A distinct lineage of giant viruses brings a rhodopsin photosystem to unicellular marine predators

Abstract

Giant viruses are remarkable for their large genomes, often rivaling those of small bacteria, and for having genes thought exclusive to cellular life. Most isolated to date infect nonmarine protists, leaving their strategies and prevalence in marine environments largely unknown. Using eukaryotic single-cell metagenomics in the Pacific, we discovered a Mimiviridae lineage of giant viruses, which infects choanoflagellates, widespread protistan predators related to metazoans. The ChoanoVirus genomes are the largest yet from pelagic ecosystems, with 442 of 862 predicted proteins lacking known homologs. They are enriched in enzymes for modifying organic compounds, including degradation of chitin, an abundant polysaccharide in oceans, and they encode 3 divergent type-1 rhodopsins (VirR) with distinct evolutionary histories from those that capture sunlight in cellular organisms. One (VirR_DTS) is similar to the only other putative rhodopsin from a virus (PgV) with a known host (a marine alga). Unlike the algal virus, ChoanoViruses encode the entire pigment biosynthesis pathway and cleavage enzyme for producing the required chromophore, retinal. We demonstrate that the rhodopsin shared by ChoanoViruses and PgV binds retinal and pumps protons. Moreover, our 1.65-Å resolved VirR_DTS crystal structure and mutational analyses exposed differences from previously characterized type-1 rhodopsins, all of which come from cellular organisms. Multiple VirR types are present in metagenomes from across surface oceans, where they are correlated with and nearly as abundant as a canonical marker gene from Mimiviridae Our findings indicate that light-dependent energy transfer systems are likely common components of giant viruses of photosynthetic and phagotrophic unicellular marine eukaryotes.

Keywords: giant viruses; host–virus interactions; marine carbon cycle; single-cell genomics; viral evolution.

Fig. 1.

A giant virus infects a predatory protist that is considered to be among the closest living unicellular relatives of metazoans. (A) Schematic tree of eukaryotes, with supergroups indicated by colors or gray branches if in contentious positions. Lineages with giant viruses (pink) known (circles) or discovered here (star) are indicated. (B) Locations of single-cell sorting where ChoanoV1 and its host, B. minor, were recovered (Station M2), where ChoanoV2 (Station 67-70) was found, and where metatranscriptomes were sequenced from unmanipulated seawater (M1, M2, 67-70; Station 67-155, 785 km from shore, not displayed on map for scale reasons). (C) Histogram showing the population (circled) of sorted choanoflagellate cells (blue dots), including the viral-infected cell (pink), based on index sorting and V4 18S rRNA gene amplicon sequencing. Other data points reflect unsorted particles in the stained seawater analyzed. The box (green) indicates the position of YG bead standards run before and after sorting at the same settings. (D) Categorized summary of the top 10 BLASTp matches for 862 ChoanoV1 proteins (e-value < 10⁻⁵) in cellular organisms and NCLDV.

Fig. 2.

Evolutionary relationships and functional aspects of the ChoanoVirus lineage. (A) Maximum likelihood phylogenomic reconstruction inferred from 10 proteins. Support >80% (500 bootstrap replicates) is indicated (LG + C20 + F + G-PMSF model) (SI Appendix, Fig. S5), and host group coloring is as in Fig. 1A. ChoanoV1 (star; from M2 single-cell sort) and ChoanoV2 (from Station 67-70; low %GC-selected DNA with metagenomics) branched together in all reconstructions adjacent to an algal stramenopile virus AaV (when included) (SI Appendix, Fig. S5), for which placement appears influenced by long-branch attraction. (B) Total number of tRNAs (Left) and orthogroup functional categorization (heat map; EggNOG categorization) of ChoanoV1 and representative giant NCLDV (Dataset S1). The frequency of each category across the viral genomes determines x-axis ordering. (C) Distribution of functional categories in ChoanoV1 (via EggNOG) for all annotated proteins. (D) ChoanoV1 proteins with no orthologs in the NCLDV representative genome set. Note that, in pies in B to D, we have omitted fractions representing the EggNOG functional category “Unknown function,” but the values are shown as text on panels along with the total number of proteins with no significant database match.

Fig. 3.

Evolution, structure, and function of viral rhodopsins. (A) Maximum likelihood phylogenetic reconstruction of bacterial, archaeal, eukaryotic, and viral rhodopsins. Viral (pink), nonviral ion-pumping (black), nonviral sensory (brown) and limited information or unclear function nonviral (gray) rhodopsins are indicated and support >80% (1,000 bootstrap replicates). Sensory rhodopsins present in the choanoflagellate S. rosetta (59) and detected here in Choanoeca perplexa and Microstomoeca roanoka, but not other choanoflagellates (27) or B. minor, have a fused phosphodiesterase region and are distant from ChoanoVirus VirR proteins. Metagenomic sequences from a sediment study reporting 30 PgV and Organic Lake virus-like VirR (62) could not be included, because they are not in GenBank, were not recovered in the IMG (Integrated Microbial Genomes) database, and, based on statistics provided, were largely partial length. This prior study recovered different VirR topologies using maximum likelihood vs. trait-informed Bayesian reconstructions that also differed from our highly supported topology, indicating that conclusions (62) regarding identification of a putative ancestor of viral rhodopsins should be revisited. Channelrhodopsins (52) were not included due to high divergence that resulted in the loss of many positions for type-1 phylogenetic analysis. Additionally, heliorhodopsins were excluded, because they are too divergent from the microbial type-1 rhodopsins. (B) Light-induced acidification of medium containing E. coli-expressing VirR_DTS in the presence of the chromophore retinal (solid line) and its abolishment by protonophore addition (i.e., carbonyl cyanide m-chlorophenyl hydrazone ([(3-chlorophenyl)hydrazono]malononitrile (CCCP)); dotted line). (C) Surface representation of the 1.65-Å resolution VirR_DTS crystal structure with electrostatic potential indicated (red, negative; blue positive) as viewed parallel to the membrane. (D) Ribbon diagram showing the retinal (light blue lines), H₂O molecules (red spheres), and 7 TM α-helices connected by 3 cytoplasmic loops, 3 extracellular loops, and short helices between TM3 and TM4. Numbers denote TM domains. (E) VirR_DTS (magenta) and H. salinarum proton-pumping BR (71) (purple; Protein Data Bank ID code 1C3W; 21% amino acid identity) structural comparison. Key residues (teal, BR; red, VirR_DTS) and H₂O molecules (spheres) are indicated. (F) VirR_DTS absorption spectrum.

Fig. 4.

Functional attributes of ChoanoViruses include chromophore biosynthesis. Shown are carotenoid pathway components and final retinal-forming cleavage step in genome data from haptophytes (Phaeocystis antarctica and Chrysochromulina representing P. globosa, which lacks genome data), choanoflagellates (M. brevicollis and S. rosetta), and relevant viruses and in metatranscriptomes. The stars indicate the two ChoanoVirus genomes and a metatranscriptome from the station where ChoanoV1 was recovered. The circle indicates the only cultured virus with a rhodopsin. *These taxa lack Blh but have RPE65 used for retinal production (e.g., in vertebrates and relatives). Detection in Pacific metatranscriptomes based on reads recruited to ChoanoV1 by BLASTx (e-value < 10⁻¹⁰); those that mapped at >95% nucleotide identity are indicated in Dataset S5. OPP, pyrophosphate group; FPP, farnesyl diphosphate; GGPP, geranylgeranyl diphosphate.

Fig. 5.

Viral rhodopsins are distributed across the world oceans. (A) Environmental VirR motifs and cluster analysis of sequences (CLANS)-based relationships between full-length proteins recruited from TARA Oceans and Station ALOHA data. (B) Normalized VirR depth distributions in the North Pacific Gyre determined by mapping metagenomic reads to VirR gene assemblies from ALOHA (60) and VirR motif distributions (pies; colors as in A). (C) VirR motifs in TARA metagenome assemblies having >300,000 contigs from 5 m (304 full-length sequences in total) and samples reflecting a true deep chlorophyll maximum (43 full-length sequences in total), which typically occurs in stratified open ocean water columns between 75 and 130 m. (D) Correlation between Mimiviridae PolB and VirR across analyzed TARA samples. (E) Normalized VirR and Mimiviridae PolB frequencies in TARA assemblies (with >300,000 contigs).