High Diversity of the Viral Community from an Antarctic Lake

See allHide authors and affiliations

Science  06 Nov 2009:
Vol. 326, Issue 5954, pp. 858-861
DOI: 10.1126/science.1179287


Viruses are the most abundant biological entities and can control microbial communities, but their identity in terrestrial and freshwater Antarctic ecosystems is unknown. The genetic structure of an Antarctic lake viral community revealed unexpected genetic richness distributed across the highest number of viral families that have been found to date in aquatic viral metagenomes. In contrast to other known aquatic viromes, which are dominated by bacteriophage sequences, this Antarctic virus assemblage had a large proportion of sequences related to eukaryotic viruses, including phycodnaviruses and single-stranded DNA (ssDNA) viruses not previously identified in aquatic environments. We also observed that the transition from an ice-covered lake in spring to an open-water lake in summer led to a change from a ssDNA– to a double-stranded DNA–virus-dominated assemblage, possibly reflecting a seasonal shift in host organisms.

Antarctica has been geographically isolated for millions of years, harboring some of the last pristine ecosystem on Earth (13). Liquid water is represented in Antarctica by temporally or perennially ice-covered lakes and streams (4). Life in Antarctic freshwater ecosystems is dominated by microorganisms that are adapted to extreme environmental conditions: low temperatures, low nutrient levels, and months of nearly complete darkness during winter (1, 57).

Viruses are important factors regulating the structure of microbial communities. Virus-mediated killing is implicated in algal bloom control, is considered as important as protozoan grazing in bacterial mortality in oceans, and thus influences aquatic food-web interactions to affect global geochemical cycles (8, 9). Moreover, microbial genetic diversity is shaped by virus-mediated gene transfer and virus host range (10, 11). Until recently, our knowledge of virus diversity has been limited by the difficulty of culturing their hosts, but now new high-throughput methods for sequencing uncultured viral communities (viromes) have been uncovering their marked genetic richness (1113).

Byers Peninsula (Livingston Island, Antarctica) contains several freshwater lakes in its inland plateau, whose environmental value led to its designation as an Antarctic Specially Protected Area (14) (fig. S1). These lakes are not substantially influenced by maritime fauna (penguins and seals), although small groups of birds such as skuas or terns sporadically visit them. Oligotrophic Antarctic lacustrine habitats such as these ecosystems receive little nutrient input and are dominated by microorganisms forming truncated food webs (5, 7). However, in spite of the relative simplicity of their ecosystems, these lakes harbor a wide range of bacteria, some algae (including mixotrophs), rotifers and protozoa, and a macrozooplankton community including the copepod Boeckella poppei and the fairy shrimp Branchinecta gainii. The benthos is dominated by an aquatic moss (Drepanocladus longifolius) and metazoans including the chironomid Parochlus steinenii and the oligochaete Lumbricillus healyae (14). Aquatic organisms populating these unique habitats experience major seasonal shifts as these lakes remain ice covered for at least 9 months per year with dim light conditions and stable inverse stratification of the water column, whereas in summer the lakes are exposed to intense ultraviolet radiation and poor water stratification owing to ice melt.

Microorganism communities are regulated by the availability of nutrients (bottom-up control), the activity of predators (top-down control), and viral lysis. The role of the top-down control in Antarctic lakes appears to be limited and, thus, the ecological role of viruses in these extreme environments may be greater than in temperate lakes (5, 7). Consistently, Antarctic lakes display the highest rate of visibly phage-infected bacteria reported (15); however, identification of Antarctic viruses is limited to some electron micrographs (16, 17). To profile the virus diversity in Antarctic ecosystems and investigate seasonal variations, water was collected from Lake Limnopolar (Byers Peninsula) before (spring) and after (summer) the ice cover melted (18). Electron microscopy (EM) revealed a large diversity of viral morphologies (Fig. 1A), and quantitative analysis showed that, in contrast to previous studies in aquatic environments, viral particles <30 nm in diameter constituted the most abundant viral morphotype in spring (fig. S2). In summer, we observed a reduction of <30-nm particles and an increase of >50-nm particles, including tailed phages and putative phycodnaviruses with ~150-nm capsids. Consistently, flow cytometry analysis of the summer sample showed a population of viruses with staining properties similar to those of large algal phycodnaviruses (fig. S2).

Fig. 1

Morphological and taxonomic diversity of viruses in an Antarctic lake. (A) Transmission electron micrographs showing the diversity of viral morphotypes. s, siphoviruses; p, podoviruses; m, myoviruses. (B) Comparison of the number of viral families in Lake Limnopolar and 32 aquatic viromes (SEED identification numbers, see description in table S2) with five or more sequences ascribed by Blast similarity to SEED_nr database.

The main constraint of Antarctic virus studies is the lack of genome sequence data. We sequenced 20 million base pairs of viral DNA purified from Lake Limnopolar. The majority of the 89,347 sequences obtained by pyrosequencing showed no significant similarity to GenBank sequences [87.6% on average (fig. S3)] or to metagenomic open reading frame [72% on average (table S1)] data sets. This phenomenon is in common with previously published viromes and reveals our limited knowledge of viral diversity, in spite of the large amount of metagenomic data collected from uncultured microorganisms. Accordingly, <3% of the Antarctic sequences showed any similarity to each of the 30 viromes obtained from aquatic environments by pyrosequencing, with freshwater lake viromes being the most similar (table S2) (12, 13). The sequences ascribed to cellular organisms (table S3) might be due to a poor representation of viruses in databases, contamination with small-size prokaryotes, unidentified prophages annotated as bacterial genomes, or host genes incorporated into viral genomes (10, 12).

There is evidence that high-latitude ecosystems may have low biological diversity (12, 19, 20). Lake Limnopolar harbored a marked diversity with viral genotypes distributed across 12 virus families (table S4), greater than that previously found in viromes from aquatic environments, most of them showing the presence of only 3 to 6 viral families (Fig. 1B and table S5). Moreover, prediction of the viral community structure estimated a species richness of 5130 viral genotypes in spring and 9730 in summer, whereas similar estimates of viral genotypes were substantially lower (ranging from 253 to 787) in the few other freshwater viromes available (table S6). This level of diversity of Antarctic viruses was in the high range of that determined for the aquatic, mostly seawater, viromes analyzed so far (12, 21). This unexpected high genetic diversity in Antarctic viruses was also found by polymerase chain reaction (PCR)–amplification within the virus genus T4-myoviruses, one of the most abundant, ubiquitous, and better-studied viruses in environmental samples (table S5) (12, 13, 22). In agreement with the variety of tail phage morphologies seen (Fig. 1A), PCR-amplified sequences of gp23 gene from T4 phages were widely distributed throughout their phylogenetic tree (Fig. 2A and table S7), and most of them had <80% amino acid sequence identity with known sequences.

Fig. 2

Genetic diversity of viruses in an Antarctic lake. (A) Phylogenetic tree including 30 independent sequences of the gp23 protein of Antarctic T4-myovirus obtained by PCR-cloning in spring and summer. The main groups of T4-phages are indicated by colored lines, and clusters of related sequences supported by internal nodes with >50% bootstrap value are represented by thicker lines. The origin of the sequences is depicted by edge circles colored as indicated. (B) Relative abundance of viral families before (spring) and after (summer) ice melt. The distribution of sequences into viral families, determined by Blast similarity to the indicated database, was corrected by the phi29 polymerase bias toward ssDNA virus genomes [estimated in 100 times (23)] and normalized by genome length to estimate the composition of the viral assemblage.

All of the viromes described to date are dominated by bacteriophages, mostly large-tailed double-stranded (dsDNA) phages (myovirus, podovirus, and siphovirus) and small single-stranded DNA (ssDNA) microvirus (table S4) (12, 21). However, the Antarctic freshwater viral assemblage reported here has a distinct composition with a large proportion of ssDNA viruses, mostly related to eukaryotic viruses. The Lake Limnopolar spring virome was dominated by ssDNA viruses (Fig. 2B), correlating with 56% of viral particles <30 nm observed by EM (fig. S2). Although some Antarctic ssDNA viruses are related to microviruses commonly found in other viromes and phylogenetically related to the chlamydiamicrovirus genus (Fig. 2B, fig. S4, and table S5) (12, 13), the Antarctic virome contains ssDNA viruses related to other viral families (circovirus, geminivirus, nanovirus, and satellites) that we have not previously seen reported for aquatic environments. Unlike microviruses that infect bacteria, these ssDNA viruses infect mammals, birds, or plants, whose virtual absence in Lake Limnopolar (14) leads us to propose that they infect other eukaryotic hosts. These sequences exclusively aligned to a highly conserved domain (pfam2407) of replication-related genes, and a phylogenetic analysis based on pfam2407 clustered most of these viruses distinctly from the related families of ssDNA viruses and from environmental sequences (Fig. 3A) (23). The abundance of ssDNA virus sequences allowed us to assemble 10 circular ssDNA genomic elements, including the replication protein and stem loop, that are highly conserved among nanoviruses, satellites, and circoviruses (Fig. 3B and table S8). The presence of genomic organizations with two ORFs transcribed unidirectionally, as well as their phylogenetic distance, suggests that these viruses may belong to previously undescribed circular ssDNA viral families. We also assembled 25 other circular DNA genomic elements with no similarity to known sequences (table S8). The host range and ecology of these small DNA viruses is unknown.

Fig. 3

Circular ssDNA viruses in Antarctica. (A) Phylogenetic tree of replication-related genes with pfam02407, including sequences assembled from Antarctic viromes (red), from rice paddy soil viromes (green) (23), and reference sequences (GenBank number indicated) from ssDNA viruses. Gray and black circles represent internal nodes with >50% and >90% of bootstrap values, respectively. The scale bar indicates the number of amino acid substitutions per residue. (B) Genome organization of 10 circular genomic elements assembled from spring and summer Antarctic lake viromes containing a putative replication gene (blue arrow) and a stem loop with a conserved nonanucleotide motif and small reiterated flanked sequences (red arrow). ORFs with no major similarities to known proteins (yellow arrow) and known ORFs not related to replication genes (orange arrow) are indicated. Representative genomes or genomic components of circular ssDNA viruses correspond to satellite (gi 18959285), nanovirus (gi 19744936), circovirus (gi 38018060), and geminivirus (gi 21426900).

In spring under ice cover, the lake virome was dominated primarily by ssDNA viruses and secondarily by dsDNA Caudovirales. The representation of ssDNA viruses decreased after the ice cap thawed in summer, and the viral assemblage became dominated by dsDNA viruses belonging mostly to the Phycodnaviridae family, but also contained Caudovirales and Mimiviridae (Fig. 2B and table S4). Nearly 87% of the phycodnavirus sequences from the summer virome matched a large portion of the genome of OtV5 (prasinovirus genus), a virus infecting the prasinophyceae Ostreococcus tauri (24) (fig. S5 and table S9). The presence of an OtV5-related prasinovirus was confirmed by PCR amplification and sequencing of a conserved region encoding its major capsid protein (fig. S4). However, the <78% amino acid identity in the major capsid protein and the fact that some regions of the genome were not identified in spite of the high sequence coverage in the virome indicate that this Antarctic prasinovirus is distinct from OtV5. The change in the viral community composition before and after ice melt, observed with five different search strategies, was consistent with the quantification of virus particles (fig. S2) and probably reflects changes in physical factors and host community (6).

Restricted metabolic activity under the ice promotes lysogeny in large Antarctic phages (15) but may allow the lytic replication of ssDNA viruses with small genomes, which are more abundant in the spring sample. The high prevalence of genes involved in photosynthesis and other cellular metabolic processes in phages and viromes has led others to propose that these genes might facilitate the expansion of their hosts into new ecological niches (2527). Similarly, the increase we observed in the Antarctic summer virome of genes involved in carbohydrate and amino acid metabolism, respiration, and stress responses suggests that they may help the infected host to better survive the changing environmental conditions after the ice cover melts (fig. S6). Host expansion will also favor virus populations; for example, the phycodnavirus expansion we have observed in the summer sample is probably a consequence of a Prasinophyceae green alga bloom. Scales of the prasinophyte Pyramimonas geledicola, the dominant phytoflagellate in some Antarctic lakes, have been observed as associated with large viral particles (28) in Antarctica. The progressive reduction of Lake Limnopolar ice cover during December provided better radiation transmission to the water column triggering the growth of phytoplankton just under the ice sheet. The algal bloom was detected in December both by a clear chlorophyll a peak and a reduction in light adsorption below 1 to 2 m (fig. S7), suggesting that it may represent the early expansion of the algal host of the phycodnavirus that dominated the lake in January (summer sample).

The viral assemblage of Lake Limnopolar reported here is unexpected by comparison with other aquatic environments. First, in contrast with other known aquatic viromes, this Antarctic lake viral community is not dominated by bacteriophages infecting prokaryotes (12, 21) but by viruses known to infect eukaryotes, including ssDNA viruses related to animal and plant viruses, and dsDNA viruses infecting algae. Second, undescribed ssDNA viruses possibly belonging to distinct viral families were found. Third, unprecedented taxonomic diversity and high genetic richness in this Antarctic lake illustrates that high virus diversity may be found in systems where biological diversity in other taxa is low (2, 3, 19). Finally, we found that melting of the ice cover in summer leads to a change in this Antarctic viral community from small ssDNA to large dsDNA viruses. Altogether, the Lake Limnopolar virome sheds light into the largely unknown community of viruses populating Antarctic freshwater ecosystems.

Supporting Online Material

Materials and Methods

Figs. S1 to S7

Tables S1 to S9


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank the Maritime Technology Unit (CSIC) and Las Palmas crew (Spanish Navy) for the logistic help and support that made this expedition possible. We also thank M. Toro, A. Camacho, and other members of the Limnopolar Project for help and discussions; A. Alejo for helpful comments and reviewing the manuscript; and M. Pignatelli for technical support. This work was funded by grants from the Spanish Ministry of Science and Innovation (CGL2005-06549-C02-01/ANT, CGL2007-29843-E/ANT, CTM2008-05134-E/ANT, and BFU2008-04501-E). Spring and summer viral metagenomes from the Antarctic Lake Limnopolar have been submitted to GenBank and assigned the genome project accession number 34669. The metagenomes are also publicly accessible from the ftp server of the SEED public database ( under the project accession numbers 4441778.3 and 4441558.3 for the spring and summer viromes, respectively. PCR-obtained sequences encoding the gp23 protein of Antarctic T4-phages and the MCP protein of Antarctic phycodnavirus have been also deposited at GenBank as “environmental sequences” and are listed under accession numbers FJ791185-247 and FJ791175-84, respectively.

Stay Connected to Science

Navigate This Article