The marine deep biosphere potentially harbors the majority of all prokaryotic cells on earth (Whitman et al., 1998). Their abundance was shown to be high in various marine sediments below the world’s oceans (Parkes et al., 1994). Even though the organisms within deep-sea sediments have to cope with severe substrate and nutrient depletion (Jørgensen and D’Hondt, 2006), a large proportion of the microbial inhabitants were shown to be alive (Schippers et al., 2005; Morono et al., 2011). Furthermore, geochemical characteristics of deeper sediments indicate microbial activity as several chemical transformations are linked to microbial processes (Parkes et al., 2000; D’Hondt et al., 2004). During the last decades we got insights into microbial community composition and diversity within the deep subsurface (Inagaki et al., 2006; Sørensen and Teske, 2006; Webster et al., 2006; Fry et al., 2008; Lipp et al., 2008; Teske and Sørensen, 2008; Durbin and Teske, 2011).

Owing to the mostly anoxic regime in highly compacted sediments, grazers are most likely absent in marine subsurface sediments. Whether viruses take over the role as the main controlling factor for indigenous microbial communities is an open question. The impact of viruses on shaping microbial communities and affecting carbon and nutrient cycling was shown by estimates of high viral activity at the seafloor (Danovaro et al., 2008). In a previous study, we found viruses in high numbers within the marine subsurface as deep as 320 mbsf (meters below seafloor) (Engelhardt et al., 2011). Nevertheless, their activity in deep subsurface sediments is not proven yet. An ongoing discussion about the activity and the origin of viruses in deeper marine sediments led to the suggestion that viruses were buried and are well preserved within sediment layers where microbial activity is low (Bird et al., 2001). A different view was given by Middelboe et al. (2011), who hypothesized that only a small fraction of the whole viral community was active in deeper sediments and that viral productivity may be restricted to microniches.

In the present study, we analyzed host populations of Rhizobium radiobacter and associated phages within various deep-sea sampling sites from the Peru margin area to the open ocean of the eastern equatorial Pacific. The sediments, drilled during Leg 201 of the Ocean Drilling Program (Supplementary Figure S1), cover representative aspects of the marine deep subsurface (D’Hondt et al., 2004). The continental margin sites 1227 and 1230 are considered to be of higher microbial activity due to a higher content of organic carbon (1227: 1.05–10.84 wt%, average 5.14 wt%; 1230: 0.71–5.21 wt%, average 2.77 wt%). Sites 1225 and 1226 are more oligotrophic (1225: 0.01–0.85 wt%, average 0.18 wt%; 1226: 0.25–3.49 wt%, average 0.92 wt%) and represent sediments with lower microbial activity. Thus, the sampling sites comprises a broad range of characteristics in terms of microbial activity, chemical profiles and abundances of prokaryotes (Parkes et al., 2000; D’Hondt et al., 2004; Schippers and Neretin, 2006). Therefore, this area of the deep biosphere is well suited to study the role of viruses for the investigated sites and also allow extrapolating their impact to other marine subsurface habitats.

R. radiobacter was chosen to analyze host-phage relationships as it is the most frequently isolated and highly abundant representative of the marine deep subsurface. This species was detected to be a dominant member of microbial communities within deep-sea sediments of the Mediterranean by both, molecular investigations (5% of the total bacterial community) and a cultivation-based study (Süß et al., 2004, 2006). In our culture collection from ODP Leg 201, R. radiobacter was the predominant isolate from various sites and sediment depths (Batzke et al., 2007). In this study, we quantified R. radiobacter in the respective sediment layers to prove if the high number of isolates correlates with their in situ abundance.

Our aim was to identify the occurrence and diversity of temperate rhizobiophages associated with our isolates and to elucidate the relevance of lysogeny for deep subsurface populations. Thus, the so far deepest sediments (381 mbsf) were analyzed for viral abundance. Two rhizobiophages (RR1-A and RR1-B) were subjected to whole genome sequencing. These phages were originally associated to a R. radiobacter-affiliated strain that was isolated from a depth of 198 mbsf at ODP site 1225. Based on the genome information of rhizobiophage RR1-A, we established a detection system to quantify their distribution with sediment depth. The characterization of this representative group of temperate phages within the host population and their quantification within the sediments will enhance our understanding of the viral impact on life and death in the marine subsurface.

Materials and methods

Origin of sediment samples

Sediment samples were collected during ODP Leg 201 of the Ocean Drilling Program from February to March 2002 with the R/V JOIDES Resolution (Supplementary Figure S1). A detailed description of the sampling sites at the Peru margin (ODP sites 1227–1230) and sites from the open Pacific ocean (ODP sites 1225, 1226 and 1231) and geochemical parameters are given by D’Hondt et al. (2002, 2004). Sediment samples for molecular analysis described in this study have been stored at −80 °C.

Enumeration of virus-like particles (VLPs)

Viruses were extracted according to the protocol described by Danovaro et al. (2001). Briefly, individual subsamples of 1 cm3 of sediment were added to a 0.02-μm pre-filtered mixture of 3.5 ml ddH2O and 1 ml sodium-pyrophosphate (55 mM). Slurries were mixed at 2,000 r.p.m. for at least for 15 min or until the sediment was completely dissolved. Samples were sonicated for 3 min with interruption of 30 s each minute and shaken at 2000, r.p.m. during interruptions. After centrifugation (700 g, 2–10 min), the supernatants were removed and 4 ml of TE-buffer (10 mM Tris, 1 mM EDTA, 0.02 μm filtered) was added to the sediment pellet. Slurries were shaken for 10 min after resuspension of sediment pellets, followed by centrifugation as described above. Supernatants of both extraction steps were pooled and filtered though polycarbonate filters (mesh 0.45 μm, Satorius, Göttingen, Germany).

For enumeration of VLPs, between 10 and 500 μl of virus extracts were filtered onto 0.02 μm anodisc filters (13 mm, Whatman, England) and washed with TE-buffer. Viruses were stained by applying 3.5 μl of a mounting solution (Lunau et al., 2005) containing SybrGreenI (final concentration 1 × ). Counting of VLPs was done in 10–20 randomly chosen fields per sample by using an Olympus BX51 microscope (Hamburg, Germany). Between 100 and 350 VLPs were counted per sample.

DNA extraction from sediments

DNA extraction was performed according to a protocol described by Lueders et al. (2004) and Gabor et al. (2003). Briefly, 0.2 ml of 1:1 mixed 0.1 mm and 0.5 mm silica beads and 700 μl PTN-buffer (120 mM NaPO4, 125 mM TRIS, 25 mM NaCl, pH 8) were added to 0.5 cm3 sediment. This slurry was vortexed for 10 s. After adding 40 μl lysozyme (50 mg ml−1) and 10 μl Proteinase K (10 mg ml−1) the mixture was incubated at 37 °C with repeated shaking. Subsequently, 100 μl SDS (20%) was added and the mixture was incubated at 65 °C for 15 min. DNA was purified by phenol/chloroform/isoamylalcohol (PCI, 25:24:1, pH 8) extraction including mechanical disruption of cells by beat beating. After addition of 100 μl of PCI, the mixture was treated by bead beating for 45 s, centrifuged (6000 g, 5 min, 4 °C) and the supernatant was collected separately. Subsequently, PTN-buffer (300 μl) was added to the slurry and after beat beating for 20 s the mixture was centrifuged again as described above. DNA was extracted from the pooled supernatants by adding one volume of PCI followed by a second step using one volume of chloroform/isoamylalcohol (24:1). Centrifugation was performed for 4 min, at 13 000 g and 4 °C. For precipitation, two volumes of PEG/NaCl (30% polyethylene glycol 8000, 1.6 M sodium chloride) were added to the final supernatant. Thereafter, the mixture was incubated for 2 h at 4 °C and the DNA pelletized (13 000 g, 30 min, 20 °C). The pellet was washed by adding 150 μl of 70% ethanol (−20 °C). After spinning (13 000 g, 5 min, 4 °C) ethanol was carefully removed. The pellet was dried and the DNA was eluted in 50 μl PCR-H2O. DNA templates were stored at −20 °C.

Quantification of R. radiobacter

Quantification of R. radiobacter was performed by quantitative PCR using R. radiobacter-specific primers as described by Süß et al. (2006). Originally, the specificity the primers was tested with 19 non-target strains from the German culture collection (DSMZ) and 9 environmental strains. Only one isolate (Bacillus sp. J105) revealed a weak positive signal that was neglectable as 106 targets of strain J105 showed the same CT-value as 10 targets of R. radiobacter. Furthermore, in mixtures with E. coli DNA, R. radiobacter was accurately quantified over a broad range of target concentrations (0.0003–30%) (Süß et al., 2006).

To generate standards for quantifying R. radiobacter, the almost complete 16S rRNA gene of strain P007 was amplified before quantitative PCR by using universal primers 8F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5′-GGTTACCTTGTTACGACTT-3′) (Overmann and Tuschak, 1997). Amplification was carried out as described by Süß et al. (2004). The PCR amplicons were diluted from 1 × 108–1 × 100 gene copies per μl. The calibration standards were additionally amplified in every qPCR run for quantification.

Quantitative PCR was performed in a Rotor-Gene-3000 cycler (Corbett Research, Sydney, New South Wales, Australia). Each reaction mixture (total volume 25 μl) contained 12.5 μl of Maxima SybrGreen qPCR MasterMix (2 × ; Thermo Scientific, Karlsruhe, Germany), BSA, respective primers. Environmental DNA templates were added in 1:10 and 1:100 dilutions to identify putative inhibitions by co-extracted substances from the sediment matrix. Quantification was performed with an initial activation at 95 °C for 15 min followed by 50 cycles with a denaturation at 94 °C of 10 s, annealing at 54 °C for 20 s, and an elongation at 72 °C for 30 s. Fluorescence was measured at 72 and 82 °C. Subsequently, a melting curve was recorded by increasing the temperature from 50–99 °C with a temperature ramp of 1 °C per 10 s.

Phylogenetic analysis of R. radiobacter isolates

In total, 36 R. radiobacter-affiliated strains were obtained from ODP Leg 201 sediments. The isolation procedure and contamination control were described in detail by Batzke et al. (2007). The 16S rRNA genes of 36 strains of R. radiobacter were amplified by using the universal primers 8F and 1492R as described above. Amplified DNA was purified (Qiagen PCR Purification Kit, Qiagen, Hilden, Germany) and subsequently sequenced by GATC Biotech AG (Konstanz, Germany). Phylogenetic analysis was performed by using the ARB software package (Versalovic et al., 1991; Ludwig et al., 2004). The database Silva 106 ( was used as a backbone-tree. Sequences obtained during this study were integrated by using the ARB Fast Aligner feature ( The accession numbers of all 16S rRNA gene sequences obtained in this study are deposited at the GenBank database under JX110581-JX110616.

The genetic diversity of the different R. radiobacter strains below the species level was determined by enterobacterial repetitive intergenic consensus (ERIC) PCR (Versalovic et al., 1991). The ERIC-PCR mixture had a final volume of 50 μl and contained: Taq polymerase (0.02 U per μl; New England BioLabs, Ipswich, MA, USA), the 10-fold polymerase buffer according to manufacturers specification, BSA (0.4 mg ml−1), dNTPs (0.2 mM each) and the ERIC primers ERIC1R (5′-ATGTAAGCTCCTGGGGATTCAC-3′) and ERIC2F (5′-AAGTAAGTGACTGGGGTGAGCG-3′) (0.2 pmol per μl each). The ERIC-PCR amplification was performed with an initial denaturation (96 °C, 4 min), followed by 37 cycles of denaturation (94 °C, 30 s), annealing (52 °C, 1 min), and elongation (72 °C, 8 min) with a single final elongation (72 °C, 10 min). PCR products were run on an acrylamid gel at 50 V for 18 h, stained with SYBRGold (Invitrogen, Carlsbad, CA, USA) and documented digitally (BioDocAnalyze, Biometra, Göttingen, Germany).

Induction of prophages from R. radiobacter isolates

Cultures of each R. radiobacter strain were grown in 250 ml LB-medium at 20 °C and shaking speed of 100 r.p.m. Growth was monitored by measuring the optical density at 580 nm. When reaching the exponential growth phase, 50 ml of the culture were separated and used as a respective control. The remaining culture was treated for 30 min with mitomycin C (final concentration: 0.5 μg ml−1), washed by centrifugation (10 000 g, 10 min, 20 °C) and resuspended in fresh media. After a drop in cell density of the induced culture in comparison with the control, the cells were pelletized (10 000 g, 10 min, 20 °C). To remove remaining cells and cell debris, the supernatant was filtered through a polycarbonate filter (mesh 0.45 μm, Sartorius stedium biotech GmbH, Göttingen, Germany). For further analyses, extracted phage particles were concentrated by a polyethylene glycol treatment.

Preparation of phage concentrates and viral DNA purification

To concentrate phage particles from induction experiments and sediment samples, a mixture of polyethylene glycol and sodium chloride (final concentration: 10% PEG 8000, 1 M NaCl) was added to the phage extract. After incubation for 1 h on ice, the phages were pelletized (10 000 g, 12 min, 4 °C). Finally, the pellet was resuspended in 400 μl of SM buffer.

To remove non-viral DNA, 88.5 μl of the phage concentrates were added to 10 μl DNAseI-buffer (10 × ) and 1.75 μl DNAseI (New England BioLabs). After incubation for 90 min at 37 °C, 1 μl ES-buffer (0.5 M EDTA, 0.5% w/v N-lauroylsarcosin) was added. To release encapsulated viral DNA, the mixture was incubated for 15 min at 75 °C. Subsequently, viral DNA was purified by using the Wizard PCR Preps DNA Purification System (Promega GmbH, Mannheim, Germany). DNA was quantified with NanoDrop 2000c Spectrophotometer (Thermo Scientific).

Pulsed-field gel electrophoresis (PFGE) to determine phages genome sizes

Pulsed-field gel electrophoresis was performed on a 0.8% agarose gel using the Rotaphor v6 system (Biometra, Göttingen, Germany). Purified viral DNA (150 ng per lane) was separated by applying a pulse ramp of 2–15 s, a linear decreasing electrophoresis rate of 5.3–4.6 V cm−1 with an included angel ramp of 120–130 °C. PFGE was performed for 22.5 h at 14 °C. The Low Range PFG Marker (New England Biolabs) was used for size assignment.

Whole genome sequencing of rhizobiophage RR1-A and RR1-B

Rhizobiophage RR1-A (Acc. no. PRJNA47553) and RR1-B (Acc. no. PRJNA60463) were induced from the host strain P007 (isolated from ODP site 1225, 198 mbsf). Sequencing, finishing and gene prediction of the viral genome was conducted in the frame of the Marine Phage Sequencing Project by the Broad Institute. For more Information see ( Manual gene annotation was performed using the BLASTp sequence analysis service of the NCBI online platform (

Establishing a detection system for temperate rhizobiophages

Two primer pairs were developed by using the pDraw32 software package ( to target the prophage antirepressor gene (ORF 65) in the genome of Rhizobiophage RR1-A. Antirepressor genes of phages are involved in the switch from lysogeny to lytic proliferation. Because of their ability to operate with non-cognate repressors, they have the capacity to coordinate a multiple prophage induction. Thus, viral antirepressor genes are unique for temperate phages. Therefore, they can be used as marker genes for temperate phages. The first primer pair RR1-A-F (5′-GGAAAAGCCAGTCCGCGC-3′) and RR1-A-R (5′-GGTGTGAGGATGCGGATGC-3′) was designed to amplify the almost complete antirepressor gene (length: 624 bp). PCR was performed with an initial denaturation at 94 °C for 10 min, followed by 30 cycles with a denaturation at 94 °C of 30 s, annealing at 56 °C for 45 s, elongation at 72 °C for 1 min and a final elongation for 10 min at 72 °C. The resulting PCR product was used to generate standards for qPCR analysis. The second primer pair, RR1As-F (5′-GCTTGCCAACCGGCTCGG-3′) and RR1As-R (5′-TTGCGGATGCCCACTGTTGG-3′), was developed to amplify a shorter fragment of the gene (124 bp) for detection and quantification of the antirepressor gene. PCR was performed with an initial denaturation at 94 °C for 10 min, followed by 30 cycles with a denaturation at 94 °C of 30 s, annealing at 56 °C for 45 s, elongation at 72 °C for 45 s and a final elongation for 10 min at 72 °C.

Quantification of temperate rhizobiophages in sediment samples

The 624 bp long fragment of the almost complete antirepressor gene was used as a standard for quantitative PCR. The amplicons were purified (Qiagen PCR Purification Kit, Qiagen) and the amount of DNA was quantified with a NanoDrop 2000c Spectrophotometer (Thermo Scientific). The number of gene targets was calculated from the length of the DNA fragment, the amount of DNA and the average weight of base pairs (1.1 × 10–21 g). A standard curve was generated by a 10-fold dilution in a range from 102–107 antirepressor gene targets.

Prior to quantification, extracts of viral particles were subjected to DNAse digestion to remove free DNA and purified as described above. Quantitative PCR was performed by using the primer set RR1As-f/RR1As-r in a Rotor-Gene-3000 cycler (Corbett Research, Sydney, Australia). Each reaction mixture (total volume 25 μl) contained 12.5 μl of Maxima SybrGreen qPCR MasterMix (2 × ; Thermo Scientific), BSA, respective primers. Viral DNA templates in 1:10 and 1:100 dilution were added to identify putative inhibitory effects. Quantification was performed with an initial activation at 95 °C for 15 min followed by 50 cycles with a denaturation at 94 °C of 10 s, annealing at 56 °C for 20 s, and an elongation at 72 °C for 30 s. Fluorescence was measured at 72 and 80 °C. Subsequently, a melting curve was recorded by increasing the temperature from 50–99 °C with a temperature ramp of 1 °C per 10 s.


Total counts of cells and viruses

The abundance of cells and viruses was determined in deep-subsurface sediments of the Peru margin (ODP sites 1227 and 1230) and the eastern equatorial Pacific (ODP sites 1225 and 1226). Cell numbers were between 107 and 108 cells cm−3 in the upper most sediment samples and continuously decreased by two orders of magnitude in the deepest sediment layer at 381 mbsf (Figure 1a). The virus counts were between 107 and 108 per cm3 at 1 mbsf and decreased with depth by one order of magnitude, only. Generally, the virus-to-cell ratios increased with sediment depth (Figure 1b). For example, at the Peru margin site 1230, the ratio between viruses and cells in the upper hundred meters were determined to be rather low (4–1.7). A higher ratio of 17 was detected for the deepest sediment layer at a depth 257 mbsf. This was also found at the open ocean site 1225, where the virus-to-cell ratios increased from about 2 at 1 mbsf to 21 at 320 mbsf (Supplementary Table S1).

Figure 1
figure 1

(a) Abundances per cm3 of cells (circles), VLP (squares) and copies of prophage antirepressor gene (triangles) within ODP Leg 201 sediments. Peru margin sites and open ocean sites are represented by black and gray symbols, respectively. (b) VLP-to-cell ratio. (c) Relative amount of R. radiobacter compared with total Bacteria based on 16S rRNA gene copies. (d) Relative amount of prophage antirepressor gene copies compared with total viral abundance.

Abundance and phylogenetic diversity of R. radiobacter

The 16S ribosomal RNA gene of R. radiobacter was specifically detected by quantitative PCR (qPCR) at all investigated sites and sediment layers except for site 1227 at 34 mbsf (Figure 1c). The proportion of R. radiobacter-related 16S rRNA gene copies increased with depth in comparison with the abundance of bacterial 16S rRNA gene copies. The relative amount of R. radiobacter ranged between 0.0004% at 1 mbsf of site 1230% and 5.6% at 198 mbsf of site 1225 (Supplementary Table S1). On an average, the relative abundance of R. radiobacter-affiliated 16S rRNA genes was 0.76%.

All 36 R. radiobacter-affiliated isolates from ODP Leg 201 (Batzke et al., 2007) were phylogenetically analyzed in detail. Based on their 16S rRNA gene similarities, all isolates showed a close affiliation to the type strain of R. radiobacter (Figure 2). Differences in 16S rRNA gene sequences were below 0.1%, only. Most strikingly, sequence similarities were highest among strains isolated from the same sampling sites. This was further proven by genome fingerprinting of the R. radiobacter-affiliated strains using ERIC-PCR. According to the ERIC-PCR pattern analysis, the different isolates arranged in three site-specific subclusters (Figure 3).

Figure 2
figure 2

Phylogenetic tree based on the 16S rRNA gene, showing the affiliation of various strains of R. radiobacter. Differences in 16S rRNA genes are indicated, bar=0.1%.

Figure 3
figure 3

Genetic similarities below species level of various R. radiobacter-affiliated strains as analyzed by enterobacterial repetitive intergenic consensus (ERIC) PCR. Site-specific clustering of R. radiobacter subpopulations regarding to their ERIC-pattern is shown. Genome sizes of inducible prophages given here are referred to the genome sizes of phages RR1-A and RR1-B. Detection of the prophage antirepressor gene of Rhizobiophage RR1-A is indicated by+(detected) and Ø (not detected).

Detection of rhizobiophages in isolates of R. radiobacter

To analyze the presence of temperate rhizobiophages, the prophage-inducing agent mitomycin C was used in stimulation experiments with all 36 isolates affiliated to R. radiobacter. A drop in cell density in the mitomycin C-treated cultures in comparison with the non-induced controls indicated a cell lysis due to the release of phage particles (Engelhardt et al., 2011). Pulsed-field gel electrophoresis (PFGE) revealed the presence of phages with two different genome sizes within the population of R. radiobacter (Figure 3). The genome sizes were 52 and 37 kb, respectively. Nine of the strains showed a multiple prophage infection as both phages were induced simultaneously. Several strains showed individual bands with 10 kb in size, which may suggest the presence of phage-like gene transfer agents, which are known to be widespread in Alphaproteobacteria (Lang and Beatty, 2007).

Genome annotation of rhizobiophage RR1-A and RR1-B

The genome of rhizobiophage RR1-A was determined to be of circular conformation and 52 641 kb in length with an average G+C content of 57.4% (Figure 4a). The genome sequence comprised a total number of 67 predicted open reading frames (Supplementary Table S2). Considering only E-values smaller than 1e−30 as significant hits, about 70% of the predicted genes were unknown. The other 30% were identified as core genes of viral genomes found in members of the order Rhizobiales. Among those, nine genes were supposed to encode for structural modules, such as capsid- and tail-forming proteins. Ten genes were related to DNA processing and phage assembly. Additionally, a detected prophage antirepressor gene was highly similar (96% query coverage, E-value 2e−91, maximum identity 55%) to that found in Rhizobium vitis (formally named Agrobacterium vitis; Young et al., 2001). Due to its function of switching from lysogeny to lytic reproduction, this gene is indicative for temperate phages.

Figure 4
figure 4

Gene cards of Rhizobiophage RR1-A and RR1-B. Predicted gene functions are indicated according to the manual annotation (Supplementary Tables S2 and S3).

Rhizobiophage RR1-B comprised 37 378 kb and 52 predicted ORFs. The average G+C−content was 58.54% (Figure 4b). 75% of the predicted genes had an unknown function (e-value >e−30). Three of these genes resulted in no hit at all. BLASTP analysis indicated six genes to be involved in DNA processing and eight predicted genes encode for structure proteins of phage particle (Supplementary Table S3). The antirepressor gene from prophage RR1-A was not detected.

Detection of the prophage antirepressor gene and quantification in sediment samples

A protein BLASTp analysis using the NCBI database revealed only one hit with a sufficient query coverage for the antirepressor gene of R. vitis strain S4 as described above. The next closest protein sequence derived from Pseudovibrio sp. strain FO-BEG1 (query coverage 70%; maximum identity 38%), other protein sequences showed query coverages below 61% and maximum identities of<57%. Available pure cultures R. radiobacter (36 strains) and virus extracts from environmental samples were tested for the presence of the antirepressor gene by using the primer set RR1As-f/RR1As-r. No unspecific amplification was observed for all investigated DNA templates (Figure 5). A diversity analysis of this gene fragment across all sites and sediment depths by denaturing gradient gel electrophoresis revealed only one band, indicating a high degree of conservation (data not shown). In total, 32 isolates of R. radiobacter were tested positively. Thirteen of these strains did not show a respective 52 kb band in the PFGE analysis (Figure 3), which might be due to defective prophages. For four other isolates, neither the PFGE band nor the the antirepressor gene was detected.

Figure 5
figure 5

Detection of prophage antirepressor gene of Rhizobiophage RR1-A. A fragment (124 bp) of the prophage antirepressor gene of RR1-A was amplified by using the primers RR1As-f and RR1As-r. DNA from the different isolates of R. radiobacter or purified viral DNA extracts from various sediment samples obtained from ODP Leg 201 were used as template DNA. R. radiobacter strain P007 served as a positive control. Ø denotes non-template control.

The BLASTp analysis, the specific PCR with isolates and environmental viral DNA extracts, and the denaturing gradient gel electrophoresis analysis of the respective PCR products indicate the specificity of the molecular approach to detect temperate rhizobiophages. This gene was used for the specific quantification of temperate rhizobiophages within the fraction of free viral particles by qPCR as a DNAse treatment of virus extracts before quantification excluded the quantification of prophages within bacterial genomes.

Gene copy numbers were 105–106 genes cm−3 of sediment. The number of free phages containing this gene varied with depths, but constituted no clear trend (Figure 1a). However, the relative amount of temperate rhizobiophages in relation to the total virus counts increased slightly with sediment depth (Figure 1d). For example, in near-surface sediments of site 1230, the relative amount of these phages accounted for 0.4% of the total viral numbers in contrast to 10.2% at 103 mbsf (Supplementary Table S1). At site 1225, a comparably high amount of temperate rhizobiophages of 8.4% was already determined in near-surface sediments and increased slightly to 12.4% at 198 mbsf.


All isolates of R. radiobacter analyzed in this study turned out to be lysogens. The large number of associated temperate phages in the sediments points towards an ongoing viral production presumably originating from natural induction events. Thus, lysogeny appears to be a predominant factor for host populations and the viral communities in the deep subsurface. We hypothesize that even the slow growing subsurface microbial communities are capable to support the production of new phages.

The high virus-to-cell ratio in deep sediment layers indicates viral production

The virus-to-cell ratio between 1 and 10 we found for the upper 100 mbsf was also described for subsurface sediments down to a depth of 110 mbsf for a high sedimentary region of the Saanich Inlet (Bird et al., 2001) and the eastern margin of the Porcupine Seabight (Middelboe et al., 2011). The latter authors interpret this ratio as a result of a better preservation of viruses within deep sediments due to the absence of exoenzymes. These enzymes are known to affect inactivation and degradation of viruses (Noble and Fuhrman, 1997). In addition, the stabilization of viral structures after absorption to inorganic particles could support this effect (Dowd et al., 1998).

In general, viral abundance in sedimentary habitats is likely the sum of preservation, degradation and ongoing production of viruses. The ratio between viruses and cells was shown to increase up to 21 in deeper sediments layers. As the deepest sediments are 11 Ma old, it appears unlikely, that this high number of viruses can be preserved over such a long time scale. Therefore, an ongoing production of phages within an active host population along the investigated sediment columns might contribute to a higher degree to the viral abundance than preservation. At ODP sites 1225 and 1230, for example, near-basement sediment layers are affected by a deep intrusion of electron acceptors from below (D’Hondt et al., 2004). For similar hydrogeological settings (IODP site 1301), it was shown that microbial communities in up to 265 m deep sediments are stimulated by fluids from the oceanic crust (Engelen et al., 2008). Such environmental conditions promote cell proliferation and probably result in enhanced viral production. Thus, the high abundance of viruses throughout all investigated sediment layers indicates that even low microbial activity may support viral production.

R. radiobacter as an abundant subseafloor inhabitant

R. radiobacter is a unique example of a highly abundant organism in the deep subsurface that can be detected by both, cultivation and molecular techniques. R. radiobacter was shown to contribute up to 5% to the bacterial community in the Mediterranean subsurface (Süß et al., 2006). In our study, their relative abundance even accounted for up to 5.6% of total bacterial 16S rRNA genes in ODP Leg 201 sediments (average: 0.75%). Furthermore, their increasing abundance with depth indicates that the environmental conditions within the deep subsurface were selective for this species. The constant degradation of organic matter during sediment accumulation leads to a limited substrate availability for indigenous microorganisms. Thus, metabolically versatile species such as R. radiobacter (Stowers, 1985; Batzke et al., 2007) might take advantage on this situation. A further adaptation factor for R. radiobacter to the conditions in the deep subsurface might derive from their associated prophages. It was shown that genes carried by prophages affect host metabolism by providing metabolic facilities (Jiang and Paul, 1998) and might increase host cell fitness and survivability. For example, the integration of prophages in bacterial genomes can shut down host genes (Brussow et al., 2004) and might help to save energy, which appears to be of critical importance for survival in subsurface sediments (Morono et al., 2011).

Site-specific development of distinct subpopulations of R. radiobacter indicates activity and evolution

Site-specific arrangement of genetically similar isolates of R. radiobacter shows the development of separate subpopulations at different sites due to geological segregation. This provides first evidence of a biogeographical distribution of an organism in the marine deep subsurface. The isolation caused by the geographical distance between continental margin and open ocean sites is the prerequisite for individual evolutionary processes. In addition, this might be promoted by the various physical and geochemical parameters contributing to the site-specific characteristics. Evolution of a subpopulation within a particular habitat only occurs when these organisms are actively growing. Nevertheless, in the deep subsurface, these processes can be expected to be very slow as the turnover times of prokaryotic communities were estimated to be several years for sediments from ODP Leg 201 (Schippers et al., 2005; Biddle et al., 2006).

Lysogeny appears to be characteristic for the deep subsurface

Previous studies addressing lysogeny in marine habitats mainly focused on pelagic environments (Weinbauer and Suttle, 1996; Cochran et al., 1998; Jiang and Paul, 1998; Leitet et al., 2006; De Corte et al., 2012). The lysogenic fraction is often shown to be rather low (Wilcox and Fuhrman, 1994; Weinbauer and Suttle, 1996). Concurrently, the contribution to total viral production by natural induction of prophages is calculated to be only 0.02% (Jiang and Paul, 1998). In contrast, unfavorable growth conditions for host organisms were shown to promote lysogeny. In the study by Williamson et al. (2002), they demonstrated lysogeny to occur in the water column mainly during winter month simultaneously to lowest microbial activity. As prokaryotes living in the marine deep subsurface are mainly prone to starvation (Jørgensen and D’Hondt, 2006), our study demonstrates the link between lysogeny and extreme environments such as the energy-poor marine subsurface. Other extreme habitats like energy-rich deep-sea hydrothermal vents can even be dominated by lysogens (Williamson et al., 2008). Previously, we found prophages in a variety of different phylogenetic groups isolated from the marine deep subsurface (Engelhardt et al., 2011). In the present study, we could show that lysogeny is characteristical for all investigated isolates of R. radiobacter. As this species accounts for a significant fraction of bacteria in this environment and rhizobiophages are found to be highly abundant, lysogeny appears to be a common feature for the marine deep biosphere.

Viral production in the deep subsurface is presumably due to a natural induction of prophages

The strict occurrence of lysogeny among R. radiobacter is reflected by the high amount of temperate rhizobiophages found in the investigated sediments. A production of phages can be assumed due to the generally high amount of free phages in deeper sediment layers. As the mobility of phages and host cells is limited in highly compacted sediments, it appears unlikely that lytic viral reproduction is a dominant mechanism for viral proliferation. In addition, the host cell density might be too low to support a lytic phage reproduction. Thus, viral production may result predominantly from a natural induction of prophages which was exemplarily shown for R. radiobacter cultures that were not treated by mitomycin C (Engelhardt et al., 2011). To what extent viral activity may have a role regarding the viral shunt and horizontal gene transfer within the deep subsurface sediments is yet unknown.