OUP user menu

Diversity of active chemolithoautotrophic prokaryotes in the sulfidic zone of a Black Sea pelagic redoxcline as determined by rRNA-based stable isotope probing

Sabine Glaubitz , Matthias Labrenz , Günter Jost , Klaus Jürgens
DOI: http://dx.doi.org/10.1111/j.1574-6941.2010.00944.x 32-41 First published online: 1 October 2010


Marine pelagic redoxclines are characterized by pronounced activities of chemolithoautotrophic microorganisms. As evidenced by the high dark CO2 fixation rates measured around the oxic–anoxic interface but also in the upper sulfidic zone, the accordant organisms participate in important biogeochemical transformations. Although Epsilonproteobacteria have been identified as an important chemoautotrophic group in these environments, detailed species-level information on the identity of actively involved prokaryotes is lacking. In the present study, active chemolithoautotrophic prokaryotic assemblages were identified in the sulfidic zone of a pelagic Black Sea redoxcline by applying rRNA-based stable isotope probing in combination with 16S rRNA gene single-strand conformation polymorphism analysis and 16S rRNA gene cloning. The results showed that a single epsilonproteobacterium, affiliated with the genus Sulfurimonas, and two different members of the gammaproteobacterial sulfur oxidizer (GSO) cluster were responsible for dark CO2 fixation activities in the upper sulfidic layer of the Black Sea redoxcline. Phylogenetically, these organisms were closely related to microorganisms, distributed worldwide, that are thought to be key players in denitrification and sulfide oxidation. Together, these findings emphasize the importance of chemolithoautotrophic members of the Sulfurimonas and GSO groups in the carbon, nitrogen, and sulfur cycles of oxic–anoxic pelagic transition zones.

  • Black Sea
  • chemolithoautotrophy
  • Sulfurimonas
  • GSO
  • redoxcline


Chemolithoautotrophic prokaryotes play an important ecological role in biogeochemical cycles of aquatic habitats. Molecular hydrogen and reduced inorganic compounds, such as nitrogen (NH4+, NO2), sulfur (e.g. H2S, S2O32−), and metal species (e.g. Fe2+, Mn2+), as well as carbon compounds (e.g. CO, CH4) can serve as electron donors for chemoautotrophic bacteria (Shively et al., 1998), whereas oxygen and nitrate mostly serve as electron acceptors. CO2 dark fixation has been determined in very different habitats, including redoxclines of anoxic marine basins (Sorokin, 1964; Jannasch et al., 1991; Taylor et al., 2001; Jost et al., 2008), marine oxygen-minimum zones (Ward et al., 1989), sulfidic caves (Engel et al., 2004), and hydrothermal vents (Karl et al., 1980; Wirsen et al., 1986).

A characteristic of chemolithoautotrophy in pelagic redoxclines of anoxic marine basins is the high fixation rate often observed below the chemocline, defined as the zone in which sulfide is first detected. This has been shown for the Black Sea (Jørgensen et al., 1991; Sorokin et al., 1995), the Cariaco Basin (Tuttle & Jannasch, 1973, 1979; Taylor et al., 2001), the Mariager Fjord (Zopfi et al., 2001), and the central Baltic Sea (Gocke, 1989; Labrenz et al., 2005; Jost et al., 2008). In most cases, the microorganisms responsible for the chemolithoautotrophic activity were identified indirectly; however, more recently, catalyzed reporter deposition (CARD)-FISH combined with microautoradiography (MICRO-CARD-FISH) was successfully used to directly identify Epsilonproteobacteria as a quantitatively important chemoautotrophic group in sulfidic zones of the central Baltic Sea and Black Sea (Grote et al., 2008). In another study, rRNA-based stable isotope probing (rRNA-SIP) analyses demonstrated the contribution of Gammaproteobacteria to dark CO2 fixation in a redoxcline of the Baltic Sea (Glaubitz et al., 2009).

The aim of this study was to gain more comprehensive information about the diversity of chemolithoautotrophic assemblages in the sulfidic zone of the world's largest anoxic basin, the Black Sea. Using the incubation-dependent rRNA-SIP method, we identified a single organism of the epsilonproteobacterial Sulfurimonas subgroup but also members of the gammaproteobacterial sulfur oxidizer (GSO) cluster (Lavik et al., 2009) as the drivers of chemoautotrophic production. Close phylogenetic relatives of these Proteobacteria are known to inhabit marine and brackish redoxcline systems worldwide (Madrid et al., 2001; Nakagawa et al., 2005; Campbell et al., 2006; Stevens & Ulloa, 2008; Glaubitz et al., 2009; Lavik et al., 2009), providing further evidence of their important role in biogeochemical cycles of oxygen-deficient pelagic systems.

Materials and methods

Sampling and in situ dark CO2 fixation

Water samples were obtained in May 2007 during cruise M72 of the research vessel ‘R/V Meteor’ and were collected in the Black Sea (station 7; 43°59.98′N, 32°01.08′E) using free-flow bottles (Hydrobios) attached to a conductivity, temperature, and depth rosette (SBE 911; Seabird). Oxygen and hydrogen sulfide concentrations were determined as described elsewhere (Grashoff et al., 1983). CO2 fixation rates throughout the redoxcline were determined according to the method of Steemann Nielsen (1952) as described in detail in Glaubitz et al. (2009) with SDs<10%. Flow cytometric prokaryotic cell counting was performed as described previously (Jost et al., 2008) with SDs of <5%.

13C-incorporation assay

Incorporation assays were performed for three independent replicates. [13C]-bicarbonate (Eurisotop, Germany) or [12C]-bicarbonate (Merck) was added anoxically to 2 L of water samples taken from a depth of 145 m to a final concentration of 4 mmol L−1, which was nearly equal to the in situ concentration of unlabeled bicarbonate. The treated water samples were incubated in the dark at the in situ temperature and under anoxic conditions. After 72 h, the water was filtered through a Durapore filter (0.22 μm pore size), which was eventually shock-frozen.

Nucleic acid extraction, isopycnic centrifugation, and 16S rRNA gene quantification

DNA-free total RNA was extracted from the frozen samples as described previously (Glaubitz et al., 2009), with minor modifications. After DNase I digestion of coprecipitated DNA, the RNA was purified with another phenol extraction (citrate-buffered phenol : chloroform : isoamylalcohol, 125 : 24 : 1, pH 4.2, Fisher Scientific). The aqueous phase was washed once with chloroform : isoamylalcohol (24 : 1) and precipitated by adding 2 volumes of absolute ethanol and 0.5 volumes of 7.5 mol ammonium acetate L−1. Before gradient preparation, the washed and dissolved RNA was quantified in an ND-1000 spectrometer (NanoDrop Technologies).

Gradient preparation, isopycnic centrifugation, and gradient fractionation were performed as described before (Lueders et al., 2004), with minor modifications as described in Glaubitz et al. (2009). The samples were centrifuged in 5.1-mL Quickseal polyallomer tubes in a VTi 65.2 vertical rotor using an Ultima L-100 XP centrifuge (all Beckman Coulter). Centrifugation was carried out at 20 °C for >65 h at 35 000 r.p.m. (105 000 gav). Quantitative reverse-transcription-PCR (RT-qPCR) of density-resolved RNA with the domain-specific primer sets Ba519f/Ba907r (Stubner, 2002) for Bacteria and Ar109f/Ar912rt (Lueders & Friedrich, 2003) for Archaea (Supporting Information, Table S1) was carried out using a one-step RT-PCR kit (Access Kit, Promega) as described previously (Glaubitz et al., 2009).

16S rRNA gene fingerprinting analyses and cloning

The procedures of single-strand conformation polymorphism (SSCP) fingerprinting and densitometric analyses of digitalized SSCP gels were performed as described previously (Glaubitz et al., 2009). Clone libraries were established from PCR products generated from one [12C]-bicarbonate and one [13C]-bicarbonate gradient. The fractions used for this experiment are given in Table 1. RT-PCR was performed using the Access Kit (Promega). Nearly full-length 16S rRNA gene RT-PCR products were generated using the primers 27f and 1492r (Lane, 1991) (Table S1). The PCR mixture (15 μL) contained 1 × PCR buffer, 1.65 mmol MgSO4 L−1, 200 μmol of each dNTP L−1, 0.2 μg bovine serum albumin μL−1 (Fermentas), 0.7 μmol of each primer L−1, and 1.5 U of AMV reverse transcriptase and Tfl DNA-polymerase each (Promega). After a reverse transcription step carried out at 45 °C for 45 min, PCR was started with an initial denaturation step at 95 °C for 5 min, followed by 30 cycles of denaturation at 94 °C for 1 min, annealing at 45 °C for 1 min, and elongation at 68 °C for 2 min. The terminal elongation step was carried out at 68 °C for 10 min. The PCR products were directly cloned into the pSC-A vector of the Strataclone system (Stratagene) and subsequently transformed into SoloPack competent cells (Stratagene) according to the manufacturer's instructions. Positive clones were selected by blue–white screening. For restriction fragment length polymorphism (RFLP) analyses, the inserts of white clones were amplified using the vector-specific primer combination T7 and T3 (Table S1). PCR mixtures (20 μL) contained 1 × PCR buffer, including magnesium, 62.5 μmol of each dNTP L−1, 0.25 μmol of each primer L−1, and 0.75 U of Taq polymerase (5Prime). Each of the clones with a specific insert was incubated with 3 U each of the restriction enzymes Hin6I and MspI (Fermentas) in 1 × Tango buffer for 150 min at 37 °C, followed by a final incubation at 65 °C for 20 min. Digested PCR products were analyzed by agarose gel electrophoresis (3% agarose in 1 × TAE) and restriction patterns were compared visually. Gene clones with identical patterns were grouped into one operational taxonomic unit (OTU).

View this table:

Descriptive and statistical parameters of all generated clone libraries

Density (g cm−3)1.7821.7901.7991.816
Clones investigated311224135103
Different RLFP patterns43432515
Coverage (%)95.892.992.693.2
Shannon–Wiener index2.863.182.181.78
Abundances (%)
Other GSO9.
Sum of all GSO57.258.073.385.4
  • The Shannon–Wiener and Chao1 indexes were calculated with the fastgroupii online tool. (http://biome.sdsu.edu/fastgroup/cal_tools.htm). C-A, clone library of the copy-number maximum; C-B, clone library of the heavier fraction.

The coverage of each of the clone libraries was calculated as follows: C=1−(n/N) × 100, where N is the total number of clones analyzed, and n the number of OTUs as identified by RFLP analyses (Rappé, 1997). The Shannon–Wiener diversity index (H′) and the Chao1 index were calculated using the fastgroup online tool (http://biome.sdsu.edu/fastgroup/cal_tools.htm) as described in Yu et al. (2006). The maximal diversity was calculated from the number of different OTUs in each clone library: Hmax=−[ln(1/s)], with s denoting the total number of species. The Shannon– Wiener diversity index (H′) and the maximum diversity index Hmax were used to calculate the evenness (E) of each clone library, with E=H′/Hmax. To clarify the differences between two density fractions of each treatment, the enrichment factor was calculated as follows: the relative abundance of each group in the clone library of a heavy fraction was divided by the relative abundance of the same group in the clone library of the fraction in which the maximal copy number occurred. Values <1 indicated a depletion and values >1 indicated an enrichment of the accordant organisms in heavier fractions within one density gradient. Enrichment factors=1 indicated that no changes were detectable.

Sequence analyses

Excised bands as well as representative clones were reamplified using the primer systems com1f/com2rpH or T7/T3 for SSCP and clones, respectively. PCR products were purified using the Nucleospin II kit (Macherey & Nagel). Sequencing was performed by Qiagen (Hilden, Germany) using the primer systems com1f/com2rpH (SSCP bands) and 27f/1492r (clones). Sequence reads were quality checked using the program seqman (dnastar). Preliminary estimation of phylogenetic affiliations of the 16S rRNA gene sequences was performed by blast (Altschul et al., 1997).

Phylogenetic analyses

The arb software package was used for the alignment and phylogenetic analyses of the obtained sequences (Ludwig et al., 2004). Sequences for analysis were reduced to unambiguously alignable positions using group-specific filters. For phylogenetic analyses, three different trees were calculated using the algorithms neighbor-joining, parsimony, and maximum-likelihood (phyml) based on nearly full-length 16S rRNA gene sequences (>1350 bp). Shorter sequences were inserted into the reconstructed tree without changing the topology. For neighbor-joining, the Jukes–Cantor correction was applied.


Physicochemical structure and prokaryotic cell abundances

The chemocline was located at a depth of 135 m. H2S increased steadily with depth, reaching 9.9 μmol L−1 at 160 m (Fig. 1a). At 126 m, the oxygen concentration was below the detection limit, resulting in a putative anoxic but nonsulfidic zone extending over about 9 m. Below 141 m, no nitrate could be detected (G. Lavik, pers. commun.). Two distinct dark CO2 fixation peaks were detected at the sampling station. At the first peak, located at a depth of 131 m, a fixation rate of 2.0 μmol L−1 day−1 was measured, while at the second peak, at 151 m, the fixation rate was 0.9 μmol L−1 day−1 (Fig. 1b). Total prokaryotic cell abundances ranged from 3.2 × 105 to 5.0 × 105 cells mL−1 (Fig. 1b).


Depth profile of station 7 (43°59.98′N, 32°01.08′E) in the Black Sea. (a) H2S and O2 concentrations. The line marks the chemocline. (b) Total cell counts as determined by flow cytometry (FACS), and dark CO2 fixation. The arrow marks the sampling depth for SIP analyses.

rRNA separation and quantification

The rRNA-SIP analyses were performed using three independent replicates obtained from a depth of 145 m and incubated with 4 mmol bicarbonate L−1 for 72 h. RT-qPCR specific for bacterial 16S rRNA gene analyses of three independent replicates were carried out after isopycnic centrifugation and subsequent gradient fractionation and revealed a density shift of the copy-number maximum of 0.0096 g cm−3 between the 12C and 13C gradients (P<0.005) (Fig. 2). More than 99% of the rRNA copies were distributed between 1.766 and 1.794 g cm−3 in the 12C gradients, and between 1.779 and 1.806 in the 13C gradients. This small difference indicates a partial labeling of the [13C]-bicarbonate-incubated bacterial community. The distribution of the [12C]-bicarbonate amendment was very similar to the bacterial copy numbers of the in situ community (data not shown).


Quantitative distribution of rRNA in CsTFA density gradients after 72 h of incubation with [12C]- and [13C]-bicarbonate. Domain-specific template distribution within gradient fractions was quantified by RT-qPCR. Data are given as the dimensionless normalization, to allow comparisons between gradients. (a) Gradients from which SSCP fingerprints were generated. (b) Replicate gradient preparation. Fractions from which the clone libraries were established are marked by an asterisk (13C) or a cross (12C).

The absolute archaeal copy numbers were about four orders of magnitude lower than the bacterial copy numbers in the same samples and undetectable after isopycnic centrifugation (data not shown). Hence, the contribution of Archaea to chemolithoautotrophic production was assumed to be negligible and was not analyzed further.

Identification of active chemoautotrophs by 16S rRNA gene SSCP fingerprinting

16S rRNA gene SSCP fingerprinting of the two density gradients revealed that three different bands, named A12-1, A12-2, and A12-3, were affected by the dark [13C]-bicarbonate incubation (Fig. 3a and b). Based on comparisons of their relative intensities, the three bands showed a clear enhancement but also a shift toward heavier fractions in the 13C gradient (Fig. 3c and d). These bands were also detected in a gradient prepared from an unamended in situ water sample, but showed minor changes in relative intensity throughout the density gradient (Fig. S1). The sequences determined for bands A12-1 and A12-3 were phylogenetically affiliated with the GSO group, and the one in band A12-2 affiliated with the epsilonproteobacterial Sulfurimonas cluster (Fig. 4). Thus, based on SSCP analyses, the identified organisms were embedded in putative chemoautotrophic clusters within the Proteobacteria (Fig. 4). By contrast, the relative intensity of a putative heterotrophic Pseudoalteromonas band (A12-4) showed a clear decrease toward heavier fractions in the 13C gradient (Fig. 3c and d).


SSCP fingerprints of density-resolved bacterial SSU-RNA templates from CsTFA density-gradient fractions of the (a) [12C]- and (b) [13C]-bicarbonate incubation experiments. The lanes of each gel represent fractions with increasing buoyant densities from left to right. Bands identified by sequencing are marked. (c) [12C]- and (d) [13C]-bicarbonate-based results of the densitometric analyses of selected bands: relative band intensities (in %) are plotted against the respective buoyant densities (in g cm−3). Open circles represent the distribution of a band affiliated with a Pseudoalteromonas sp. (A12-4) across the density gradient. Closed symbols mark sequences belonging to potentially chemoautotrophic organisms (see text).


Unrooted maximum-likelihood tree of sequences generated in this study that are affiliated phylogenetically with the Epsilonproteobacteria and sulfur-oxidizing Gammaproteobacteria. Bold letters denote sequences generated in this study. ●, Validation of subtree by neighbor-joining and parsimony; ◻, validation of subtree by parsimony; ○, validation of subtree by neighbor-joining; *, identified as chemolithoautotroph in this study; †, identified as chemolithoautotroph in pelagic redoxclines in previous studies. 1Detected in a Cariaco Basin redoxcline (Madrid et al., 2001). 2Detected in a Black Sea redoxcline (Vetriani et al., 2003). 3Detected in a oxygen minimum zone of the African shelf (Lavik et al., 2009). 4Detected in a oxygen-minimum zone of the eastern tropical South Pacific (Stevens & Ulloa, 2008). 5Detected in a central Baltic Sea redoxcline (Brettar et al., 2006; Grote et al., 2007; Glaubitz et al., 2009). 6Based on metagenome analyses from a Saanich Inlet fjord oxygen-minimum zone (Walsh et al., 2009). Scale bar=10 substitutions per 100 nucleotides.

Identification of active chemoautotrophs by 16S rRNA gene cloning

To investigate the bacterial community structure in greater detail, 16S rRNA gene clone libraries were established from the copy-number maxima (12C-A and 13C-A) and from heavier fractions (12C-B and 13C-B) (Fig. 2b). According to the descriptive statistical parameters of these clone libraries, coverage was >90%, indicating that the major part of the diversity was detected by this approach (Table 1). In general, both the Shannon–Wiener index and the evenness factor demonstrated that the 16S rRNA gene clone libraries of the copy-number maximum were more diverse and more even than those of the heavier fractions (Table 1). All 16S rRNA gene clone libraries included Gammaproteobacteria of the Alteromonadales, Pseudomonadales, Chromatiales, Oceanospirillales, Methylococcales, and representatives of the GSO group, as well as members of the Alphaproteobacteria and the Planctomycetales (for a detailed list of all sequenced clones, see Table S2).

The 12C and 13C 16S rRNA gene clone libraries of the copy-number maxima were dominated by several members of the GSO group (Table 1), with BS-GSO1 and BS-GSO2 being most abundant. Phylogenetically, the gammaproteobacterial sequences deriving from clone and SSCP analyses were very similar (Fig. 4). A comparison of the 12C and 13C clone libraries from the heavier fractions (Fig. 2b) showed that the BS-GSO1 and especially the BS-GSO2 cluster were more abundant in the 13C clone library (Fig. 5). About 99% of all GSO sequences in this library were affiliated with one of these clusters. Other GSO sequences were hardly detectable in the 13C-fraction, indicating that these organisms did not incorporate heavy bicarbonate. Unexpectedly, epsilonproteobacterial clones could not be generated by any of the 16S rRNA gene clone libraries.


Relative changes in the contribution of selected clones to the composition of the clone libraries. The dimensionless enrichment factor was calculated by dividing the relative abundance of a clone in the clone library of heavier fractions by the relative abundance of the respective clones in the clone library of the copy-number maximum. Values <1 indicate depletion and values >1 indicate enrichment of the particular group in the heavier fractions.


The aim of this study was to directly link dark CO2 fixation with the identity of the responsible microbial assemblages in a sulfidic CO2 fixation maximum of a Black Sea pelagic redoxcline. Using rRNA-SIP and SSCP fingerprinting, we were able to attribute chemolithoautotrophic production to one epsilonproteobacterium affiliated to the genus Sulfurimonas and two populations related to the GSO cluster, with clone libraries confirming the role of GSO for the investigated pelagic Black Sea redoxcline.

Chemosynthesis in sulfidic areas of Black Sea pelagic redoxclines is a well-known phenomenon (Sorokin, 1964; Jørgensen et al., 1991). In recent years, several potentially chemolithoautotrophic bacterial organisms or groups have been identified in the Black Sea. For instance, using terminal RFLP analyses in combination with 16S rRNA gene cloning, Vetriani et al. (2003) identified Epsilon- as well as Gammaproteobacteria below the chemocline. Phylogenetically, these bacteria are closely related to the chemolithoautotrophic epsilonproteobacterium SSCP band A12-2 and the BS-GSO1 cluster of the present study (Fig. 4.). Thus, these proteobacterial groups could be widely involved in chemolithoautotrophy in the Black Sea.

Epsilon- and Gammaproteobacteria have been quantified in other studies using class-specific gene probes, and cellular abundances of approximately 20% and 6%, respectively, were determined for zones below the chemocline in the Black Sea (Lin et al., 2006; Wakeham et al., 2007). Using CARD-FISH, Grote et al. (2008) detected epsilonproteobacterial abundances of 11–35% at comparable physicochemical depths. By MICRO-CARD-FISH, these authors demonstrated in situ the important role of Epsilonproteobacteria in chemolithoautotrophy, with 24–100% of all chemolithoautotrophs phylogenetically belonging to the Epsilonproteobacteria. However, specific organisms were not identifiable by these analyses and the rRNA-SIP approach of the present study indicated that the diversity of chemolithoautotrophic Epsilonproteobacteria in Black Sea redoxclines can be low, and potentially reduced to only one representative related to the genus Sulfurimonas.

A similar phenomenon was observed in a central Baltic Sea redoxcline, where the cellular abundance of the specific Sulfurimonas subcluster GD17 may account for 15% (Grote et al., 2007) or occasionally even as high as 30% (unpublished data) of the total cell count. Subcluster GD17 is closely related to Sulfurimonas denitrificans and has been proposed as the dominant player within the sulfur and nitrogen cycle of the central Baltic Sea (Brettar et al., 2006; Grote et al., 2008). Thus, the abundance of a single chemolithoautotrophic epsilonproteobacterial taxon may well be characteristic of pelagic redoxclines of the Black Sea and the central Baltic Sea. Currently, however, this conclusion is based only on random analyses and needs further investigation; nonetheless, it is certainly possible that the presence of unusual or even extreme nutrient or redox conditions in pelagic redoxclines favors specific and well-adapted members of the genus Sulfurimonas.

In the above-mentioned MICRO-CARD-FISH study, only Epsilon- rather than Epsilon- and Gammaproteobacteria-specific gene probes were used to quantify chemolithoautotrophic assemblages in sulfidic zones of Black Sea redoxclines (Grote et al., 2008). In this study and in similar pelagic redoxclines, besides Epsilonproteobacteria, no chemolithoautotrophs other than Gammaproteobacteria were detected (Glaubitz et al., 2009). By simply subtracting chemolithoautotrophic epsilonproteobacterial cell numbers from the total number of dark CO2 fixing cells in the study by Grote et al. (2008), the contribution of Gammaproteobacteria to dark CO2 fixation can be estimated, resulting in proportions usually well below 20% on the cellular level. This estimate is in accordance with the FISH data of Lin et al. (2006), who determined total gammaproteobacterial abundances of 2–6% of the total prokaryotic community in the sulfidic zone of a Black Sea redoxcline. Overall, in the present study, GSO members were found to be more diverse than the Epsilonproteobacteria, but only two closely related representatives showed chemolithoautotrophic activity (Figs 4 and 5). One organism, identified by 16S rRNA gene SSCP (band E8-1, see Fig. 3a and b) and cloning (OTU 18, see Table S2) and nearly identical to gammaproteobacterium band A, previously shown to be chemolithoautotrophic in the Baltic Sea (Glaubitz et al., 2009), did not incorporate [13C]-bicarbonate in the Black Sea sample. By SSCP and cloning, this organism was determined to be abundant, but its physiological and ecological role in the Black Sea redoxcline is as yet unclear. The coexistence of putative chemolithoautotrophic Epsilonproteobacteria and members of the GSO cluster is presumably reasonable in the different adaptation strategies. The low diversity of epsilonproteobacterial autotrophs and the broad depth distribution suggests a versatile and generalist lifestyle as hypothesized by Grote et al. (2007, 2008), whereas the considerably enhanced diversity, but reduced activity of GSO might indicate more specialized ecological niches at the species level.

Despite a potentially lower abundance of chemolithoautotrophic GSO in Black Sea redoxclines, GSO members are in general widely distributed and have been detected in similar habitats as Epsilonproteobacteria, such as hydrothermal vents (Ruby et al., 1981), oxygen minimum zones (Stevens & Ulloa, 2008; Walsh et al., 2009), and pelagic redoxclines of the Cariaco Basin (Madrid et al., 2001) and central Baltic Sea (Labrenz et al., 2007). Recently, members of this cluster, which are phylogenetically closely related to the cluster BS-GSO1, were shown to be involved in sulfide oxidation in an oxygen-minimum zone in the Namibia upwelling system (Lavik et al., 2009). There, the abundances of these organisms coincided with the occurrence of sulfide, the formation of colloidal sulfur, and denitrification, which emphasizes the impact of these bacteria on important biogeochemical cycles.

Indirect evidence for the chemolithoautotrophic growth of symbiotic GSO has accumulated from genome and proteome analyses, as well as from the δ13C values of host organisms, and from enzyme activities (Karl et al., 1980; Felbeck, 1981; Markert et al., 2007; Nakagawa & Takai, 2008; Robidart et al., 2008; Walsh et al., 2009). With the experimental setup in the present study, we were not able to distinguish between free-living and symbiotic lifestyles of the identified GSO, but most metazoans that harbor chemolithoautotrophic endo- or ectosymbionts live in the benthos (reviewed by Stewart et al., 2005) and the probability of occurrence of these hosts in the pelagic, sulfidic habitat is rather low. A similar approach was used to demonstrate the transfer of chemolithoautotrophically fixed 13C-bicarbonate to Euplotes sp. (Glaubitz et al., 2009), but a potential symbiotic relationship, conceivable with unicellular free-living eukaryotes, has not been demonstrated thus far.

Methodological considerations

One major point of criticism concerning isotope-labeling experiments is related to secondary effects, cross-feeding, and overlabeling of the microbial community (reviewed in Neufeld et al., 2007), all of which have to be taken into consideration in correctly interpreting the obtained results. In our experiments, the probability of the 13C tracer signal being transferred from lysed autotrophic bacteria to heterotrophic prokaryotes that feed on dissolved or particulate organic matter was considered to be low. Compared with a previously described signal transfer from chemolithoautotrophs to the potential grazer Euplotes sp. (Glaubitz et al., 2009), carbon-source fluxes between bacteria should be negligible due to dilution effects and fractionation within the individual biosynthetic processes. The distribution of 12C-labeled 16S rRNA gene copies in the density gradients of our study was comparable to that in previously described SIP experiments, whereas the banding of the 13C-labeled rRNA species was indicative of 13C isotope enrichment, but to a lesser extent than expected for completely labeled rRNA species, which usually results in a shift in the copy-number maximum of approximately 0.04 g CsTFAcm−3 (Lueders et al., 2004). The enhanced abundance of the BS-GSO1 cluster in the 12C-B library can be explained by the distribution of the specific 16S rRNA gene copies in the density gradient. The maximal relative intensity of this organism in the 12C-SSCP gel was detected at 1.807 g cm−3, which is heavier than the fraction the clone library originates from. Using 16S rRNA gene clone libraries and SSCP fingerprints, we also detected typical heterotrophs, such as Pseudoalteromonas sp. and Alteromonas sp. (Table S2), which were less abundant in the heavier fractions of the [13C]-bicarbonate-labeled samples (Fig. 3c and d and Fig. 5), implying that in our SIP experiments secondary effects such as cross-feeding can be neglected. Thus, we conclude that the bulk of the observed density shifts were due to the incorporation of [13C]-bicarbonate into chemolithoautotrophs and that other factors, for example, anaplerotic reactions and cross-feeding, accounted for an insignificant portion of the observed effects. The analysis of band intensities and clone frequencies has to be considered as semi-quantitative due to PCR and cloning biases, differential amount of rRNA operons and copies per cell. The dominance of the GSO and the Epsilonproteobacteria in the SSCP gel is presumably due to methodological biases, and thus potentially overestimated. Comparing fingerprints of the augmented samples with that from the negative control, a stimulation of the chemoautotrophs was visible. However, whether it was due to the bicarbonate addition or to potentially introduced electron acceptors cannot be elucidated yet.

Despite the laborious procedure of SIP, we analyzed three replicate density gradients by qPCR. This effort provided evidence for the practicability and reproducibility of this approach, but no additional information was obtained. The shift of the copy-number maximum toward higher buoyant densities was clear in all replicates. For powerful statistical analyses, more gradients have to be investigated in detail, but this was not practicable for this study. Hence, the SIP method should be considered as a purely qualitative, nonquantitative approach.

Although we detected an abundant SSCP band affiliated to the Sulfurimonas cluster, the accordant 16S rRNA gene sequence was absent in our clone libraries. This was unexpected because the primers we used (27f/1492r) are supposed to be universal (Lane, 1991), and in several habitats, Sulfurimonas spp. have already been successfully detected using this primer combination (Engel et al., 2003; Nakagawa et al., 2005; Grote et al., 2007). However, in a previous Black Sea study, the reverse primer 1517R (Vetriani et al., 2003) was used, which holds two different bases at the 5′-position as compared with 1492r. This could explain why we were unable to detect Epsilonproteobacteria based on cloning analogously.

In conclusion, this study provides evidence that chemolithoautotrophic assemblages in marine pelagic redoxclines consist of phylogenetically very similar Epsilon- and Gammaproteobacteria. This is an indication that these organisms are well-adapted to comparable physicogeochemical conditions within such habitats, underlining the importance of Proteobacteria in carbon, nitrogen, and sulfur cycles.


Nucleotide sequence accession numbers: 16S rRNA gene sequences of the detected organisms were deposited in GenBank database under the accession numbers GU108512GU108572.

Supporting Information

Additional Supporting Information may be found in the online version of this article:

Fig. S1. SSCP fingerprint of density-resolved 16S rRNA templates of the in situ community.

Table S1. PCR primers used in this study.

Table S2. Summary of all sequenced clones in all clone libraries.

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.


The authors are very grateful to the captain and crew of the ‘R/V Meteor’ for their excellent support during the cruise. Chemical analyses during fieldwork were performed by Gaute Lavik and Evelyn Marschall. The excellent technical assistance of Christian Meeske, Katja Becker, Bärbel Buuk, and Annett Grüttmüller during the analysis of the samples at the IOW is greatly appreciated. We are indebted to Jana Grote and Claudia Wylezich for helpful discussions. This work was funded by the DFG grants LA 1466/4-1, 4-2 to M.L.


  • Editor: Alfons Stams


View Abstract