Phylogeny and distribution of the soxB gene among thiosulfate-oxidizing bacteria

Authors


*Corresponding author. Tel.: +49 (431) 5973850; Fax: +49 (431) 565876, E-mail: jimhoff@ifm.uni-kiel.de

Abstract

A PCR protocol for the detection of sulfur-oxidizing bacteria based on soxB genes that are essential for thiosulfate oxidation by sulfur-oxidizing bacteria of various phylogenetic groups which use the ‘Paracoccus sulfur oxidation’ pathway was developed. Five degenerate primers were used to specifically amplify fragments of soxB genes from different sulfur-oxidizing bacteria previously shown to oxidize thiosulfate. The PCR yielded a soxB fragment of approximately 1000 bp from most of the bacteria. Amino acid and nucleotide sequences of soxB from reference strains as well as from new isolates and environmental DNA from a hydrothermal vent habitat in the North Fiji Basin were compared and used to infer relationships of soxB between sulfur-oxidizing bacteria belonging to various 16S rDNA-based phylogenetic groups. Major phylogenetic lines derived from 16S rDNA were confirmed by soxB phylogeny. Thiosulfate-oxidizing green sulfur bacteria formed a coherent group by their soxB sequences. Likewise, clearly separated branches demonstrated the distant relationship of representatives of α-, β-, and γ-Proteobacteria including representative species of the former genus Thiobacillus (now Halothiobacillus–γ-Proteobacteria, Thiobacillus–β-Proteobacteria and Starkeya–α-Proteobacteria). This general picture emerged although apparent evidence for lateral transfer of the soxB gene is indicated and comparison of soxB phylogeny and 16S rDNA phylogeny points to the significance of this gene transfer in hydrothermal vent bacterial communities of the North Fiji Basin.

1Introduction

Processes of the oxidative sulfur cycle are widely distributed in nature and occur wherever reduced sulfur compounds are available from the activity of sulfate-reducing bacteria or from geological sources in hydrothermal vent environments. Reduced sulfur compounds can be oxidized by a great variety of bacteria, including archaea and representatives of chemotrophic and phototrophic eubacteria. Sulfur oxidation is a common property of anoxygenic phototrophic purple and green sulfur bacteria, but also found in some purple non-sulfur bacteria and even in cyanobacteria [1–4]. In addition, numerous chemolithotrophic and chemoheterotrophic bacteria are able to use sulfur compounds as an energy source and can use oxygen or nitrate as terminal electron acceptors [5–9]. All kinds of reduced sulfur compounds such as sulfide, elemental sulfur, polythionates and thiosulfate can be oxidized to sulfate as the final oxidation product [5–8]. Thiosulfate is used by most of the sulfur-oxidizing bacteria including chemolithotrophic bacteria, but also by chemoorganotrophic bacteria which oxidize thiosulfate to tetrathionate and use thiosulfate as a supplemental but not as the sole energy source, e.g. some Pseudomonas and Halomonas species. This latter group of bacteria can make an important contribution to the oxidation of thiosulfate in some natural environments [10–13]. Thiosulfate is a stable sulfur compound of intermediate oxidation state and fulfils an important role in the natural sulfur cycle [10–13]. Thus, thiosulfate oxidation is of importance for the physiology of sulfur-oxidizing bacteria and for their ecological function in nature and deserves attention in the analysis of sulfur-oxidizing bacteria.

Considering the phylogenetic diversity of sulfur-oxidizing bacteria [3,5,6,8,14], the selective analysis of natural communities of these bacteria cannot be achieved by methods based on 16S rDNA sequences except for a few phylogenetically coherent groups as has been demonstrated for Thiomicrospira species [15]. Therefore, alternative approaches which use specific gene sequences of functional genes of the sulfur oxidation pathway are much more suited to analyze this functional group in nature. Such functional approaches have already been applied for other physiological groups of bacteria such as denitrifying or sulfate-reducing bacteria [16–19], but not for sulfur-oxidizing bacteria.

At least two major biochemical pathways of sulfur oxidation could be identified [7]. The ‘S4intermediate’ pathway (S4I) includes the formation and oxidation of tetrathionate and is found in obligate and facultative chemolithotrophic bacteria such as Halothiobacillus neapolitanus, Acidithiobacillus ferrooxidans and Thiobacillus aquaesulis[20]. The ‘Paracoccus sulfur oxidation’ (PSO) pathway has been studied in detail in Paracoccus versutus[7] and the corresponding genes have been identified recently [21–24]. One enzyme of the thiosulfate-oxidizing multi-enzyme complex is soxB containing a prosthetic manganese cluster in the reaction center and being essential for thiosulfate oxidation by Paracoccus pantotrophus GB 17 [21,23,25].

The present study reports on the development and application of a primer system based on soxB as a functional marker for sulfur-oxidizing bacteria. Several degenerate primers for soxB were designed and used to amplify gene sequences of approximately 1000 bp from a variety of sulfur-oxidizing bacteria and environmental DNA. The sequences obtained from these PCR products allowed a first view of the distribution of this gene among sulfur-oxidizing bacteria and on its phylogenetic relationships.

2Materials and methods

2.1Bacterial strains

Reference strains of sulfur-oxidizing bacteria (listed in Table 2) were obtained from the DSMZ and grown on media recommended by DSMZ. Isolates and environmental DNA from the North Fiji Basin were obtained during the SO 134 cruise with the research vessel FS Sonne. The isolates were obtained according to [10] and identified by 16S rDNA sequencing and comparison with the EMBL database using FASTA searches (Table 2). Green sulfur bacteria were taken from the culture collection maintained in the authors’ laboratory.

Table 2.  PCR amplification of genomic DNA from new isolates and reference organisms
Species or isolate Identification of isolates according to 16S rDNA sequence (% similarity)Thiosulfate oxidationLength of obtained soxB sequenceAccession no. of soxB gene
  1. +: positive PCR amplification of the expected length, −: no PCR product or PCR products of wrong length were obtained, n.d.: not determined. Accession numbers in brackets are obtained from the EMBL database, all other sequences were obtained during this study. TIGR: sequences were obtained from the Institute of genomic research. DOE: sequences were obtained from DOE biological and environmental research program.

Thiosulfate-oxidizing reference strains
1A. aeolicus VF5database sequence+database sequence(AE000757)
2C. limicola f. thiosulfatophilum DSM 249Tdatabase sequence+1009AJ 294319
3Chlorobium vibrioforme f. thiosulfatophilum DSM 263database sequence+1059AJ 294320
4C. vibrioforme f. thiosulfatophilum DSM 265database sequence+1062AJ 294321
5C. tepidum ATCC 49652database sequence+database sequenceTIGR
6P. pantotrophus GB 17database sequence+database sequence(X79242)
7P. versutus DSM 582Tdatabase sequence+965AJ 294324
8Pelodictyon phaeoclathratiforme DSM 5477database sequence+1061AJ 294323
9R. palustris CGA 009database sequence+database sequenceDOE
10H. hydrothermalis DSM 7121database sequence+653AJ 294325
11H. neapolitanus DSM 581Tdatabase sequence+992AJ 294332
12S. novella DSM 506Tdatabase sequence+database sequence(AF139113)
13T. thioparus DSM 505Tdatabase sequence+1034AJ 294326
14T. crunogena ATCC 700270Tdatabase sequence+
15Thiomicrospira denitrificans DSM 1251Tdatabase sequence+
16Thiomicrospira pelophila DSM 1534Tdatabase sequence+
      
New thiosulfate-oxidizing isolates from North Fiji Basin
17H. hydrothermalis HY-66Halothiobacillus hydrothermalis DSM 7121 (99.7%)+
18T. crunogena HY-62T. crunogena ATCC 35932T (99.7%)+982AJ 294327
19α-Proteobacterium HY-103S. mediterraneus CH-B427 (95.4%)+1016AJ 294328
20Marinobacter sp. HY-106Marinobacter sp. DS 40M8 (98.7%)+642AJ 294329
21environmental clone HY-90n.d.+929AJ 294330
22environmental clone HY-86/2n.d.+968AJ 294331
      
Tetrathionate-forming strains
23P. stutzeri DSM 5190Tdatabase sequence+
24H. variabilis HY-51H. variabilis SW 48 (98.7%)+
25E. citreus HY-6E. citreus RE35F/1 (98.3%)+
      
Non-thiosulfate-oxidizing strains
26C. limicola DSM 245Tdatabase sequence
27Chlorobium phaeovibrioides DSM 269Tdatabase sequence
28Prosthecochloris aestuarii DSM 271Tdatabase sequence
29Pelodictyon luteolum DSM 273Tdatabase sequence

2.2Enrichment cultures of sulfur-oxidizing bacteria

Enrichment cultures were prepared from water samples from a hydrothermal vent habitat of the North Fiji Basin immediately after sampling. Five mM sodium thiosulfate was added to each 50-ml fluid sample. After 3 weeks of incubation under a CO2-enriched atmosphere at room temperature on a rotary shaker (100 rev min−1)the bacteria were harvested by centrifugation (10 min at 5000 rpm).

2.3Extraction of genomic DNA from pure cultures and environmental samples

DNA from pure cultures and enrichment cultures was extracted with spin columns of the QIAamp DNA extraction kit according to the manufacturer's instructions (Qiagen, Hilden, Germany). Twenty μl of the crude DNA extract was purified with spin columns from the QIAquick PCR purification kit (Qiagen, Hilden, Germany) according to the kit's manual. The rest of the extracted DNA was stored at −70°C for later purification and amplification.

2.4PCR amplification

PCR amplifications were performed as a two-step PCR in a total volume of 25 μl using ReadyToGo PCR beads (Amersham-Pharmacia, Freiburg). After a denaturation step of 2 min at 94°C, 10 cycles with an annealing temperature of 55°C (Progene-Cycler, Thermo-Dux, Wertheim, Germany) consisting of 30 s elongation at 72°C, 40 s annealing and 30 s denaturing at 94°C were performed. Then, additional 25 cycles were performed with an annealing temperature of 47°C. The amplification products were analyzed on 1% (w/v) agarose gels (Biozym, Hess, Oldendorf, Germany) followed by 10 min ethidium bromide staining (0.5 mg l−1). Eubacterial primers (5′-Start-9-27 and 3′-1387) were used for amplification of 16S rDNA. The primer sequences and positions according to the E. coli enumeration for the 16S rDNA and according to P. pantotrophus GB 17 enumeration for soxB are listed in Table 1.

Table 1.  Sequences and relative positions of amplification and sequencing primers for soxB and 16S rDNA
PrimerPrimer sequenceRelative position
  1. Sequences and positions of primers used for soxB and 16S rDNA amplification. The primer positions for 16S rDNA are enumerated according to E. coli sequence, soxB priming sites are enumerated according to the soxB sequence of Paracoccus pantotrophus GB 17 [35].

soxB432F5′-GAYGGNGGNGAYACNTGG-3′432–450
soxB693B5′-TANGGRAANGCYTGNCCGAT-3′713–693
soxB693F5′-ATCGGNCARGCNTTYCCNTA-3′693–713
soxB1164B5′-AARTTNCCNCGNCGRTA-3′1181–1166
soxB1164F5′-TAYCGNCGNGGNAAYTT-3′1166–1181
soxB1403B5′-TTRTCNGCNACRTCYTC-3′1403–1386
soxB1446B5′-CATGTCNCCNCCRTGYTG-3′1446–1428
5′-Start5′-GTTTGATCMTGGCTCAG-3′11–27
3′-13875′-CCCGGGAACGTATTCACCGT-3′1387–1368

2.5Sequencing

Sequences of soxB and 16S rDNA amplicons were obtained from QIAquick-purified products by cycle sequencing with the ABI Prism sequencing kit (Perkin-Elmer, Weiterstadt, Germany) and the chain termination reaction [26] using a capillary sequencer (Perkin-Elmer, Weiterstadt, Germany). The cycle sequencing reactions were performed with an annealing temperature of 48°C according to the manufacturer's manual.

2.6Phylogeny of soxB

For a phylogenetic evaluation of the soxB nucleotide sequences the dataset was treated in different ways. Generally, sequences were aligned using ClustalW [27] with subsequent manual control. Missing data and gaps in more than one sequence were treated as missing information. Nucleotide distances were calculated from the dataset according to the algorithm of Jukes and Cantor [28] by using DNADIST from the PHYLIP program package [29]. Maximum likelihood analysis was performed by using the PAUP program written by D.L. Swofford [30] with the Hasegawa–Kishino–Yano model [31] including a stepwise addition of species to the tree with bisection reconnection rearrangement within the heuristic search. Phylogenetic trees were inferred from the distance data with global rearrangements from FITCH [29].

The deduced amino acid sequences were aligned using the ClustalW program package from Washington State University, Seattle, WA, USA [27]. Gaps and missing sequence information present in more than one sequence were excluded from calculation of distances and trees yielding datasets of 378 amino acids. Protein distance matrices were calculated by using the PROTDIST program from the PHYLIP 3.57 program package [29] with the Dayhoff PAM matrix as amino acid replacement model [32]. Phylogenetic trees were inferred from the distances by using FITCH with the global rearrangement option. Maximum parsimony analysis of protein data was performed by using the protein parsimony algorithm from the PROTPARS program.

Bootstrap analyses of nucleotide sequences (maximum likelihood algorithm with 100 bootstrap resamplings) and amino acid sequences (maximum parsimony algorithm with 500 bootstrap resamplings) were performed with the PAUP program.

3Results

3.1PCR and sequence analysis

At the beginning of this study, the soxB sequences of Paracoccus pantotrophus GB 17 (originally known as Thiosphaera pantotropha, [21–23,25]) and Chlorobium tepidum ATCC 49652 from the genome sequencing project [33] were available in databases. An alignment of both sequences indicated five conserved regions within the gene. Five degenerate primers were designed to amplify a partial stretch of the soxB gene from a variety of thiosulfate-oxidizing reference strains and new isolates from a hydrothermal vent habitat (North Fiji Basin). Recently, sequences of soxB genes from Aquifex aeolicus VF5 (AE000757), Starkeya novella DSM 506T (AF139113) and Rhodopseudomonas palustris CGO 009 [34] also became available in the databases. (Note that Thiobacillus novellus has been reclassified as S. novella[35].) These sequences were used to refine the primer sequences. A detailed analysis of the primary and secondary structure of the amino acid sequences revealed that the selected priming sites are highly conserved and have only a few amino acid changes throughout all sequences. The conserved regions are present in the same relative order and in identical regions of all soxB genes demonstrating an orthologous relationship of the genes from the various bacteria included in this study.

SoxB genes from different thiosulfate-oxidizing bacteria and from natural samples were successfully amplified and yielded PCR products of the expected length in high quantity which could be purified and sequenced. The primer pair soxB432F/soxB1446B offered the most successful and reliable amplification results. This primer pair yielded PCR products from a number of thiosulfate-oxidizing bacteria and new isolates, but not from Halothiobacillus hydrothermalis DSM 7121 or from the new Marinobacter isolate HY-106. The primer pair soxB432F/soxB1446B was also used to amplify soxB genes from several hydrothermal vent water samples to which thiosulfate was added immediately after sampling and which were incubated after the addition of thiosulfate for 3 weeks. When soxB amplification failed with these primers, others were used in different combinations and in a multiplex PCR in order to obtain positive amplification results.

DNA from sulfide-oxidizing bacteria which lack the capability of thiosulfate oxidation or which oxidize thiosulfate by a different process were used as a negative control to test for the specificity of the primers and their discriminative capacity. No PCR amplification product was obtained with these strains under the described PCR conditions (Table 2). In addition, some of the thiosulfate oxidizers did not yield amplification products with any primer combination. No soxB amplification product was obtained from three Thiomicrospira species and an isolate of Halothiobacillus hydrothermalis HY-66 from the North Fiji Basin (numbers 14–17 in Table 2), all of which were shown to oxidize thiosulfate. These negative PCR results can be ascribed to the fact that the obligate chemolithotrophic Thiomicrospira species oxidize thiosulfate via the ‘S4I’ pathway [7,36,37] which is enzymatically different from the ‘PSO’ pathway of Paracoccus pantotrophus and does not involve the soxB gene. Also, three bacteria (Pseudomonas stutzeri, Halomonas variabilis, Erythrobacter citreus) that form tetrathionate in addition to sulfate apparently lack the soxB gene.

In two cases different results were obtained with closely related strains. The chemolithotrophic hydrothermal vent isolate Thiomicrospira crunogena HY-62 yielded a PCR product that could be identified as soxB by sequence and database comparison. The sequence revealed 75% amino acid similarity to the soxB sequence of Paracoccus pantotrophus GB 17 (Table 3) and all five conserved priming sites were found within the sequence undoubtedly demonstrating that a stretch of soxB was amplified. This contrasts with the results obtained for the type strain of this species from which no amplification product could be obtained but which was 99.7% similar in 16S rDNA sequence to the new isolate HY-62. Additionally, from two strains of H. hydrothermalis (DSM 7121 and HY-66) isolated from the same hydrothermal vent habitat ([38] and this work) and with 99.7% 16S rDNA sequence similarity, a soxB PCR product was obtained from the former but not from the latter one (Table 2). A systematic screening of the PCR conditions did not yield any amplification product for T. crunogena ATCC 35932T and H. hydrothermalis HY-66 and a stepwise decrease of the annealing temperature led to unspecific PCR products of different lengths.

Table 3.  Sequence similarities of amino acid and nucleotide sequences from soxB gene
 New isolates and reference speciesSequence similarity (%)
  123456789101112131415161718
  1. Distance data for amino acid sequences presented in the upper triangle were calculated with PROTDIST from the PHYLIP program package using Dayhoff PAM matrix. Distance data for nucleotide sequences presented in the lower triangle were obtained with DNADIST from PHYLIP program package using maximum likelihood algorithm.

1H. neapolitanus DSM 581T 82.151.952.250.351.649.855.048.849.547.449.151.049.250.350.755.443.5
2H. hydrothermalis DSM 712177.6 58.958.456.858.155.357.750.951.452.652.852.552.553.653.960.146.4
3C. limicola DSM 249T54.062.3 90.887.587.780.462.445.651.348.047.548.047.748.949.055.442.7
4C. tepidum ATCC 4965254.761.287.9 90.090.382.764.046.352.148.147.548.347.548.849.555.942.6
5C. vibrioforme DSM 26353.862.883.384.0 95.380.963.045.951.448.647.248.048.149.249.955.441.1
6C. vibrioforme DSM 265T55.163.585.884.389.5 80.361.745.752.548.047.148.147.748.849.455.941.9
7P. phaeoclathratiforme DSM 547752.655.874.174.274.473.7 64.745.849.747.846.846.746.548.849.254.642.4
8Marinobacter sp. HY-10657.660.861.662.062.663.462.6 52.054.754.153.254.854.855.355.363.144.3
9R. palustris CGA 00954.361.454.353.553.755.250.855.3 76.158.759.962.960.461.761.550.841.5
10S. novella DSM 506T53.459.657.658.257.360.154.355.277.2 64.365.767.067.066.666.651.745.8
11Unidentified α-Proteobacterium HY-10351.758.853.451.152.152.651.660.862.163.7 83.072.474.278.077.652.741.6
12T. crunogena HY-6254.962.752.652.054.054.051.159.764.964.977.4 71.172.075.875.851.641.9
13Environmental sequence HY-9057.357.654.152.652.953.352.662.264.966.573.271.7 87.674.574.660.446.6
14Environmental sequence HY-86/255.257.852.552.553.052.550.962.064.366.373.672.492.0 79.279.150.443.3
15P. versutus DSM 582T55.962.257.056.357.557.153.259.666.071.074.273.673.376.1 98.251.644.5
16P. pantotrophus GB 1755.963.658.057.156.656.853.860.067.271.174.874.474.977.394.8 51.944.4
17T. thioparus DSM 505T59.467.659.659.860.962.256.762.560.361.158.859.570.459.759.560.3 45.1
18A. aeolicus VF545.947.446.746.944.845.445.948.445.247.145.142.947.945.746.047.046.3 

3.2Evolutionary relationship of soxB

All amplified PCR products were sequenced on both strands and translated into amino acid sequences. Phylogenetic trees were calculated from aligned amino acid and nucleotide sequences to infer reliable branching orders (Fig. 1). The soxB amino acid and nucleotide alignments yielded nearly identical trees and the overall topologies of both trees were strongly substantiated by high bootstrap values (Fig. 1).

Figure 1.

Phylogenetic tree of thiosulfate-oxidizing reference strains and new isolates from the North Fiji Basin based on stretches of approximately 1000 bp of the soxB gene. The tree was calculated from deduced amino acid sequences using the PHYLIP program package based on a manually controlled Clustal alignment. Bootstrap analyses utilized PAUP and were performed on amino acid data with parsimony algorithm and 500 resamplings and on nucleotide data with maximum likelihood algorithm and 100 resamplings. The bootstrap values upon the branches were obtained from deduced amino acid sequences, the values below the branches were retrieved from nucleotide sequences. Only values higher than 60% are shown.

Within the phylogenetic tree based on soxB sequences, thiosulfate-oxidizing strains of the green sulfur bacteria formed a separate cluster with strong bootstrap support and similarity values of 80–95% for the amino acid data (Fig. 1, Table 3). The soxB sequence of A. aeolicus, an early diverging hyperthermophilic chemolithotroph [39], formed a separate phylogenetic branch with the lowest amino acid sequence similarities (42–47%) to all other sequences which was congruent to the 16S rDNA derived phylogeny. Another major branch of related soxB sequences is represented by the α-Proteobacteria and includes two Paracoccus strains, a new isolate related to Sulfitobacter mediterraneus (strain HY-103), and two environmental clones obtained from thiosulfate-fed enrichment cultures of the hydrothermal vent samples which share amino acid sequence similarities of 72–78%. Also S. novella and R. palustris are included in this group, but somewhat more distant (Fig. 1). Although the North Fiji isolate HY-62 revealed 99.7% sequence similarity of 16S rDNA to the type strain of T. crunogena, which belongs to the γ-group of the Proteobacteria, its soxB sequence is associated to this branch of α-Proteobacteria (Fig. 1).

Members of the γ-Proteobacteria, H. neapolitanus and H. hydrothermalis[20], represent a separate phylogenetic branch which is distantly related to all other soxB sequences with similarity values of 50 and 60% for amino acid and nucleotide sequences, respectively (Table 3). The soxB sequence of Thiobacillus thioparus, a representative of the β-Proteobacteria, also forms an independent phylogenetic branch distantly affiliated with the other sequences (Fig. 1, Table 3).

Generally, the genetic distances listed in Table 3 demonstrated a high percentage of sequence similarity within the major phylogenetic branches but significantly lower similarity values between the different branches for both amino acid and nucleotide data. The average values between the distantly related branches of Proteobacteria were 45–55% for amino acid and 55–65% for nucleotide sequences. Among the closely related sequences these values are well above 70% and reach 95% for nucleotide data (98% for amino acid data).

The branching order of the phylogenetic trees calculated from the soxB sequences and the corresponding 16S rDNA data are basically congruent concerning the separation of green sulfur bacteria and the different branches of representatives of the α-, β-, and γ-Proteobacteria. Nevertheless, differences exist in two important details which primarily concern sequences obtained from isolates and clones from the North Fiji Basin. Firstly, soxB sequences of isolates and environmental samples from the North Fiji Basin with the exception of the Marinobacter isolate HY-106 are closely related to each other and associated with the branch of the α-Proteobacteria although the T. crunogena isolate HY-62 belongs to the γ-group of Proteobacteria (Fig. 1). Secondly, the isolate HY-106 belongs to the γ-group of Proteobacteria and is closely related (98.7% 16S rDNA sequence similarity) to a Marinobacter species but the soxB sequence of this strain is only distantly related to the other γ-Proteobacteria (Fig. 1, Table 3).

4Discussion

The aim of this study was to develop a PCR assay and a probe for sulfur-oxidizing bacteria based on a functional gene essential for sulfur oxidation and to screen its distribution among recognized sulfur-oxidizing bacteria, new isolates and environmental samples from a hydrothermal vent habitat. Because of the diversity of sulfur-oxidizing bacteria and the variety of phylogenetic lines in which sulfur oxidation is possible, genes of the sulfur oxidation pathway are the preferred choice for environmental analyses of these bacteria. In view of the variation of biochemical reactions of sulfur oxidation pathways and the wide distribution of sulfur-oxidizing genes throughout the bacterial world [4–6], it is not likely that a universal PCR primer set will be available for all sulfur- and thiosulfate-oxidizing bacteria. Because thiosulfate oxidation is a property common to most sulfur-oxidizing bacteria and is an important step in the biogeochemical sulfur cycle [10,13], this study established a PCR protocol for the detection of soxB genes.

The level of conservation of the soxB amino acid sequences ranged from 41 to 98% and is comparable to other functional genes [16–19] but the distribution of similarity values within the distance matrix (Table 3) is quite conspicuous. Amino acid sequences of different clusters revealed low similarity values (45–55%), whereas closely related sequences within the clusters shared much higher similarities (>70–98%) reflecting a strong separation between the different lines of descent of the soxB gene. This observation may indicate that soxB may have a long evolutionary history and/or an increased substitution rate leading to a strict separation of major phylogenetic lines and therefore supports the idea of widespread dissemination of sulfur-oxidizing bacteria due to an early origin of this physiological trait as put forward by Lane et al. [8].

All of the strains of green sulfur bacteria that are able to oxidize thiosulfate yielded PCR products of the soxB gene, whereas those strains lacking thiosulfate-oxidizing capacity did not (Table 2). Although different enzyme activities of thiosulfate metabolism including thiosulfate sulfur transferase, thiosulfate oxidoreductase and thiosulfate reductase have been described in Chlorobium species [40], the soxB gene so far was not detected in these bacteria, except for the genome sequence of C. tepidum. It is known that during sulfur oxidation in green sulfur bacteria polythionates are not detectable [41] which is characteristic for the PSO pathway. Another interesting observation is that although the sox gene cluster has been found genomically encoded in P. pantotrophus GB 17 there is evidence for a plasmid-encoded capability of thiosulfate oxidation in Chlorobiaceae. A transformation experiment with a 14-kb plasmid from Chlorobium limicola f. thiosulfatophilum into another C. limicola strain which was unable to grow on thiosulfate led to the acquisition of a thiosulfate-oxidizing capacity by this latter strain [42,43].

Another striking result of this investigation was the evidence for soxB in chemolithotrophic bacteria such as T. crunogena, H. neapolitanus and H. hydrothermalis[20]. These findings demonstrate the presence of enzymes of the ‘PSO’ pathway in obligate chemolithotrophic sulfur-oxidizing bacteria which were generally considered to use the ‘S4I’ pathway for sulfur oxidation [7]. The distribution of soxB genes among these sulfur-oxidizing bacteria is complicated because of two aspects: (i) the existence of soxB seems to be strain dependent and (ii) the distribution of soxB genes among thiosulfate-oxidizing bacteria could be the result of a lateral gene transfer. Assuming a transferable soxB gene, the successful amplification of soxB gene sequences from T. crunogena HY-62 contrary to the negative PCR results obtained with T. crunogena ATCC 35932T and two other Thiomicrospira species may become plausible. Thus, if T. crunogena HY-62 has obtained the soxB gene through lateral gene transfer from an α-Proteobacterium from the environment, this could explain the affiliation of the soxB sequence of this γ-Proteobacterium to soxB sequences from other bacteria (α-Proteobacteria) from this environment. Different results for the PCR assay of two H. hydrothermalis strains isolated from the same habitat (Table 2) may indicate another example of lateral gene transfer of the soxB gene towards strain DSM 7121, could however also be explained by the loss of soxB in strain HY-66.

Besides lateral gene transfer there may be two other possible explanations for phylogenetic discrepancies: a convergent evolution of different enzymes or a functional divergence of proteins which may have evolved from the same origin. Concerning the high level of sequence similarity which is comparable to other orthologous genes [16–18] and the high homology of the soxB amplicons demonstrated by the relative order and distance of the conserved regions, an orthology of the soxB genes can be assumed. Therefore, a convergent evolution of different enzymes is quite unlikely. On the other hand, a divergent evolution leading to functional differences in similar proteins in general cannot be excluded although we could not detect sequential similarities or homologies to any other protein. Neither the tetrathionate reductase nor the eukaryotic sulfite oxidases available in the databases yielded any significant similarity or homology to the soxB gene products. However, a functional divergence of soxB amplicons needs approval by biochemical experiments that clearly demonstrate the functional role of these enzymes in these bacteria.

The most striking result of this work is the strong separation of soxB sequences from major phylogenetic lines that results in clearly separated clusters, i.e. of green sulfur bacteria and of different branches of representatives of α-, β-, and γ-Proteobacteria including representative species of the former genus Thiobacillus (now H. neapolitanus–γ-Proteobacteria, T. thioparus–β-Proteobacteria and S. novella–α-Proteobacteria [20,35]). This result is of particular interest because strain-specific distribution of soxB genes and apparent evidence for lateral transfer of this gene would be expected to seriously spoil the evolutionary record. Future work has to demonstrate the extent to which gene transfer has left footprints in the evolutionary record.

Acknowledgements

Part of this work was supported by a Grant (no. 03G0134E) from the Bundesminister für Bildung, Wissenschaft, Forschung und Technologie.

Ancillary