilizing thiosulfate from tetrathionate. In the next step we set out to investigate whether DsrE3A-Cys-S-thiosulfonate could serve as a thiosulfate donor for TusA.
DsrE3A-Cys-S-thiosulfonate and TusA were mixed at molar ratio of 3:2 and incubated at 65°C for 30 min. Visualization of TusA with 1,5-I-AEDANS was not successful (Fig. 5A). MALDI-TOF MS confirmed thiosulfate transfer from DsrE3A-Cys-S-thiosulfonate to TusA (Fig. 5B). Not only were the peaks representing TusA and DsrE3A-Cys-S-thiosulfonate observed, but a newly emergent peak matched the mass of TusA-Cys 18 -S-thiosulfonate, which was deduced to be a product resulting from the transfer of a thiosulfate group from DsrE3A-Cys-Sthiosulfonate to TusA.
Mutants of DsrE3A were also assayed. The results (Table 3) showed that the mutated protein carrying a single replacement  of cysteine residues, i.e. DsrE3A-C 93 S or DsrE3A-C 101 S, retained its ability to transfer thiosulfate to TusA.
TusA-Cys 18 -S-Thiosulfonate Does Not Transfer Thiosulfate to DsrE3A, and DsrE3A Cleaves TusA-Cys 18 -S-Thiosulfonate-We further determined whether TusA-Cys 18 -S-thiosulfonate was able to transfer its thiosulfate group to DsrE3A. The thiosulfate group of TusA-Cys 18 -S-thiosulfonate did not transfer to DsrE3A, but no trace of it was detected in the MALDI mass spectrum after incubation (Fig. 6A). Thus, we deduced that DsrE3A cleaved the TusA-Cys 18 -S-thiosulfonate and released free TusA. Additional experiments (Fig. 6B) showed that the double replacement mutant DsrE3A-C 93 S/C 101 S lost the ability to cleave the thiosulfate group from TusA-Cys 18 -S-thiosulfonate, suggesting that cysteine residues of DsrE3A played an important role during cleavage.
We also tested whether DsrE3A-Cys-S-thiosulfonate and TusA-Cys 18 -S-thiosulfonate were able to transfer their thiosulfate groups to DsrE2B. Thiosulfate transfer to DsrE2B was not observed (data not shown).
DsrE3A and TusA Interact Physically with Each Other and Form a Heterocomplex-As demonstrated above, DsrE3A and TusA both reacted with tetrathionate resulting in DsrE3A-Cys-S-thiosulfonate and TusA-Cys 18 -S-thiosulfonate. In addition, DsrE3A cleaved TusA-Cys 18 -S-thiosulfonate. These results invoked the idea that DsrE3A and TusA might interact physically with each other. Pulldown assays were conducted for DsrE3A plus Strep-tagged TusA with Strep-Tactin Superflow gravity flow columns. In a control experiment, DsrE3A did not bind to the affinity matrix, whereas a considerable portion was retained on the column in the presence of Strep-tagged TusA and co-eluted with it (data not shown). DsrE3A did not interact with TusA-C 18 S, in which the active site cysteine was replaced by serine, showing that Cys 18 of TusA is indispensable for the interaction. Analysis by gel permeation chromatography showed that DsrE3A and TusA formed a heterocomplex after they were incubated together (Fig. 7). These results suggested a specific interaction between DsrE3A and TusA. Using the same methods, interactions were not observed between DsrE2B and TusA or DsrE3A.
Moreover, surface plasmon resonance was applied to quantitatively evaluate the affinity between DsrE3A and TusA (

DISCUSSION
This work demonstrated that proteins DsrE3A and TusA from the acidothermophilic archaeon M. cuprina Ar-4 have the ability to mobilize thiosulfate from tetrathionate. Moreover, thiosulfate transfer from DsrE3A-Cys-S-thiosulfonate to TusA was shown. To our knowledge, proteins that react with tetrathionate and form protein-Cys-S-thiosulfonates have thus far not been reported in sulfur oxidizers. Thus, DsrE3A and TusA from M. cuprina Ar-4 represent the first pair of proteins with such novel properties. Although both DsrE3A and TusA were able to react with tetrathionate, we observed that DsrE3A-Cys-S-thiosulfonate further transferred its thiosulfate group to TusA. This observation might imply that DsrE3A functions as a thiosulfate donor to TusA in vivo. Thiosulfonated TusA could then serve as the substrate for other enzymes such as the hdrC1B1AhyphdrC2B2-encoded proteins (Fig. 9). We envision  that the sulfonate group is first released, either hydrolytically as sulfate or reductively as sulfite, by an as yet unknown mechanism. The sulfane group remaining on TusA could then be oxidized and finally also released.
The close genomic linkage of the TusA-encoding gene (Mcup_0683) with hdr-like genes and the DsrE3A-encoding gene (Mcup_0681) with a lipoamide dehydrogenase-encoding gene, not only in M. cuprina but also in other sulfur-oxidizing archaea and bacteria (Figs. 1 and 2B), opens the possibility of a functional linkage of these systems. This idea is corroborated by the tight interaction of archaeal TusA and DsrE3A proven in this work by several independent experimental approaches. Involvement of a lipoamide-binding protein as a potential sulfur carrier and linkage to lipoamide dehydrogenase could even result in transfer of some of the electrons arising from sulfane sulfur oxidation to NAD ϩ .
In addition to the reaction with tetrathionate that leads to binding of a thiosulfonate group, we showed that DsrE3A is able to release thiosulfate from TusA-Cys 18 -S-thiosulfonate. This reaction requires two electrons. In principle, such a mechanism resembles the reverse of the oxidative binding of thiosulfate to a cysteine of the SoxYZ protein, which occurs as the first step of thiosulfate oxidation catalyzed by the Sox system (42). We envision that the electrons required stem from the formation of an intramolecular disulfide between the two conserved cysteine residues of DsrE3A or from the formation of an intermolecular disulfide between two molecules of DsrE3A. The formation of an intermolecular disulfide is supported by the observation that single cysteine replacement mutants DsrE3A-C 93 S and DsrE3A-C 101 S retained the ability to release thiosulfate from TusA-Cys 18 -S-thiosulfonate.
Our results demonstrated that TusA is involved in thiosulfate transfer. Based on the finding that TusA reacts with tetrathionate and that DsrE3A cleaves a thiosulfate group from TusA-Cys 18 -S-thiosulfonate, we propose that TusA is involved in dissimilatory oxidation of tetrathionate in M. cuprina Ar-4 when grown with tetrathionate as energy source, a role distinct from the function of TusA as a sulfurtransferase in tRNA modification (43) and molybdenum cofactor biosynthesis (44) in E. coli. TusA of M. cuprina Ar-4 functions as a dissimilatory protein, just as has been reported for TusA from the purple sulfur bacterium A. vinosum (24).
As an alternative function, or even in addition to the model elaborated above, archaeal TusA could generate thiosulfate in the cytoplasm, which would be further oxidized by TQO (Fig.  9). TQO oxidizes thiosulfate and transfers electrons via caldariellaquinone (14). Genes coding for TQO are present in the genome of M. cuprina Ar-4 (11), and the active site of TQO in A. ambivalens has been suggested to face toward the cytoplasm (45). Tetrathionate produced by TQO is thus released to the cytoplasm, where DsrE3A is located, and further converts tetrathionate (Fig. 9).