Virophage control of antarctic algal host-virus dynamics

Abstract

Viruses are abundant ubiquitous members of microbial communities and in the marine environment affect population structure and nutrient cycling by infecting and lysing primary producers. Antarctic lakes are microbially dominated ecosystems supporting truncated food webs in which viruses exert a major influence on the microbial loop. Here we report the discovery of a virophage (relative of the recently described Sputnik virophage) that preys on phycodnaviruses that infect prasinophytes (phototrophic algae). By performing metaproteogenomic analysis on samples from Organic Lake, a hypersaline meromictic lake in Antarctica, complete virophage and near-complete phycodnavirus genomes were obtained. By introducing the virophage as an additional predator of a predator-prey dynamic model we determined that the virophage stimulates secondary production through the microbial loop by reducing overall mortality of the host and increasing the frequency of blooms during polar summer light periods. Virophages remained abundant in the lake 2 y later and were represented by populations with a high level of major capsid protein sequence variation (25-100% identity). Virophage signatures were also found in neighboring Ace Lake (in abundance) and in two tropical lakes (hypersaline and fresh), an estuary, and an ocean upwelling site. These findings indicate that virophages regulate host-virus interactions, influence overall carbon flux in Organic Lake, and play previously unrecognized roles in diverse aquatic ecosystems.

Fig. 1.

Transmission electron micrographs of negatively stained virus-like particles from Organic Lake. (A) Virus-like particles resembling the size and morphology of PVs, (B) Sputnik virophage, and (C) bacteriophages.

Fig. 2.

Phylogeny and genomic maps of OLPVs. Neighbor-joining tree of B family DNA polymerase (A) and MCP (B) sequences of OLPV, and NCLDV sequences from GenBank. (C) Maps of OLPV-1 and OLPV-2 scaffolds and comparison of the location of genes; single-copy conserved orthologs and MCP (red); regions with identity to OLV (green); proteins identified in the metaproteome (yellow); ribosomal nucleotide reductase β (1), VV A32 packaging ATPase (2), VV VLTF3 transcription factor (3), VV D5 replicative helicase (4), PbCV-1 A482R-like putative transcription factor (5), ribonucleotide reductase α (6), and DPOB (7). Lines connect homologous regions between OLPV-1 and OLPV-2 scaffolds in the same orientation (red) and reverse orientation (blue).

Fig. 3.

Genomic map of OLV. From the outside in, circles represent (i) predicted coding sequences on the forward strand, (ii) predicted coding sequences on the reverse strand, (iii) GC skew, and (iv) GC plot. Predicted coding sequences are colored: Sputnik homologs (red), OLPV homologs (green), non-Sputnik National Center for Biotechnology Information nonredundant homologs (purple), GOS peptide database homologs (blue), and ORFan (cyan). Sequences identified in the metaproteome are marked with an asterisk. Descriptions of the predicted coding sequences from both strands are shown clockwise from position zero.

Fig. 4.

Extended Lotka-Volterra model of host–OLPV–OLV population dynamics. The classic Lotka-Volterra model (37) is based on a pair of first-order, nonlinear, differential equations that can be used to describe the periodic oscillation of the populations of a predator and its prey (38). The extended model shown is based on three equations describing the interactions of host (prey), virus (predator), and virophage (predator of predator) interactions. (A) Time course of host (red line) and virus (blue line) populations. (B) Orbit plot between host and OLPV populations in the absence of OLV, with the host and OLPV populations approaching zero during an equilibrium cycle. (C) Time course describing the effect of the addition of OLV (green line) on OLPV–host population dynamics, resulting in a higher minimal number of hosts and OLPVs (D) compared with in the absence of OLV (B). The orbit plot of OLPV and OLV is also shown (E). The model shows that (independent of the parameters used) although the absolute number of hosts does not increase greatly as a result of OLV preying on OLPV, the frequency of host blooms increases in the presence of OLV. The increased frequency of blooms would increase secondary production (from lysed hosts) and hence overall carbon flux through the system. The increased turnover in the microbial loop during the extended light periods of the polar summer may help to maintain microbial populations in the lake.

Fig. 5.

Abundance and diversity of virophage capsid proteins in environmental samples. (A) Number of ORFs from metagenomic reads that match to OLV MCP (BLASTp e-value cutoff 1e-5, abundance normalized to 100,000), and the proportion of virophage capsid types, for the Organic Lake 0.1-μm and Ace Lake 0.1-, 0.8-, and 3.0-μm fractions. (B) Maximum likelihood tree of a conserved 103-aa region of the MCP from Organic Lake, Ace Lake and GOS metagenome data, and Sputnik genome.