Two different pathways for thiosulphate oxidation are present in the purple sulphur bacterium Allochromatium vinosum: oxidation to tetrathionate and complete oxidation to sulphate with obligatory formation of sulphur globules as intermediates. The tetrathionate:sulphate ratio is strongly pH-dependent with tetrathionate formation being preferred under acidic conditions. Thiosulphate dehydrogenase, a constitutively expressed monomeric 30 kDa c-type cytochrome with a pH optimum at pH 4.2 catalyses tetrathionate formation. A periplasmic thiosulphate-oxidizing multienzyme complex (Sox) has been described to be responsible for formation of sulphate from thiosulphate in chemotrophic and phototrophic sulphur oxidizers that do not form sulphur deposits. In the sulphur-storing A. vinosum we identified five sox genes in two independent loci (soxBXA and soxYZ). For SoxA a thiosulphate-dependent induction of expression, above a low constitutive level, was observed. Three sox-encoded proteins were purified: the heterodimeric c-type cytochrome SoxXA, the monomeric SoxB and the heterodimeric SoxYZ. Gene inactivation and complementation experiments proved these proteins to be indispensable for thiosulphate oxidation to sulphate. The intermediary formation of sulphur globules in A. vinosum appears to be related to the lack of soxCD genes, the products of which are proposed to oxidize SoxY-bound sulphane sulphur. In their absence the latter is instead transferred to growing sulphur globules.
Thiosulphate (S2O32–) is a rather stable and environmentally abundant sulphur compound of intermediate oxidation state and fulfils an important role in the natural sulphur cycle (Jørgensen, 1990; Podgorsek and Imhoff, 1999; Sorokin et al., 1999). Most phototrophic and chemotrophic sulphur oxidizers are able to use thiosulphate. In many of these organisms, tetrathionate is the end-product of thiosulphate oxidation. It is formed by oxidative condensation of two thiosulphate anions catalysed by thiosulphate dehydrogenase (EC 18.104.22.168; thiosulphate:acceptor oxidoreductase) (Trudinger, 1961). ΔG0′ for the electron-donating, formally hydrogen-forming reaction is +84.5 kJ mol−1 (Thauer et al., 1977). This pathway has been especially well studied in a group of chemo-organotrophic bacteria that use thiosulphate as a supplemental but not as the sole energy source, e.g. some Pseudomonas and Halomonas species (Jørgensen, 1990; Podgorsek and Imhoff, 1999; Sorokin et al., 1999). More widespread than tetrathionate formation from thiosulphate, however, is complete oxidation of thiosulphate to sulphate (ΔG0′ for the electron-donating, formally hydrogen-forming reaction is +210 kJ mol−1; Thauer et al., 1977). Two different pathways appear to exist: in the first, both sulphur atoms of thiosulphate are oxidized to sulphate without the appearance of sulphur deposits as intermediates. It occurs in a number of facultatively chemo- or photolithotrophic organisms like Paracoccus pantotrophus or Rhodovulum sulfidophilum (Appia-Ayme et al., 2001; Friedrich et al., 2001; 2005). The second involves the formation of conspicuous globules of polymeric, water-insoluble sulphur and occurs in many environmentally important sulphur oxidizers including the phototrophic green and purple sulphur bacteria and free-living and symbiotic chemotrophic sulphur oxidizers like Beggiatoa or Thiothrix (Dahl, 1999; Dahl and Prange, 2006).
In organisms that do not form sulphur deposits from thiosulphate, a periplasmic thiosulphate-oxidizing multienzyme complex (Sox) has been found and characterized through pioneering work by the groups of Kelly (Lu et al., 1985) and Friedrich (Wodara et al., 1997; Friedrich et al., 2000; 2001; Rother et al., 2001) in Paracoccus versutus and P. pantotrophus. This enzyme complex appears to be of importance in many thiosulphate-oxidizing bacteria (Appia-Ayme et al., 2001; Friedrich et al., 2001; Kappler et al., 2001; Petri et al., 2001). In P. pantotrophus the Sox complex is essential for thiosulphate oxidation in vivo and catalyses a reduction of cytochrome c coupled to the oxidation of thiosulphate, sulphide, sulphite and elemental sulphur in vitro. The proposed mechanism for thiosulphate oxidation requires four different proteins: SoxB, SoxXA, SoxYZ and SoxCD (Friedrich et al., 2001). The heterodimeric SoxYZ has been identified as the substrate-binding molecule of the complex (Quentmeier and Friedrich, 2001). SoxXA is a heterodimeric c-type cytochrome that is reduced while oxidatively coupling the sulphur compound to SoxYZ. The monomeric, manganese-containing SoxB has been proposed to act as a sulphate thiol esterase or sulphate thiol hydrolase (Friedrich et al., 2005) and is responsible for hydrolytic cleavage of a sulphate group from the bound sulphur substrate. SoxCD oxidizes the remaining sulphane sulphur, acting as a sulphur dehydrogenase. Further action of SoxB releases a second sulphate molecule and thereby restores SoxYZ (Friedrich et al., 2001).
Despite intensive research over several decades, much less is known about thiosulphate oxidation involving the intermediate deposition of sulphur either inside (e.g. Beggiatoa, Chromatiaceae) or outside (e.g. green sulphur bacteria, Ectothiorhodospiraceae) of the cells. Studies with radioactively labelled thiosulphate in purple sulphur bacteria demonstrated very clearly that the more reduced sulphane and the more oxidized sulphone sulphur atoms are oxidized by different pathways (Smith and Lascelles, 1966; Trüper and Pfennig, 1966). Only the sulphane sulphur accumulates as stored sulphur [S0] before further oxidation, whereas the sulphone sulphur is rapidly converted into sulphate. This led to the conclusion that thiosulphate oxidation is initiated by a – probably reductive – cleavage of the molecule. It has been a very long-held conviction that the observed cleavage is brought about by the action of rhodaneses or thiosulphate reductases (Brune, 1989; 1995a; Dahl, 1999; Brüser et al., 2000). Both enzymes can function as thiosulphate sulphur transferases and release sulphite and sulphide when suitable reduced thiol acceptors like glutathione or dihydrolipoic acid are present. However, the physiological role of these enzymes remained dubious. Enzymes exhibiting thiosulphate sulphur transferase activity occur in many bacteria and eukarya not able to oxidize thiosulphate (Westley et al., 1984; Le Faou et al., 1990) and genetic studies were lacking. During the past several years data accumulated that cast further doubt on an important role of rhodanese or thiosulphate reductase during thiosulphate oxidation in sulphur-storing bacteria: Clusters of sox genes were identified in thiosulphate-oxidizing green sulphur bacteria (Eisen et al., 2002; Vertéet al., 2002) and PCR analyses proved the existence of the highly conserved soxB gene also in other thiosulphate-utilizing but not in non-thiosulphate-utilizing strains of green sulphur bacteria (Petri et al., 2001).
Allochromatium vinosum, a purple sulphur bacterium of the family Chromatiaceae, is an ideal model organism for studying thiosulphate oxidation, as it is metabolically versatile and genetically accessible (Pattaragulwanit and Dahl, 1995; Dahl, 1996). Under neutral to slightly acidic growth conditions A. vinosum oxidizes part of the thiosulphate present to tetrathionate. In a second pathway, thiosulphate can also be oxidized to sulphate with the formation of sulphur globules as an obligate intermediate (Trüper and Pfennig, 1966; Pattaragulwanit et al., 1998; Pott and Dahl, 1998; Prange et al., 2002). In this work we purified proteins essential for both pathways. Furthermore, we identified sox genes in A. vinosum, the mutational analysis of which unambiguously showed that earlier models for thiosulphate oxidation via intermediate sulphur formation have to be revised. Clusters of sox genes very similar to those of A. vinosum have been identified in the only distantly related green sulphur bacteria. This suggests that the mechanism of thiosulphate oxidation via sulphur deposition is evolutionarily highly conserved and that studies in A. vinosum have broad implications on the thiosulphate oxidation pathways in other phototrophic and chemotrophic sulphur-storing bacteria.
Characterization of thiosulphate dehydrogenase
The purification procedure outlined in Experimental procedures resulted in an approximately 2500-fold purification with a recovery of 15% (Table S1, Supplementary material). The apparent molecular mass of the enzyme was determined to be 30 kDa by gel filtration chromatography on Sephadex 75. The apparent isoelectric point was determined to be 4.2 by chromatofocussing on MonoP (Amersham-Pharmacia Biotech). Native PAGE revealed the presence of one strong band that was identified as thiosulphate dehydrogenase by activity staining (Fig. S1, Supplementary material) and one weak band that did not show activity. SDS-PAGE analysis yielded two protein bands of 15 and 30 kDa. The N-terminal sequence AIERTLSIIKPNAVAKDAIG of the 15 kDa protein representing the weak band in native PAGE exhibited significant similarity to bacterial nucleoside diphosphate kinases. We conclude that this protein is a contamination of thiosulphate dehydrogenase that we could not completely remove from our preparations. An N-terminal sequence of EEPPYVALTVPAAALLPDGALGGSIVRGRYLSDTPAQLPRDFVGN was determined for the 30 kDa protein. It exhibits similarity to amino acids 28–67 of a hypothetical c-type cytochrome from Ralstonia metallidurans (Accession No. ZP_00022700; 20 of 45 amino acids are identical). The first 24 amino acids of the latter protein constitute a typical Sec-dependent leader peptide mediating transport across the cytoplasmic membrane (Fekkes and Driessen, 1999). Accordingly, A. vinosum thiosulphate dehydrogenase was also proven to be a periplasmic c-type cytochrome. Haem staining after SDS-PAGE revealed that the 30 kDa but not the 15 kDa band contained haem (Fig. S1, Supplementary material). The UV-visible absorption spectrum of the thiosulphate-reduced enzyme gave maxima at 278, 419 (γ-band), 523 (β-band) and 554 nm (α-band) (Fig. 1) indicating the presence of c554 haem. The enzyme was reducible by thiosulphate at pH 5.0 but not at pH 7.0. Taken together, these results indicate that A. vinosum thiosulphate dehydrogenase is a monomeric 30 kDa c-type cytochrome.
The end-product of thiosulphate oxidation in the presence of ferricyanide as artificial electron acceptor by thiosulphate dehydrogenase was tetrathionate. HPLC analysis revealed the stoichiometric conversion of 0.54 ± 0.02 mM thiosulphate into 0.27 ± 0.02 mM tetrathionate. Thiosulphate was not formed after the addition of excess tetrathionate (up to 4 mM) to a mixture of enzyme and ferrocyanide at pH values of 4.25, 5.0 and 7.0 indicating that the reaction is practically irreversible under these conditions. The pH optimum for the enzyme was determined in 100 mM acetate buffer over a range of pH 3.5–5 in 0.25 pH unit increments and found to be 4.25 where activity was 5 times higher than at pH 5.0. At the latter pH, activity was identical in 100 mM acetate and 100 mM phosphate buffer. At pH 7.0, activity in phosphate buffer further decreased to only one tenth of that at pH 5.0.
We initiated an examination of the kinetic properties of A. vinosum thiosulphate dehydrogenase. In the case of enzymes that use two molecules of the same substrate (here: thiosulphate), the primary v versus [S] plots provide the best way to examine data (Segel, 1993). The family of experimental v versus [thiosulphate] plots for thiosulphate dehydrogenase is shown in Fig. 2A. The Vmax obtained from each curve fit is an apparent maximal velocity (Vmax,app) for the fixed subsaturating ferricyanide level used. At 0.8 mM ferricyanide the apparent Vmax is about 22 000 units (mg protein)−1 (on the basis of a molecular mass of 30 kDa kcat is about 1.1 × 104 s−1), the apparent [S]0.5 is about 1 mM. At none of the experimental levels of ferricyanide does thiosulphate display strong substrate inhibition. A plot of Vmax,app versus [ferricyanide] is shown in Fig. 2B. Fitting this replot to the Hill equation allowed an estimate of the following constants: the limiting Vmax is about 34 000 units (mg protein)−1 (corresponding to a kcat of 1.7 × 104 s−1), the [S]0.5 for ferricyanide is about 0.5 and n is about 1.4. For an enzyme that binds two molecules of the same substrate either randomly or in order before any product is formed an n-value between 1.3 and 1.5 is indeed expected when there is no strong substrate inhibition (Segel, 1993). The family of v versus [ferricyanide] plots does show substrate inhibition (Fig. 2C). The fact that the peak velocity appears to move to higher ferricyanide levels as the fixed thiosulphate concentration is increased is consistent with competition between ferricyanide as an inhibitor and thiosulphate as a substrate. Thiosulphate dehydrogenase activity was significantly inhibited by sulphite. In 100 mM acetate buffer, pH 4.25, in the presence of 8 mM thiosulphate and 1 mM ferricyanide 50% inhibition of activity was observed at 80 μM. Cytochrome c from yeast is used as electron acceptor by the enzyme, under optimized assay conditions (20 mM acetate buffer, pH 5.5, 0.4 mM thiosulphate, 25 μM yeast cytochrome c) the activity amounted to 230 μmol cytochrome c reduced min−1 (mg protein)−1. Horse heart cytochrome c is not accepted. Further detailed investigations are necessary to clarify the kinetic mechanism of the enzyme and will be subject of a future communication.
Nucleotide sequence analysis of A. vinosum sox gene loci
Sequencing of two independent DNA fragments (11 015 and 3395 bp) containing sox genes from A. vinosum revealed the presence of a total of 15 open reading frames (Fig. 3). Five of these resemble genes encoding polypeptides that are part of the thiosulphate-oxidizing multienzyme complex of P. pantotrophus (Friedrich et al., 2001). The genes soxB and soxXA are located on the longer fragment and transcribed divergently. In the 325 bp intergenic region two putative promoter sequences were identified: the putative soxB promoter is situated 72 bp upstream of the corresponding start codon, the putative soxX promoter is located 244 bp upstream of the corresponding start codon. An inverted repeat with a potential for formation of a hairpin loop structure (free energy of formation −118.9 kJ mol−1) was found inside of ORFc. This structure could function as transcription terminator; however, poly(T) sequences located directly downstream of such hairpin loops in typical eubacterial rho-independent transcription terminators (Reynolds et al., 1992) are not present. The genes soxYZ are located on the shorter sequenced fragment. A promoter search revealed a potential promoter responsible for the transcription of soxYZ inside ORFd. The −10 box is located 299 bp upstream of the soxY start codon. Genes with similarity to soxC or soxD are not present in the vicinity of the other A. vinosum sox genes. We were furthermore unable to detect these genes by low stringency hybridization with the respective genes from P. pantotrophus (U. Kappler and C. Dahl, unpublished).
The results of the sequence analyses of the proteins deduced from all sequenced potential genes are summarized in Table 1. For the soxB gene three different possible start codons were detected at nt 4095, 4083 and 4074. A potential ribosome binding site AGGAGG is situated at 4084–4089 making the ATG at 4074 the most likely translational start codon. A start codon could also not unambiguously be identified for the gene rhd. The presence of a likely ribosome binding site and the more suitable length of 24 aa (versus 49 aa for the alternative start codon) for the putative signal peptide make a translational start at nucleotide 6061 most probable. All Sox proteins except SoxZ as well as the products of ORF9 and the rhd gene are predicted to be synthesized as precursors carrying signal peptides for transport into the periplasm. As also proposed for P. pantotrophus (Friedrich et al., 2001) SoxZ is very likely co-transported together with SoxY.
Table 1. Features of proteins encoded in the sox loci of A. vinosum.
Calculated mol wt +/– signal peptide
Predicted cellular localization transport pathway
Conserved domains, comments
Highest sequence similarity, accession number
Signal transduction: COG0642: histidine kinase + COG 2202: PAS/PAC domain
PAS, Shewanella amazonensis, EAN40471
65 046/62 242
Periplasm, soluble Tat-dependent
COG0737: 5′-nucleotidase/2′,3′-cyclic phosphodiesterase and related esterases
SoxB, Chlorobaculum tepidum, AAM72255
13 629/11 119
Periplasm, soluble Sec-dependent
pfam00034, cytochrome c, 1 × haem c (CxxCH)
SoxX, Chlorobaculum tepidum, AAM72250
31 730/29 680
Periplasm, soluble Sec-dependent
None, 1 × haem c (CxxCH)
SoxA, Chlorobaculum tepidum, AAM72253
12 194/9 232
Periplasm, soluble Sec-dependent
Hypothetical protein C. tepidum, AAM72254
27 275 or 24 644/22 170
Periplasm, soluble Sec-dependent
cd01521: rhodanese homology domain, conserved active site cysteine present
Rhodanese-like protein T. denitrificans, YP_315408
Conserved hypothetical protein Methylococcus capsulatus, AAU91450
In addition to the proteins with significant similarity to Sox proteins from other organisms, the polypeptides deduced from two of the other sequenced genes deserve special attention: the protein encoded by ORF9 shows similarity to hypothetical proteins from the green sulphur bacteria Chlorobium limicola (strain Tassajara, DSM 249) as well as from Chlorobaculum (formerly Chlorobium; Imhoff, 2003) tepidum (strain TLS, DSM 12025T). In both organisms the corresponding open reading frame is located in a sox gene cluster (Eisen et al., 2002; Vertéet al., 2002). The rhd gene product contains a conserved domain typical for rhodaneses, enzymes responsible for sulphur group transfer that are found in all three domains of life. As sulphur group transfer could well be part of the thiosulphate oxidation pathway, the presence of a rhodanese in the vicinity of sox genes is noteworthy. While the product of ORFa could theoretically be involved in regulation of sox genes, the ccmA/ccmB gene products (Thöny-Meyer et al., 1995) might play a role in maturation of the haem c-containing Sox proteins SoxX and SoxA.
Identification and purification of Sox proteins in A. vinosum
With the exception of the soxA-encoded cytochrome c551 from C. limicola (Vertéet al., 2002) biochemical information about sox-encoded proteins from sulphur-storing organism has so far not been available. Our first aim was therefore to find whether the proteins predicted by the sox genes are indeed produced in A. vinosum and, if so, to obtain information about their structure and properties.
Using a specific antiserum raised against a synthetic SoxA peptide (see Experimental procedures), the major part of this protein was detected in the soluble fraction of thiosulphate-grown A. vinosum crude cell extracts as a 29 kDa band probably representing a processed form (after signal peptide cleavage) of the complete soxA-encoded polypeptide (31.7 kDa) (Fig. 4). After gel filtration, SoxA was detected in fractions corresponding to an apparent molecular weight of 40 kDa. This is equivalent to the predicted heterodimer of SoxA (29 kDa) and SoxX (11 kDa). Indeed, both subunits were detected in gels after SDS-PAGE (Fig. 4). As also reported for P. pantotrophus (Friedrich et al., 2000), the SoxX protein band exhibited a weaker Coomassie stain than SoxA, even though both subunits are present in a 1:1 ratio. As predicted from the nucleotide sequence, haem staining proved the presence of covalently bound haem in both subunits of SoxXA (Fig. 4). An UV-visible spectrum of purified SoxXA revealed a typical spectrum of a reduced c-type cytochrome (Fig. S2). As the purification was performed under aerobic, non-reducing conditions, this was quite surprising, even more so as all other SoxXA proteins described so far have been isolated in the oxidized state (Friedrich et al., 2000; Cheesman et al., 2001; Kappler et al., 2004). The α band at 550 nm characterizes the A. vinosum SoxXA as a cytochrome c550[β: 522 nm, γ (soret band): 415.5 nm]. Analysis of the deduced A. vinosum SoxA amino acid sequence revealed the presence of only one haem binding site as has also been found for the proteins from C. limicola (Vertéet al., 2002) and Starkeya novella (Kappler et al., 2004). This site corresponds to the carboxy-terminal of the two haem binding sites in P. pantotrophus and R. sulfidophilum SoxA (Friedrich et al., 2000; Appia-Ayme et al., 2001; Bamford et al., 2002). Whether this difference leads to a different function is not yet known.
In contrast to SoxA, SoxB could not be detected in crude extracts. However, after several purification steps the definite detection of the protein in SDS-PAGE gels was possible as a 62 kDa band matching the predicted size (Fig. 4). Gel filtration yielded an apparent molecular mass of 38 kDa indicating that the protein is present as a monomer. The low apparent molecular mass determined by gel filtration as compared with that observed upon denaturing gel electrophoresis is probably due to interaction between protein and column material. Unambiguous identification of the isolated protein as SoxB was obtained by MALDI-TOF analysis: the masses of 15 tryptic fragments fitted those predicted for the soxB-encoded polypeptide (Table S2, Supplementary material). Additionally, one of the obtained fragments (mass: 1860.7369) confirmed the predicted protein processing, as it starts with the first amino acid after the predicted signal peptidase cleavage site.
As also found for SoxB, detection of SoxYZ with the serum directed against the respective P. pantotrophus protein was not possible in crude extracts of A. vinosum but required several purification steps. From P. pantotrophus SoxYZ is purified as a heterodimer (Friedrich et al., 2000; Quentmeier et al., 2003). A SoxYZ heterodimer was also found for A. vinosum as the protein eluted in fractions from gel filtration chromatography corresponding to an apparent molecular mass of 24 kDa. However, only one protein band was detectable upon SDS-PAGE and immunoblot analysis (Fig. 4). This is most probably due to the similar size of the two polypeptides (12.7 kDa for processed SoxY and 11.2 for SoxZ). MALDI-TOF analysis of tryptic peptides provided proof for the existence of both polypeptides in our preparation (Table S2, Supplementary material). While one of the detected tryptic fragments clearly arose from SoxZ, two fragments could unequivocally be attributed to SoxY. The mass of a third fragment (775.38739 Da) matches the predicted tryptic SoxY peptide VTIGGCGG (663.3130 Da) when the covalent attachment of a thiosulphate molecule (average molecular mass 112.12 Da) is assumed. The thiosulphate most probably originates from the thiosulphate-containing stabilizing buffer used during protein purification. The tentative thiosulphate-binding A. vinosum peptide represents the highly conserved carboxy-terminus of SoxY and the strictly conserved cysteine residue of the P. pantotrophus protein has already been shown to bind inorganic sulphur compounds including thiosulphate (Quentmeier and Friedrich, 2001).
Regulation of sox gene expression
We studied the presence of SoxA in A. vinosum grown photoorganoheterotrophically on malate (RCV medium) or photolithoautotrophically on thiosulphate and sulphide (thiosulphate medium). A faint but detectable signal was present in the soluble fraction of malate-grown cells. The signal substantially increased in thiosulphate-grown cells (data not shown). This suggests that Sox proteins in A. vinosum are formed at a low basic level even in the absence of sulphur compounds but that an increased production is specifically induced by thiosulphate.
Inactivation of sox genes
To determine the importance of the sox genes and the encoded Sox proteins for thiosulphate oxidation in A. vinosum, different mutants were constructed (Table 2) and examined in batch culture. The concentrations of the substrate thiosulphate and the two enzymatically produced final products sulphate and tetrathionate are depicted in Fig. 5. It should be noted that the sum of these three sulphur compounds is not always the same and up to 1 mM sulphur atoms cannot be accounted for at certain time points. This is due first to the formation of up to 0.2 mM internally stored sulphur as an intermediate by the wild type which is very similar to sulphur concentrations reported earlier for the wild type on 1.5 mM thiosulphate under comparable experimental conditions (Dahl, 1996; Pott and Dahl, 1998). For clarity, sulphur is not shown in Fig. 5. Second, we proved the formation of trithionate and pentathionate in the cultures by HPLC. These additional poylthionates originate from abiotic reaction of tetrathionate and thiosulphate (Kelly et al., 1969; Suzuki, 1999) and were not quantitatively determined in all growth experiments.
Table 2. Bacterial strains and plasmids used in this study.
Kmr, XbaI fragment of PCR-amplified genome region around soxY with deletion of 404 bp of the soxY sequence
Apr, 4.5 kb NcoI fragment (soxB to ORFc) in pGEM5 Zf(+)
Apr, 4.5 kb SphI/SmaI fragment (ORFa to soxB) in pGEM7 Zf(+)
Apr, 5.5 kb ClaI fragment (soxX to ORFc) in pGEM7 Zf(+)
Apr, 5.5 kb XhoI fragment (ORF9 to ccmB) in pGEM7 Zf(+)
Apr, 1.5 kb EcoRI fragment (ORFd to soxZ) in pGEM7 Zf(+)
Apr, 2.5 kb ClaI fragment (soxY to ORFf) in pGEM7 Zf(+)
Apr, 800 bp PCR fragment of soxA (NdeI/BamHI) in NdeI/BamHI pET-11a
Apr, 1.8 kb PCR fragment of soxB (NdeI/XhoI) in NdeI/XhoI of pET-22b
Apr, 840 bp PCR fragment of soxYZ (NdeI/HindIII) in NdeI/HindIII of pET-22b
Kmr, Kmr cartridge (BamHI) from pHP45 ΩKm in BglII of pGEM-SoxB
Kmr, Kmr cartridge (EcoRI) from pHP45 ΩKm in NcoI of pDHSS192
Kmr, Kmr cartridge (EcoRI) from pHP45 ΩKm between EcoRI sites of pGEM-SoxB
Apr, Kmr, ClaI/BglII fragment from pDHCl7 in ClaI/BamHI of pSUP202, Kmr cartride (EcoRI) from pHP45 ΩKm in XhoI of ClaI/BglII insert
Cmr, Emr, 4.5 kb ApaI/SpeI fragment from pGEM-SoxB in ApaI/SpeI of pBBR1-MCS, Emr cartridge (SmaI) in EcoRV of construct
Kmr, 1.5 kb PCR fragment of soxYZ (XbaI) in XbaI of pBBR1 MCS2
Allochromatium vinosum wild type readily oxidized added thiosulphate, resulting in the simultaneous production of sulphate and tetrathionate. The soxXΩKm, soxBΩKm and soxBXΩKm mutants carrying insertions of a polar Ω kanamycin resistance cassette all exhibited a significantly reduced rate of thiosulphate oxidation (Fig. 5A). While tetrathionate formation remained unaffected (Fig. 5C), the lack of sulphate production (Fig. 5B) indicated that the pathway from thiosulphate to sulphate had been blocked completely in the mutant strains. Taking the polar effect of the inserted resistance cassette into consideration, the observed phenotypes cannot be assigned to one single deleted sox gene. This is demonstrated by the lack of SoxA encoded downstream of the ΩKm insertion site in the soxXΩKm and ΔsoxBXΩKm mutants (Fig. 6C). The soxY gene was inactivated by in-frame deletion, therefore soxZ should still be expressed. As SoxZ is not synthesized with a signal peptide but instead very probably co-transported with SoxY, it is likely that SoxZ accumulates in the mutant cytoplasm and is eventually degraded. The observed phenotype is therefore probably due not only to the lack of SoxY but to the lack of the complete substrate-binding molecule SoxYZ. Again, sulphate was not produced from thiosulphate by the ΔsoxY mutant (Fig. 5B), showing that SoxYZ, like SoxXA and SoxB, is absolutely essential for the oxidation of thiosulphate to sulphate. Tetrathionate formation was still possible in mutant ΔsoxY, but started late compared with the other mutant strains (Fig. 5C). The inactivation of ORF9/rhd, situated downstream of soxA, had no detectable effect on thiosulphate degradation (data not shown). Therefore, the gene products do not seem to play an essential role in the oxidation of thiosulphate under the chosen experimental conditions.
Complementation of sox mutant strains
To prove that the observed lack of sulphate formation from thiosulphate was indeed caused by inactivation of sox genes, two A. vinosum mutants, soxXΩKm and ΔsoxY, were complemented in trans. The soxXΩKm complementing plasmid pΔsoxX+X contained a 4.5 kb ApaI/SpeI fragment of A. vinosum DNA including besides part of soxB, the complete soxXA, ORF9, and rhd genes as well as the soxBX intergenic region with the potential promoter region. As shown in Fig. 6A and B, thiosulphate oxidation in the complemented mutant soxXΩKm+X was restored to the wild-type phenotype and sulphate was again the major product. In addition, immunoblot analysis revealed the renewed presence of SoxA in the soxXΩKm+X strain (Fig. 6C), further confirming: (i) the functionality of the proposed promoter and (ii) the correct expression in trans of the re-introduced genes. The ΔsoxY complementing plasmid pΔsoxY+Y contained both complete soxYZ genes and the region upstream of soxY with the potential natural promoter. In the complementation strain ΔsoxY+Y the principal capability to oxidize thiosulphate to sulphate was clearly re-established (Fig. 6A and B). However, the thiosulphate oxidation rate of the complemented mutant was significantly lower than that of the wild type. Tetrathionate formation was comparable to that of the wild type for both complemented mutant strains (not shown).
Two pathways for thiosulphate oxidation in A. vinosum
In the present study, two distinct enzyme systems for thiosulphate oxidation in A. vinosum have clearly been identified. A thiosulphate dehydrogenase is responsible for the oxidation to tetrathionate, while a sox gene-encoded multienzyme complex is essential for the oxidation to sulphate. In conjunction with its localization in the bacterial periplasm, the low pH optimum of 4.25 for the A. vinosum thiosulphate dehydrogenase is in full agreement with the observation that whole cells preferably form tetrathionate from thiosulphate at pH values below 7.0 while sulphate is the main product under alkaline conditions (Smith, 1966; Smith and Lascelles, 1966).
Tetrathionate-forming thiosulphate dehydrogenase is an acidophilic c-type cytochrome
The properties of thiosulphate dehydrogenase from A. vinosum characterized during this study are compatible with older data presented by Smith (1966) and Fukumori and Yamanaka (1979). In both reports a tetrathionate-forming activity with a pH optimum in the acidic range was described. With our current analysis we cannot confirm the presence of any tetrathionate-forming enzyme operating at pH 8.0 in A. vinosum as has been claimed earlier by Knobloch et al. (1981) and Schmitt et al. (1981). Of the tetrathionate-forming enzymes characterized so far, thiosulphate dehydrogenase from Acidithiobacillus thiooxidans (Nakamura et al., 2001) most closely resembles the enzyme from A. vinosum. Both organisms belong to the γ-Proteobacteria. The protein from Acidithiobacillus has been described as a monomeric 27.9 kDa c-type cytochrome with a pH optimum at 3.5. Thiosulphate dehydrogenases from other sources show remarkable heterogeneity concerning structural properties and catalytic characteristics (Kusai and Yamanaka, 1973; Knobloch et al., 1981; Then and Trüper, 1981; Visser et al., 1996) which has been interpreted as indicating convergent rather than divergent evolution (Visser et al., 1996).
Sox proteins are present in A. vinosum
Three soluble Sox proteins SoxXA, SoxB and SoxYZ were purified from thiosulphate-grown A. vinosum. The proteins are located in the periplasm, all except of SoxZ are predicted to be synthesized as precursors carrying signal peptides. Processing by signal peptidase was experimentally confirmed for SoxB. The thiosulphate-oxidizing multienzyme system of P. pantotrophus and some other α-Proteobacteria is also located in the periplasm, while the Sox system in S. novella is thought to be anchored to the periplasmic side of the cytoplasmic membrane (Kappler et al., 2001).
SoxB is predicted to be transported via the Tat pathway implying transport of a mature, folded protein probably containing a cofactor. We assume that the protein resembles SoxB from P. pantotrophus in that it contains two manganese atoms per monomer (Friedrich et al., 2000). SoxXA was purified as a heterodimer; the presence of haem in both subunits was proven experimentally. Even under the oxic purification conditions applied, the reduced form of A. vinosum SoxXA was purified and seems to be quite stable. A. vinosum belongs to the group of Sox-containing organisms that contain a mono-haem SoxA while (among others) the proteins from P. pantotrophus and R. sulfidophilum bind two c-type haems. The mono-haem SoxA proteins lack the N-terminal haem which does not seem to be involved in electron tunnelling (Bamford et al., 2002). A. vinosum SoxYZ was isolated as a heterodimer. For P. pantotrophus SoxYZ additional homodimeric and heterotetrameric states have been described (Quentmeier et al., 2003). In P. pantotrophus SoxY is believed to be the active site of protein-bound sulphur oxidation (Quentmeier and Friedrich, 2001). MALDI-TOF analysis provided evidence for a covalent attachment of thiosulphate to the carboxy-terminal tryptic peptide of A. vinosum SoxY, thereby yielding further proof for the correctness of this assumption.
Formation of SoxA in A. vinosum appeared to be thiosulphate-inducible above a low constitutive level, consistent with a role in the utilization of thiosulphate. Thiosulphate-dependent formation of SoxA has also been reported for Chlorobium limicola (Vertéet al., 2002). In P. pantotrophus thiosulphate-dependent induction of gene expression mediated by the proteins SoxRS has been demonstrated for soxXYZABCD and soxFGH (Friedrich et al., 2001; Rother et al., 2005). In S. novella the production of SoxXA appears to be inducible by thiosulphate (Kappler et al., 2000; 2004) while SoxC is detectable independent of the presence of thiosulphate (Kappler et al., 2001). As soxYZ are located independently from soxBXA in A. vinosum, the expression of these genes could be regulated in a different way. In C. tepidum the inactivation of a gene encoding a Rubisco-like protein (RLP) resulted in an inhibition of thiosulphate oxidation, due to a lack of SoxY (Hanson and Tabita, 2003). The expression of soxA and soxB, however, remained undisturbed. How RLP influences the expression of soxY in C. tepidum is not yet known. A. vinosum also contains a gene encoding a RLP that could play a role in Sox system regulation.
In A. vinosum sox genes are essential for the oxidation of thiosulphate to sulphate
Inactivation of sox genes and the complementation of two of the mutant strains revealed the absolute necessity of Sox proteins for thiosulphate oxidation to sulphate in A. vinosum. The products of the soxB, soxXA and soxYZ genes are equally essential for this pathway. Tetrathionate formation from thiosulphate, however, remained largely unaffected. The reason for the delayed formation of tetrathionate in the ΔsoxY mutant has to be further elucidated and an involvement of SoxYZ in tetrathionate formation can currently not be completely ruled out. When soxX was supplied in trans in a mutant carrying an interposon in soxX, the mutant fully regained its ability to form sulphate from thiosulphate with wild-type rates. While the ΔsoxY+Y complementation strain principally regained its ability to produce sulphate from thiosulphate, the observed metabolic rates were significantly slower than those of the wild type. Currently we cannot fully explain this phenotype, although it might be related to an unbalanced formation of SoxYZ from a plasmid present in several copies as compared with one copy of the genes on the chromosome.
Gene arrangement and probable functions
In P. pantotrophus and most of the other examined Sox-system-containing organisms the sox genes are arranged in one single cluster (Friedrich et al., 2005). This situation enables a joint expression of the genes soxXYZABCD, all of which encode proteins essential for Sox-dependent thiosulphate oxidation in the organism (Friedrich et al., 2001). In contrast, in A. vinosum soxBXA and soxYZ are located on two independent sites on the genome, thereby preventing co-transcription.
The Sox system is found in green sulphur bacteria like C. tepidum as well as in different groups of Proteobacteria and the thermophilic bacterium Aquifex aeolicus (Petri et al., 2001; Friedrich et al., 2005). However, gene presence or gene arrangement can differ quite distinctly from the ‘model organism’P. pantotrophus. Most conspicuously, the genes for SoxCD are less abundant than the genes soxXYZAB. In P. pantotrophus SoxCD are believed to catalyse the oxidation of SoxY-bound sulphane sulphur to the oxidation level of sulphate (Friedrich et al., 2001). We were not able to detect soxCD in A. vinosum and in this combination the genes are also absent from the genomes of the magnetotactic bacteria Magnetococcus sp. MC-1 and Magnetospirillum magnetotacticum MS-1 as well as the green sulphur bacterial genomes searchable at the DOE Joint Genome institute (http://img.jgi.doe.gov). Interestingly, for all these organisms the formation of sulphur globules as intermediates of reduced sulphur compound oxidation is well established (Dahl and Prange, 2006). Furthermore, soxD does not occur in the genome of Thiobacillus denitrificans ATCC 25259 (Beller et al., 2006). It appears noteworthy that T. denitrificans strain RT (DSM 807) has been reported to accumulate elemental sulphur during the oxidation of thiosulphate although not as microscopically visible but as finely dispersed membrane-associated sulphur (Schedel and Trüper, 1980). We can therefore conservatively state that the difference between organisms with and without soxCD appears to be the absence or presence of elemental sulphur as a metabolic intermediate during Sox-dependent thiosulphate oxidation to sulphate. In A. vinosum the proteins encoded in the dsr (dissimilatory sulphite reductase) locus are required for further oxidative metabolism of stored sulphur (Dahl et al., 2005). Indeed, sulphur formation not only appears to correlate with the absence of soxCD genes but also occurs concomitantly with the presence of relevant dsr genes in the sequenced magnetotactic and green sulphur bacterial genomes as well as in T. denitrificans.
The A. vinosum soxBXA and soxYZ genes products exhibit higher sequence identities to those of the sulphur-forming green sulphur bacteria (62–65% for SoxB, 50–51% for SoxA, 46–51% for SoxX, 48–52% for SoxY, 37–45% for SoxZ) and to those of the β-Proteobacterium T. denitrificans (57% for SoxB, 43% for SoxA, 47% for SoxX, 55% for SoxY, 55% for SoxZ) than to those of the chemotrophic α-proteobacterial model organism P. denitrificans (53% for SoxB, 30% for SoxA, 36% for SoxX, 41% for SoxY, 35% for SoxZ) and the phototrophic α-Proteobacterium R. sulfidophilum (50% for SoxB, 30% for SoxA, 39% for SoxX, 43% for SoxY, 32% for SoxZ). The latter two organisms do not form sulphur globules. This finding is in complete agreement with earlier observations by Beller et al. (2006) who pointed out a higher similarity of the T. denitrificans sox gene products with those of C. tepidum than with those of the Proteobacteria Paracoccus, Starkeya and Pseudoaminobacter. Although it appears premature to draw definite conclusions, the high similarity of Sox proteins from only distantly related sulphur-storing organisms as opposed to the comparatively low similarity of Sox proteins from much closer related organisms that differ by their capacity to form sulphur globules as intermediates, is highly conspicuous.
In this work we have shown unambiguously that sox-encoded proteins are absolutely essential for the oxidation of thiosulphate to sulphate in A. vinosum. We have also demonstrated that the Sox system in this sulphur-forming purple sulphur bacterium operates differently from that of the well-studied chemotrophic and phototrophic α-Proteobacteria (Mukhopadhyaya et al., 2000; Appia-Ayme et al., 2001; Friedrich et al., 2001; 2005; Kappler et al., 2001). On the basis of our current results and previous studies on localization and oxidation of sulphur globules in A. vinosum (reviewed in Dahl and Prange, 2006) we suggest a model for thiosulphate oxdiation in sulphur-storing organisms: the initial oxidation and covalent binding of thiosulphate to SoxYZ would be brought about by SoxXA and sulphate would then be hydrolytically released by SoxB just as proposed for Paracoccus (Friedrich et al., 2001). However, in organisms like A. vinosum that lack the ‘sulphur dehydrogenase’ SoxCD the sulphane sulphur atom still hooked up to SoxY cannot be directly further oxidized. We suggest that the sulphur is instead transferred to growing sulphur globules. Such a suggestion is feasible as the sulphur globules in A. vinosum and in many if not all other organisms forming intracellular sulphur deposits reside in the bacterial periplasm (Pattaragulwanit et al., 1998; Dahl and Prange, 2006) and therefore in the same cellular compartment as the Sox proteins. How the transfer of SoxY-bound sulphur to the sulphur globules is achieved is currently unclear. The potential sulphur transferase encoded by the rhd gene could well play a role in this process however, its inactivation did not lead to a detectable phenotype. Possibly, other sulphur transferases present in the cells function as a back-up system.
Bacterial strains, plasmids and growth conditions
The bacterial strains and plasmids used in this study are listed in Table 2. A. vinosum strains were cultivated and harvested as described earlier (Prange et al., 2004). Pfennig's medium (Pfennig and Trüper, 1992) lacking sulphide is referred to as ‘0 medium’. When large amounts of A. vinosum cell material were required for protein purification the medium (‘thiosulphate medium’) was prepared as follows (for 10 l of medium): 100 ml of 100× macro element solution [100 g of KH2PO4, 70 g of NH4Cl, 40 g of MgSO4 × 7 H2O, 10 g of CaCl2 × 2 H2O, 100 ml of 10× trace element solution SL12 (Pfennig and Trüper, 1992), and 193 ml of 37% HCL in a final volume of 500 ml] diluted to 9500 ml to give solution 1, and 500 ml of solution 2 containing 26.5 g of Na2CO3, 21 g of NaHCO3, 31 g of Na2S2O3 × 5 H2O and 25 ml of sulphide solution (74 g of HNaS × 1 H2O l−1). Both solutions were autoclaved separately. After cooling, solution 2 was added to solution 1 under stirring and nitrogen atmosphere. The medium pH was 7.5 without further titration. Escherichia coli was cultivated in LB medium (Sambrook et al., 1989). Antibiotics used for mutant selection were applied at the following concentrations (in μg ml−1): for E. coli: ampicillin 100, kanamycin 50, erythromycin 100, chloramphenicol 50; for A. vinosum: rifampicin 50, streptomycin 50, ampicillin 10, kanamycin 10–25, erythromycin 10.
Recombinant DNA techniques
Standard methods were used for molecular biological techniques (Sambrook et al., 1989; Ausubel et al., 1997). Chromosomal DNA of A. vinosum strains was obtained, Southern hybridizations and PCR amplifications with Taq DNA polymerase or Pfu DNA polymerase were performed as described previously (Dahl, 1996; Pott and Dahl, 1998). The genotypes of the A. vinosum recombinants used in this study were confirmed by Southern hybridization and/or PCR. DNA was sequenced by Sequiserve (Vaterstetten, Germany). Nucleotide sequences were compiled and analysed with Clone Manager (SES central) software. Promoter prediction was performed using bprom at http://www.softberry.com. Similarity searches were performed using the blast algorithm (Altschul et al., 1997). Protein sequences were analysed using resources available at the ExPASy molecular biology server (http://www.expasy.ch). Signal peptides were predicted by the psort program package.
Cloning of sox gene loci
The primers soxB-forward (5′-gacggtggtgatacctg-3′) and soxB-reverse2 (5′-catgtcgccgccctgctg-3′) were derived from sequences conserved in soxB genes from different organisms. The obtained PCR fragment was used as a probe to identify positive clones with a 4.5 kb NcoI insert in a library of 4–5 kb NcoI fragments of chromosomal A. vinosum DNA in the pGEM5 Zf(+) vector. This was the first of four overlapping clones covering the first set of sox genes. Using the same soxB probe further positive clones with a 4.5 kb SphI/SmaI insert were selected from a library of 4–5 kb SphI/SmaI fragments of chromosomal DNA in the pGEM7 Zf(+) vector. A rhd probe was produced using the primers RhdAv forward (5′-cttccccgcccatgggctcgatcggtc-3′) and RhdAv reverse (5′-ttcccggcgaggatccgggctggacgc-3′) and applied to select two positive clones with a 5.5 kb ClaI insert and a 5.5 kb XhoI insert, respectively, from the corresponding 5–6 kb fragments in the pGEM7 Zf(+) vector. A heterologous DNA probe derived from the soxYZ sequence of C. limicola DSM 249 (Vertéet al., 2002) using the primers ClimYfor (5′-tttcgtgccagtaacggt-3′) and ClimZrev (5′-agcatgtcgcctgccttg-3′) was used for the successive construction of two further overlapping gene libraries, one with a 1.5 kb EcoRI fragment and the other with a 2.5 kb ClaI fragment. Both fragments were cloned into a correspondingly digested pGEM7 Zf(+), yielding the two plasmids pDHEcoYZ and pDHClaYZ respectively. This led to the detection and sequencing of the second set of sox genes in A. vinosum.
Expression of sox genes in E. coli
The pET series of plasmids was used for the expression of the sox genes soxA, soxB and soxYZ in the E. coli strain BL21(DE3) to produce the proteins as positive controls for antisera directed against synthetic peptides (SoxA) or purified proteins from other organisms (SoxB, SoxYZ). Suitable restriction sites were introduced at the 5′ and 3′ ends of the genes by PCR amplification with modified primers.
Construction of A. vinosum recombinants
The plasmids used for the construction of A. vinosum strains with mutations in sox genes are listed and described in Table 2. The inactivation of soxY in the corresponding mutant ΔsoxY was achieved by in-frame mutagenesis, making use of splicing by overlap extension PCR (SOEing PCR; Horton, 1995), with the primers Yforward (5′-aggccgtctagaatttccgtgacacattgc-3′), Ysoe-reverse (5′-tcttatagagcttgttgatttatctcctct-3′), Ysoe-forward (5′-agaggagataaat caacaagctctataaga-3′) and Yreverse (5′-tgcgcctctagaggctggtttcgaattcta-3′). The final PCR product was digested with XbaI and ligated into the mobilizable suicide vector pK19mobsacB (Schäfer et al., 1994) that had previously been cut with XbaI. The resulting plasmid was transferred into A. vinosum and the gene exchange was achieved and verified as described in Lübbe et al. (2006).
To obtain plasmids for complementation of the soxXΩKm and ΔsoxY mutant, respectively, fragments of genomic DNA of A. vinosum containing the corresponding gene and potential upstream promoter sequences were cloned into the broad-host-range vector pBBR1 MCS-2. The plasmid for soxXΩKm complementation was obtained by cloning a 4.5 kb ApaI/SpeI fragment from pGEM-SoxB into pBBR1 MCS-2. In addition, the erythromycin resistance cassette from pHP45ΩEm (SmaI) was inserted into the EcoRV site, resulting in pΔsoxX+X. For complementation of ΔsoxY a 1.5 kb PCR fragment containing soxYZ (primers: Yforward/Yreverse, see above) was amplified. Both PCR fragment and the vector pBBR1 MCS-2 were digested with XbaI and ligated resulting in pΔsoxY+Y.
Transfer of plasmid DNA to A. vinosum
Electrotransformation was established during this work and used to generate the mutant strains soxXΩKm+X, soxBΩKm and soxBXΩKm. To produce electrocompetent A. vinosum cells, cultures were grown photoheterotrophically in RCV medium to an optical density of approximately 0.6 at 690 nm. Cell material equivalent to a total cell number of 2 × 1010 (approximately 45 ml) was harvested at 4°C and 5900 g for 15 min. The pellet was washed three times in 2 ml of ice-cold dimethyl sulphoxide (DMSO) solution [10% (v/v)], precipitated again (4°C, 5900 g, 10 min) and finally resuspended in 1 ml of the same solution. This cell material was used directly for electrotransformation, as further storage significantly decreased the transformation efficiency. Plasmid transfer was performed in electrotransformation cuvettes of 2 mm width using a Gene Pulser II with a Pulse Controller Plus (Bio-Rad, München). Each reaction contained 50 μl of cell solution and 0.1–0.5 μg of purified plasmid DNA. After mixing and incubation for 15 min on ice the electrical pulse was applied (200 Ω, 25 μF; equally good results with 9 kV cm−1 and 12.5 kV cm−1 respectively). The time constant τ should be about 5 ms under the chosen conditions. Directly after exposure to the electric pulse the cuvette was put on ice. The cell material was resuspended in 2.5 ml of RCV medium supplied with 10 mM sodium acetate and 4 mM sodium thiosulphate and incubated in the light for 18 h in brimful, tightly closed glass vials. For selection of transformants A. vinosum was plated on selective RCV solid medium (using phytagel as gelling agent), containing the appropriate antibiotic and incubated anaerobically in the light. Conjugation was performed as described in Pattaragulwanit and Dahl (1995), using either E. coli SM10 or E. coli S17-1 as donor organisms. This method was used for production of the mutant strains rhd/ORF9 ΩKm and ΔsoxY and both complementation strains (soxXΩKm+X and ΔsoxY+Y).
Allochromatium vinosum wild-type and mutant strains were characterized in batch culture experiments essentially as described in Prange et al. (2004). A photoheterotrophically grown stationary-phase culture (250 ml) was harvested (5900 g, 20 min) and the cell material was used to inoculate 1.5 l of ‘0 medium’ in a thermostated fermenter. Experiments were started by the addition of 2 mM thiosulphate from a sterile filtered stock solution (0.4 M). The pH was maintained at pH 7.0 ± 0.2. Protein concentrations were determined using Bradford reagent (Sigma, #B6916) as specified by the manufacturer.
Denatured proteins were separated by SDS-PAGE using the method of Laemmli (1970), followed by Coomassie blue or silver staining. Immunoblot (Western) analysis was performed by the ‘semidry’ procedure, using the Transblot SD semidry transfer apparatus (Bio-Rad, Munich) and nitrocellulose membranes (Protean BA 85, Schleicher and Schuell, Dessel, Germany) with antibodies raised from rabbits. The antibody against SoxA was raised against a potentially highly immunogenic oligopeptide deduced from the nucleotide sequence (C+TYMNNGLELNGPGARK) and was produced at the facility of Eurogentec in Seraing, Belgium. The antibodies against SoxB and SoxYZ were raised against the corresponding protein purified from P. pantotrophus (Friedrich et al., 2000) and were generously provided by Professor Cornelius G. Friedrich, Dortmund, Germany. Antisera were used at a 1:1000 dilution. The secondary anti-rabbit horseradish peroxidase antibody conjugate was used at a 1:5000 dilution. The binding of Anti-SoxA and Anti-SoxB was detected using 4-chloro-1-naphthol, the binding of Anti-SoxYZ was detected using the SuperSignal West Pico Chemiluminescent Substrate detection system (Pierce).
For MALDI-TOF analysis enriched proteins were concentrated 10-fold by ultrafiltration centrifugation (Centriplus YM10; Millipore, Bedford, Massachusetts) and separated by SDS-PAGE. The protein bands corresponding to SoxB and SoxY were excised from the Coomassie-stained polyacrylamide gels and further analysed at the laboratories of Seqlab (Göttingen, Germany).
Haem staining was performed either directly in gels after SDS-PAGE (McDonnel and Staehelin, 1981) or after transfer to PVDF membranes (Vargas et al., 1993). N-terminal amino acid sequences were determined from protein bands in polyacrylamide gels by automated Edman degradation as described previously (Brune, 1995b; Dahl et al., 2005). UV-visible spectra of samples in 1 ml quarz cuvettes were recorded using a Perkin Elmer UV/vis Spectrometer Lambda 11.
Purification of Sox proteins from A. vinosum
The purification of SoxXA, SoxB and SoxYZ from A. vinosum DSM 180T was monitored using the appropriate antisera. Thawed cell material was resuspended in stabilizing buffer [50 mM potassium phosphate buffer with 2 mM sodium thiosulphate, 1 mM magnesium sulphate and 1 μM phenylmethylsulphonylfluoride (PMSF), pH 7.5] at a ratio of 3 ml of buffer per gram wet weight. After homogenization and cell disruption by ultrasonic treatment (1 min ml−1), the cell extract was subjected to centrifugation (25 000 g, 4°C, 30 min). The supernatant was subjected to ultracentrifugation (145 000 g, 4°C, 3 h). This supernatant was brought to 40% saturation with (NH4)2SO4. After stirring at 4°C overnight, the supernatant from a centrifugation at 25 000 g, 4°C, for 30 min was applied to hydrophobic interaction chromatography on a low-substitution Phenyl-Sepharose matrix (Amersham Pharmacia Biotech, diameter 2.6 cm, 80 ml gel volume), equilibrated with 40% saturated (NH4)2SO4 in stabilizing buffer. Bound protein was eluted by decreasing the (NH4)2SO4 concentration in a linear gradient of 550 ml (2.5 ml min−1). SoxYZ eluted at ∼22%, SoxXA at ∼7% and SoxB at ∼2% saturation with (NH4)2SO4. For further purification of SoxXA and SoxB the appropriate protein containing fractions were combined, dialysed overnight against MonoQ-stabilizing buffer (10 mM TrisHCl, 2 mM sodium thiosulphate, 1 mM magnesium sulphate, 1 μM PMSF, pH 7.5) and loaded onto MonoQ HR 5/5 equilibrated with the same buffer. The column was washed with MonoQ-stabilizing buffer containing 100 mM NaCl, and the proteins were eluted with a linear gradient from 100 to 600 mM NaCl (1 ml min−1). SoxXA and SoxB eluted at ∼300 mM NaCl. The corresponding fractions were combined, concentrated to keep the total volume below 2 ml, and further purified by gel filtration on Superdex TM200 equilibrated with gel filtration-stabilizing buffer (50 mM TrisHCl, 2 mM sodium thiosulphate, 1 mM magnesium sulphate, 1 μM PMSF, pH 7.5) containing 150 mM NaCl (flow rate 0.5 ml min−1). The apparent molecular mass of the eluted proteins was determined by calibrating the column with standard proteins (Sigma Marker low-range; molecular mass 6500–66 000 Da). For the purification of SoxYZ, chromatography on MonoQ was omitted. The combined fractions after hydrophobic interaction chromatography were dialysed against gel filtration-stabilizing buffer containing 150 mM NaCl, and concentrated to approximately 2 ml using centrifugal ultrafiltration before gel filtration as described above.
Purification of thiosulphate dehydrogenase from A. vinosum
Thawed cell material was resuspended in 20 mM TrisHCl, pH 7.5, containing 20 mM MgCl2 and some grains of DNase, disrupted by ultrasonic treatment and centrifuged as described above. The supernatant of ultracentrifugation was brought to a pH of 4.6 by dropwise addition of 1 M acetic acid. After stirring for 5 min on ice denatured protein was removed by centrifugation (25 000 g, 4°C, 20 min) and the pH was increased back to pH 7.5 by addition of a 1 M Tris base solution followed by addition of (NH4)2SO4 to 1.2 M. After incubation on ice for 10–16 h, precipitated protein was removed by centrifugation (25 000 g, 4°C, 30 min). The supernatant was subjected to a Phenyl-Sepharose matrix (diameter 2.6 cm, 50 ml gel volume) equilibrated with 1.2 M (NH4)2SO4 in 20 mM TrisHCl, pH 7.5. The column was washed with 840 and 660 mM (NH4)2SO4 in the same buffer. Thiosulphate dehydrogenase was eluted with 588 mM (NH4)2SO4 in 20 mM TrisHCl, pH 7.5 (flow rate 4 ml min−1). Combined fractions containing the enzyme were dialysed against 20 mM TrisHCl, pH 7.5 and loaded onto DEAE Sephacel (Amersham Pharmacia Biotech, Uppsala, diameter 2.6 cm, 30 ml gel volume) equilibrated with the same buffer. The column was washed with 20 mM TrisHCl, pH 7.5 containing 8 mM NaCl and thiosulphate dehydrogenase was eluted with 14 mM NaCl in the same buffer (flow rate 3 ml min−1). Combined fractions containing thiosulphate dehydrogenase were subjected to chromatography on Macroprep ceramic hydroxyapatite type I 20 μm (Bio-Rad, Munich, diameter 1.6 cm, 10 ml gel volume) equilibrated with 1 mM potassium phosphate buffer, pH 5.8. Thiosulphate dehydrogenase was eluted by a stepwise increase of phosphate buffer concentration to 2 mM. Combined fractions containing the enzyme were loaded onto a pre-packed MonoQ HR 5/5 column equilibrated with 20 mM TrisHCl, pH 7.5. Proteins were eluted with a linear gradient from 0 to 40 mM Na2S2O3 (flow rate 1 ml min−1). Thiosulphate dehydrogenase eluted at 20–30 mM Na2S2O3. The apparent molecular mass of the enzyme was determined by gel filtration on Sephadex 75 (Amersham Pharmacia Biotech, Uppsala, HiLoad 16/60) equilibrated with 50 mM TrisHCl containing 150 mM NaCl, pH 7.5 (flow rate 0.5 ml min−1). The column was calibrated using standard proteins (see above).
Thiosulphate dehydrogenase assay and activity stain
The standard reaction mixture (1 ml) contained 100 mM acetate buffer, pH 4.25, 8 mM Na2S2O3, 1 mM K3Fe(CN)6 and enzyme. The measurements were performed at 25°C and started by the addition of the enzyme. Reduction of ferricyanide was measured at 420 nm in a Perkin Elmer UV/vis Spectrometer Lambda 11 using an extinction coefficient of 1.09 mM−1 cm−1. One unit of activity is defined as 1 μmol ferricyanide reduced per min. All activity determinations were taken as the mean of three measurements. Data were examined according to Segel (1993) via primary v versus [S] plots and fitted to the empirical Hill Eqn 1 using DeltaGraph Pro 4.05c/Mac version. The Hill equation resembles the classical Henri–Michaelis–Menten equation; however, the n term allows to account for non-hyperbolic shapes.
For curves with n > 1 a substrate concentration [S]0.5 can be reported that yields half maximal velocity and is characteristic of the process. The constant K, which is not equivalent to Km, characterizes enzyme–substrate interaction. The relationship between K and [S]0.5 is K = [S]0.5n. For activity staining, gels after native PAGE were incubated in substrate solution [100 mM acetate buffer, pH 5.0, 8 mM Na2S2O3, 1 mM K3Fe(CN)6] for 30 min at room temperature, rinsed with destilled water and incubated in 10 mM FeCl3. The reaction was stopped by transfer of the gels into 7% acetic acid.
Analysis of sulphur compounds
Thiosulphate, sulphite and sulphate were determined by HPLC analysis using the methods of Rethmeier et al. (1997). Elemental sulphur and tetrathionate were determined colorimetrically by cyanolysis (Kelly et al., 1969). For kinetic analysis of thiosulphate dehydrogenase thiosulphate and tetrathionate were determined by ion pair HPLC after Steudel and Holdt (1986).
Accession number (GenBank) for sox gene sequence
The nucleotide sequences for the two A. vinosum sox gene loci are available at GenBank under DQ441405 and DQ441406.
We thank the group of Cornelius Friedrich in Dortmund for supplying antisera against SoxB and SoxYZ, and SoxCD as well as for a gene probe directed against soxCD. We are deeply indebted to Irwin H. Segel from the University of California, Davis, who provided essential help for interpretation and plotting of the enzyme kinetic data. We also thank Monika Kräling and Birgitt Hüttig for technical assistance. This study was supported by the Deutsche Forschungsgemeinschaft (Da 351/4-1)