|Home | About | Journals | Submit | Contact Us | Français|
During cruises in the tropical Atlantic Ocean (January to February 2000) and the southern North Sea (December 2000), experiments were conducted to monitor the impact of virioplankton on archaeal and bacterial community richness. Prokaryotic cells equivalent to 10 to 100% of the in situ abundance were inoculated into virus-free seawater, and viruses equivalent to 35 to 360% of the in situ abundance were added. Batch cultures with microwave-inactivated viruses and without viruses served as controls. The apparent richness of archaeal and bacterial communities was determined by terminal restriction fragment length polymorphism (T-RFLP) analysis of PCR-amplified 16S rRNA gene fragments. Although the estimated richness of the prokaryotic communities generally was greatly reduced within the first 24 h of incubation due to confinement, the effects of virus amendment were detected at the level of individual operational taxonomic units (OTUs) in the T-RFLP patterns of both groups, Archaea and Bacteria. One group of OTUs was detected in the control samples but was absent from the virus-treated samples. This negative response of OTUs to virus amendment probably was caused by viral lysis. Additionally, we found OTUs not responding to the amendments, and several OTUs exhibited variable responses to the addition of inactive or active viruses. Therefore, we conclude that individual members of pelagic archaeal and bacterial communities can be differently affected by the presence of virioplankton.
Viruses now are recognized as abundant and dynamic members of marine microbial communities, influencing the cycling of organic matter and nutrients in the sea (for a recent review, see reference 46). Viral infection is a stochastic process and depends on the abundance of viruses and hosts (24). Theoretical considerations regarding this mechanism suggest that viruses might selectively kill the most abundant members of the prokaryotic community and, in so doing, influence species richness (42). A number of studies support this hypothesis (38, 39, 44, 47). More recently, virus-mediated genetic exchange through generalized transduction as an additional mechanism potentially involved in shaping prokaryotic community structure has received attention as well (6, 34).
Although the overall role of bacterioplankton in the cycling of organic carbon and nutrients in the sea has been fairly well studied (2, 3, 10), current knowledge of the functional role of planktonic Archaea is rather rudimentary. Archaea are ubiquitously distributed throughout the world's oceans (11, 19) and constitute a numerically dominant group of the microbial community in the meso- and bathypelagic zones (12). It has been reported that marine planktonic Archaea take up amino acids (30) and bicarbonate (48) and therefore are metabolically active in the oceanic water column. Thus far, marine pelagic Archaea have withstood all attempts at cultivation in pure cultures, making it impossible to isolate viruses infecting these Archaea. Thus, the question arises as to whether marine Archaea are controlled by viruses in a manner similar to that reported for Bacteria. Previous studies of interactions of viruses and Archaea focused on viruses of extremophiles (4, 29, 49); however, the influence of viruses on marine pelagic Archaea has not been studied yet.
For decades, microbial ecologists used batch cultures to investigate the responses of complex prokaryotic communities to experimental manipulation. However, conclusions based on such incubations must be drawn cautiously, as first noted by ZoBell (50). Recently, studies on the effects of confinement of prokaryotic assemblages found shifts in the phylogenetic compositions of bacterial communities as a consequence of the exclusion of grazers (37, 40) and increased metabolic activity of at least some prokaryotes compared to the results obtained under in situ conditions (36). Massana and Pedrós-Alió (20) showed that bacterial communities in confinement are altered significantly within a few days of incubation and that these alterations are accompanied by dramatic decreases in community richness. However, incubation of microorganisms in micro- and mesocosms remains an invaluable tool for answering specific questions, especially when dealing with otherwise unculturable microorganisms such as marine Archaea.
The objective of this study was to monitor the responses of complex bacterial and archaeal communities to virus amendments by using terminal restriction fragment length polymorphism (T-RFLP) analysis of PCR-amplified 16S rRNA gene fragments. We present the results of batch culture experiments carried out with samples from two contrasting environments, the tropical Atlantic Ocean and the North Sea.
(This work was carried out in partial fulfillment of the requirements for a Ph.D. degree from the University of Groningen by C. Winter.)
We conducted a cruise in the tropical Atlantic Ocean off the coast of Africa (27 January to 14 February 2000) and an additional cruise in the southern North Sea (11 to 16 December 2000) with RV Pelagia (Fig. (Fig.1).1). Water samples (140 to 220 liters) were obtained from 25- and 75-m depths in the tropical Atlantic Ocean (stations AF1, AF2, and AF3) and from a 5-m depth in the North Sea (station NS1) (Fig. (Fig.1).1). The samples were collected with 10-liter NOEX bottles mounted on a rosette also holding the conductivity, temperature, and depth sensors. Immediately after collection, the water samples were processed as described below.
A total volume of 140 to 220 liters of seawater was filtered through 0.8-μm-pore-size filters (Isopore ATTP, 142-mm diameter; Millipore) by using a Teflon filter holder (Sartorius) and an air pressure pump (CONTEX-SOT; Verder). The prokaryotic fraction was concentrated to a final volume of 400 to 550 ml by using tangential-flow filtration (Pellicon filter cassette PTGVPPC05, 0.22-μm-pore size, stored in 1 N HCl; Millipore). The 0.22-μm-pore-size filtrate containing viruses was concentrated to a final volume of 500 to 800 ml by using a spiral-wound ultrafiltration cartridge with a molecular mass cutoff of 30 kDa (S10Y30, stored in 0.1 N NaOH; Amicon). Both ultrafiltration devices were operated by using peristaltic pumps (604S; Watson and Marlow) at a maximum pressure of 2 bars (1 bar = 105 Pa). Five to 10 liters of each sample was used to rinse the filtration units. Between samples, the filtration units were rinsed first with 1 N HCl and then with 40 liters of virus-free ultrafiltrate obtained from the second ultrafiltration step (30-kDa molecular mass cutoff). The filtration procedure was performed at the in situ temperature within 5 h after sample collection. Large water samples were needed to obtain enough viral concentrate for the viral enrichment experiments.
Each experiment consisted of four batch cultures in 10-liter glass bottles (Duran glass; Schott), soaked in 1 N HCl overnight and rinsed several times with virus-free ultrafiltrate before use. After an aliquot (100 to 150 ml) of the prokaryotic concentrate was dispensed into the culture flasks, virus-free ultrafiltrate from the same sampling site was added to a total volume of 500 ml. An aliquot (95 to 200 ml) of the viral concentrate was heated three times nearly to boiling in a microwave oven and chilled on ice for 10 min between the heating steps to inactivate the viruses (32). The viral concentrate was distributed as follows: one flask received no viruses; the second flask received one aliquot of the heat-inactivated viral concentrate; and the third and fourth flasks received one and two aliquots, respectively, of the untreated viral concentrate. Thus, each experiment consisted of two negative controls, corresponding to the treatments without virus amendment and with inactivated viruses, and two virus treatments. Finally, the culture flasks were filled with virus-free ultrafiltrate from the same sampling site to a final volume of 10 liters. The treatments are referred to here according to the initial viral abundance relative to the in situ abundance (e.g., 100%Vir-inactive indicates treatment with a microwave-treated aliquot of the viral concentrate equivalent to 100% of the in situ viral abundance, 0%Vir indicates treatment without the addition of viruses, and 100%Vir indicates treatment with an aliquot of the viral concentrate equivalent to 100% of the in situ abundance). Prokaryotic abundance at the start of the incubations ranged from 10 to 100% of the in situ prokaryotic abundance. The label used for the stations (AF1, AF2, AF3, and NS1) was also used to denote the experiments performed with water from these stations. The batch cultures were incubated in the dark at the in situ temperature for up to 96 h. Subsamples for prokaryotic and viral abundance determinations were taken from each batch culture at regular intervals (4 to 6 h). Additionally, prokaryotic cells from 1-liter subsamples of each batch culture were collected on 0.22-μm-pore-size filters (Isopore GVWP, 100-mm diameter; Millipore) every 24 h. These filters were shock frozen in liquid nitrogen and stored at −80°C until further processing for nucleic acid extraction and T-RFLP analysis.
In order to obtain an estimate of the viral impact on prokaryotes during the incubations, we assumed that increases in viral abundance typically detected in 4- to 6-h intervals in the incubations are a measure of net viral production. For each peak in the development of viral abundance, the net increase in viral abundance per time was divided by an assumed burst size of 30 and by the bacterial abundance at the start of the peak. The lysis rates expressed as a percentage of the prokaryotic standing stock lysed per hour were averaged over all detected peaks in an incubation.
Samples (3 ml) for the enumeration of prokaryotic cells and viruses were fixed with formaldehyde (final concentration, 2%) and filtered through 0.02-μm-pore-size filters (Anodisc, 25-mm diameter; Whatman). Prokaryotic cells and viral particles on the filters were stained with SYBR-Green I (Molecular Probes) and enumerated by epifluorescence microscopy by the procedure of Noble and Fuhrman (27).
Extraction of the nucleic acids from the filters was carried out as described by Winter et al. (45). Briefly, the procedure consisted of four freeze-thaw cycles (−196°C to +37°C) and subsequent treatment with lysozyme (Sigma catalog no. L-7651) and proteinase K (Fluka catalog no. 82456) in 1% sodium dodecyl sulfate. No intact cells were found after the final incubation step, as determined by epifluorescence microscopy for selected samples from all experiments, indicating complete lysis of the cells. The liquid phase was extracted once with an equal volume of a mixture of phenol, chloroform, and isoamyl alcohol (25:24:1) buffered with 10 mM Tris-HCl (pH 8.0)-1 mM disodium EDTA and with chloroform-isoamyl alcohol (24:1) (35). Ethanol precipitation overnight at −20°C was used to concentrate and clean the extracted nucleic acids. The pellet was redissolved in 100 μl of ultrapure water (Sigma catalog no. W-4502). Fifty microliters of this crude nucleic acid extract was purified further by using a QIAEX II gel extraction kit (Qiagen) as recommended by the manufacturer for DNA fragments of larger than 10 kbp. The nucleic acids were recovered in a final volume of 20 μl of elution buffer (Qiagen) and used for subsequent PCR amplification. Two microliters of this nucleic acid extract were loaded onto a 1% agarose gel to check the integrity of the DNA.
The PCR conditions and chemicals used here were the same as those described by Moeseneder et al. (23). Briefly, 1 to 2 μl of the cleaned nucleic acid extract was used as a template in a 50-μl PCR mixture. We used primer 27F, specific for Bacteria, in combination with universal primer 1492R (14) and, in a second reaction, primers 21F and 958R, specific for Archaea (8), to amplify ca. 1,480-bp and ca. 920-bp fragments of the bacterial and archaeal 16S rRNA genes, respectively. For T-RFLP analysis, forward and reverse primers were fluorescently 5′-end labeled with phosphoramidite fluorochrome 5-carboxy-fluorescein and 6-carboxy-4′,5′-dichloro-2′,7′-dimethoxyfluorescein, respectively (both from Interactiva, Ulm, Germany). After PCR amplification, excess fluorescently labeled primer was removed by ethanol precipitation and subsequent gel purification in a 1% agarose gel with TAMRA loading dye (Perkin-Elmer Applied Biosystems). Properly sized bands were cut from the gel, and the PCR fragments were recovered in 20 μl of elution buffer by using a QIAquick gel extraction kit (Qiagen). Finally, 1 μl of the purified PCR fragments and 5 μl of SmartLadder (Eurogentec, Seraing, Belgium) were run on a 1% agarose gel for quantification purposes.
The restriction digests contained 50 ng of purified PCR fragments as well as 20 U of HhaI restriction enzyme and the recommended buffer (both from Amersham Pharmacia Biotech) in a total volume of 100 μl. Restriction was performed at 37°C for 12 h to ensure complete digestion. The DNA fragments from the restriction digests were recovered in a final volume of 2 μl of ultrapure water by performing linear polyacrylamide precipitation (22). T-RFLP analysis was performed with an ABI Prism 310 automated capillary sequencer (Perkin-Elmer Applied Biosystems) as previously described (23).
Cloning of PCR-amplified fragments of the archaeal 16S rRNA gene was performed by using a pMOSBlue blunt-ended cloning kit (Amersham Pharmacia Biotech catalog no. RPN 5110) according to the manufacturer's instructions. Based on the uniqueness of the T-RFLP patterns, 11 samples from the tropical Atlantic Ocean were chosen for cloning. Inserts were obtained by PCR amplification with the archaeal primer pair as described above, except that the primers were not 5′-end labeled. We used 40 ng of insert per cloning reaction, corresponding to a molar ratio of vector to insert of 1:2.5. Thirty colonies containing an insert were randomly picked from each of the 11 transformations, resuspended in 200 μl of ultrapure water, and incubated at 95°C for 5 min. The clone libraries were screened by using 1 μl of this suspension as a template in PCRs with the fluorescently labeled archaeal primer pair and performing T-RFLP analysis as described above.
Sequencing of the clones harboring unique archaeal 16S rRNA gene fragments was performed by using an ABI Prism BigDye terminator cycle sequencing ready reaction kit (Perkin-Elmer Applied Biosystems) in a PCR consisting of 25 cycles (denaturation at 96°C for 10 s, annealing at 55°C for 5 s, and elongation at 60°C for 4 min). We sequenced from both ends of the fragments by using forward primer 21F and reverse primer 958R. Amplicons were subsequently purified by isopropanol precipitation, and sequences were obtained by using an automated sequencer (ABI Prism 310). Contiguous sequences were assembled by using the software package AutoAssembler (version 2; Perkin-Elmer Applied Biosystems). First, the nucleotide sequences were evaluated with the program Chimera_Check (version 2.7; Ribosomal Database Project at Michigan State University; http://rdp.cme.msu.edu/html/) (18) to identify possible chimeric artifacts. Further analysis involved alignment of the sequences with ARB (version 071101-7; Technical University of Munich; http://www.arb-home.de) and comparison to the 16S rRNA gene sequences in the GenBank nucleotide library by BLASTN searching (version 2.2.5; National Center for Biotechnology Information; http://www.ncbi.nlm.nih.gov/BLAST/) (1). The sequences from this study, together with 16 additional sequences retrieved from GenBank, were used to construct a phylogenetic tree with maximum likelihood (PHYLIP; version 3.5; http://evolution.genetics.washington.edu/phylip.html/) (9). The additional sequences from GenBank were Thermus aquaticus, Cenarchaeum symbiosum, AEGEAN_52, AEGEAN_58, AEGEAN_59, AEGEAN_66, AEGEAN_70, DCM65231, PVA_OTU_1, DCM875, SBAR16, ME-450_P9, 19a-27, DCM858, PVA_OTU_3, and WHARQ. The branching of the tree was evaluated by bootstrapping with the programs Seqboot and Consense (PHYLIP).
Due to the variability in the discriminatory power of the forward and reverse primers for the Bacteria and the Archaea, the results obtained with both primers were combined for each group, serving as a measure of community richness. Additionally, the frequency of appearance of each individual peak in the T-RFLP patterns, subsequently referred to as OTUs (25) and obtained at three time points during incubation in experiment NS1 and at four time points in experiments AF1, AF2, and AF3, was calculated for each treatment to identify potential effects of virus amendment on individual OTUs.
Prokaryotic production was measured as the incorporation of [3H]leucine (final concentration, 20 nM) in duplicate samples and one formaldehyde (final concentration, 2%)-killed blank sample (13). We used a conversion factor of 0.07 × 1018 cells produced per mol of leucine incorporated (33) and a carbon content of 20 fg of C per cell to convert cell production into C biomass production (15).
The concentration of organic carbon in duplicate samples was measured after high-temperature combustion (5) by using an automated analyzer (TOC-5000A; Shimadzu). Unfiltered samples (8 ml; total organic carbon [TOC]) were dispensed directly into combusted (450°C for 4 h) glass ampoules (except for North Sea seawater, which was filtered through combusted [450°C for 4 h] Whatman GF/F filters [nominal pore size, 0.7 μm]; dissolved organic carbon [DOC]) and acidified with 3 or 4 drops of concentrated HCl before the ampoules were sealed. The samples were stored frozen at −20°C until analysis.
The 16S rRNA gene sequences obtained in this study were submitted to GenBank and are available under the accession numbers listed in Table Table11.
The stations in the Atlantic Ocean differed from the North Sea station with respect to physicochemical parameters, fluorescence, and the concentration of organic carbon (Table (Table2).2). However, prokaryotic and viral abundance and prokaryotic production were similar at all of the stations. Prokaryotic abundance varied from 0.3 × 106 to 1.0 × 106 ml−1, and viral abundance varied from 5.2 × 106 to 8.8 × 106 ml−1. Prokaryotic production ranged from 0.3 to 6.5 μg of C liter−1 day−1; thus, productivity was rather low.
The development of prokaryotic abundance in experiment AF1 (initial prokaryotic abundance, 10% of in situ abundance) showed a sharp increase after the first 20 h of incubation and remained stable for the remainder of the experiment with all four treatments (Fig. (Fig.2).2). In experiments AF2 and AF3, prokaryotic abundance with all of the treatments displayed fluctuations throughout the incubation period and increased at similar rates despite the difference in initial prokaryotic abundance (AF2, 20% of in situ abundance; AF3, 100% of in situ abundance) (Fig. (Fig.2).2). In experiment NS1 (initial prokaryotic abundance, 90% of in situ abundance), prokaryotic abundance increased slowly during the first 60 h and then increased sharply with all of the treatments during the final 20 h of the incubation period (Fig. (Fig.2).2). The average prokaryotic abundance (data not shown) in the samples receiving the highest virus amendment was significantly higher than that in the other samples in all of the experiments (P value determined by paired t test, 0.0114).
The development of viral abundance with the various treatments is shown for experiment AF1 (Fig. (Fig.3).3). Similar results were obtained for the other experiments (data not shown). With all treatments and in all experiments, viral abundance displayed fluctuations over time. Overall, viral abundance decreased during the course of the incubation in experiment AF2 and increased in experiments AF3 and NS1 with all of the treatments (data not shown). Viral lysis of prokaryotes was lower (<1.5% of prokaryotic abundance h−1) with the 0%Vir and Vir-inactive treatments than with the virus treatments and was highest (6 to 28% of prokaryotic abundance h−1) in the samples receiving the highest virus amendment (Table (Table33).
In experiments AF1 and AF2, the development of bacterial richness was similar, and the same pattern was found for archaeal richness (Fig. (Fig.4).4). Initial bacterial richness with all of the treatments in experiments AF1 and AF2 decreased, on average, by about 50 and 90%, respectively, within the first 24 h of the incubation. However, in both experiments, a similar decrease in archaeal richness, on average, by about 60% compared to initial values, was detected only after 72 h (Fig. (Fig.4).4). In experiment AF3, bacterial richness declined, on average, by about 40% during the first 24 h with all of the treatments and showed little variation for the remainder of the incubation period (Fig. (Fig.4).4). In contrast to the findings in experiments AF1 and AF2, archaeal richness in experiment AF3 decreased, on average, by about 60% during the first 24 h of the incubation (Fig. (Fig.4).4). In experiment NS1, bacterial richness increased slightly within the first 43 h; however, after 67 h, it decreased by about 50% compared to values at the start of the experiment (Fig. (Fig.4).4). The major difference between experiments NS1 and experiments AF1, AF2, and AF3 was that no archaeal PCR product could be obtained at the North Sea station (a sample from experiment AF1 served as a positive control in the reaction). In general, bacterial and archaeal richness decreased during the course of the incubation with all of the treatments, and differences between samples receiving the various treatments were rather small compared to the overall decrease in richness (Fig. (Fig.44).
The addition of viruses caused distinct responses of bacterial and archaeal OTUs. As an example, portions of the T-RFLP patterns obtained with the archaeal reverse primer (958R) in experiment AF1 after 96 h of incubation are shown in Fig. Fig.5.5. Similar results were obtained for Bacteria in the same experiment and for both groups in the other experiments. Several OTUs were present in samples receiving the 0%Vir and Vir-inactive treatments but were negatively affected by the addition of intact viruses (e.g., 249 and 442 bp; Fig. Fig.5).5). A second group of OTUs appeared to be unaffected by the presence or absence of viruses and was detected in the presence of all of the treatments (e.g., 327, 363, 538, and 559; Fig. Fig.5).5). In a third group, OTUs showed variable responses to the addition of inactivated or active viruses. These OTUs appeared to be stimulated by the addition of heat-inactivated and active viruses compared to the results obtained with the 0%Vir treatment and remained detectable in the samples receiving high virus amendments (data not shown) or responded negatively only to high virus concentrations (e.g., 172, 179, 507, and 547 bp; Fig. Fig.5).5). Individual OTUs, detectable in more than one experiment, as was the case for a number of OTUs in experiments AF1, AF2, and AF3, occasionally exhibited differential responses to the treatments in different experiments. To further investigate the effects of treatments on individual OTUs, the frequency of appearance of each OTU in the presence of the treatments during the incubation was calculated. Based on this conservative comparison, which integrates treatment effects over the entire incubation period, a total of six OTUs were identified as being negatively affected by virus amendment (Table (Table44).
The nucleotide sequences of 17 unique archaeal OTUs detected by T-RFLP analysis were obtained from samples in experiments AF1, AF2, and AF3. The program Chimera_Check indicated that OTU AFRICA_3/13-21 was possibly of chimeric origin. However, the sequence was not discarded, since searching of GenBank for close relatives of this OTU retrieved sequences missing in the database used by Chimera_Check. Phylogenetic analysis confirmed the archaeal origin of the 17 nucleotide sequences obtained in this study (Fig. (Fig.6).6). The majority of the OTUs belonged to marine group II Archaea, five OTUs were members of marine group I Archaea (Crenarchaeota), and one OTU was affiliated with marine group III Archaea. The restriction fragment sizes of the OTUs varied between 77 and 920 bp and between 116 and 920 bp for the forward and reverse primers, respectively (Table (Table1).1). OTU AFRICA_1/1-2 did not contain a restriction site for HhaI, as confirmed in silico (data not shown). The OTUs could be distinguished from each other only by using the forward and reverse primers together. For instance, OTUs AFRICA_1/6-5, AFRICA_1/6-10, AFRICA_2/12-24, AFRICA_3/12-19, AFRICA_3/13-2, and AFRICA_3/19-17 could be distinguished in the patterns obtained with the forward primer but yielded only one restriction fragment (249 bp) in the patterns obtained with the reverse primer (Table (Table1).1). In contrast, OTUs AFRICA_1/1-4, AFRICA_2/12-24, and AFRICA_3/19-8 yielded fragments of 128, 249, and 138 bp, respectively, when the reverse primer was used but were indistinguishable when the forward primer was used (211 bp) (Table (Table11).
We performed batch culture experiments to investigate the possible influence of viruses on complex microbial communities in the absence of grazers and under non-steady-state conditions. Since the efficiency of recovery of prokaryotic cells and viral particles during the ultrafiltration steps varied between stations and could not be tested on the ship, the initial conditions in terms of prokaryotic and viral abundance relative to in situ abundance differed between the experiments. Additionally, due to the large volume of water needed (140 to 220 liters), it was not feasible to run the experiments in duplicate.
The use of different volumes of virus concentrate in the treatments might have increased the fraction of bioavailable organic carbon for the confined prokaryotic community compared to the fraction present under in situ conditions. However, the virus fraction of DOC typically is only a small portion of the total DOC (26). Moreover, in experiment AF1, TOC concentrations were measured at the start of the experiment with the different treatments and were lower than the TOC concentrations in situ (data not shown), suggesting that there was no contamination with organic carbon in the ultrafiltration steps. Therefore, it seems unlikely that virus amendment substantially increased the fraction of bioavailable organic carbon at the start of the incubations compared to the fraction present under in situ conditions.
Since no viruses could be detected in the <30-kDa filtrate (data not shown), the initial viral abundance in the 0%Vir treated-samples (Fig. (Fig.3)3) was the result of viruses added with the prokaryotic concentrate. A microwave treatment was used to inactivate the viruses (32), but this treatment did not physically disintegrate all of the viruses, as indicated by the higher viral abundance seen with the Vir-inactive treatments than with the 0%Vir treatments at the start of the experiment (Fig. (Fig.3).3). Due to the inability to differentiate between inactivated viral particles and intact viruses by epifluorescence microscopy, the actual number of intact viruses in the Vir-inactive treatments probably was grossly overestimated in the initial phase of the incubation. However, within the first ca. 30 h of the experiments, viral abundance in the Vir-inactive-treated samples decreased to values similar to those in the 0%Vir-treated samples, likely due to the decay of the inactivated viruses. Furthermore, viral lysis of bacterial cells already infected at the start of the incubations and the potential induction of lysogenic cells could have contributed to viral numbers in all of the treatments.
A number of studies showed that the quantitative interpretation of the results of fingerprinting techniques based on PCR amplification, such as T-RFLP analysis, potentially is biased (7, 31, 41). Thus, only the presence or absence of OTUs was used for subsequent numerical analysis. Due to the variation in the discriminatory power of the forward and reverse primers for Bacteria and Archaea, we found it necessary to combine the results of both primers for each group. Therefore, the results obtained by this approach represent a relative measure of community richness (Fig. (Fig.44).
To prove the archaeal origin of the PCR products obtained from the archaeal primers, cloning and sequencing were performed. The archaeal origin of 17 unique OTUs obtained directly from the experiments at different time points was confirmed by phylogenetic analysis (Fig. (Fig.6).6). Although not all of the detected archaeal OTUs could be identified by the cloning procedure, we are confident that the T-RFLP patterns obtained with the archaeal primers reflected the development of the archaeal assemblages, since all 330 screened clones were of archaeal origin.
The major loss factor for prokaryotic cells in the experiments likely was viral lysis, since grazers were excluded by filtration. Although the initial prokaryotic abundance in experiment AF1 was similar to that in experiment AF2 (AF1, 0.11 × 106 ml−1, equivalent to 10% of the in situ abundance; and AF2, 0.16 × 106 ml−1, equivalent to 20% of the in situ abundance), the prokaryotic abundance increased sharply after a lag phase of ca. 20 h in experiment AF1 but increased at a lower rate in experiment AF2 (Fig. (Fig.2).2). This difference cannot be explained by the higher virus amendments used in experiment AF1 (100 and 200% of in situ viral abundance) than in experiment AF2 (35 and 55% of in situ viral abundance), since the 0%Vir treatment led to the same sharp increase (Fig. (Fig.2).2). It is possible that prokaryotic abundance at station AF1 was grazer controlled and that, after elimination of the grazing pressure by filtration, prokaryotic cells responded with elevated net growth. Station NS1 (13 December 2000) differed in many aspects from the stations in the tropical Atlantic Ocean (Table (Table2).2). The winter situation in the North Sea is characterized by lower temperatures and lower biological activity than in the summer months (17). These unfavorable conditions could explain the unusually long lag phase of ca. 60 h before a sharp increase in prokaryotic abundance was detectable with all of the treatments in experiment NS1 (Fig. (Fig.22).
The development of viral abundance in our experiments was a consequence of viral lysis and decay. The strong fluctuations of viral abundance in the experiments (Fig. (Fig.3)3) suggest that viral lysis was changing dynamically over time, even in the samples with strongly reduced viral abundance (0%Vir and Vir-inactive treatments). Among the factors which potentially influenced the development of viral abundance in our experiments were changes in prokaryotic community composition during the incubations (Fig. (Fig.4;4; see below also), the release of virucidal organic matter during viral lysis, prophage induction, and changes in prokaryotic activity.
Proctor and Fuhrman (32) reported that the addition of viruses to incubations of unfiltered seawater (including grazers) reduced prokaryotic abundance by about 25% within 24 h compared to the results for controls without virus amendments. However, in experiments without grazers, the addition of viruses resulted in increases in bacterial abundance (28). This result is similar to the findings of this study, where the average prokaryotic abundance was highest in the presence of the highest virus amendments. In the absence of grazers, the stimulation of prokaryotic growth likely was caused by the increased release of products of viral lysis in the samples with a high viral abundance (28).
Generally, confinement had a strong influence on the prokaryotic communities in our incubations and resulted in dramatic decreases in bacterial and archaeal community richness. Such confinement effects were reported previously for Bacteria but not for archaeal communities and can be more important than treatment effects (20). The removal of phytoplankton and grazers likely contributed to the changes in prokaryotic community composition as well (40).
Bacterial and archaeal communities appeared to react at different times in the incubations. In experiments AF1 and AF2, confinement resulted in decreased bacterial community richness already by 24 h, while the strongest effect of confinement on archaeal community richness was detected in these experiments only after 72 h (Fig. (Fig.4).4). Higher rates of growth of Bacteria than of Archaea in surface waters could explain this pattern by causing a faster response of the bacterial community to changing conditions. Changes in community composition as a result of prokaryotic growth likely occurred in experiment NS1. The unfavorable conditions during winter resulted in a long lag phase with few initial changes in the bacterial community composition. The decrease in bacterial community richness after 67 h coincided with the dramatic increase in prokaryotic abundance, indicating the development of prokaryotic populations which were adapted to the specific experimental conditions (Fig. (Fig.22 and and4).4). The effects of confinement on the bacterial and archaeal communities in experiment AF3 differed from those in experiments AF1 and AF2. While bacterial community richness decreased only slightly compared to initial conditions (to ca. 90%), archaeal community richness decreased by about 60% already by 24 h (Fig. (Fig.4).4). The water used to perform experiments AF1 and AF2 was sampled below the thermocline, whereas the water for experiment AF3 was sampled above the thermocline (data not shown). If indeed differences in growth rates were the cause for the different responses of Bacteria and Archaea to confinement, then our results suggest that Archaea might have been more active in the experiments with water sampled from above the thermocline than from below the thermocline. Alternatively, dilution of the prokaryotes in the experiments, which ranged from 10 to 20% of in situ abundance in experiments AF1 and AF2 and from 90 to 100% of in situ abundance in experiments AF3 and NS1, could have resulted in changes in community composition. If this notion holds true for our experiments, Archaea would react most rapidly to confinement at low dilution rates and Bacteria would do so at high dilution rates (Fig. (Fig.44).
Thingstad and Lignell (42) suggested that viruses selectively infect the most abundant members of the prokaryotic community and therefore might be a driving force in maintaining prokaryotic diversity. However, in our incubations, differences in bacterial and archaeal richness between the negative controls and the virus-treated samples were small compared to the effects of confinement and were not consistent over time (Fig. (Fig.4).4). Essentially the same results were obtained when similarity values calculated from the T-RFLP patterns were used (data not shown).
A prediction of the “killing the winner” hypothesis is that the species composition of viral communities should reflect the composition of the host community. Thus, the most abundant viral populations should be infective for the most abundant host populations. It is conceivable that the viral communities in our incubations were not sufficiently adapted to the changing prokaryotic communities; hence, the overall effect of virus amendment on the level of prokaryotic communities as detected by T-RFLP analysis might have been rather small. Another possible explanation for the small differences between the negative controls and the virus-treated samples could be that most members of the prokaryotic communities were resistant to viral infection. Resistance could be caused by a reduction in the number and/or a change in the structure of virus receptors at the surface of the cells or by lysogenic viruses (46). Previous studies demonstrated the ability of prokaryotes to quickly acquire resistance to cooccurring viruses, especially under favorable growth conditions (16, 43). At times when prokaryotes entered logarithmic growth (Fig. (Fig.2),2), resistance to viral infection might have been established in some prokaryotic populations. However, a substantial fraction of the prokaryotic cells receiving the virus treatments in our experiments was infected with viruses (Table (Table3).3). Thus, resistance to viral infection and the changing prokaryotic communities in the incubations cannot explain the small differences in community composition between the samples. Middelboe (21) demonstrated that viruses can influence the ratio of sensitive to resistant clones in a prokaryotic population. Therefore, if viruses influence the clonal composition rather than the community composition of prokaryotes, then the effect of viruses might not be detectable by 16S rRNA gene-based community fingerprinting.
Although the differences seen with the treatments were small and variable over time at the level of archaeal and bacterial communities (Fig. (Fig.4),4), individual OTUs in the T-RFLP patterns of Archaea and Bacteria showed distinct responses to the presence of infectious viruses (Fig. (Fig.5).5). The negative response of some of the OTUs to virus amendment might be explained best by viral lysis and would be consistent with the finding that viral infection increased with virus amendment (Table (Table3).3). Another group of OTUs showed no response to virus amendment, possibly as a consequence of resistance of these OTUs to viral infection. Furthermore, the number of lytic viruses infecting these OTUs might have been too low to cause detectable differences in the treatment groups.
It has been argued that the noninfected prokaryotic community can benefit from lysis products (21); consequently, one might expect that some prokaryotic populations are stimulated by the presence of viruses. However, we found no case in which an archaeal or bacterial OTU was detected more frequently in the presence of the virus treatments than in the presence of the 0%Vir and Vir-inactive treatments (Table (Table44 and Fig. Fig.5).5). OTUs which were not detected or were hardly detected with the 0%Vir treatments were found with the Vir-inactive treatments, indicating the presence of a stimulatory factor in the virus concentrate (e.g., 172, 179, and 547 bp; Fig. Fig.5).5). When only the Vir-inactive and virus treatments were compared, these OTUs either were not affected by the infectious viruses or responded negatively only at a high viral abundance. The nature of the stimulatory factor is not known, but it does not seem to be temperature sensitive, since its effect could be detected when the Vir-inactive treatments (microwave-treated virus concentrate) and virus treatments (untreated virus concentrate) were used.
By comparing the frequencies of appearance of the OTUs in the T-RFLP patterns obtained with the different treatments, we found that six OTUs were more frequently detected in the negative controls throughout the entire incubation period (Table (Table4).4). In a comparison between the Vir-inactive treatments and the two virus treatments, we found that 44 OTUs (Archaea and Bacteria) were more frequently detected in the presence of the Vir-inactive treatments (data not shown). These results indicate that there is a negative effect of viruses on some OTUs and that viral infection and lysis affect a few abundant hosts, notions that are compatible with the “killing the winner” hypothesis.
On the level of archaeal and bacterial community richness, the effects of virus amendment were rather small compared to the strong influence of confinement and, moreover, were variable over time. Differences in the responses of individual OTUs to virus amendment were detected for both groups, Archaea and Bacteria. These results are augmented by the finding that six OTUs were detected more frequently in the negative controls throughout the incubation period than in the samples receiving virus treatments. This negative response to the addition of intact viruses probably was due to viral lysis of Archaea and Bacteria. We conclude from this study that marine Archaea respond to confinement and to virus amendment. This conclusion further substantiates the emerging notions that marine pelagic Archaea are metabolically active in the surface layer of the sea and that viral lysis potentially influences the community composition of pelagic Archaea.
We thank the captain and the crew of RV Pelagia for their support at sea. We thank Geert-Jan Brummer and Martien Baars for their organizational skills as cruise leaders and willingness to accommodate our large sampling volumes during the cruises in the Atlantic Ocean and the North Sea. We acknowledge Txetxu Arrieta for sharing the prokaryotic production measurements and Geraldine Kramer for analyzing TOC and DOC. We acknowledge two anonymous reviewers for improving the manuscript with their comments.
Funding for this study was provided by the Dutch Science Foundation (NWO-ALW grant 809.33.004 and grant 835.20.004).