Quantitative analysis of bacterial communities from Mediterranean sapropels based on cultivation-dependent methods

Abstract

Microbial communities of ancient Mediterranean sapropels, buried sediment layers of high organic matter, were analyzed by most probable number (MPN) approaches. Mineral media containing different carbon sources in sub-millimolar concentrations were used. MPN numbers were elevated in sapropels and at the sediment surface, which mirrored total cell count distributions. Highest MPN counts were obtained with a mixture of different monomeric and polymeric substrates, with amino acids or with long-chain fatty acids as sole carbon sources. These values reached up to 2 × 10⁷ cm⁻³, representing 3.3% of the total cell count. A total of 98 pure cultures were isolated from the highest positive dilutions of the MPN series, representing the most abundant microorganisms culturable by the methods used. The strains were identified by molecular biological methods and could be grouped into 19 different phylotypes. They belonged to the α-, β-, γ-, and δ-Proteobacteria, to the Actinobacteria and the Firmicutes. However, about half of the number of isolates was closely related to the genera Photobacterium and Agrobacterium. Regarding the high cultivation success, these organisms can be assumed to be typical sapropel bacteria, representing a substantial part of the culturable indigenous microbial community.

Introduction

In sediments of the eastern Mediterranean Sea, conspicuous dark sediment layers, called sapropels, occur repeatedly in the sediments. They are characterized by elevated organic carbon contents, ranging from 2 to over 30% of dry weight [1,2] and were formed periodically approximately every 20,000 years, following the Milankovic cycles [1,3,4]. It is thought that sapropels were formed in response to climate changes leading to elevated nutrient input, higher productivity and increased sedimentation rates [1,3,4]. The resulting formation of anoxic bottom water led to the enhanced preservation of organic material in the sediments. After restoration of oxic conditions in the bottom water, the sapropels were buried under low organic carbon calcareous sediments[5].

It has been estimated that subsurface sediments contain more than 90% of the microbial biomass on earth [6,7]. However, these habitats exhibit only very low microbial activities [8,9] due to their depletion in easily degradable organic material. Sapropels, on the other hand are elevated in organic carbon, which despite its age and kerogenic state, still provides an energy source for indigenous microorganisms [10,11]. Recent studies showed that sapropels have elevated numbers of microorganisms compared to intervening low organic carbon layers, and that these can even reach those found at the sediment surface [10,11]. Even in almost 220,000-year-old sapropel layers active microbial populations were detected by the use of fluorescent substrate analogues and ¹⁴C-labelled substrates[11]. Molecular analyses of the indigenous microbial communities indicated the dominance of sequences related to the Green nonsulfur bacteria (GNSB) and to Crenarchaeota[11], both of them so far without cultured counterparts.

Although sapropels contain active microbial communities, these appear to have low viability since cultivation approaches employing standard media, yielded no more than 1500 cells g⁻¹ sediment[12]. Nevertheless, to investigate and understand the ecology of the sapropel microbiota and their physiological adaptations, it is necessary to obtain typical sapropel microbes in culture for study. In order to enrich a broad diversity of microorganisms, in particular GNSB, whose metabolism is so far unknown, we used a range of different media for cultivation. Besides media considered selective for terminal oxidizers such as sulfate-reducing bacteria, substrate combinations were applied, which should allow a broad variety of different microorganisms to grow. Since it is likely that environmental microorganisms are exposed to a ‘substrate shock' after transfer into substrate-rich medium lowering their viability [13,14], all substrates were provided in sub-millimolar concentrations. Additionally, we used a ‘natural' sapropel extract medium, since it was reported previously that sediment extract media achieved higher MPN counts than synthetic media[15].

Material and methods

Sampling

Sediment samples were obtained during cruise M51/3 of the R/V Meteor in November 2001 from three stations (562, 567, 575, Table 1) in the Eastern Mediterranean Sea. The temperature at the sediment surface was around 14 °C. Surface sediment was taken by multicorer, whilst for deeper sediment (up to 12 m) a gravity corer was used. The recovery speed of the sampling devices ranged from 0.3 to 0.5 m s⁻¹. Aseptic sampling of sediment was performed as described by Coolen et al. [11, supplementary material]. From the different sediment layers, 1 cm³ was taken by use of cut-off sterile plastic syringes, and was transferred into 9 ml sterile anoxic artificial seawater. Until processing, these sediment slurries were kept under a nitrogen atmosphere in the dark at 4 °C, although part of each slurry was fixed immediately with glutaraldehyde (final concentration 2%) for determination of total cell counts. These samples were stored at 4 °C in the dark.

Determination of total cell counts

Fixed sediment slurry (5–7 μl) was added to an equal volume of Tween 80 (0.5%, sterile-filtered) and 50 μl of particle-free sterile PBS buffer (130 mM NaCl, 5 mM NaH₂PO₄, 5 mM Na₂HPO₄, pH 7.2) in a sterile 1.5 ml Eppendorf reaction tube. After short ultrasonic treatment (5 × 5 s), the solution was filtered through a white polycarbonate membrane (0.1 μm pore size, 25 mm diameter, Anodisc 25, Whatman, Maidstone, UK). Bacterial cells were stained by the method described by Jaspers et al.[16] using a staining solution consisting of 70 μl DAPI (4′,6-diamidino-2-phenylindole, Sigma, 10 μg ml⁻¹), 1 ml of fixing solution (paraformaldehyde final conc. 4%, Triton 100 final conc. 0.1% in PBS buffer) and 930 μl ddH₂O. For comparison, samples were also stained with acridine orange as described by Coolen and Overmann[17].

The air-dried filter membranes were mounted in DABCO (25 mg Diazabicyclo-octan, 1 ml PBS, 9 ml glycerol) under a coverslip. Counting was performed using an epifluorescence microscope (Zeiss Axiolab, Oberkochen, Germany) equipped with filter sets for DAPI (BP365, FT395, LP397) and acridine orange (BP459-490, FT510 and LP515). For each layer, three replicates were counted, each up to at least 200 cells.

Growth media

An artificial seawater medium was used for preparing sediment slurries, MPN series and isolation of pure cultures. This medium contained (in g l⁻¹): NaCl (24.3), MgCl₂.6H₂O (10), CaCl₂.H₂O (1.5), KC1 (0.66), Na₂SO₄ (4), KBr (0.1), H₃BO₃ (0.025), SrCl₂.6H₂O (0.04), NH₄Cl (0.021), KH₂PO₄ (0.0054), NaF (0.003). The medium was supplemented with 1 ml l⁻¹ trace element solution SL10[18] and 0.2 ml l⁻¹ of a selenite and tungstate solution[19]. After autoclaving, the medium was cooled under N₂/CO₂ (80/20, v/v). To the cold medium, 10 ml of a solution of 10 vitamins[20] and 30 ml l⁻¹ of a 1 M NaHCO₃ solution were added from sterile stocks. Finally, the medium was reduced by addition of Na₂S and acid FeCl₂ solutions to final concentrations of 1.5 and 0.5 mmol l⁻¹, respectively. The pH of reduced medium was set to 7.2–7.4 with sterile HC1 or Na₂CO₃.

For oxic MPN counts and isolation of aerobic microorganisms, a slightly modified medium was used. Instead of bicarbonate and CO₂ the medium was buffered with Hepes (2.4 g l⁻¹) and the pH was adjusted to 7.2–7.4 with NaOH before autoclaving. After autoclaving, the oxic medium was cooled under air and then supplemented with vitamins and sodium bicarbonate (final concentration 0.2 g l⁻¹).

A diluted yeast extract–peptone medium was used for maintenance of aerobic pure cultures and for tests of aerobic growth. It consisted of the HEPES-buffered oxic seawater described above, amended with yeast extract (0.03 g l⁻¹), peptone (0.06 g l⁻¹), sodium lactate (5 mmol l⁻¹), glucose (1 mmol l⁻¹), sodium thiosulfate (1 mmol l⁻¹), vitamins and sodium bicarbonate (0.2 g l⁻¹).

Substrates used for MPN series

Six different substrate combinations were used as electron and carbon sources for MPN series: (1) AS (amino acids), containing 20 different l-amino acids (final concentration 0.1 mmol l⁻¹ each), (2) ALC (n-alcohols), containing methanol, ethanol, n-propanol and n-butanol (final concentration 0.1 mmol l⁻¹, each), (3) LFA (long chain fatty acids), containing saturated straight chain fatty acids with 14 to 18 carbon atoms (final concentration 0.1 mmol l⁻¹, each), (4) ARO (aromatic compounds), containing salicylate, p-OH benzoate and the methoxylated aromatic compounds vanillate, coumarate, ferulate, sinapinate and syringate (final concentration 0.1 mmol l⁻¹, each), (5) MKS (defined substrate mixture), containing a broad range of different carbon compounds, in particular the amino acids and alcohols as listed above, but also the short chain fatty acids formate, acetate, propionate, butyrate, valerate and caproate, and additionally glycerol, glucose, lactate, fumarate, malate, succinate (0.1 mmol⁻¹ final concentration, each). The MKS medium was additionally supplemented with the polysaccharides laminarin, xylan and chitin (0.1 g l⁻¹ each). (6) SED (sediment extract medium). Equal amounts of freeze–dried sapropel material and oxic mineral medium were thoroughly mixed and boiled for at least 2 h. After coarse particles had settled, the turbid supernatant was filtered to remove particles and was finally autoclaved. After autoclaving, the sapropel extract medium was cooled under air. The cooled medium was supplemented with vitamins and bicarbonate (0.2 g l⁻¹) as described above. For the preparation of sapropel extract medium, only a limited amount of sapropel material was available. Therefore, for the last isolation steps sediments from a tidal flat from the North Sea were used. In order to guarantee that older sediments with a low content of easily degradable organic matter were used, material was dug out from a depth of approximately 50 cm. Additionally, MPN series without added substrates were inoculated as a blank.

Preparation and incubation of MPN series

Viable counts were determined by MPN series that were inoculated on shipboard using the sediment slurry described above. All equipment necessary for preparation of the MPN series was placed into a polyethylene chamber (AtmosBag, 280 1, Aldrich, Milwaukee, WI), which was then flushed with nitrogen gas, evacuated and filled with N₂ again. The procedure was repeated up to five times to remove atmospheric oxygen. The MPN series were incubated in polypropylene 96-deep-well plates (Beckman, Fullerton, CA). Every well contained 450 μl medium, to which 50 μl inoculum were added. After inoculation the plates were covered-with sterile lids (CAPMAT, Beckman, Fullerton, CA), sealing each well separately. Each plate contained four different MPN series with three replicates and six 10-fold dilutions. Furthermore, in each plate four uninoculated dilution series were prepared as control.

The inoculated and sealed MPN plates were put into gas-tight plastic bags equipped with a gas generating and catalyst system for anoxic conditions (Anaerocult C mini, Merck, Germany). Incubation proceeded normally for 12 weeks at 20 °C, but long-chain fatty acid medium was incubated for 6 months.

Analysis of MPN series

The MPN plates were analyzed for microbial growth by epifluorescence microscopy for higher sensitivity. Cell densities in positive cultures were generally not very high, and many of the microorganisms found were attached to particles, e.g. FeS, hence detection by phase contrast microscopy proved difficult.

The gas-tight plastic bags containing the MPN plates were carefully transferred into an anaerobic hood. The lid was cut into small pieces and carefully lifted using sterile tweezers. From each well, 10 μl of culture was transferred into a cavity of a diagnostica slide. To each sample, 2 μl of a solution of SybrGreen II (1:100, Molecular Probes, Leiden, Netherlands) was added and mixed. The stained samples were incubated for at least 1 h in the dark prior to examination. For epifluorescence microscopy, an UV excitation filter set for SybrGreen II (BP450-490, FT510 and LP515) was used. The MPN counts were calculated as described by de Man[21] and corrected for the values obtained in substrate-free MPN series.

Isolation of pure cultures

For the isolation of pure cultures, the highest positive dilutions were used. All subculturing and isolation procedures were performed with the same media as in the MPN series. Isolation of strains from the oxic MPN series was done on agar plates. For the isolation of anoxic cultures, the deep-agar dilution method was used. Purity of cultures was checked by microscopy and by denaturing gradient gel electrophoresis (DGGE)[22].

Molecular biological screening of pure cultures by signature PCR

In order to affiliate the pure cultures with major bacterial groups, a signature PCR (SIG-PCR) following the protocol of Uphoff et al.[23] was carried out for all isolates. The applied primer mixture contained the universal primers for Eubacteria (BAC8f and UNI1493r) and the group specific primers for α-, β- and γ-Proteobacteria (αδP688r, βε P774r and γβP1403r), Actinobacteria (HGC1153r), Cytophaga-Flavobacterium-Bacteroides phylum (CFB947r) and for marine phototrophic α-Proteobacteria belonging to the Roseobacter and Erythrobacter clusters (αER247r). The PCR mixture had a final volume of 50 μl and contained: 1 U RedTaq polymerase (Sigma), the 10×polymerase buffer according to manufacturers specification, BSA (0.4 μg), dNTPs (200 μmol l⁻¹ each) and the SIG-PCR primer mix (0.1 μmol l⁻¹ each). The MgCl₂ concentration was adjusted to 2.1 mmol l⁻¹. The SIG-PCR followed a touch down protocol with decreasing annealing temperatures (59–53 °C, 0.2 °C step) as described by Uphoff et al.[23]. PCR products were analyzed by agarose gel electrophoresis. According to the applied primer combination, PCR products of the following lengths were expected: 100 bp for γ-Proteobacteria, 350 bp for Actinobacteria, 650 bp for α-Proteobacteria, 700 bp for (β-Proteobacteria, 1000 bp for the Cytophaga Flavobacteria phylum, and 1400 bp for Eubacteria.

Denaturing gradient gel electrophoresis (DGGE)

For DGGE analysis 630-bp long fragments of the 16S rDNA of each strain were amplified using the universal Eubacterial primer set GC357f (containing a 40-bp long GC clamp) and 907r. The PCR was carried out according to the protocol of Sass et al.[22]. DGGE was performed using an INGENYphorU-2 system (Ingeny, Leiden, The Netherlands). PCR products were loaded onto polyacrylamide gels (6% wt/vol) stored in 1×TAE (40 mmol l⁻¹ Tris, 20 mmol l⁻¹ acetate, 1 mmol l⁻¹ EDTA), with a denaturing gradient from 50% to 70% (100% denaturant correspond to 7 mol l⁻¹ urea and 40% formamide). Electrophoresis was performed at a constant voltage of 100 V and at a temperature of 60 °C for 20 h. After electrophoresis, the gels were stained for 2 h with 1× SybrGold in 1×TAE (Molecular Probes, Leiden, The Netherlands), washed for 20 min in distilled water, and documented digitally (BioDocAnalyze, Biometra, Göttingen, Germany).

Amplification and sequencing of the 16S rDNA

The almost complete 16S rRNA gene was amplified using the Eubacteria-specific primer set 8f and 1492r (10 pmol l⁻¹ each). A step down PCR was performed. Initial denaturation was done at 96 °C for 4 min; followed by 10 cycles with denaturation at 94 °C for 30 sec, annealing at 59 °C for 45 sec and extension at 72 °C for 1 min., and another 20 cycles with the same denaturation and extension parameters, but with annealing for 45 s at 54 °C. A terminal post-extension was done for 10 min at 72 °C. After amplification, PCR products were purified using the QIAquick PCR Purification Kit (Qiagen GmbH, Hilden, Germany). The DNA concentration of the purified products was quantified fluorometrically using the PicoGreen® dsDNA Quantitation Kit (Molecular Probes, Leiden, Netherlands) according to the manufacturers protocol. For sequence analysis, the DNA content of the template was set to 4 ng μl⁻¹.

The DNA sequencing reactions were performed using the DYEnamic Direct Cycle Sequencing Kit (Amersham Biosciences Europe GmbH, Freiburg, Germany) following the manufacturers instructions with a modified protocol for the PCR mixture, containing 1 μl 7deaza dNTPs, 2 μl IRD-labelled primer (0.5 pmol μl⁻¹) and 2 μl template. The 16S rDNA sequences were determined by cycle sequencing with the LiCor DNA Sequencing System 4000 (MWG Biotech, Ebersberg, Germany) as previously described by Overmann and Tuschak[24].

Sequences most closely related to those obtained in the present study were retrieved from GenBank using the BLASTN tool[25]. The nucleotide sequences of representative strains of each phylogentic group presented in this study are available at the GenBank database[24] under the Accession Nos. AJ630143 to AJ630199 and AJ748349 to AJ748351.

Results

Total cell counts

Total cell counts were determined at the sediment surface (Surf), in low carbon intermediate layers (Zl to Z6) and in the organic-rich sapropels (S1 to S7) at the three stations studied (Table 2). In order to get reliable counts, sediment slurries had to be treated by ultrasound and by the addition of detergents for cell detachment. This treatment turned out to be necessary since the major part of the cells appeared to be stuck to sediment particles, and therefore could not be reliably distinguished from the background fluorescence of the sapropel material. After the treatment described above, most cells were detached from particles and could be counted easier.

Total cell counts after DAPI-staining were in the range of 2 × 10⁸–1 × 10⁹ cells cm⁻³. The acridine orange direct counts revealed almost identical results (data not shown). These values are up to tenfold higher than in previous reports [10,11], most likely due to the modified staining procedure. Cell counts showed only a limited decrease with sediment depth. In sapropels, cell numbers were elevated compared to intermediate layers (Table 2).

Viable counts and cultivation efficiency

MPN counts were generally high and reached maximum values of 2 × 10⁷ cells cm⁻³ in the surface layer (stations 567 and 575) and in the sapropel layers S1 and S5 at station 567 (Table 2). In general, MPN counts decreased with depth but were elevated in sapropels compared to the intermediate sediment layers. However, with some substrate combinations MPN counts remained more or less constant or even increased with depth. This was in particular the case for aromatic compounds, that produced MPN counts orders of magnitude lower than those with other substrates at sediment surface. In the deeper layers, however, they were as high as or even exceeded MPN counts with the other substrate combinations.

Surprisingly, MPN counts obtained after oxic and anoxic incubation were similar, indicating the presence of high numbers of facultatively anaerobic microorganisms. Sediment extract medium yielded lower MPN counts than the mineral media used (Table 2). In substrate-free controls, counts were below the detection limit.

From total cell numbers and MPN counts cultivation efficiencies were calculated. The highest values reached 3.3% and were obtained at the surface, in sapropel 1, but also in sapropel 5. The substrates with the highest cultivation efficiencies were the substrate mixture MKS, amino acids but also long chain fatty acids. Considering the 95% confidence levels given in Table 2 the highest cultivation success was in the range of 0.8–15% of the total cell count.

Isolation of pure cultures

To identify the most abundant microorganisms growing in the various media, pure cultures were isolated from the highest positive MPN dilutions. In total, 144 subcultures were inoculated. However, some of these failed to grow after repeated transfers or did not form visible colonies. In particular, no pure cultures were obtained with aromatic compounds and long chain fatty acids as substrates. Finally, 98 pure cultures were obtained (Table 3). Most of these (71 strains) were isolated under anoxic conditions, compared to 27 aerobic strains. Screening for aerobic growth revealed, that only six out of the 71 anaerobic isolates were strict anaerobes, having no visible colonies developed within two months under oxic conditions.

Generally, growth under anoxic conditions was very slow. In most cases, the bacteria in deep-agar dilutions needed at least six, but sometimes up to 12 weeks to form small visible colonies. Furthermore, anaerobically grown colonies were not very large in size. However, the size of the colonies was depending on the growth substrate as well as on the phylogeny of each strain. Especially, on substrates which are difficult to ferment like alcohols, and also on sapropel extract medium, colonies were rarely larger than 0.2 mm in diameter. Whilst on amino acids or the defined substrate mixture MKS, colonies could reach up to 4 mm in diameter.

Molecular biological screening and phylogenetic affiliation of pure cultures

A two-step screening was applied to avoid the expensive and time-consuming procedure of 16S rDNA analysis of all isolates. In the first step, a signature PCR (SIG-PCR) with a mixture of six different primer pairs was used to roughly affiliate the isolates with the most common phylogenetic groups (Fig. 1). With this approach 42% of the isolates belonged to the γ-Proteobacteria, 21% to the α-Proteobacteria and 17% to the Actinobacteria. Twelve isolates yielded none, or only the 1400-bp long PCR product for Eubacteria. No SIG-PCR products indicative for β-Proteobacteria or members of the Cytophaga-Flavobacterium-Bacteroides phylum were obtained. The strictly anaerobic isolates were not screened by SIG-PCR.

After screening the isolates via SIG-PCR, denaturing gradient gel electrophoresis (DGGE) was applied to subdivide the isolates based on sequence differences of the 16S rDNA. In total, the DGGE analysis produced nineteen different melting types: five in the γ-Proteobacteria, four in each of the Actinobacteria, the α-Proteobacteria and the strains that produced only the eubacterial band in the SIG-PCR screening, and two in the strictly anaerobic strains. Two to twelve representatives of each melting type were identified by partial sequencing of the 16S rDNA (400–850 bp long fragments). The grouping by molecular methods was supported by similarities in the phenotypical appearances of colonies and cells. With only one single exception, strains belonging to a single melting type shared sequence similarities of at least 98%.

The BLAST analysis of the target group revealed that the majority of the γ-Proteobacterial strains (33) was related to Photobacterium profundum^T (Table 3). Within this group 16S rDNA sequence differences of up to 4% were found. Closest relatives of the other γ-Proteobacteria were Marinobacterium jannaschii^T, Vibrio splendidus^T, and Vibrio halioticoli IAM 14598. One strain affiliated with the γ-Proteobacteria by SIG-PCR, turned out to be a β-Proteobacterium with Janthinobacterium sp. An8 as closest relative. The second-largest group of isolates (21 strains) belonged to the α-Proteobacteria. Of these, seventeen were closely related to Agrobacterium sp. JS71. The other four were affiliated with Erythrobacter citreus^T, Paracoccus marcusii^T, and Roseobacter sp. PIC-68. The strains identified as Actinobacteria (17 strains) were related to the genera Brachybacterium (nine strains) and Micrococcus (eight strains). All strains that could not be classified after signature PCR turned out to be Firmicutes. Primers specific for Firmicutes were not in the SIG-PCR primer set. Firmicutes isolated under anoxic conditions were affiliated with Bacillus jeotgali^T and Sporosarcina macmurdoensis^T, while those isolated as aerobes were related to Bacillus licheniformis^T and Bacillus barbaricus^T. Of the six strictly anaerobic strains directly analysed by partial 16S rDNA sequence analysis, four were closely affiliated with Desulfofrigus fragile^T within the δ-Proteobacteria and two were identified as Firmicutes, most closely related to an uncultured member of the Clostridiales.

Discussion

Cultivation efficiency

In the present study the first quantitative cultivation-based data on ancient sapropels from the Eastern Mediterranean Sea are published and remarkable cultivation efficiencies of more than 3% of the total cell numbers were obtained, in considering the age of these sapropels (8000–195,000 years old). Considering the large confidence levels of MPN results (Table 2, in the range of a 1/4 of to 5-times the MPN count), the MPN counts are probably in a range between 0.8% and 15% of the total cell count. These values are clearly higher than those typically found in marine sediments (up to 0.1% of the total cell count, Amann et al.[26]), and those obtained during a previous investigation on sapropels (0.01% or less [Sass et al., unpublished]).

Up to now, there are only very few cultivation-based quantitative studies on the microbial communities in marine subsurface sediments [2,27–29]. Although in a few cases enrichment of up to 30% of the total cell counts was reported (Table 4), in most cases the cultivation efficiencies were far below 1%. However, in the present study high MPN counts were achieved in almost all sediment layers investigated. The media applied here seem, therefore, to support not only a relatively high cultivation success but also consistently elevated results. This might be due to the use of media with low substrate concentrations compared to standard media. Low nutrient media have recently been shown to increase the number and diversity of cultured bacteria from pelagic habitats [30–33]. Microorganisms enriched and isolated using low substrate concentrations should typically be oligocarbophilic or oligotrophic bacteria [34,35], which are assumed to represent the majority of the environmental microorganisms. In nutrient-rich media these generally slow growing organisms can easily be outcompeted by copiotrophic bacteria [34,35], and are likely to be irreversibly damaged after exposure to high substrate concentrations (substrate shock [13,14]). Besides the application of low substrate concentrations, a second strategy to prevent the isolation of copiotrophic bacteria was the use of highly diluted samples as inoculum or of dilution series cultures. In this respect the use of MPN series and the isolation from the highest positive dilutions might be advantageous. Microorganisms growing in the highest positive dilutions represent theoretically the most abundant microbes that can be cultured in the medium used.

The use of “natural media” was reported to yield high MPN counts in recent studies[15]. It was supposed that this medium contained the electron donors present also in the natural environment. This is debatable, since the heating and autoclaving most likely will lead to the release of low molecular weight organic material[36]. This would not be available for the microorganisms in situ. However, in our case the applied sapropel extract medium was outmatched by the artificial but defined seawater medium (Table 2).

Another improvement in the media was the use of FeS as reducing agent for anoxic media. FeS is common in marine sediments and was present in the sapropel layers[37]. Therefore, indigenous microbes should be adapted to it. FeS is less reactive than other reducing agents, like dithionite or titanium chloride and should, therefore, be less aggressive towards cells. Furthermore, the concentration of free hydrogen sulfide was decreased. The presence of FeS particles seemed also to support growth of several strains, as these preferentially grew attached to particles.

The cultivation success in this study appeared to be independent from the type of substrate provided, since relatively high MPN counts (10⁵–10⁷ cells cm³) were obtained with nearly all substrate combinations, particularly at the sediment surface. Even the use of aromatic compounds seemed to be useful to stimulate the growth of microbes from very old sediment layers. Obviously, microbes in the old sediments are accustomed to deal with these compounds that are recalcitrant under anoxic conditions. In most cases the number of microorganisms stimulated to grow on the substrate mixture MKS was as high as on the single substrates or even higher. This result was to be expected, since MKS represents a combination of the single substrates used. Why the MPN numbers with MKS from sapropel S5 (#567) are lower than those with amino acids alone can only be speculated.

Despite the small volume of the inocula, the deep-well plates turned out to be well suited for MPN analyses. Because of their small size and because all steps can be carried out with multichannel pipettes, they allow a large number of incubations at small volume and relatively quickly. A possible negative effect of the small volume of inocula on cultivation success is unlikely. Zhang et al.[38] did not find any correlation between viable counts from subsurface sediments and sample size.

It the present study total cell counts were higher than in other studies on Mediterranean sapropel-bearing sediments [10,11]. A possible explanation is the treatment of fixed samples with ultrasound and the addition of detergents. Both procedures were addressed to the detachment of cells from sediment particles and it is possible that these were therefore easier to detect. It is unlikely that the use of DAPI caused the elevated total counts, for example by unspecific staining of sediment particles, because acridine orange counts produced almost the same total cell numbers.

Diversity of isolated microorganisms

In the present study the MPN series were not only used for quantitative investigations, but also to isolate abundant microorganisms. The application of media containing different carbon sources was aimed at the enrichment of a broad variety of different sapropel microorganisms. Because the MPN counts reached a relatively high percentage of the total cell count, pure cultures obtained from the highest positive MPN dilutions represent the most abundant culturable members of the microbial community. With 98 different strains isolated, most of them under anoxic conditions, it is by far one of the most extensive strain collections obtained from marine subsurface habitats so far. A few reports described extensive aerobic culture collections that were obtained from surface sediments [39,40] or marine chemoclines[22], but not subsurface marine sediments. Despite the age of most of our sediments, the diversity (indicated as Shannon–Weaver index[41], data not shown) was in the same range as described for cultures from surface freshwater and marine sediments [39,40] or chemoclines[22]. A rarefaction analysis indicated that almost all bacteria from the sediments that were capable of growth in our media were present in the culture collection (Fig. 2). However, Green non sulfur bacteria (GNSB), which were identified as dominant bacterial group in sapropels[11] and subsurface sediments from the Nankai Trough[42], were not present in the culture collection. So far, there is no counterpart of this phylogenetic group in culture and their metabolism and physiology can only be speculated. However, as long as the needs of these organisms are unknown, it is hard to design specific cultivation medium. On the other hand, it could also be that they are strict barophiles and are, therefore, either unable to grow at atmospheric pressure or that they were damaged by decompression during sampling.

Predominance of facultative anaerobes within culture collection

One remarkable finding was the predominance of aerobes among our isolates. This finding was supported by the high MPN counts obtained after oxic incubation. Even most of the strains isolated under anoxic conditions turned out to be only facultative anaerobes. Nonetheless, these bacteria were growing without oxygen and it has to be assumed that most of these microorganisms are anaerobically active in situ. Conversely, the Eastern Mediterranean Sea is a highly oligotrophic environment with a low productivity and little sedimentation. As a consequence, the input of organic matter into the sediment is low and only little microbial activities can be detected. At some sites oxygen penetrated down to a depth of 210 mm[12] and nitrate and oxidized manganese or iron species were present along the whole surface sediment penetrating the upper layers of the youngest sapropel S1[43]. These oxidized but oxygen-free conditions would be favourable for facultative anaerobes. Only a minority of the isolates were spore formers, and hence might have been dormant in the sediment.

The low number of sulfate-reducing bacteria in the culture collection is surprising. In the sapropels and intermediate layers sulfate is still present and it should be expected that sulfate reduction is the dominant terminal oxidation process. Sulfate reduction was shown to occur in sapropels[37], although rates were relatively low and no sulfate decrease within sapropels could be detected. However, despite their importance in the biogeochemical cycles, in most sediments sulfate reducers are not the numerically dominating group of microorganisms. Even in sediments with high sulfate reduction, sulfate-reducing bacteria account for less than 12% of the total cell count[44]. On the other hand it could be possible that some fermenting microorganisms within the sapropels are using the kerogenic organic material as an electron sink as it was described for humic acids [45,46]. In this case they would not rely on the cooperation with terminal oxidizers.

Phylogenetic affiliation of isolates

Members of 19 different operational taxonomic units (OTU) were isolated. Most of these are known to be typical marine bacteria frequently found in the deep sea or in marine sediments. Some of them have also been detected in subsurface habitats, e.g. Micrococcus sp., Rhodococcus sp. or Pseudomonas sp. [47–49]. During a previous study on sapropels some of the OTUs found in the present study were also detected (e.g. Photobacterium sp., Vibrio sp., Erythrobacter sp. [Sass et al., unpublished]). In previous studies on sapropels almost exclusively Green non-sulfur bacteria were detected by culture-independent methods[11]. However, in newer investigations also several other sequences were retrieved [Overmann et al., unpublished], some of them related to Sphingomonas sp. and Marinobacter sp., genera present in our culture collection.

Strikingly, Agrobacterium-related organisms were repeatedly isolated from the different sediment layers and formed a dominant group within the culture collection. This finding seems surprising since Agrobacterium sp. are typical soil bacteria, some of them known to be even plant pathogens. Nevertheless, Agrobacterium-related 16S rDNA fragments were retrieved from sapropels by PCR and DGGE [Overmann et al., unpublished]. Commonly, only 16S rDNA sequences that constitute more than 1% of all template molecules in a natural bacterial community are detectable by the PCR-DGGE approach[50]. Therefore, these organisms seem to be quite common in the sapropels and intermediate layers. In addition, organisms with Agrobacterium-related 16S rDNA fragments have been isolated from the terrestrial deep subsurface [47,49]. Despite their high similarity on basis of 16S rDNA the Agrobacterium-like isolates from the Mediterranean Sea and the ‘normal' Agrobacterium sp. do not necessarily have the same metabolism. The 16S rDNA gene is normally exposed to a limited environmental selection[51] and it might be possible that it can remain almost unchanged over a long period while the rest of the genome is probably undergoing adaptive changes. How these organisms thrive in their natural habitat and what metabolism they possess is a task for future research.

Phylogenetic screening of the culture collection by SIG-PCR and DGGE

At present time, analysis of the 16S rRNA gene sequence is the principle approach for the phylogenetic classification of microorganisms. However, the sequence analysis of a large number of isolates is time-consuming and expensive. For that reason, a pre-screening step was used to sort the isolates into different operational taxonomical units (OTU). Of each of these OTUs, two or a few representatives were analyzed, avoiding an unnecessary large sequencing effort. SIG-PCR was chosen for the sorting strains into major bacterial phyla and the correspondence of bands in DGGE, together with the phenotypical appearance of cells and colonies, was used for grouping the strains into the different OTUs. In general, this procedure worked well despite the fact that no primers for δ-Proteobacteria and Firmicutes were available. This is due to the fact that SIG-PCR was invented for pelagic systems[23] but for future approaches the system can be adapted. The fact that strains related to the β-Proteobacteria were first assigned to the γ-Proteobacteria was caused by γ-Proteobacteria specific primer also fits to some β-Proteobacteria[23].

Acknowledgements

The support of the scientific party of RV Meteor cruise M51/3, with Christoph Hemleben as chief scientist is gratefully acknowledged. We thank Herbert Hoffellner, Karin Schubert and Jörg Overmann for onboard preparation of sediment samples, and Imke Prölß for help with microscopic analysis of MPN plates and John Parkes for discussion and suggestions. This work was supported by a grant of the Deutsche Forschungsgemeinschaft.

References