Correspondence: Marco J.L. Coolen, Marine Chemistry and Geochemistry Department, Woods Hole Oceanographic Institution, 360 Woods Hole Road, Woods Hole, MA 02543, USA. Tel.: +1 508 289 2931; fax: +1 508 457 2164; e-mail: firstname.lastname@example.org
Recent DNA-based phylogenetic studies have reported high eukaryotal diversities in a wide range of settings including samples obtained from anoxic environments. However, parallel RNA-based surveys are required in order to verify whether the species detected are in fact metabolically active in such extreme environments. The Black Sea is the World's largest anoxic basin but remains undersampled with respect to molecular eukaryotic diversity studies. Here, we report the distribution of active eukaryotes (18S rRNA-based survey) along a vertical nutrient and redox gradient in the water column and surface sediments of the Black Sea. A wide variety of eukaryotes were active in suboxic deep waters. Notably, certain species were active but escaped detection during a parallel 18S rDNA survey. The 18S rDNA survey from surface sediments yielded taxa of pelagic origin but none of these were identified from the water column at the time of sampling. Our data also indicate that gene transcripts do not always provide unequivocal proof that active microorganisms are indigenous to a specific position in an environmental gradient, because certain zoo- and phytoplankton species were still viable with detectable 18S rRNA in up to 300-year-old sulfidic sediments that underlie ∼830 m of sulfidic waters.
The identification of active microbial populations within environmental samples can be addressed by targeting SSU rRNA gene transcripts directly instead of DNA. This approach has mainly been used to identify active prokaryotes in environmental samples (MacGregor et al., 2001; Norris et al., 2002; Mills et al., 2004) and only recently to detect active microbial eukaryotes (Coyne & Cary, 2005; Stoeck et al., 2007; Alexander et al., 2009). rRNA is an essential structural component of the ribosome, which is the organelle responsible for protein synthesis in all prokaryotes and eukaryotes. The number of cellular ribosomes, and thus also the rRNA content, increases with growth rate and decreases with starvation (Wagner, 1994; Muttray & Mohn, 1998; Buckley & Szmant, 2004; Chicharo & Chicharo, 2008; Hawkins et al., 2008). Therefore, rRNA extracted from environmental samples serves as a phylogenetic marker for the identification and relative abundance of metabolically active members within complex microbial communities. Stoeck et al. (2007) constructed rRNA- and rDNA-based clone libraries derived from an anoxic water sample from the Framvaren Fjord (Norway) to compare the diversity and relative abundance of active vs. inactive marine microeukaryotes. This study revealed 84 unique phylotypes of which 27% occurred in both libraries, 25% occurred exclusively in the rRNA library and 48% occurred exclusively in the rDNA library. Furthermore, phylotypes of phototrophic dinoflagellates, uncultured marine alveolates group I, and various parasites were exclusively detected in the rDNA library and represented nonindigenous members of the anoxic microeukaryote community below the sulfidic chemocline (Stoeck et al., 2007). Even though the analyzed clone library was extensive (600 clones) and revealed many unique phylotypes, only one depth was analyzed in the above study. In addition, although the community structure below the chemocline in stratified waters is likely to remain stable as long as the stratification is not disturbed, pelagic samples provide only a snapshot of photic zone-derived cells in transit to the sediment (Coolen et al., 2007). Benthic rRNA libraries in these settings would provide a more complete picture of the capacity of organisms stemming from the oxygenated part of the photic zone to remain active or viable in suboxic or fully anoxic environments for substantial periods of time. For example, another recent study has shown that viable dinoflagellate cysts in anoxic Delaware's Inland Bays surface sediments could be identified using PCR-based approaches from 18S rRNA libraries and even from much more labile mRNA libraries (Coyne & Cary, 2005). This suggests that such gene transcripts could remain detectable in resting stages of planktonic species not indigenous to anoxic water layers for substantial periods of time.
The Black Sea, the world's largest permanently stratified basin, is devoid of oxygen and contains abundant sulfide from a depth of about 100 m to the seafloor at 2200 m. A 20–30-m-deep suboxic layer, depleted in both O2 and sulfide, overlies the sulfide zone (Jørgensen et al., 1991). This stable, vertically expanded redox gradient provides an opportunity to study the distribution of microbial communities within different chemical environments, but thus far remains undersampled with respect to eukaryotic diversity. In the present study, we performed a high spatial resolution analysis of the eukaryotic diversity (18S rDNA survey) and sought to determine whether these organisms were metabolically active (18S rRNA-based library) along a vertical gradient of O2, H2S, and nutrients in the western Black Sea. In addition, we performed a parallel survey on underlying surface sediments to identify active benthic eukaryotes and, more importantly, to establish whether eukaryotes derived from the photic zones remain viable in this highly sulfidic depositional setting.
Materials and methods
Water column chemistry and physical conditions
A Conductivity-Temperature-Depth profiler (CTD) was deployed to measure conductivity, temperature, salinity, fluorescence, dissolved oxygen (SBE-43 oxygen sensor; Sea-Bird Electronics, Bellevue, WA) and to collect water to measure N-species (nitrite, nitrate, ammonia), phosphate, and sulfide. This effort revealed the exact depths of the chlorophyll maximum, the upper and lower nitrite peaks, the suboxic zone with <0.2 mL L−1 of oxygen, as well as the sulfidic chemocline. This information was used to select the depths for the collection of particulate organic matter (POM) for the molecular survey.
Nutrients (nitrite, nitrate, ammonia, and phosphate) were measured spectrophotometrically according to Grasshoff et al. (1983). Hydrogen sulfide was measured photometrically according to Cline (1969).
Water samples were collected from 19 depths in the stratified Black Sea (42°46.569″N : 28°40.647″E) aboard the R/V Akademik (Institute of Oceanology, Bulgarian Academy of Sciences; IO-BAS) using a SBE911 plus CTD (Sea-Bird Electronics) equipped with twelve 5-L Niskin bottles. These depths included the chlorophyll maximum at 22, 40, 60–115 m (5-m intervals along the oxygenated mixolimnion), 120–135 m (5-m intervals within the suboxic zone with <0.2 mL L−1 oxygen), and within the sulfidic chemocline (150 and 160 m). We took precautions to prevent contact of the oxygen-depleted and sulfidic waters with the atmosphere by keeping the water inside the Niskin bottle until cells in 3–5 L of water were collected onto 142-mm, 0.2-μm pore size, polycarbonate filters (Millipore, Billerica, MA). The water was directly filtered from the Niskin bottles via a closed in-line tubing system connected to the lower outlet of each Niskin bottle under gentle pressure (<50 mL min−1) provided by a peristaltic pump (Masterflex, Cole and Palmer, Vernon Hills, IL). The total water depth at this location was 971 m.
Fifty-centimeter-long undisturbed cores of laminated sediments (Fig. 1b) were obtained at the same location using a MC800 multicorer (Ocean Instruments, San Diego, CA). One subcore (Core MC06) was subsectioned in 2-cm intervals aboard ship and sections from the upper 8 cm of undisturbed laminated sediments, including the top fluffy layer, were used for subsequent extraction of nucleic acids as described below.
Simultaneous DNA and RNA extraction and cDNA synthesis
Directly after filtration, the filters were placed in a 15-mL sterile conical centrifuge tube with a mixture of 4-mL denaturing buffer (Hurt et al., 2001), 50 μL 2-mercaptoethanol, and 2 mL of nucleases-free low-binding zirconium beads with a diameter of 200 μm (OPS Diagnostics, Lebanon, NJ). Cells on the filter were mechanically disrupted to release the nucleic acids by bead beating for 5 min at maximum speed using the Vortex Genie 2® Vortex and adapter (MoBio, Carlsbad, CA). The presence of the denaturing buffer and 2-mercaptoethanol ensured instant inactivation of the nucleases present. Then the mixtures were immediately stored at −80 °C or colder until further analysis. At the Woods Hole Oceanographic Institution (WHOI), all steps involved in the extraction and purification of nucleic acids were performed inside a HEPA-filtered fume hood, which was UV-sterilized and cleaned with Eliminase (Decon Laboratories, King of Prussia, PA) to destroy nucleases before use. Additional precautions, such as using filter tips during pipetting, using only RNAse-free reagents and disposable plastic ware were also taken. Vinyl gloves were worn at all times and cleaned in between steps with Eliminase. Then, each deeply frozen filter/buffer/bead mixture was transferred to precombusted (225 °C for 8 h) and deeply frozen (−80 °C) mortars and ground with a cold pestle upon the thawing of the mixture. Circular movements during grinding were omitted to minimize shearing of the nucleic acids. The mixture was transferred into 50-mL-sterile conical centrifuge tubes and an equal volume (∼4 mL) of phenol : chloroform : isoamylalcohol (PCI), 25 : 24 : 1, pH 8 (Ambion, Austin, TX) was added. After vigorously vortexing, the mixture was placed on ice for 15 min to ensure the complete disassociation of nucleoprotein complexes, and centrifuged at 4 °C for 15 min at 10 000 g. The aqueous phase was transferred to a new sterile tube and the nucleic acids were precipitated with 0.1 volume of 5 M NaCL (Ambion) and 1 volume of isopropanol. After centrifugation at 10 000 g for 20 min at room temperature, the air-dried pellet was dissolved in RNA Storage Solution (Ambion).
Total nucleic acids were extracted at 2-cm intervals from the top 8 cm of sediment. Directly after sectioning, 1.5 mL of sediment was transferred to 5-mL cyrovials and mixed with equal volume of denaturing buffer (Hurt et al., 2001), 20 μL of 2-mercaptoethanol, and 0.75 mL of zirconium beads. Cells in the sediment were then immediately mechanically disrupted to release the nucleic acids and to inactivate the nucleases by bead beating as described above. The cyrovials were stored in liquid nitrogen on board the R/V Akademik, transferred on dry ice to WHOI, and stored at −80 °C until further analysis. At WHOI, the frozen mixture underwent the same grinding procedure in sterile and deeply frozen mortars as described above, but followed an additional extraction step with 8 mL of extraction buffer as described by Hurt et al. (2001). Following centrifugation (10 min, 10 000 g), total nucleic acids were extracted from the supernatant using PCI and the additional steps described above. The quality of the RNA and DNA extracts was verified by agarose gel electrophoresis. The total nucleic acid extracts from the POM and sediments were used for the 18S rDNA survey.
Preparation of DNA-free RNA and cDNA synthesis
For the 18S rRNA survey, 0.1 volume of the extracted total nucleic acids was treated for 30 min at 37 °C with 1 U DNAse and 1 × buffer using the DNA-free™ kit (Ambion) followed by inactivation of the DNAse using 0.1 volume of the provided DNAse inactivation reagent. Before PCR, DNA-free RNA samples were randomly transcribed into cDNA using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA). The 20 μL reaction mixtures contained RNAse-free water, 1 U iScript reverse transcriptase, 1 × iScript Reaction Mix, 50 μg bovine serum albumin (BSA; Ambion), and 0.1 volume of DNA-free RNA sample. The reverse transcriptase (RT) reactions were performed in a Mastercycler (Eppendorf, Westbury, NY) and included incubation for 5 min at 25 °C, 30 min at 42 °C, and 5 min at 85 °C. The latter cDNA served as template for PCR to study the diversity of the metabolically active eukaryotes in the Black Sea.
PCR amplification of eukaryote 18S rDNA and reverse transcribed 18S rRNA
Each amplification reaction contained 0.25 mM of each deoxynucleotide (dNTP), 4.5 mM of MgCl2, 0.5 × (from 10 000 × concentrated stock) SYBR®Green I (Invitrogen, Carlsbad, CA), 50 μg of BSA (Ambion), 5 μL of 10 × PicoMaxx™ reaction buffer (Stratagene, LaJolla, CA), 2.5 U of DNA polymerase (PicoMaxx™ high fidelity PCR system; Stratagene), 0.2 μM of primers (Thermo-Fisher, Ulm, Germany), and 10 ng of DNA or cDNA template. Partial (560-bp long) eukaryote 18S rDNA (or reverse transcribed 18S rRNA) was PCR amplified using primer set Euk1A and Euk516r-GC (Set A) as described previously (Díez et al., 2001) The reaction mixtures were adjusted to a final volume of 50 μL with nucleic acids and nuclease-free water (Ambion). Each PCR amplification series included a reaction without a DNA template, which served as a control for contaminations during the pipetting of the reaction mixture components. For each cDNA sample, a parallel reaction with 0.2 vol% of digested RNA was subjected to PCR as a control for the presence of traces of incomplete digested DNA in the various RNA templates used for cDNA synthesis.
All reactions were performed in a Realplex quantitative PCR cycler (Eppendorf) and involved initial denaturing (4 min at 95 °C), followed by 38 cycles including denaturing (30 s at 94 °C), primer annealing (40 s at 64 °C), primer extension (60 s at 72 °C), and imaging of newly formed fluorescent (SYBR®Green I labeled) double-stranded DNA (80 °C for 20 s). All DNA and cDNA samples were then subjected again to real-time quantitative PCR but this time all reactions were stopped at the end of the exponential phase (i.e. after 25–32 cycles). This prevented overamplification of individual 18S rDNA fragments to allow semi-quantitative analysis by denaturing gradient gel electrophoresis (DGGE; Schäfer & Muyzer, 2001) as outlined below.
Phylogenetic analysis of sequenced DGGE bands
For the majority of the samples, 100 ng of PCR-amplified partial eukaryote 18S rRNA and RT-PCR amplified 18S rRNA (600 bp including the 40-bp-long GC clamp) was separated by DGGE (Schäfer & Muyzer, 2001). The polyacrylamide gels (6%, w/v) contained a denaturing gradient of 20–50% (with 100% denaturant equaling 7 M urea and 40% formamide). Gels were run for 15 h at 5 V cm−1 and 60 °C using a PhorU2 system (Ingeny, Leiden, the Netherlands). Afterwards, the gels were stained for 20 min by covering the gels twice with 10 mL of 1 × TAE buffer (pH 8.3), containing 2 μL undiluted SYBR®Gold (Invitrogen) followed by destaining for 60 min in 1 × TAE buffer (pH 8.3). In order to prevent DNA damage by UV, we used a Dark Reader (Clare Chemicals Research Inc., Dolores, CO), which uses visible light instead of UV in order to visualize the SYBR®Gold-stained DNA. Digital gel images were made using a Foto/Analyst® Express System (Fotodyne, Hartland, WI) and imagej software. totallab tl100 v2006 1D-gel analysis software (Nonlinear Dynamics, Durham, NC) was used to determine the pixel density and vertical position of each band. This information was used to determine the relative abundance of each band within a given sample and to identify the exact vertical position of each band in order to characterize unique vs. identical bands between samples (e.g. Díez et al., 2001). DGGE bands were sliced from the gel with a sterile scalpel and the DNA of each sliced gel fragment was eluted in 75 μL sterile 10 mM Tris-HCl (pH 8.0) by incubation for 48 h at 2 °C. One microliter of the eluted 18S rDNAs (approximately 107 copies) was reamplified using 25 cycles and the primer combinations listed above, but this time without the GC clamp. These amplicons served as DNA for the subsequent cycle sequencing reactions. PCR reagents were as described above but without the addition of SYBR®Green I or extra MgCl2. All reamplification reactions were performed in a Master cycler (Eppendorf) and involved initial denaturing (4 min at 95 °C), followed by 25 cycles including denaturing (30 s at 94 °C), primer annealing (40 s at 58 °C), and primer extension (12 s at 72 °C). A final extension step was set at 72 °C for 30 min. The quality of each sequencing template was checked by agarose gel electrophoresis. The DNA concentration of each amplicon was measured fluorometrically using picogreen staining (e.g. Coolen et al., 2006). Thirty nanograms of reamplified DGGE bands were submitted to Agencourt (Beverly, MA) for subsequent bidirectional cycle sequencing with primers EukA and Euk516r.
Sequences have been analyzed using the arb software package (version December 2007) (Ludwig et al., 2004) and the corresponding SILVA SSURef 96 database (Pruesse et al., 2007). After importing, all sequences were automatically aligned according to the SILVA SSU reference alignment. Manual refinement of the alignment was carried out taking into account the secondary structure information of the rRNA. Then, the phylogenetic bootstrap trees (1000 replications) were first reconstructed based on 1200-bp-long available sequences of the closest relatives using the neighbor-joining method (arb). The shorter aligned environmental 18S rDNA sequences from this study were inserted afterwards using the parsimony interactive tool implemented in the arb software package without changing overall tree topology. Sequences obtained in this study have been deposited in the NCBI sequence database under accession numbers GQ402464–GQ402489.
Results and discussion
Physical parameters and location of the suboxic zone, nutrients, and onset of sulfide
The upper 160 m of the Black Sea are brackish with salinities between 17.2 psu at the surface and 20.6 psu toward the sulfidic chemocline (Fig. 2). The surface water temperature was 22.5 °C at the time of sampling, with a steep drop to 15.3 °C at the thermocline at 25 m, followed by relatively constant temperatures (8.1±0.6 °C) between 31 and 160 m (Fig. 2). Fluorescence data show that a chlorophyll maximum was present just above the thermocline at 22 m, whereas chlorophyll concentrations dropped close to zero values <80 m (Fig. 2). A small increase in fluorescence was observed at the top of the suboxic zone at 120 m.
The suboxic zone with <0.2 mL L−1 (i.e. <10 μM) oxygen and undetectable sulfide concentrations (<0.2 μM) (Murray et al., 1989; Codispoti et al., 1991) at Station 5 was found between 120 m (στ=15.62) and 139 m (στ=15.90). At 140 m sulfide became detectable, showing 0.28 μM, and increased to 6 μM at 150 m (στ=16.00) and 12.5 μM at 160 m (στ=16.10) (Fig. 2).
Ammonia concentrations dropped from 8.5 μM at ∼160 m to undetectable levels (<0.2 μM) at the base of the suboxic zone. A small nitrite maximum (0.15 μM) was found in the suboxic zone at 130 m (Fig. 2). At depths between ∼100 and 120 m where oxygen concentrations reached up to 5 mL L−1, a second (upper) nitrite maximum occurred, which agrees well with previous findings (Ward & Kilpatrick, 1990; Murray & Yala-del-Rio, 2006). A nitrate peak with up to 4 μM of nitrate was observed between ∼80 and 120 m. Phosphate concentrations were highest at the bottom of the suboxic zone and in the sulfidic waters down to 160 m (∼4 μM) (Fig. 2).
Distribution of metabolically active eukaryotes along the environmental gradients
Metabolically active eukaryotes were identified within 19 POM samples collected along a vertical gradient (salinity, oxygen, hydrogen sulfide, and nutrients), and within the upper 8 cm of sulfidic sediments that underlie 830 m of sulfidic water at the sampling location in the Black Sea. The active population was studied from reverse transcribed and PCR-amplified (two-step RT-PCR) 18S rRNA whereas combined molecular signatures of both metabolically active and inactive populations were studied from PCR-amplified (genomic) 18S rDNA (i.e. 18S rRNA- vs. 18S rDNA-based libraries). All amplicons were subjected to DGGE (Fig. 3a). Unfortunately, we lost the DNA extract from 80 m during the purification procedure and, as a result, no 18S rDNA amplicons could be generated from that sample (Fig. 3a). The corresponding DNAse-treated 18S rRNA extract from the 80 m sample could subsequently be analyzed by RT-PCR (Fig. 3a). Total DNA and RNA was successfully extracted from 115, 75, 135, and 160 m, but the samples from these depths contained fewer 18S rDNA (i.e. 115a) or 18S rRNA (75b, 135b, 160b) template for PCR and yielded less amplicons for subsequent DGGE analysis as compared with other samples within the environmental gradient (Fig. 3a). Fifty-six numbered DGGE bands were excised from the DGGE gels and 53 of the excised bands were successfully sequenced. These sequences represented 18 unique phylotypes (based on a 99% sequence similarity cutoff) (Fig. 4). The DGGE bands that represented unique phylotypes are numbered 1–18 in the DGGE (Fig. 3a). These phylotypes comprised copepods (four phylotypes; 12_BSA6S5Euk, 16–18), tunicates (phylotype 11), haptophytes (phylotype 1), dinoflagellates (eight phylotypes; 3, 6, 7, 8, 9, 10, 13, 14), and ciliates (phylotypes; 2 and 4) as well as a sequence (phylotype 15), which did not fall into established taxonomic groups but which showed up to 96% sequence similarity to 18S rDNA sequences recovered from the sulfidic waters of Framvaren Fjord (Fig. 4, Table 1) (Behnke et al., 2006).
Table 1. General information about the eukaryotic phylotypes recovered from the Black Sea
Closest (named) relative(s)
Sequence similarity (%)
Phyl. no., phylotype number depicted from Fig. 4; Tax. grp, taxonomic group; Cop, copepods; Tun, tunicates; Hap, haptophytes; Din, dinoflagellates; Cil, ciliates; Unc, unclassified; Rot, rotifers. Light- and dark-gray shaded part of the table represent phylotypes recovered from, respectively, the water column and the sulfidic sediments of the Black Sea.
Present between 4 and 8 cm; activity not confirmed
Sulfidic waters Mariager Fjord
Clone BS_DGGE_Euk-1 (DQ234281); Black Sea sulfidic sapropel Unit II
Present between 6 and 8 cm; activity not confirmed
A copepod (phylotype 12) with 100% sequence similarity to partial 18S rDNA of Pseudocalanus elongatus (Fig. 4, Table 1) was identified from both 18S rDNA/rRNA libraries in the oxygenated part of the mixolimnion (between 40 and 95 m) (Fig. 3a) and represented up to 40% of the total reverse-transcribed and PCR-amplified 18S rRNA pool (Fig. 2). Pseudocalanus elongatus is known to be highly abundant throughout the Black Sea (Niermann & Greve, 1997) and makes up ∼40% of the zooplankton in autumn within the offshore regions (Unal et al., 2006). It is a cold-water species with an upper temperature limit of occurrence of 13 °C, which would explain its absence in the chlorophyll maximum (Figs 2 and 3a) that was 22.2 °C at the time of sampling.
The sequence of phylotype 16 was identical to that of both Calanus pacificus as well as Calanus helgolandicus (Fig. 4, Table 1) and was present at depth intervals throughout the oxygenated part of the water column (Figs 2 and 3a). Most likely, phylotype 16 represents the endemic Calanus euxinus (Hulsemann) (Besiktepe et al., 1998) instead, for which no 18S rDNA sequence is available. Both C. helgolandicus and C. euxinus are morphologically similar and show <0.5% sequence differentiation in the faster evolving mitochondrial cytochrome oxidase subunit I gene (mtCOI). This recently raised the question of whether C. euxinus is a different species (Unal et al., 2006). Calanus helgolandicus is an epipelagic copepod and typically inhabits the saline waters of world's oceans (Fleminger & Hulsemann, 1977) with salinity values of 32–39, whereas C. euxinus is a key component of the Black Sea pelagic ecosystem (Besiktepe et al., 1998) and has adapted to the lower salinity of the Black Sea (∼18 psu). Furthermore, the predominance of phylotype 16 in both 18S rDNA/rRNA libraries within the deeper waters, notably the suboxic zone, corroborates with the lifestyle of C. euxinus in the Black Sea. Previous studies have shown that C. euxinus descends during the day to the cold suboxic zone thereby significantly decreasing its total metabolism as well as enabling it to utilize the energy of consumed food for growth and lipid accumulation (Svetlichny et al., 2000; Svetlichny et al., 2006).
The copepod represented by phylotype 17 was metabolically active at most depths between 40 and 80 m, whereas the closely related phylotype 18 was identified only from the narrow interval within the oxygenated upper nitrite maximum between 105 and 110 m (Figs 2, 3a, and 4, Table 1). The closest relative of both phylotypes was an unassigned copepod found at a depth of 180 m in the Arctic Baffin Bay (Fig. 4, Table 1) with low in situ temperatures (1 °C) and low chlorophyll concentrations (Hamilton et al., 2008). A parallel identification of these unassigned copepod species based on morphological characteristics was no longer possible because the filters were completely used up by the nucleic acids extraction.
Phylotype 11 represented Oikopleura dioica, the only recorded gelatinous tunicate (appendicularian) in the Black Sea (Shiganova, 2005) (Fig. 4, Table 1). Its 18S rRNA (DGGE band 11, Fig. 3a) was mainly found in the chlorophyll maximum at 22 m but a faint DGGE band 11 was also present in the lower part of the upper nitrite maximum (between 110 and 115 m) (Figs 2 and 3a). The predominance of its 18S rDNA at 22 m is in accordance with the knowledge that most individuals of this euryhaline and eurythermic tunicate are present in the thermocline, whereas fewer individuals appear to be present in the deeper oxygenated waters of the Black Sea (Shiganova, 2005 and references therein). The coinciding presence of Oikopleura 18S rRNA implies that viable Oikopleura individuals are present in the chlorophyll maximum as well as in the deeper oxygenated waters of the Black Sea (Figs 2 and 3a). Oikopleura dioica populations grow exceptionally fast (Seo et al., 2001) and this species ingests phytoplankton (5%) but its main food is bacteria and detritus (95%) (Shiganova, 2005). The specialized filter apparatus of O. dioica retains particles no larger than 15 μm and this feature allows feeding in deep, detritus-based ecosystems. This utilization of small size classes decreases competition with other zooplankton species (Tiselius et al., 2003). The presence of viable O. dioica in the nitrite maximum just above the suboxic zone with low chlorophyll levels (Figs 2 and 3a) suggests that O. dioica could be actively involved in filtering mainly bacteria and/or detritus at this depth.
Our molecular survey revealed a previously overlooked haptophyte (i.e. phylotype 1), which is affiliated with calcifying haptophytes and Cruciplacolithus neohelis as its closest named relative (Fig. 4, Table 1). Whereas 18S rRNA of this active haptophyte represented up to 30% of the total RT-PCR-amplified 18S rRNA pool (Fig. 2), the corresponding genomic 18S rDNA of this species was often below detection on DGGE (Fig. 3a) and most likely also escaped prior identification from an 18S rDNA library constructed for additional locations in the Black Sea (Coolen et al., 2006). This discrepancy can be caused by the fact that DGGE is insensitive to less dominant species (e.g. Coolen et al., 2007), or those with low genomic rRNA gene copy numbers.
The presence of active cells of this haptophyte down to 115 m could point to a capacity for heterotrophic growth, given the extremely low light intensity and narrow light spectrum supporting photosynthetic activity at this depth (Overmann et al., 1992). Many haptophytes from early diverging, noncalcifying lineages (Prymnesiales, Phaeocystales) have retained heterotrophic behavior such as predation on bacteria and small algae (i.e. phagotrophy) (Kawachi et al., 1991; Tillmann, 1998). Phagotrophy is restricted to motile haptophyte cells with an emerging prey-catching haptonema (Kawachi et al., 1991; Jones et al., 1994; Tillmann, 1998). Also the motile haploid flagellate phase of certain calcifying haptophytes were found to be phagotrophic (Houdan et al., 2006). A phagotrophic life style for the haptophyte related to calcifying species is thus plausible and could be verified as soon as a culture from the Black Sea's deep suboxic zone would become available.
The widespread marine bloom-forming calcareous haptophyte Emiliania huxleyi, which is also common in the Black Sea (Eker-Develi & Kideys, 2003), was not identified in our molecular surveys. A test with parallel tag-encoded FLX amplicon pyrosequencing (e.g. Sogin et al., 2006) of a 130-bp-long fragment of the V9 region revealed that 18S rDNA of haptophytes of the order Isochrysidales to which E. huxleyi belongs, represented only 2% of the total eukaryotic 18S rDNA in the chlorophyll maximum at 22 m (M.J.L. Coolen & C. Davis, unpublished data). As outlined earlier, DGGE is biased to identify only phylotypes that represent >∼2% of the total pool of PCR-amplified DNA (e.g. Coolen et al., 2007), and would explain why E. huxleyi was missed in our general eukaryotic 18S rDNA PCR/DGGE approach.
Our sampling set revealed eight unique dinoflagellate phylotypes, which were found to exhibit high vertical stratification, and their presence often coincided with the presence of specific nitrogen species or suboxic/anoxic conditions. For example, the dinoflagellates represented by phylotypes 9 and 3 were present and active at specific depths between 40 and 95 m (Figs 2 and 3a), and were closely affiliated with uncultivated dinoflagellates from the deep chlorophyll maximum in the Sargasso Sea and/or phagotrophic Gyrodinium spp. (Fig. 4, Table 1).
In addition, nucleic acids of some unassigned dinoflagellates were found only in the suboxic or anoxic waters of the Black Sea. The dinoflagellates represented by phylotypes 10 and 13 (Fig. 4, Table 1) were rare in the water column and only their 18S rDNA was identified from the upper part of the suboxic zone (Figs 2 and 3a). Therefore, their activity or their indigeneity to suboxic waters could not be confirmed. On the other hand, phylotype 14 (Fig. 4, Table 1) was clearly active at the oxic/suboxic transition at 120 in the presence of 3 μM of oxygen (Figs 2 and 3a). The DNA of this dinoflagellate was most abundant at 125 m, whereas its RNA was below the detection limit at this depth, underlying the value of comparative analysis of gene transcripts for verification of metabolic activity of microbiota within environmental gradients. Phylotype 8, related to a clone from the Sargasso Sea deep chlorophyll maximum, represented the only dinoflagellate to be active in the deeper part of the suboxic zone with undetectable levels of oxygen, as well as at 150 m in the presence of 6 μM of hydrogen sulfide (Fig. 2).
The sulfidic chemocline furthermore contained transcripts of the 18S rRNA gene of an unclassified eukaryote (i.e. phylotype 15) with up to 96% sequence similarity to 18S rDNA sequences recovered from sulfidic waters of Framvaren Fjord (Fig. 4, Table 1) (Behnke et al., 2006). This phylotype was not detected from the oxygenated waters above, but a more sensitive PCR approach using species-specific primers would be required to verify whether the presence of this eukaryote was restricted to the slightly sulfidic part of the stratified water column of the Black Sea.
Information about the species composition of ciliates in the Black Sea is sparse and mostly relates to coastal areas (Kovaleva & Golemanski, 1979; Bouvier et al., 1998). Our survey revealed only two unique phylotypes of ciliates throughout the upper 160 m of the water column. Phylotype 4 (Fig. 4, Table 1) is closely related to ciliates of the class Oligohymenophorea and was found in the RNA-based (metabolically active) library from fully oxygenated waters at a depth of 40 m (Figs 2 and 3a). Environmental sequences from DNA-based libraries of highly sulfidic Framvaren Fjord waters (Behnke et al., 2006) were the closest relatives of phylotype 4 (Fig. 4).
A ciliate with Strombidium styliferum as its closest relative (phylotype 2) (Fig. 4, Table 1) was only found to be metabolically active at the oxic/suboxic zone transition at 120 m and just present at 125 m. However, more precise genotyping would be required to identify this ciliate because a total of 63 morphotypes have been described from this genus (Agatha, 2004) and the estimates of genetic diversity is much greater (Song & Packroff, 1997).
Eukaryotes in the Black Sea's sulfidic sediments
Nine out of 17 successfully sequenced DGGE bands recovered from the upper 8 cm of sulfidic sediments that underlie 830 m of sulfidic waters represented unique phylotypes and grouped with copepods (phylotype 27), rotifers (phylotype 24), haptophytes (phylotype 22), dinoflagellates (phylotypes 21 and 23), and ciliates (four phylotypes; 19, 20, 25, 26) (Fig. 4, Table 1). Figure 3b displays the position of the sequenced DGGE bands corresponding to these phylotypes. The phylogenetic affiliation with other eukaryotes, their relative abundance, and whether they represent species indigenous to the sediments or allochthonous species derived from the water column above, will be discussed next.
The upper 4 cm of sediment contained both DNA and RNA from a copepod distantly related to Tortanus sp. New Caledonia (phylotype 27; Fig. 4, Table 1). In the surface sediment (0–2 cm), 18S rRNA of this copepod represented 40% of the total RT-PCR-amplified 18S rRNA pool (Fig. 5). Most likely, the 18S rRNA of this Tortanus-related copepod was extracted from resting eggs derived from the overlying oxygenated water column. Assuming an absence of a postdepositional redistribution of the exquisitely laminated sulfidic sediments (Fig. 1b) and an average sedimentation rate of 0.2 mm year−1 (M.J.L. Coolen, unpublished data), the sediment interval at 4-cm depth with detectable Tortanus 18S rRNA gene transcripts would correspond to a depositional age of 200 years before present. This finding indicates that copepod eggs (depending on the species) could remain viable with detectable levels of 18S rRNA gene transcripts in highly sulfidic environments for up to 200 years after sedimentation. This is significantly longer than the previous reported recovery of viable copepod eggs from up to 40-year-old sulfidic sediments from the stratified Pettaquamscutt River Basin (RI) in the presence of 2 mM of hydrogen sulfide (Marcus et al., 1994), which is approximately seven times the concentration of sulfide in the Black Sea surface sediments (Zopfi et al., 2004). Because the primary targets of sulfide are aerobic respiration enzymes, for example cytochrome c oxidase, it was suggested that copepod eggs may gain protection from sulfide by undergoing anaerobic respiration (Marcus et al., 1997).
The rotifer Mytilina mucronata was the closest relative of phylotype 24 (Fig. 4). Its 18S rRNA, probably extracted from viable resting eggs, comprised up to 20% of the total 18S rRNA pool within up to 300-year-old (i.e. the upper 6 cm) sulfidic sediments (Figs 3b and 5). 18S rDNA of the rotifer was also identified but resulted only in a faint DGGE band (band 24; Fig. 3b) at 2–4 cm, which was also the sample with the most intense corresponding band with RT-PCR amplified 18S rRNA. Our data suggest that rotifer eggs in the Black Sea sediments remained longer viable than copepod eggs because the copepod 18S rRNA rapidly declined and reached undetectable levels <4 cm (i.e. ∼200-year-old sediments; Fig. 5). This is in agreement with the results from Marcus et al. (1994) who found that a higher percentage of eggs from a rotifer of the genus Brachionus could be hatched from up to 40-year-old sulfidic Pettaquamscutt River Basin sediments as compared with copepod eggs.
Interestingly, a unique haptophyte (phylotype 22; Figs 3b and 4, Table 1) with E. huxleyi as its closest relative was found to be active in the top 2 cm of sediment, whereas its DNA was too faint to be detected by DGGE. This haptophyte was not detected from the water column at the time of sampling. These results imply that this haptophyte has developed a strategy to remain metabolically active for substantial periods of time (at least weeks and up to decades) in the presence of ∼175 μM of sulfide. Previously, it was shown that haptophytes were among various protists that were found to proliferate from incubated Swedish coastal sediments (Persson, 2002) but it is unclear to what extent those protists were exposed to hydrogen sulfide.
Nucleic acids of two dinoflagellate phylotypes were recovered from the sediments. Phylotype 23 is related to dinoflagellates of the cyst-forming genus Scrippsiella with an identical sequence to that of Scrippsiella sp. MBIC11164 (Fig. 4, Table 1). Phylotype 23 comprised up to 40% of the total 18S rRNA pool in the upper 6 cm of sulfidic sediment (Fig. 5), which suggests that cysts of this species remain viable for up to 300 years.
On the other hand, phylotype 21 with 99.8% sequence similarity to species of the genus Pentapharsodinium (Fig. 4, Table 1) was only recovered from the sedimentary 18S rDNA pool. Its DNA was most likely protected inside organic-walled cysts of Pentapharsodinium, which have been described from recent Black Sea sediments (Marret et al., 2004). However, the absence of Pentapharsodinium RNA suggests that cysts of Pentapharsodinium were no longer viable in the sulfidic sediments.
Phylotypes 19, 20, 25, and 26 were recovered from up to 8 cm of sulfidic Black Sea sediments (Figs 3b and 5) and were closely related to 18S rRNA clones from unnamed ciliates previously retrieved from the sulfidic waters (i.e. ∼50 μM of sulfide) of the Danish Mariager Fjord (Fig. 4, Table 1) (Zuendorf et al., 2006). The absence of 18S rRNA from the four ciliates implied that they were not viable in the subsurface Black Sea sediments, which exhibit a sulfide concentration three times higher than in the Mariager Fjord waters. Whether the sedimentary ciliate DNA was derived from cells stemming from the less sulfidic chemocline could not be confirmed because the same phylotypes were not detected in the two analyzed sulfidic waters at 150 and 160 m.
Summary and conclusions
We have shown that analysis of gene transcripts has enabled the identification of species that were active but escaped identification in a parallel 18S rDNA survey. In addition, the above examples also indicate that gene transcripts do not always provide unequivocal proof that metabolically active microorganisms are indigenous to a certain position in an environmental gradient, because certain species stemming from the oxygenated part of the photic zone can remain viable in the presence of high sulfide concentrations. This is especially true when environmental sequences represent novel taxonomic clades with unknown phylogenetic affiliation. The 18S rRNA gene survey from surface sediments yielded taxa of pelagic origin, but none of these were identified in the water column at the time of sampling. Vice versa, none of the species, which were thriving in the water column at the time of sampling, could be detected in the fossil DNA fraction. There are various possible explanations for this discrepancy: (1) the filtered POM only represents a snapshot of the total annual plankton community, (2) not all cellular material is equally well transported to the sediment, or (3) there are species- or cell-specific variations in the level of (post)depositional degradation of intracellular DNA. Although the DGGE method represents a relatively fast approach in screening a large number of samples along the environmental gradient for major shifts in the eukaryote community structure, it is insensitive to less dominant species, or those with low genomic rDNA copy numbers. Future research will require application of other molecular biological approaches, notably tag-encoded FLX amplicon pyrosequencing, in order to determine the diversity and relative abundance of total vs. active eukaryotes.
We would especially like to thank Ognyana Hristova and Tatyana Nikolova, IOBAS, for the analysis of the water column chemistry and Dr Cornelia Wuchter, Dr Angela Dickens, Alan Gagnon, Daniel Montluçon, Chris Ward (WHOI), and the R/V Akademik staff, in particular Delcho Solakov, for their extensive organizational and participatory help with the cruise. We thank Dr Timothy Eglinton (WHOI) and two anonymous reviewers for suggestions to improve the manuscript. We are grateful for the financial support from the US National Science Foundation grant OCE 0602423, as well as funding from WHOI's Access to the Sea program, and a grant from the Andrew W. Mellon Foundation Endowed Fund for Innovative Research.