Correspondence: Christian Rinke, School of Biological Sciences, Washington State University, Pullman, WA 99163, USA. Tel.: +1 509 335 7281; fax: +1 509 335 3184; e-mail: firstname.lastname@example.org
The giant marine ciliate Zoothamnium niveum (Ciliophora, Oligohymenophora) is obligatorily covered by a monolayer of putative chemoautotrophic sulfur-oxidizing (thiotrophic) bacteria. For Z. niveum specimens from the Caribbean Sea it has been demonstrated that this ectosymbiotic population consists of only a single pleomorphic phylotype described as Candidatus Thiobios zoothamnicoli. The goal of our study was to identify and phylogenetically analyse the ectosymbiont(s) of a recently discovered Z. niveum population from the Mediterranean Sea, and to compare marker genes encoding key enzymes of the carbon and sulfur metabolism between the two symbiont populations. We identified a single bacterial phylotype representing the ectosymbiont of Z. niveum from the Mediterranean population showing 99.7% 16S rRNA gene (99.2% intergenic spacer region) similarity to the Caribbean Z. niveum ectosymbiont. Genes encoding enzymes typical for an inorganic carbon metabolism [ribulose 1,5-bisphosphate carboxylase/oxygenase (RuBisCO)] and for sulfur metabolism (5′-adenylylsulfate reductase, dissimilatory sulfite reductase) were detected in both symbiotic populations. The very high amino acid sequence identity (97–100%) and the high nucleic acid sequence identity (90–98%) of these marker enzymes in two geographically distant symbiont populations suggests that the association of Z. niveum with Cand. Thiobios zoothamnicoli is very specific as well as temporally and spatially stable.
Symbioses with chemoautotrophic, sulfur-oxidizing (thiotrophic) Bacteria are widespread among invertebrate and protist hosts and have been reported from diverse marine, shallow-water to deep-sea habitats (Fisher, 1996; Polz et al., 2000; Bright & Giere, 2005). A putative thiotrophic symbiosis hosted by the sessile, colonial, giant ciliate Zoothamnium niveum (Ciliophora, Oligohymenophora) has been reported from subtropical, tropical, and temperate shallow subtidal areas (Bauer-Nebelsick et al., 1996a, b; Ott et al., 1998; Rinke et al., 2006, 2007) (Fig. 1). The ciliate builds a colonial formation that consists of a central stalk with alternate branches bearing three different cell types: the microzooids are feeding types, the macrozooids are dispersal types, capable of leaving the colony to form an entire new colony after settlement, and the terminal zooids are responsible for asexual reproduction by longitudinal fission (Bauer-Nebelsick et al., 1996a, b). Zoothamnium niveum is obligatorily covered by a dense coat of ectosymbiotic bacteria (Bauer-Nebelsick et al., 1996a; Clamp & Williams, 2006; Rinke et al., 2006), whereby the microorganisms establish a monolayer composed of two morphotypes. Stalk, branches, terminal zooids and macrozooids bear rod-shaped symbionts, whereas microzooids are covered by rods on the aboral side as well as by coccoid rods on the oral side. A series of intermediate shapes between these two bacterial morphotypes on the microzooids was noted (Bauer-Nebelsick et al., 1996b). Phylogenetic 16S rRNA gene analyses and FISH of specimens from the Belize Barrier Reef (Caribbean Sea) revealed a single pleomorphic phylotype named Candidatus Thiobios zoothamnicoli, which clustered with thiotrophic free-living and symbiotic Gammaproteobacteria (Rinke et al., 2006). Based on the hydrogen sulfide-rich habitat of the symbiosis (Ott et al., 1998) and the white color of the symbiont (Bauer-Nebelsick et al., 1996a), which was attributed to membrane-bound vesicles containing elemental sulfur, a chemoautotrophic sulfur-oxidizing nature of the ectosymbiotic bacteria was suggested (Bauer-Nebelsick et al., 1996a). This assumption is supported by recent cultivation experiments which revealed that the symbiosis is not able to survive without sulfide (Rinke et al., 2007). However, further studies that would support these putative physiological capabilities of the symbiont, such as screening for genes involved in carbon fixation or sulfur oxidation metabolic pathways, have not been undertaken so far.
The reductive pentose phosphate cycle (Calvin–Benson cycle) represents the CO2 fixation pathway in almost all aerobic autotrophic bacteria (Watson & Tabita, 1997). One of the key enzymes involved in the Calvin–Benson cycle is ribulose 1,5-bisphosphate carboxylase/oxygenase (RuBisCO, EC 220.127.116.11), which catalyzes the assimilation of CO2 to organic carbon. The enzyme exists in two main forms. Form I consists of large and small subunits (LnSn, typically L8S8), and form II contains only large subunits (Ln) (Kellogg & Juliano, 1997); however, only the large subunits are responsible for carbon fixation (Miziorko & Lorimer, 1983). The genes coding for the large subunits of form I and form II are cbbL and cbbM, respectively.
Thiotrophic bacteria require a reduced sulfur source, usually hydrogen sulfide or thiosulfate, as electron donor and oxygen, or in some cases, nitrate as electron acceptor. The electrons derived from sulfur oxidation are used for energy transformation via the respiratory chain and for carbon dioxide fixation. However, a common mechanism for bacterial sulfur oxidation does not exist (Friedrich et al., 2005). An enzyme system present in some but not all photo and chemotrophic microorganisms that oxidize reduced sulfur species was found to be encoded in the dsr gene cluster, which was originally identified in the phototrophic sulfur-oxidizing bacterium Allochromatium vinosum (Pott & Dahl, 1998; Dahl et al., 2005). The dsrAB gene duo encode the α and β subunits of a dissimilatory sulfite reductase (dsr; EC 18.104.22.168), a characteristic enzyme used by sulfate-/sulfite-reducing microorganisms for generating energy via anaerobic respiration (Zverlov et al., 2005; Stahl et al., 2007; Loy et al., 2008a) and which is operating in the reverse direction in sulfur-oxidizing bacteria. Molecular and (meta)genomic studies indicate that dsrAB genes are widespread among sulfur oxidizers of the phyla Proteobacteria (the classes Alpha-, Beta-, and Gammaproteobacteria) and Chlorobi (green sulfur bacteria) (Dahl et al., 1999; Sabehi et al., 2005).
In the sulfur metabolism of sulfur-oxidizing Bacteria and Archaea studied so far the main intermediate in the oxidation of reduced sulfur compounds to sulfate is the highly reactive sulfite. Two main pathways of sulfite oxidation are known: the direct oxidation to sulfate, or the indirect oxidation catalyzed by the enzymes adenosine-5′-phosphosulfate (APS) reductase (EC 22.214.171.124) and ATP sulfurylase (EC 126.96.36.199) with APS as an intermediate. APS reductase, consisting of two subunits (α and β) encoded by the genes apsA and apsB, respectively, is also found in sulfate reducers catalyzing the two-electron reduction of APS to sulfite and AMP, and was therefore proposed as a phylogenetic marker for bacteria involved in oxidative and reductive sulfur metabolism (Hipp et al., 1997). Lateral gene transfer was, however, reported for apsA of sulfate-reducing as well as sulfur-oxidizing bacteria (Friedrich, 2002; Meyer & Kuever, 2007a, b).
The main objectives of this study were to identify and phylogenetically analyse the ectosymbiont(s) of a recently discovered Z. niveum population from the Mediterranean Sea (Rinke et al., 2007), and to compare the genetic makeup (with respect to three key enzymes involved in autotrophic carbon fixation, RuBisCO, and in sulfur metabolism, dsr and APS reductase) between the Mediterranean symbiont(s) and a recently described symbiont of Z. niveum from the Caribbean Sea (Rinke et al., 2006). A single bacterial phylotype could be indentified representing the ectosymbiont of Z. niveum from the Mediterranean population showing 99.7% 16S rRNA gene similarity to the Caribbean Z. niveum ectosymbiont. We detected genes encoding enzymes typical for inorganic carbon (RuBisCO –cbbL) and sulfur metabolism (dsrAB, apsA) in both symbiotic populations. Comparing two geographically distant symbiont populations we found very high amino acid sequence identity (97–100%) of these marker enzymes, suggesting that the association of Z. niveum with Cand. Thiobios zoothamnicoli is very specific and highly stable over temporal and spatial scales. However, a universal free-living population of the symbiont serving as reinfection source cannot be ruled out completely.
Materials and methods
Specimen collection and preparation
Colonies of Z. niveum (Hemprich & Ehrenberg, 1831) were collected by scuba diving in the bay of Calvi (42°34′N, 8°43′E) in Corsica, France (Mediterranean Sea) on decaying Posidonia oceanica leaves at a depth of 13.8 m in 2005 (site 3, Rinke et al., 2007). Colonies of the Caribbean Z. niveum population were sampled in 2006 at Cuda Cut (16°83′N, 88°10′W) at the island Twin Cays belonging to the Belize Barrier Reef. Specimens were cut from a mangrove peat wall, with a vertical elongation of about 15 m, at 1–3 m depth and combined during scuba diving. Immediately after collection, colonies from the respective sampling site were randomly picked and fixed in 99% ethanol for molecular biology experiments or in 50% Bouin's solution for FISH and subsequent confocal laser scanning microscopy.
Gene amplification, cloning and sequencing
From each sampling site (Caribbean and Mediterranean Sea) 20 colonies were randomly chosen, pooled and used for PCR-mediated amplification of gene fragments of interest. To avoid contamination with other bacteria occurring on the lower parts of large colonies (Rinke et al., 2006), the basal third of each colony was cut off. Subsequently, samples were transferred to phosphate-buffered saline (PBS; 0.1 M, pH 7.4), followed by two short pulses of sonication for 10 s each (HD 2070 sonicator, Bandelin, Berlin, Germany), to mechanically disintegrate the large colonies. For each PCR, 1 μL of these PBS solutions containing a mixture of host tissue and symbionts was directly used as a template without prior DNA extraction, except for the PCR amplification of the cbbL, cbbM, and apsA genes of the Caribbean samples, for which DNA was isolated as described by Groot et al. (2005). PCR reactions amplifying the 16S rRNA genes and the intergenic spacer region (ISR) between the 16S and the 23S rRNA genes were performed with a standard PCR cycling program using an annealing temperature of 50 and 54 °C, respectively. PCR reactions targeting cbbL and cbbM genes were carried out according to Elsaied & Naganuma (2001), those targeting apsA genes according to Blazejak et al. (2006), and those targeting dsrAB genes according to Kjeldsen et al. (2007). All primers used in this study are shown in Table 1. Positive controls applied were: Ralstonia metallidurans DNA extract for cbbL; Polaromonas naphthalenivorans DNA extract for cbbM; Riftia pachyptila trophosome extract for cbbM and apsA. Each assay contained a negative control, in which no DNA was added to the reaction mixture.
PCR products were separated by electrophoresis using 1.5% agarose gels, which were stained with ethidium bromide and visualized by UV transillumination. Amplification products of the desired size (c. 1500 bp for the 16S rRNA gene, 1200 bp for ISR, 800 bp for cbbL, 400 bp for apsA, and 1900 bp for dsrAB) were extracted from the gel and used for cloning reactions with the TopoTA cloning kit (Invitrogen Life Technologies, Lofer, Austria), except for cloning of the dsrAB, which was carried out using the TopoXL cloning kit (Invitrogen Life Technologies).
Clones of cbbL and aprA genes were analysed with a restriction fragment length polymorphism (RFLP) applying a combination of the restriction enzymes AluI and MspI at 0.67 U μL−1 each (Fermentas GmbH, St Leon-Rot, Germany), whereas dsrAB clones were analysed applying MspI only. The enzyme sample mix was incubated for 90 min at 37 °C and subsequently visualized using 3% agarose gels, which were thereafter stained with ethidium bromide and visualized by UV transillumination.
Nucleotide sequences of cloned DNA fragments were determined on an ABI 3130 XL genetic analyser using the BigDye Terminator kit v3.1 (ABI, Vienna, Austria).
Sequence analysis and phylogenetic reconstruction
Maintenance and phylogenetic analysis of the obtained sequences were conducted using the arb program package (Ludwig et al., 2004). New sequences were added to the corresponding 16S rRNA gene, ISR, RuBisCO, apsA and dsrAB nucleotide databases. Automatic alignment was performed with the fast aligner of the arb package and subsequently corrected by visual inspection. The cbbL, apsA, and dsrAB sequences were translated to amino acids. The phylogenetic relationships of 16S rRNA gene and ISR nucleic acid sequences were inferred applying the following methods: neighbor-joining (Jukes–Cantor correction), maximum parsimony (Phylip DNAPARS 3.573, with and without 100 resamplings), maximum likelihood (AxML), and treepuzzle 5.0. The results of all applied treeing methods were combined in a consensus tree. Phylogenetic calculations of RuBisCO, APS reductase, and Dsr were applied with the deduced amino acid sequences using the following treeing methods: neighbor-joining (using the PAM matrix for amino acid substitution), maximum parsimony (Phylip protpars 3.573, with and without 100 resamplings), and maximum likelihood (protml). APS reductase and dsr amino acid sequence phylogenies were additionally inferred using treepuzzle 5.0. Obtained results were combined in a consensus tree.
For 16S rRNA gene phylogeny, an outgroup of 34 sequences covering the main phyla of the domain Bacteria was used. A filter considering only those alignment positions that were conserved in at least 50% of all Gammaproteobacteria sequences was applied, and only sequences with a length of more than 1400 nucleotides were used, resulting in 1288 positions for calculations. Phylogenetic analyses of RuBisCO were performed without a conservation filter based on an alignment of 271 amino acids, reflecting the sequence length obtained in this study. An outgroup consisting of five RuBisCO form II sequences was used. Amino acid sequences shorter than that indicated above or with only partial overlaps to the sequence region used for calculation were individually added to the different trees using the parsimony interactive tool of arb. The phylogeny of APS reductase was calculated from partial sequences consisting of 130 amino acids. Amino acid sequences shorter than indicated above were added to the different trees using the parsimony interactive tool of ARB. Dsr trees were calculated based on sequences >592 amino acids and using a filter that excludes sequence regions affected by insertions and deletions (indels), leaving a total of 551 amino acid positions for phylogenetic analyses. ISR sequences were screened for tRNA genes with the online software tool aragorn (Laslett & Canback, 2004). The retrieved nucleotide sequences have been deposited in the public databases GenBank/EMBL/DDBJ under accession numbers EU439003 (16S rRNA gene and ISR); EU439000, EU439002 (cbbL); EU439004, EU439001 (apsA); and EU817177, EU817178, EU817179 (dsrAB).
For FISH, specimens from the bay of Calvi (Mediterranean Sea) were fixed with 50% Bouin's solution (EMS, Hatfield, PA) for 15 min, rinsed in PBS three times for 5 min each, dehydrated in 30% and 50% ethanol for 5 min each, and stored in 70% ethanol until further treatment up to a year later. Specific probes for Bacteria (EUB338, EUB338II, EUB338III; Amann et al., 1990; Daims et al., 1999), Gammaproteobacteria (GAM42a; Manz et al., 1992), Betaproteobacteria (BET42a; Manz et al., 1992) as well as two Cand. Thiobios zoothamnicoli-specific probes ZNS196 (5′-CTG ATA GTG ACC GAA GTC T-3′) and ZNS1439 (5′-GAG CGC CCA TTA TTA AGC TA-3′) (Rinke et al., 2006) were labeled on their 5′ end with the fluorescent dyes Cy3, Cy5, or fluorescein (Fluos) (Thermo Hybaid, Ulm, Germany). Probe NON338 (Wallner et al., 1993) complementary to EUB338 was applied as negative hybridization control. Detailed information on the oligonucleotide probes used in this study is available at probeBase (Loy et al., 2007). FISH on whole-mounted Z. niveum colonies was performed as described previously (Rinke et al., 2006).
Results and discussion
16S rRNA gene phylogeny and FISH
For the Calvi (Mediterranean Sea) ciliate sample, five 16S rRNA genes and four ISR clones were randomly chosen and sequenced completely. The 16S rRNA gene sequences were all 1453 nucleotides in length and of identical composition. The ISR sequences were also identical with each other (416 nucleotides in length) and contained two tRNA genes (isoleucine and alanine; Fig. 3). The ISR could be unambiguously attributed to the 16S rRNA gene due to the sequence identity of 372 overlapping nucleotides.
The 16S rRNA gene sequence analysis revealed that the symbiont of Z. niveum from Calvi shows 99.7% similarity to Cand. Thiobios zoothamnicoli, the ectosymbiont of Z. niveum colonies from the Caribbean Sea (island Twin Cays, Belize) (Rinke et al., 2006). According to Stackebrandt & Ebers (2006) a 16S rRNA gene sequence similarity below a threshold range of 98.7–99% is recommended for the unambiguous differentiation of two species. However, a 16S rRNA gene sequence similarity of over 99%, as found in this study, can still mask high genome heterogeneity (Thompson et al., 2005). It is thus not possible to decide unambiguously whether the two symbiont populations belong to the same species. Tentatively, we will refer thereafter to the 16S rRNA gene sequence obtained in this study from Calvi, France, as ‘strain Calvi’, and to the sequence from Twin Cays, Belize (Rinke et al., 2006) as ‘strain Twin Cays’ (Table 2). All treeing methods used to resolve the phylogenetic relationships of the obtained 16S rRNA gene sequence to other Gammaproteobacteria in our dataset provided highly congruent results. The Mediterranean Z. niveum symbiont formed a monophyletic group together with Cand. Thiobios zoothamnicoli from the Caribbean Sea, the free-living sulfur-oxidizing bacterium ODIII6 (Kuever et al., 2002), a symbiont (named ‘scaly snail endosymbiont’) of a recently discovered hydrothermal vent gastropod (family Peltospiridae) from the Indian Ocean (Goffredi et al., 2004) and three uncultured marine Gammaproteobacteria retrieved from presumably sulfide-rich habitats (Fig. 2). Furthermore, all treeing methods supported the monophyly of a symbiont-containing group comprising the gammaproteobacterial ectosymbiont of the nematode Laxus oneistus of the subfamily Stilbonematinae (Polz et al., 1994), the endosymbiont of the nematode Astomonema sp. (Musat et al., 2007), and endosymbionts of the oligochaete subfamily Phallodrillinae (Dubilier et al., 1999).
Table 2. Sequences obtained from the two geographically distinct Candidatus Thiobios zoothamnicoli strains from Twin Cays (Caribbean Sea) and Calvi (Mediterranean Sea)
The obtained ISR sequences from strain Calvi showed a sequence similarity of 99.2% to strain Twin Cays. In previous studies the ISR has been a powerful tool for comparing bacterial strains and populations (Otsuka et al., 1999; Boyer et al., 2001), because ISR are highly variable due to frequent DNA recombination and mutation events (Gürtler, 1999). The ISR sequence similarity between strain Calvi and strain Twin Cays is surprisingly high, showing a variation of only five-nucleotide position, three occurring within tRNA coding regions (Fig. 3). Nevertheless, the available ISR sequences unambiguously distinguish the two Cand. Thiobios zoothamnicoli strains.
The very high sequence similarity of the 16S rRNA gene and the ISR points to a relatively recent separation of the Caribbean and the Mediterranean Cand. Thiobios zoothamnicoli populations in geological times. Using molecular clocks of bacteria, assuming that bacterial 16S rRNA gene sequences diverge at a substitution rate of 1–2% per 50 million years (Ochman & Wilson, 1987; Moran et al., 1993), the diversity observed between strain Calvi and strain Twin Cays of 0.36% results in an estimated separation of the two lineages 9–18 million years ago. Since the Atlantic Ocean was already established 65 million years ago, the Z. niveum symbiosis must have found ways of bridging far distances of what is nowadays over 9200 km (5700 miles) to achieve such a wide distribution.
Macrozooids, which are covered by a monolayer of symbionts (Bauer-Nebelsick et al., 1996a), leave the colony as swarmers and survive a maximum of 24 h as motile stages, in which they cover a maximum distance of 400 m before settling (Ott & Bright, 2004). Therefore it is more likely that drifting sulfide sources, such as mangrove tree trunks or other massive macrophyte debris, serve as vessels to explore new distant habitats. Hydrogen sulfide, generated by a diverse community of microorganisms using sulfate as an oxidizing agent, is known to be a major product of anaerobic degradation of cellulose in marine systems (Leschine, 1995).
The Mediterranean Sea dried up for the last time 5–6 million years ago (Hsü, 1983), and thus it is most likely that the Z. niveum symbiosis (re)colonized this habitat from the Atlantic Ocean. Furthermore, the land connection between North and South America started developing around 5 million years ago, which makes it possible that the Z. niveum symbiosis also swept out into the Pacific Ocean.
Alternatively, one cannot rule out the possibility that the symbiont has its own additional way of distribution by co-occurring in a motile, free-living population in the water column interacting with the sessile symbiotic population, as suggested for the thiotrophic symbionts of Rimicaris exoculata, a hot vent shrimp from the Mid-Atlantic Ridge (Polz & Cavanaugh, 1995). However, considering the scattered distribution of sulfide sources in the Z. niveum habitat (Ott et al., 1998) and results showing that the swarmers actively find and colonize new sulfide minivents (Pöhn, 2002) within a few 100 m day−1 (Ott & Bright, 2004), it seems rather unlikely that a free-living population of the bacterial symbiont could keep up with this speed of distribution.
FISH of whole-mounted Z. niveum colonies strain Calvi using Cand. Thiobios zoothamnicoli-specific probes ZNS196 and ZNS1439 under stringent conditions (35% formamide in the hybridization buffer) demonstrated that both ectosymbiotic morphotypes, the rods as well as the coccoid rods, hybridized with the symbiont-specific probes, the probe for Gammaproteobacteria, and the probe for Bacteria (Fig. 1b–e), and therefore belong to one phylotype. These results are consistent with those reported for the Caribbean Z. niveum population from the mangrove island Twin Cays, Belize (Rinke et al., 2006).
cbbL, apsA and dsrAB gene diversity
From the Caribbean Z. niveum sample, 26 clones of the amplified cbbL gene, encoding the large subunit of RuBisCO, were analysed with RFLP, resulting in identical patterns. Thereafter six clones were randomly chosen, sequenced and found to be identical. From the Mediterranean sample, six cbbL clones were retrieved, which were fully sequenced and also found to be identical. The obtained cbbL gene sequences of both symbionts (Caribbean and Mediterranean) were 812 nucleotides in length (Table 2) and showed 99% nucleic acid sequence similarity. The deduced amino acid sequences (271 amino acids) were 100% identical. The closest relative of both symbiont RuBisCo sequences was the RuBisCO of an uncultured bacterium, retrieved from suboxic mobile muds from French Guyana (V. Madrid, pers. commun.), showing 97% amino acid identity (Fig. 4). All amino acid identities to other RuBisCO sequences in public databases were below 88.5%. Although multiple PCR assays under various conditions were performed, form II RuBisCO, encoded by cbbM, was not detected in the Z. niveum samples. All applied phylogenetic treeing methods placed the RuBisCO sequences of both Z. niveum strains in a monophyletic group together with thiotrophic symbiotic (Inanidrillus leukodermatus and Inanidrillus makropealos endosymbiont) and free-living (A. vinosum) Gammaproteobacteria, as well as with an uncultured bacterium from suboxic mobile mud and Cyanobacteria of the Prochlorococcus and Synechoccus spp. clade (PS clade), all bearing form 1A RuBisCO (Fig. 4).
Amplified apsA sequences, encoding the APS reductase α subunit, were cloned and the RFLP analyses of 28 clones from the Caribbean sample resulted in two distinct, equally distributed patterns. Thereafter five clones of each pattern were randomly chosen and sequenced, revealing identical sequences, showing only three nucleotide sequence differences at the primer binding sites, most likely due to the use of degenerate primers applying PCR. From the Mediterranean sample, five clones were sequenced and found to be identical. The apsA genes of both symbionts were 389 nucleotides in length (Table 2) and shared 95.7% nucleic acid sequence identity. The deduced amino acid sequences (130 amino acids) were 98.4% identical. Phylogenetic analysis of apsA amino acid sequences of strain Twin Cays and strain Calvi revealed the highest amino acid sequence identity to Thiobacillus denitrificans apsA2 (95.1/95.1%), A. vinosum strain DSM 180 (93.4/92.6%), and to Candidatus Vesicomyosocius okutanii (93.4/91.8%), a thiotrophic intracellular symbionts of the deep-sea clam Calyptogena okutanii (Fig. 5). All treeing methods applied placed the apsA gene of the two Z. niveum ectosymbiont strains in a well-supported group together with sulfur-oxidizing symbiotics (Inanidrilus exumae endosymbiont, Cand. Vesicomyosocius okutanii, Candidatus Ruthia magnifica, Idas sp. endosymbiont, Bathymodiolus brevior endosymbiont) and free-living (A. vinosum) Gammaproteobacteria, with T. denitrificans, a sulfur-oxidizing Betaproteobacteria, and with an unclassified environmental clone S8CL5 from metalliferous peat soils. This group clusters within the apsA lineage I, which was proposed to contain the ‘authentic’apsA gene loci not affected by lateral gene transfer (Meyer & Kuever, 2007b). A second group of symbiotic and free-living sulfur-oxidizing bacteria was found to possess apsA genes showing close affiliation to sulfate-reducing bacteria. These sulfur-oxidizing Gamma- and Betaproteobacteria were recently placed into apsA lineage II, where acquisition of apsA genes was hypothesized to be due to lateral gene transfer from sulfate-reducing bacteria (Meyer & Kuever, 2007b). Despite these lateral gene transfer events, apsA phylogeny is still instrumental in distinguishing between bacteria involved in the oxidative vs. the reductive sulfur metabolism.
For dsrAB, encoding dsr α and β subunit, two evenly abundant distinct RFLP patterns were obtained by screening 30 clones from the Caribbean symbiont. For each pattern, five clones were randomly chosen, sequenced and two distinct clone sequences TZdsrAB1 and TZdsrAB2 were obtained. From the Mediterranean symbiont, two clones were retrieved, sequenced and found to be identical. The dsrA genes from both Z. niveum symbionts were 1080 nucleotides (360 amino acids) in length, separated by a spacer region of 97 nucleotides from the dsrB gene of 731 nucleotides (242 amino acids) length, resulting in a total of 1908 nucleotides (602 amino acids) for dsrAB (Table 2). The Caribbean sequences TZdsrAB1/TZdsrAB2 showed 90.3% nucleic acid similarity and the deduced amino acid sequences 97.7% identity to each other, and 98.2/90.1% nucleic acid similarity as well as 99.4/97.1% amino acid sequence identity to the Mediterranean clone. Phylogenetic analysis revealed the sulfur-oxidizing Gammaproteobacteria A. vinosum as most similar relative, with 82.2–83.3% DsrAB amino acid identity to strain Twin Cays and strain Calvi. All applied phylogenetic treeing methods grouped the dsr sequences of the two Z. niveum ectosymbiont strains next to A. vinosum within a well-supported cluster of sulfur-oxidizing Gammaproteobacteria (Fig. 6). The latter were clearly separated from sulfur-oxidizing Alpha- and Betaproteobacteria, and Chlorobi, as well as from sulfate/sulfite-reducing microorganisms in all treeing methods applied (Fig. 6). In contrast to sulfate-/sulfite-reducing microorganisms (Zverlov et al., 2005), there is currently no hint that the distribution of dsrAB from sulfur oxidizers was affected by LGT (lateral gene transfer) among major bacterial taxa (phyla/classes). However, only a limited number of dsrAB sequences from cultivated sulfur oxidizers are available (Loy et al., 2008). Interestingly, two distinct dsrAB sequences were obtained from the Twin Cays sample. These two dsrAB types showed only moderate nucleic acid similarity (90.3%), but high amino acid identity (97.7%) to each other. So far only two microorganisms are known to harbor two dsrAB copies in their genomes. The two dsrAB sequences from the photoautotrophic sulfur-oxidizer Chlorobaculum tepidum show 95.5% nucleic acid similarity (94.8% amino acid identity). However, one copy contains an authentic frame shift within dsrB, suggesting that this copy is nonfunctional. The sulfite-reducing archaeon Pyrobaculum aerophilum possesses two dsrAB copies with only 74% nucleic acid similarity (70.6% amino acid identity). Nevertheless, whether both dsrAB sequences retrieved from the Caribbean sample derive from the same organism or from two closely related microorganisms remains to be determined. However, an organism's ability to cope with varying environmental conditions might be extended by the presence of multiple copies of a functional gene in one strain, considering that the different gene versions may code for functional enzymes with slightly different characteristics (Tchawa Yimga et al., 2003).
Our results show that the colonial ciliate Z. niveum is obligatorily covered by the single polymorphic bacterial phylotype Cand. Thiobios zoothamnicoli, whereby symbionts sampled in the Caribbean and the Mediterranean Sea show high 16S rRNA gene similarity, suggesting a differentiation on the strain level. Detection of sulfide was reported from both sampled Z. niveum habitats (Ott et al., 1998; Rinke et al., 2007), and in vivo experiments revealed that the Z. niveum symbiosis requires sulfide for its survival (Ott & Bright, 2004; Rinke et al., 2007). Consistent with these observations the detection of highly similar RuBisCO, APS reductase, and dsr-encoding genes in two samples from different oceanic sites provides the first molecular evidence that Cand. Thiobios zoothamnicoli is indeed a chemoautotrophic sulfur-oxidizing (thiotrophic) Gammaproteobacteria. Observations of membrane-bound vesicles, most likely containing elemental sulfur, by transmission electron microscopy (Bauer-Nebelsick et al., 1996a), the white color of the bacteria, an attribute to intracellular stored sulfur, and our confirmation of dsr-encoding genes support the hypothesis that Cand. Thiobios zoothamnicoli generates intracellularly stored sulfur vesicles as essential intermediates in the sulfur metabolism, as reported for A. vinosum (Pott & Dahl, 1998).
Because a variety of thiotrophic symbionts are known to form sulfur deposits, dsr could very well be one of the mayor pathways of sulfur-oxidizing symbiotic bacteria from shallow water to the deep sea. This concept is supported by available metagenome data, showing dsrAB genes for the Olavius algarvensis and the Calyptogena magnifica symbiont (Cand. Ruthia magnifica; Fig. 6).
In general, microbial symbioses with eukaryotic hosts are characterized by the physiological capabilities of the symbionts. Thus this study is a crucial step in illuminating the thiotrophic metabolism of two geographically separated Cand. Thiobios zoothamnicoli populations.
We would like to thank U. Kloiber, S. Katz, and J. Ott for their contribution during the field work. Christian Baranyi and Stephan Duller are acknowledged for excellent technical assistance. The present study has been supported by Austrian Science Foundation grants P16840-BO3 (M.B.), Y277-B03 (M.H.), P18836-B17 (A.L.) and the Caribbean Coral Reef Ecosystems Program (CCRE), Smithsonian Institution, supported in part by the Hunterdon Oceanographic Research Endowment (grant no. 838).