Virus Succession Observed during an Emiliania huxleyi Bloom

Abstract

Denaturing gradient gel electrophoresis was used as a molecular tool to determine the diversity and to monitor population dynamics of viruses that infect the globally important coccolithophorid Emiliania huxleyi. We exploited variations in the major capsid protein gene from E. huxleyi-specific viruses to monitor their genetic diversity during an E. huxleyi bloom in a mesocosm experiment off western Norway. We reveal that, despite the presence of several virus genotypes at the start of an E. huxleyi bloom, only a few virus genotypes eventually go on to kill the bloom.

Blooms of the unicellular marine phytoplankton Emiliania huxleyi are known to affect the oceanic carbon pump (9) and climate (6). Vast coastal and midocean populations of this organism, which are readily visualized by satellite imagery due to their reflective calcium carbonate coccoliths, often disappear suddenly, causing substantial fluxes of calcite to the seabed (29) and cloud-forming dimethyl sulfide to the atmosphere (14). Until recently, the mechanisms of E. huxleyi bloom disintegration were poorly understood, but it is now accepted that viruses are intrinsically linked to these sudden crashes (13, 26). Viruses are ubiquitous in the marine environment, and they exert significant control over bacteria and phytoplankton populations, influencing diversity, nutrient flow, and biogeochemical cycling (11, 28).

Over the last decade, significant advances have been made in understanding the dynamics of viruses and their effects on marine eukaryotic phytoplankton communities. The majority of the work in this area has dealt with viruses that infect Micromonas pusilla. These viruses are widespread, genetically diverse, dynamic, and highly virulent (7, 8, 22). Other researchers have looked at viruses that infect Heterosigma akashiwo. These viruses were shown to play an important role in determining the clonal composition and maintaining the clonal diversity of H. akashiwo populations (23). However, the understanding of the dynamics and effects of viruses on E. huxleyi is still in its infancy. Several studies have shown that virus numbers increase dramatically following the demise of E. huxleyi-dominated blooms (1-4, 13). E. huxleyi viruses have been isolated from such blooms (5, 26) and have recently been assigned to a new genus, Coccolithovirus, based principally on the phylogeny of their DNA polymerase genes (21). Coccolithoviruses belong to the family Phycodnaviridae, a diverse family of icosahedral double-stranded DNA viruses that infect eukaryotic algae (25).

Recently, we reported the cloning and sequencing of amplified segments of the major capsid protein (MCP) gene from viruses that infect E. huxleyi (21). Significant sequence variation was observed between virus strains, revealing the potential of using this gene as a genetic tool to differentiate viral genotypes in natural communities (21). Denaturing gradient gel electrophoresis (DGGE) is a powerful yet simple technique that can separate genotypes amplified from mixtures of sequences in natural samples and is widely used in microbial ecology (18). DGGE has recently been successfully used to determine the genetic diversity within a wide range of naturally occurring algal viruses (22). In this study, we used PCR and DGGE to differentiate known E. huxleyi virus isolates based on the sequence variation in the MCP gene. This technique was used to monitor the progression of the E. huxleyi virus community during an E. huxleyi-induced bloom in a mesocosm experiment conducted in Norway during June 2000.

MATERIALS AND METHODS

Study site and flow cytometric analysis.

The mesocosm experiment was carried out at the Marine Biological Field Station adjacent to Raunefjorden, 20 km south of Bergen, Norway. The experimental design and the flow cytometric analysis of E. huxleyi and its natural viral communities from enclosure 1 (a nutrient enrichment regime with a 15:1 N/P ratio [1.5 μM NaNO₃-0.1 μM KH₂PO₄]) are described by Jacquet et al. (13).

Collection and preparation of samples for DGGE.

Samples were collected daily at 0900 h throughout the mesocosm experiment. Prior to concentration of the virus fraction, 1-liter volumes of seawater were gently filtered through 0.45-μm-pore-size Supor-450 47-mm-diameter filters (PALL Corp.). The resulting filtrates were concentrated 1,000-fold in two steps, first by tangential-flow ultrafiltration using 50-kDa-cutoff Vivaflow 50 units (Sartorius) to a volume of 20 ml. The viruses were further concentrated in a second step by centrifugal filtration using Macrosep paddle filters (PALL Corp.) to a final volume of 1 ml. The concentrates were stored at −80°C prior to use.

PCR and DGGE.

PCR of the E. huxleyi virus isolates and the concentrated virus samples were conducted in two stages. One microliter of each of the clonal E. huxleyi virus lysates (21) and 1,000-fold virus concentrates was added to a 49-μl first-stage PCR mixture containing 1 U of Taq DNA polymerase (Promega), 1× PCR buffer (Promega), 0.25 mM deoxynucleoside triphosphates (dNTPs), 1 mM MgCl₂, and 10 pmol of each E. huxleyi virus-specific primer, MCP-F and MCP-R (21). Viral lysates of M. pusilla (M. pusilla virus PB5; courtesy of Steven Short, University of British Columbia, Vancouver, Canada) and Phaeocystis globosa (P. globosa virus 102, a partly characterized strain isolated from the English Channel) and mixtures with no template added served as controls. The PCR was conducted with a PTC-100 cycler (MJ Research) as described by Schroeder et al. (21). A 2.5-μl subsample from the first-stage PCR was treated with ExoSAP-IT (U.S. Biochemical Corp.) to remove unutilized dNTPs and primers.

A second-stage PCR was conducted to amplify the variable region within the MCP genes by adding each of the ExoSAP-IT-treated subsamples to a 46.5-μl PCR mixture containing 1 U of Taq DNA polymerase (Promega), 1× PCR buffer (Promega), 0.25 mM dNTPs, 1 mM MgCl₂, and 10 pmol of each E. huxleyi virus-specific primer, MCP-F2 and MCP-R2. The two oligomers, MCP-F2 (5′-CGC CCG GGG CGC GCC CCG GGC GGG GCG GGG GCA CGG GGG GTT CGC GCT CGA GTC GAT C-3′; the underlined sequence represents the GC clamp) and MCP-R2 (5′-GAC CTT TAG GCC AGG GAG-3′), were designed based on the alignment described by Schroeder et al. (21). The PCR was conducted with a PTC-100 cycler (MJ Research) with an initial denaturing step of 95°C (3 min) followed by 34 cycles of denaturing at 95°C (30 s), annealing at 55°C (60 s), and extension at 74°C (90 s), and finally one cycle of denaturing at 95°C (30 s), annealing at 55°C (300 s), and extension at 74°C (300 s). The PCR products were resolved in a 1.5% (wt/vol) agarose gel in Tris-borate-EDTA buffer (20). The gels were stained with ethidium bromide, visualized on a UV transilluminator, and photographed with the Gel Doc 2000 system (Bio-Rad). DGGE of second-stage PCR products was conducted using 30 to 60% linear denaturing gradient 10% polyacrylamide gels, where 100% denaturant is a mixture of 7 M urea and 40% deionized formamide. Ten microliters of PCR product was loaded into wells with 10 μl of 2× gel loading dye (70% [vol/vol] glycerol, 0.05% [vol/vol] bromophenol blue, and 0.05% [vol/vol] xylene cyanol). Electrophoresis was carried out for 3.5 h in 1× TAE (20) at 200 V and a constant temperature of 60°C using the D-code electrophoresis system (Bio-Rad). The gels were stained in a 0.1× SYBR Gold (Molecular Probes) solution for 20 min, and the visualized bands were photographed as described above.

Sequencing and sequence analysis.

Individual bands were excised, reamplified, verified by DGGE, and sequenced directly. Twenty-seven excised gel fragments representing 27 DGGE bands were placed in sterile microcentrifuge tubes with 150 μl of 1× Tris-EDTA buffer (20) and heated to 95°C for 5 min. Two microliters of the resultant eluant was PCR amplified as described for the second-stage reactions. Ten microliters was loaded on a DGGE gel as described above to verify purity and confirm the correct mobility of the excised bands. In addition, 7 μl was used for direct sequencing of these bands using the fluorescently labeled oligomers MCP-F2_700 (5′-TTC GCG CTC GAG TCG ATC-3′) and MCP-R2_800 (5′-GAC CTT TAG GCC AGG GAG-3′) as sequencing primers. Sequencing was performed with the SequiTherm EXCEL II DNA Sequencing Kit-LC (Epicentre Technologies) with a LI-COR automated DNA sequencer (DNA Analyzer; Gene Reader 4200). The sequenced data were analyzed using e-Seq release 1.1 software (LI-COR). Homology searches were carried out using the BLAST algorithm provided by the Internet service of the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/).

Phylogenetic analysis.

DNA and protein sequences of MCP genes were obtained from the GenBank database. The DNA and amino acid sequences of the conserved region I (24) within the MCP genes were aligned using ClustalW (http://www.clustalw.genome.ad.jp/). Phylogenetic trees of the amino acid alignment were constructed using the various programs in PHYLIP (Phylogeny Inference Package) version 3.57c (10), and the robustness of the alignments was tested with the bootstrapping option (SeqBoot). Genetic distances, applicable for distance matrix phylogenetic inference, were calculated using the Protdist program in the PHYLIP package. Phylogenetic inferences based on the distance matrix (Neighbor) and parsimony (Protpars) algorithms were applied to the alignments. In both trees, the best tree or the majority-rule consensus tree was selected using the consensus program (Consense). The trees were visualized and drawn using TREEVIEW software version 2.1 (http://taxonomy.zoology.gla.za).

The abbreviations and GenBank accession numbers of the virus sequences used in the phylogenetic analysis are as follows: Invertebrate iridescent virus, IIV-6 (NP_149737), IIV-22 (P22166), and IIV-1 (P18162); Lymphocystis disease virus, LCDV-1 (NP_044812); Frog virus, FRGV3 (AAB01722); Paramecium bursaria Chlorella virus CVT2 (AB006978), CVK2 (AB011506), G1 (AF076921), and 1 (M85052); Ectocarpus siliculosus virus V-1 (NP_077601); E. huxleyi virus 84 (AF453849, 86 (AF453848), 88 (AF453850), 163 (AF453851), 201 (AF453857), 202 (AF453856), 203 (AF453855), 205 (AF453847), 207 (AF453853), and 208 (AF453852); and E. huxleyi virus operational taxonomic unit 1 (EhVOTU1) (AY144374), EhVOTU2 (AY144375), EhVOTU3 (AY144376), EhVOTU4 (AY144377), EhVOTU5 (AY144378), EhVOTU6 (AY144379), EhVOTU7 (AY144380), and EhVOTU8 (AY144374).

RESULTS AND DISCUSSION

The aim of this study was to determine whether DGGE could be successfully employed to exploit the variations previously observed in the MCP gene of E. huxleyi virus isolates (21), and consequently, to genetically fingerprint E. huxleyi virus communities in natural samples.

DGGE of E. huxleyi virus isolates.

MCP gene fragments were amplified, yielding products of 284 bp in the first-stage PCRs (data not shown) and 175 bp in the second-stage PCRs (Fig. 1A) from all 10 E. huxleyi virus isolates. DGGE analysis of the second-stage PCR fragments resulted in the differentiation of 7 out of the 10 E. huxleyi virus isolates (Fig. 1B). Isolates E. huxleyi virus 201, E. huxleyi virus 202, and E. huxleyi virus 84 could not be differentiated from E. huxleyi virus 205, E. huxleyi virus 207, and E. huxleyi virus 88, respectively. The denaturing gradient used in this analysis was too broad to separate the individuals that differed by 1 bp (21). However, a significant proportion of the isolates could be differentiated. Hence, we were able to demonstrate that this technique could be applied as a tool to genetically fingerprint E. huxleyi virus communities as long as a significant number of bands are sequenced to confirm their identities.

FIG. 1.

Images of gel electrophoresis of PCR fragments amplified in second-stage PCR from a range of virus isolates (21). (A) Agarose gel. Lane 1, lambda DNA digested with PstI; lane 2, E. huxleyi virus 84; lane 3, E. huxleyi virus 86; lane 4, E. huxleyi virus 88; lane 5, E. huxleyi virus 163; lane 6, E. huxleyi virus 201; lane 7, E. huxleyi virus 202; lane 8, E. huxleyi virus 203; lane 9, E. huxleyi virus 205; lane 10, E. huxleyi virus 207; lane 11, E. huxleyi virus 208; lane 12, M. pusilla virus PB5; lane 13, P. globosa virus 102; lane 14, no DNA. (B) DGGE gel. Lane 1, E. huxleyi virus (EhV) 84; lane 2, E. huxleyi virus 86; lane 3, E. huxleyi virus 88; lane 4, E. huxleyi virus 163; lane 5, E. huxleyi virus 201; lane 6, E. huxleyi virus 202; lane 7, E. huxleyi virus 203; lane 8, E. huxleyi virus 205; lane 9, E. huxleyi virus 207; lane 10, E. huxleyi virus 208; and lane 11, all E. huxleyi virus isolates.

Mesocosm study.

The progression of different microbial populations, including E. huxleyi and its associated viruses (E. huxleyi viruses), was monitored by flow cytometry during the mesocosm experiment (13). Flow cytometry has been routinely used for the analysis of different phytoplankton populations (19), bacterial communities (17), and virus communities (15, 26, 27) in marine samples and is commonly accepted as a reference technique in oceanography (15).

E. huxleyi numbers increased by 2 orders of magnitude, from 10³ to 10⁵ cells · ml⁻¹, between 9 and 18 June 2000 prior to the onset of the bloom crash on 19 June 2000 (Fig. 2A). The E. huxleyi numbers returned to prebloom levels of 3 × 10³ cells · ml⁻¹ at the end of the study (24 June 2000). There was an increase of ∼2 orders of magnitude in E. huxleyi virus numbers, from 4.7 × 10⁵ to 3.5 × 10⁷ E. huxleyi viruses · ml⁻¹, during the collapse of the E. huxleyi bloom, suggesting it was terminated by viral infection (Fig. 2A).

FIG. 2.

(A) Time series obtained for E. huxleyi (⧫) and E. huxleyi virus (EhV) concentrations (▪) using flow cytometry between 6 and 24 June 2000 in enclosure 1. B, D, and A refer to before, during, and after the bloom, respectively. (B) DGGE gel of PCR fragments amplified in second-stage PCR from samples collected during the mesocosm experiment. The symbols indicate the 27 bands excised from the gel, and identical symbols indicate identical nucleotide sequences. The sequences represented by solid triangles, open circles, solid circles, solid squares, open triangles, small open square, large open squares, and open diamond were designated E. huxleyi operational taxonomic unit numbers 1 to 8 (EhVOTU1 to -8), respectively. Solid circles and solid triangles, two DGGE band profiles observed before, during, and after the bloom; open symbols, five DGGE band profiles observed before the bloom; solid squares, one band profile observed during and after the bloom. S, standards (same as Fig. 1B, lane 11). The format for dates is day/month/year.

Flow cytometry was used to monitor the total abundance of E. huxleyi viruses during the E. huxleyi bloom, and it revealed a classic lytic virus response to a susceptible host population. However, DGGE analysis of the E. huxleyi virus community during this bloom revealed a more dynamic sequence of events. In the 8 days prior to the onset of the bloom, a genetically diverse community of E. huxleyi viruses was observed, with a different DGGE profile each day (Fig. 2B; 6 to 13 June 2000). A succession from this diverse and variable E. huxleyi virus community to a more stable E. huxleyi virus community, with the same DGGE profile each day, was observed from the onset of the E. huxleyi bloom (Fig. 2B; 14 to 24 June 2000).

The authenticity of the DGGE bands was confirmed by excising 27 bands (Fig. 2B), verifying their purity (data not shown), and finally sequencing them. A BLAST search of the GenBank database revealed that these sequences showed high identity to MCP gene sequences (data not shown). Phylogenetic analysis clearly showed that they belonged to the family Phycodnaviridae, and they clustered together with E. huxleyi virus isolates (Fig. 3). This cluster exhibited high bootstrap values with both the neighbor-joining and parsimony phylogeny inferences, indicative of a strong association. The evidence collectively confirms that the E. huxleyi virus-specific PCR primers used in this study amplified only E. huxleyi virus sequences.

FIG. 3.

Phylogenetic inference based on a distance matrix algorithm between conserved domain I (24) within MCP among a few members of the family Phycodnaviridae and a few members of a distantly related family, Iridoviridae. The numbers at the nodes indicate bootstrap values retrieved from 100 replicates for both the parsimony and neighbor-joining analyses. The bar represents 1 base substitution per 10 amino acids. Abbreviations: EhV, E. huxleyi virus; PBCV, Paramecium bursaria Chlorella virus; IIV, Invertebrate iridescent virus; LCDV, Lymphocystis disease virus; FRGV; Frog virus; EsV, Ectocarpus siliculosus virus.

The eight sequences obtained from the excised DGGE band profiles differed from the sequences obtained from the isolates collected from the English Channel off the coast of Plymouth, United Kingdom (Fig. 4). In contrast, the DGGE bands that migrated like E. huxleyi virus 163 had sequences identical to that of E. huxleyi virus 163 (Fig. 4). This was not surprising, since E. huxleyi virus 163 was isolated from this mesocosm experiment (21). Two DGGE band profiles were observed before, during, and after the bloom. However, five DGGE band profiles were observed before the bloom, while one was observed during and after the bloom (Fig. 2B). Therefore, despite the presence of a diverse viral community prior to the development of the E. huxleyi bloom, only three dominant DGGE bands were observed during and after the bloom, suggesting that they were responsible for its collapse.

FIG. 4.

Multiple sequence alignment of conserved domain I (24) within MCP among members of the genus Coccolithovirus (21). EhV, E. huxleyi virus. Solid triangles, DGGE bands that migrated like E. huxleyi virus 163; see the legend to Fig. 2 for other symbols. Conserved bases are identified as a dash underneath the corresponding base from EhV203.

Similar DGGE profiles (unpublished data) were obtained in samples collected from mesocosms that were either P or N depleted (described by Jacquet et al. [13]). Interestingly, nutrient availability was considered important, as it was thought to affect virus-host interactions (1, 13). However, contradictory evidence exists concerning the roles of P and N limitations in these virus-host interactions. Bratbak et al. (1) concluded that viruses might be more sensitive to P than to N limitation, because viruses have high nucleic acid-to-protein ratios. Jacquet et al. (13) reported that E. huxleyi virus production was delayed in N-depleted enclosures. However, our data suggest that nutrient availability had no effect on the molecular succession dynamics of E. huxleyi viruses, i.e., the same viruses were responsible for the termination of the bloom irrespective of the growth conditions. If the nutrient conditions had affected host-virus interactions, then one would expect that a different host(s) would be selected to succeed under these conditions. Hence, a different virus profile is expected, i.e., either an additional band(s), the disappearance of a band(s), or a combination of both. If the same host(s) is successful under all conditions, then the same virus profile is expected. This is what was observed. Hence, the nutrient conditions did not appear to play a role in the host-virus interactions. In addition, if the viral infection cycle were susceptible to the conditions tested, then either the growth rate of the host(s) would be affected or no infection would occur. Neither was the case (13). A more likely hypothesis is that nutrient availability may simply affect the growth rate of the host and consequently the severity of infection. A similar conclusion was reached by Jacquet et al. (13).

In summary, it is clear from the DGGE profiles produced prior to the onset of the bloom that a variety of virus genotypes is present in the water column of the fjord. We hypothesize that these viruses are remnants of previous bloom lysis events. The enrichments of the unfiltered seawater, which was collected from the fjord for the mesocosm experiment, resulted in the formation of an E. huxleyi bloom. Prior to the development of the bloom, various virus isolates appeared and disappeared daily. A possible explanation for this might be that as the indigenous host populations become metabolically active, they are infected by their respective virus(es) and hence are “removed” from the water column. As a consequence, the virus(es) might reappear one or a few days later depending on the success of the infection(s). EhVOTU3 is a good example of this. We see that this isolate is present the first 3 days of the experiment (6 to 8 June 2000), disappears for 3 days (9 to 11 June 2000), reappears for 1 day only to disappear again (12 and 13 June 2000, respectively), and finally reappears throughout the progression and subsequent collapse of the bloom (14 to 24 June 2000) (Fig. 2B). An alternative explanation for this phenomenon could be that it is a direct consequence of the threshold concentration and the relative template abundance in the water samples. This could explain the absence of EhVOTU4 prior to the bloom (Fig. 2B, 6 to 13 June 2000).

As the bloom developed, three dominant virus isolates were detected. This observation suggests that these three isolates infected the bloom-forming host population. The makeup of this host population is unknown, i.e., we cannot be sure whether the bloom was the product of one or many E. huxleyi strains. Host range analysis revealed that certain virus isolates (separated based on differences in their MCPs) can infect a number of host strains, while others were more host specific (21). Therefore, the three different virus isolates represented by the three DGGE bands could infect the one E. huxleyi strain that caused the bloom. Similarly, each virus isolate could infect its respective host strain, and thus the bloom could be the product of at least three different E. huxleyi strains. Ultimately, we cannot be certain unless a genetic marker is found to answer this question. The tools presently available for the enumeration of E. huxleyi diversity are randomly amplified polymorphic DNA (16) and microsatellites (12). However, both of these techniques would be extremely difficult or even impossible to use for evaluating community structure, as each strain is represented by multiple banding patterns.

The inability to determine the host composition of the bloom does not negate the significance of our observations. We clearly demonstrated the usefulness of the MCP gene in conjunction with DGGE to distinguish between different virus isolates and hence to give an idea of the diversity of isolates present in a bloom dynamic. However, a note of caution is warranted. The variability in the MCP gene as revealed by DGGE is not necessarily an absolute measure of virus functional diversity. However, we were able to show that a few base pair changes in the MCP gene can be correlated with different host range profiles and hence are significant enough to serve as markers for diversity (21).

In conclusion, we were able to show for the first time that changes occurred at the genotypic level within the E. huxleyi virus community throughout the progression of an E. huxleyi bloom. This observation highlights what marine virologists had suspected for a long time—that we were seriously underestimating the level of genotypic diversity in marine viral communities (11). We believe that our understanding of the extent of virus diversity in natural communities and their effects on their hosts needs to be revised. These new data also raise further questions regarding host susceptibility, particularly since genetic diversity has been observed within E. huxleyi bloom populations (16).