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

A Wild Predatory Protist in the North Pacific Ocean and Its Virus.

To capture uncultured heterotrophic protists, we used high-purity fluorescence-activated cell sorting (FACS) of single cells with acidic vacuole staining to discriminate protists from prokaryotes and an additional exclusion gate against photosynthetic organisms to select heterotrophic protists only (SI Appendix). In a FACS survey in the eastern North Pacific, we recovered a coherent population of choanoflagellates (Fig. 1B andC), heterotrophic predators belonging to the supergroup Opisthokonta that are considered to be among the closest living unicellular relatives of metazoans (27). Choanoflagellates comprised 99% of the 198 wells for which V4 18S ribosomal RNA (rRNA) gene amplicons were recovered after initial multiple displacement amplification of DNA from single cells, and the remaining 3 wells harbored amplicons with highest identity to uncultured syndiniales (putative parasites) and 2 different uncultured cercozoans (bacterivores), respectively. Choanoflagellates are widespread bacterivorous protists that we expected to be targeted by our staining protocol, because they contain an acidic food vacuole.

From one choanoflagellate cell, we assembled an 875-Kb viral genome after eukaryotic single-cell metagenomic sequencing (SI Appendix, Figs. S1 and S2). The virus, ChoanoV1, represents the largest pelagic marine giant virus genome sequenced yet; its genomic DNA base composition (GC content) was low (22%), rivaled only by nonmarine Hokoviruses (21%) and CroV (23%), whereas other giant viruses range to 64% GC (10,12,21) (SI Appendix, Fig. S2 andDataset S1). The ChoanoV1 genome encoded 862 predicted proteins, and its gene content suggested that it belonged to the NCLDV (Fig. 1D andSI Appendix, Fig. S2), a diverse group of eukaryotic viruses (10,11).

Presence of a eukaryotic virus coassociated with a single choanoflagellate cell could reflect several possible ecological interactions: first, that the virus had infected the choanoflagellate and replicated there; second, that the virus had been consumed by the predator as a prey item as reported in 2 prior culture-based studies on viral-feeding by predatory protists (28,29); and third, that the virus had infected a prey item of the choanoflagellate (before that prey was consumed). Multiple lines of evidence support the first scenario. The average sequencing depth of the viral genome (215 ± 157×) and other assembly statistics (SI Appendix) suggested the virus was highly replicated (30) in the sorted choanoflagellate, implying there were many ChoanoV1 genomes present in the host cell. Among nonviral reads in the well, more than half belonged to the uncultured choanoflagellateBicosta minor. This was determined by mapping reads against an 87-Mb partialB. minor genome that we generated from 4 other sort wells (SI Appendix, Fig. S1B), each containing single identical 18S rRNA gene sequences (assembled from metagenomic data and in V4 18S rRNA gene amplicons) (Dataset S2) that had 99% identity toB. minor as identified, handpicked, and sequenced in a prior field study (31). Contigs from bacterial prey (and phages) were also present in the choanoflagellate–virus-containing well but had a lower N50 (i.e., the minimum contig length needed to cover 50% of the genome; specifically, 13,326 vs. 86,624 for the virus genome), and none had genomes close to completion. These results suggest that the bacteria present were diverse and potentially in a degraded state as would occur in the choanoflagellate food vacuole. Additionally, the N50 ofB. minor contigs (2,098) was lower than in wells where the virus was not detected (8,546), suggesting that it (as host) was also being degraded (SI Appendix, Fig. S1B andC). While these statistics point to an active infection, it is hypothetically possible that many of the same virus had been ingested, leading to the high-coverage statistics for ChoanoV1. However, traditional metagenomic data from the same Pacific Ocean site and sort date showed that prokaryotes (prey) were >50,000 times more abundant than ChoanoV1 based on the relative numbers of bacterial 16S rRNA gene reads (a gene that is often single copy in marine bacteria) and ChoanoV1 DNA Polymerase B (PolB) reads (a single-copy gene in viral genomes). Hence, if choanoflagellates were to feed on giant viruses, the predator–prey encounter rate would strongly favor consumption of bacterial cells such that consumption of more than 1 ChoanoV1 virion is improbable. The other mechanism by which many of the virus could have been ingested is consumption of an infected small eukaryotic prey item. We did not detect sequences in the sort wells from any of the picoeukaryotes that are abundant in marine waters, including those in prior reports on this region (32,33). Additionally, for the encounter rate of algal prey to be sufficiently higher than bacteria, one might presume that a bloom is necessary. However, Chlorophylla concentrations at the depth sampled for sorting and others from the same water column and date were not indicative of a bloom; rather, the spring bloom seemed to be initiated later in the season (Dataset S3) as is typical for the region (34). Furthermore, the gene content of ChoanoV1 is highly distinct from the many available genome sequences from viruses of picoeukaryotes (35,36) or other known algal viruses (18–20,22,23,37) (SI Appendix, Fig. S3). Collectively, these results point to us having recovered an actively infectedB. minor host cell in which ChoanoV1 had already replicated. After Canarypox virus, which infects birds (38), ChoanoV1 represents just the second giant virus identified with an opisthokont host (Dataset S1).

We next sought to recover a ChoanoVirus genome from another field site. Therefore, we exploited the low GC content observed in ChoanoV1 to sequence and assemble a related virus in an eastern North Pacific sample collected 200 km offshore 7 y before theBicosta single-cell study (Fig. 1B). This sample was chosen for low %GC DNA enrichment on a density gradient, followed by deep sequencing, because environmental clone libraries showed that theB. minor 18S rRNA gene was present (100% identity) and vintage metagenomic data from the sample (7) contained ChoanoV1-like reads. The resulting ChoanoV2 assembly contained 89% of ChoanoV1 genes (average 94% amino acid identity), despite its fragmented nature resulting from traditional metagenome assembly limitations (SI Appendix, Fig. S4A). Our discovery poised us to investigate the evolution, function, and importance of specific metabolic traits in viruses of a key group of opisthokonts or more generally, heterotrophic marine protists and broader ecological implications.

Evolutionary Analyses Establish a Distinct NCLDV Giant Virus Lineage.

Preliminary analyses suggested the ChoanoViruses were NCLDVs, with about 20% of the ChoanoV1 predicted proteins and 23% of the more fragmented ChoanoV2 proteins showing highest BLASTp affiliations to NCLDV proteins (Fig. 1D andSI Appendix, Fig. S4B). For proteins that had BLASTp affiliations primarily to cellular life, most of those closest to eukaryotic proteins seemed to be opisthokont derived, suggesting acquisition from hosts in past time (SI Appendix, Fig. S4C). Unfortunately, the paucity of genomic resources for marine eukaryotic viruses and marine protists themselves precludes statistically valid examination of potential horizontal or host-to-virus gene transfer (HGT) at a genome wide scale, and hence, we did not examine questions of origin globally. The other half of the ChoanoVirus proteins have not been seen in cellular organisms or viruses sequenced to date. Overall, these observations, including ∼50% of proteins being unknown, are quite typical of newly sequenced NCLDV genomes (13,39), at least at this stage in time, in which relatively few have been sequenced. Of these ChoanoVirus orphan genes, 70% were detected in metatranscriptomes that we sequenced from the eastern North Pacific, demonstrating expression (SI Appendix, Fig. S2).

To reconstruct evolutionary relationships, we used phylogenomic approaches to analyze proteins considered core to NCLDV genomes (40). We reexamined presence, absence, and copy number patterns for the 47 proteins previously proposed to be core (40). We next excluded, for example, fast-evolving proteins and proteins for which unclear paralogs existed within a single NCLDV genome, and thereby expanded the set of NCLDV proteins suitable for phylogenomics used in recent reconstructions (12) from 5 to 10 (Fig. 2A andDataset S2). Phylogenomic reconstructions with the 2 protein sets provided similar topologies, with higher statistical node support in the 10 protein phylogeny (SI Appendix, Fig. S5). These reconstructions showed the ChoanoViruses belong to the extendedMimiviridae, comprising a divergent clade from those already established (12,13). PolB reconstructions highlighted a large group of marine viral PolB, distinct from nonmarineMimiviridae (Mimiviruses, Tupanviruses, Klosneuviruses) and CroV, when assembled metagenomic sequences from TARA Oceans (41) and Global Ocean Survey (GOS) (42) were searched and included (SI Appendix, Fig. S6). Within this broad marine group, the ChoanoViruses formed a supported clade that incorporated Pacific Ocean, Atlantic Ocean, and Southern Ocean sequences for which the viral hosts remain unknown. These analyses demonstrated the value of recovering viral genomes from uncultured hosts, which exposed here the unique ChoanoVirus lineage and its presence in multiple oceans.

ChoanoVirus Auxiliary Metabolic Genes and Biogeochemical Implications.

AMGs are host-derived genes carried by viruses that are not directly involved in viral replication but rather supplement or augment cellular functions within infected cells (5,6). An important example in marine bacteriophages is oxygenic photosynthesis proteins that augment cyanobacterial photosynthetic machinery during infection (4). Although oxygenic photosynthesis-related proteins have not been found in eukaryotic viruses sequenced to date, the giant viruses encode a plethora of AMGs that augment cellular processes. These include proteins involved in, for example, translation, transcription, lipid biosynthesis, and transport of phosphate or ammonium (6–8,43). Systematic analyses of ChoanoVirus metabolic potential revealed a broad repertoire of such proteins, several types being enriched or unique in ChoanoViruses relative to other NCLDVs (Fig. 2B–D andSI Appendix, Figs. S3 and S7A). Like other giant viruses, the ChoanoVirus genomes encode proteins for augmenting host processes, including aminoacyl-tRNA synthetases, photolyases, and proteins involved in signal transduction, replication, recombination and repair, cell wall biogenesis, and posttranslational modifications (Fig. 2B andSI Appendix, Fig. S7A) (6–13,18–23). The ChoanoViruses also encode 22 tRNAs (Fig. 2B) such that tRNA numbers seem to roughly scale with genome size, with more being found in the larger genome-sized Tupanvirus from deep sea sediment (43) and less in the smaller genome-sized pelagic marine giant viruses TetV, CroV, PgV, and CeV (18,20–22). Furthermore, the ChoanoVirus tRNAs correspond to amino acid usage, suggesting preferential retention of those optimized for amino acid usage of virus over host, and 17 tRNAs are collocated in a single genomic region (SI Appendix, Fig. S7B andC). Hence, the large ChoanoVirus genomes encoded many proteins once considered unique to cellular life, that now seem to be held in common across disparate giant viruses (10–13,18–21).

Clustering based on presence and absence patterns of orthologous protein groups in NCLDV placed ChoanoV1 adjacent to the only other sequenced marine pelagic virus with a host that is a heterotrophic predator, CroV (SI Appendix, Fig. S3) (21). These 2 viruses were part of a broader cluster incorporating marine algal giant viruses, which appeared more similar to each other in their orthogroup presence and absence patterns than to nonmarine giant viruses or smaller viruses that infect marine algae. Many of the proteins making up these orthogroups lack characterized functions or have only broad functional classification. Combined with the limited overall representation of giant virus lineages, these findings call for a major initiative to expand viral taxonomic sampling so that the significance of the presence and absence pattern observations could be estimated.

Comparison of ChoanoV1 with other genome-sequenced viruses shows an enrichment in NCLDV orthologs involved in transport and metabolism of nucleotides, amino acids, and carbohydrates (Fig. 2B andSI Appendix, Fig. S7A). ChoanoV2 shows the same trend, although its more fragmented state precludes robust global ortholog comparisons. Even among ChoanoVirus proteins lacking orthologs in other NCLDV, these functional categories are prominent (Fig. 2D andSI Appendix, Fig. S7D) and include a chitinase new to marine viruses that is present in both ChoanoViruses (SI Appendix, Fig. S8). Chitinase degrades the polysaccharide chitin, a component of zooplankton, some algae, and many other organisms, to labile saccharides readily consumed by marine microbes (44). This enzyme has been reported in a virus of the freshwater algaChlorella (45) and viruses that infect insects, specificallyLepidoptera (46). Our phylogenetic analyses placed moth virus chitinases in a clade with sequences from theirLepidoptera hosts within bacterial chitinases (potentially a complex series of transfer events), while Chlorella virus and fungal chitinases grouped together (SI Appendix, Fig. S8). The ChoanoVirus chitinase branched with opisthokont chitinases, suggesting potential acquisition from a host of an ancestral opisthokont virus. Collectively, these results suggest that acquisition by each of the 3 types of viruses occurred in independent events. From a functional perspective, release of viral chitinase inLeptidoptera larvae is necessary for liquefaction, but the mechanism and overall roles during infection are unclear (46). The Chlorella virus chitinase has hypothesized roles in degrading the chitin-rich host cell wall (45). However, in contrast to moths andChlorella, which have chitin as an abundant structural component, choanoflagellates lack known chitin-based structures, although they possess chitin synthase (47). Thus, ChoanoVirus chitinase activity, potentially on prey material, alongside activities of viral carbohydrate metabolism proteins may supply hosts with nutrition when choanoflagellate feeding is impacted by the infection or other factors. Alternatively, a structural feature of choanoflagellate cells, such as the theca, may have an as yet unrecognized chitin-containing composition, in which case, the viral chitinase may operate in host degradation. Regardless, the organic matter released from the lysed host will provide more readily available carbon sources, such as labile saccharides, to marine microbes than will hosts infected and lysed by viruses that lack these enzymes or other forces of mortality. As such, in addition to release of cellular substrates on lysis, viral infection may “prime” substrates to be accessed more readily, potentially altering the microbial loop (48) in terms of rate and fate of the cellular material remineralization in the ocean.

Viral Rhodopsin Sequence Characterization.

Strikingly, we also identified 3 distinct putative rhodopsins in each ChoanoVirus genome (Dataset S4). Rhodopsins are integral membrane proteins that capture or sense sunlight using a bound retinal chromophore in cellular organisms (49). Microbial (type-1) rhodopsins include a variety of light-driven ion pumps (including H⁺, Cl⁻, Na⁻) (SI Appendix, Table S1) and sensory receptors involved in signal transduction (including Sensory Rhodopsins I and II, which have been shown to regulate phototaxis in some protists) (50–52). Additionally, heliorhodopsins are considered distantly related family members and are thought to have light-sensing activities (53). Type-1 proton-pumping rhodopsins are widespread in heterotrophic marine bacteria (54,55), increasing survival during starvation when illuminated (50), and homology-based studies postulate that some eukaryotic algae have similar systems (56). Phylogenetic analyses show that the ChoanoVirus rhodopsins split into 2 type-1 groups composed primarily of metagenomic sequences, which collectively exhibit distinct phylogenetic histories from those in cellular organisms (Fig. 3A). Among viruses with known hosts, the only other rhodopsin reported is in the giant virus PgV, which infects the marine haptophyte algaP. globosa (18,57), and belongs to a clade that includes 1 of the 3 ChoanoVirus rhodopsins (Fig. 3A). We term these 2 groups (clades) that have this distinct history from those of cellular organisms VirR Group-I and VirR Group-II. Importantly, all VirR are highly diverged from a microbial rhodopsin clade harboring the fusion protein Rho-PDE that is present in the genome-sequenced choanoflagellateSalpingoeca rosetta, wherein it exhibits light-dependent phosphodiesterase activity (58,59). While we identified homologs of Rho-PDE in 2 transcriptome-sequenced choanoflagellate species (Fig. 3A), it is absent from genome-sequencedMonosiga brevicollis and is not found in transcriptome assemblies from 17 other choanoflagellate species or in theBicosta 4-well partial genome assembly. Overall, the ChoanoVirus VirR proteins do not seem to be derived from extant opisthokonts. Indeed, the tree topology and additional testing (SI Appendix) suggest that rhodopsin may have been present in an ancestral virus before host-range expansion into disparate algae and heterotrophs (Fig. 3A).

Several marine studies have now reported putative viral rhodopsins in traditional metagenomic data—for which the viral hosts are by default unknown (57,60–62). The function of these is not clear, since often, they lack the amino acid motifs that have been shown through biochemical characterization of various type-1 rhodopsins to generally confer functional differences. Indeed, the function of type-1 rhodopsins can sometimes be inferred from 3 key amino acid residues (referred to as motif sequences), such as the proton (DTD, DTE) and chloride (TSA, NTQ) pump motifs (49). In bacteriorhodopsin (BR), the residues that make up the motif are at positions 85, 89, and 96. BR has been biochemically characterized to function as a proton pump, wherein the D85 acts as a proton acceptor, T89 forms a hydrogen bond with D85, and D96 acts as a proton donor in this DTD motif rhodopsin (49); other motifs have proton pumping or other functions (SI Appendix, Table S1). Previously detected VirR sequences in PgV and GOS were hypothesized to have sensory roles in host phototaxis (57) or to be involved in light sensing in the host (61), because some lack the retinylidence Schiff base proton donor carboxylate, which has been taken to be essential for proton transport, similar to sensory rhodopsins (63). However, recent work has shown that some rhodopsins lacking the proton donor carboxylate do pump protons (64). Based on in silico transmembrane predictions (TMHMM, a method for prediction of transmembrane domains based on hidden markov models), the 3 different rhodopsin proteins in the ChoanoViruses each have 7 transmembrane (TM) domains, as expected (49), and we detected transcripts for 2 of 3 in eastern North Pacific metatranscriptomes from Stations M1 and M2 (Fig. 1B), demonstrating their expression (SI Appendix, Fig. S2). The Viral Group-I rhodopsin present in each ChoanoVirus and in PgV has a DTS motif (VirR_DTS) (Fig. 3A). Prokaryotic DTS-motif rhodopsins have been reported in proton-pumping clades (e.g., the proteorhodopsin [PR] clade and DTG-motif clade) and the xenorhodopsin clade (e.g.,Anabaena sensory rhodopsin, ASR) of sensory rhodopsins, indicating that information on the motif sequence alone is not enough to predict function (65,66). The motifs of the ChoanoVirus Group-II rhodopsins, DTV and YML, are not present in functionally characterized rhodopsins (SI Appendix, Fig. S9). The bacteriumThermochromatium tedium has a YTM motif, with some similarity to YML, that is predicted to be sensor type but as yet not functionally characterized (67). Unlike the observed YML motif, the DTS and DTV motifs have been observed in environmental sequences inferred to come from viruses at Station ALOHA in the North Pacific Gyre (60), in the Red Sea (61), and in coastal sediments (62). Our results provided evidence for VirR proteins being in viruses of heterotrophic protists and for a single virus having both Group-I and Group-II viral rhodopsins. However, the amino acid differences for all VirR from biochemically characterized proteins alongside their long-branch lengths (Fig. 3A) left uncertainty regarding function, as is the case for many proteins identified in marine metagenomic studies.

Viral Rhodopsin Activity and Structure.

Because of the presence of VirR_DTS in the only pelagic marine giant viruses with known hosts (i.e., the uncultured ChoanoViruses and the cultured algal virus PgV), we next turned to laboratory experiments to examine the structure and function of this VirR protein. Heterologous expression inEscherichia coli of the homolog from PgV caused substantial light-induced acidification of retinal-amended medium up on illumination, demonstrating that it has proton-pumping capabilities (Fig. 3B). This clear pH change was abolished by protonophore addition. VirR_DTS predominantly possessed all-trans retinal (SI Appendix, Fig. S10A). At neutral pH, the Schiff base linkage was protonated (pK_a = 7.8), and a counterion residue was deprotonated (pK_a = 3.6) (SI Appendix, Fig. S10B andC). We analyzed the photocycle of VirR_DTS, demonstrating that time constant of recovery from the O540 intermediate to the original state was 386 ms (SI Appendix,SI Results and Discussion and Fig. S10D andE). This recovery time is longer than that of BR fromHalobacterium salinarum (BR,t = 10 ms), an archaeal proton-pumping rhodopsin, but similar to proton-pumping rhodopsins from other taxa, such as BR fromHaloquadratum walsbyi (∼300 ms), thermophilic rhodopsin fromThermus thermophilus (277 ms), and PRs from a number of marine bacteria (PRs; ∼250 ms) (SI Appendix, Table S2) (68–70).

Because VirR_DTS is divergent from characterized light-driven proton-pumping rhodopsins and no viral rhodopsin structure is known, we next dissected how it pumps protons. The crystal structure of the cell-free synthesized VirR_DTS was determined at 1.65-Å resolution, revealing broad-scale similarities to BR (Fig. 3C andD andSI Appendix,SI Results and Discussion and Fig. S11A–E) (71). The root-mean-square deviation (RMSD) was 1.83 Å, while adoption of a different structure from BR was observed in the loops, especially the TM3–TM4 short helix. The pentagonal cluster formed by 3 water molecules (Wat401, -402, and -406), Asp81, and Asp207, corresponding to the most important region for BR proton pumping, did have a similar structure to that of BR (Fig. 3E). Electron densities around the retinal showed that it is in all-trans conformation, covalently attached to Lys211. We then examined several residues that hold key positions in VirR_DTS and other opsins (SI Appendix, Fig. S11B–I), including Asp81 and Ser92, which are similarly positioned to Asp85 and Asp96 of the BR DTD-motif group (71) (Fig. 3E). Mutation analyses of these and other residues established their essentiality for proton-pumping activity, especially the proton acceptor residue Asp81 (SI Appendix, Fig. S11J). In addition, we showed that maximal VirR_DTS absorption is in the green wavelengths (Fig. 3F).

Finally, we compared the VirR_DTS structure with 2 typical structures of sensory rhodopsins: ASR (from Anabaena) and SRII, theNatronomonas pharaonis sensory rhodopsin II (SI Appendix, Fig. S11F andG) (66,72). Given our data, it seems that VirR_DTS is a proton-pumping opsin; however, it is possible that it could have a sensory function as previously proposed based on sequence data (61). There is much debate about interpretation of sequence data alone as well as photocycle data and its comparability when conducted using different conditions. Hence, ultimately, in vivo manipulation in the proper cell biological context will be needed to determine overall function. Our in silico comparisons show that the overall structures of ASR and SRII have similarities to that of VirR_DTS, with RMSDs of 1.94 and 2.22 Å, respectively. While the positions of Ser92 (corresponding to Ser86 in ASR) are similar between VirR_DTS and ASR, the water molecule and amino acid positions around the retinal adopt quite different structures (SI Appendix, Fig. S11F). Likewise, these aspects of ASR positions are different from BR (SI Appendix, Fig. S11H). However, the corresponding portion of SRII is similar to that of VirR_DTS and BR (SI Appendix, Fig. S11G andI). Our searches for the proteins required for signal transduction by sensory rhodopsins using queries known to fulfill this function (e.g., HtrI and HtrII [73]) did not recover related proteins in either theBicosta 4-well assembly or the ChoanoVirus genomes. The viral rhodopsins also lack fusions of known transducer-related domains that occur in eukaryotic sensory rhodopsins (74), although notably, VirR_YML has an N-terminal domain of unknown function. Furthermore, a fusion protein integrating a rhodopsin and phosphodiesterase (RhoPDE; also discussed above,Fig. 3A) was recently discovered inS. rosetta, which, like other choanoflagellates, lacks an eyespot or other known light-sensory structures (58,59,75). While we found phosphodiesterases inBicosta, again, no rhodopsin (or related fusion protein) was recovered, and we did not find these proteins inM. brevicollis or 17 of 19 transcriptome-sequenced choanoflagellates (27). Thus, if the viral rhodopsin was a sensory rhodopsin, the potential mechanisms by which it operates remain elusive as are the biological implications. These observations indicate that motifs, monomeric structures, or photocycle data are individually not enough to determine whether a rhodopsin functions as a pump or sensor. Collectively, our results show that VirR_DTS is a green light-absorbing proton pump that has a structure similar to that of BR and transfers light energy in a manner that substantially changes medium pH when expressed in a cell.

A Viral Chromophore Biosynthesis Pathway.

Demonstration of VirR_DTS proton-pumping activity on illumination raises questions regarding the natural source of the carotenoids needed to produce the light-harvesting chromophore, retinal (50,51), especially in a nonphotosynthetic host, likeBicosta. Most algae, including PgV’s hostPhaeocystis, biosynthesize the required pigment, β-carotene (and related carotenoids), as well as the retinal-producing carotenoid cleavage oxygenase (Blh) (Fig. 4). However, most heterotrophic eukaryotes, including animals, do not biosynthesize β-carotene, instead acquiring carotenoids through diet. As expected, cultured genome-sequenced choanoflagellates encode only early steps that overlap between sterol and carotenoid biosynthesis and a final cleavage enzyme (Dataset S5). Likewise, BLASTx searches against theBicosta 4-well partial genome assembly failed to recover carotenoid biosynthesis enzymes. Remarkably, the ChoanoVirus genome analyses exposed both the β-carotene biosynthesis pathway and Blh, with 4 proteins being adjacent to one another, similar to the pathway in bacteria (76) (Fig. 4,SI Appendix, Fig. S12A, andDataset S5). Eastern North Pacific metatranscriptomes confirmed expression of all components (Fig. 4). Thus, while the algal virus relies on its host to biosynthesize the pigment used in light-energy transfer, ChoanoViruses encode the complete rhodopsin-based photosystem.

The evolutionary origins of the retinal biosynthesis proteins in the ChoanoViruses remain unclear. They seem to derive from archaea (phytoene synthase) or marine bacteria (phytoene desaturase) or are too divergent for robust phylogenetic conclusions (lycopene cyclase, Blh) (SI Appendix, Fig. S12B–E). In each case, the respective ChoanoV1 and ChoanoV2 proteins clustered together, indicating their common origin. Rhodopsin-bearing bacterial or archaeal lineages with retinal biosynthesis-related genes are each thought to have acquired them together as a unit by HGT (77). However, despite the 4 ChoanoVirus retinal biosynthesis genes being colocated in the genome, long branch lengths and incomplete taxonomic sampling make it unclear whether these proteins were accumulated over time or acquired in a single HGT event, although the latter scenario seems most likely.

Viral Rhodopsins in the Global Ocean.

Our studies now provided the structure and function of VirR_DTS, but the frequency of VirR genes as a whole in nature remained unclear. Prior analyses of viral rhodopsins in traditional metagenomic data focused on individual locations, specifically the Red Sea (61) and Station ALOHA (60), or had relatively shallow sequencing depth, such as GOS (57). It should be noted that one other metagenomic study of coastal sediments reported 30 VirR (62) that were similar to PgV VirR_DTS and to the VirR metagenomic sequences from Organic Lake that have been suggested to come from another (currently unknown) haptophyte algal virus. These partial metagenomic sequences (62) may well, therefore, represent remnants of a senesced (infected) haptophyte bloom exported to sediments at 11- to 50-m bottom depth. Our searches of TARA metagenomic assemblies greatly expanded the global VirR repertoire (Fig. 5A andDataset S6). Assembled VirR proteins were recovered at 37 of 39 TARA photic-zone sampling sites examined, and only at photic-zone depths in Station ALOHA profiles that included deep ocean sampling (Fig. 5B), as expected for a sunlight-dependent energy transfer system. Motifs were diverse; however, the DTS motif was the most common vertically and globally (Fig. 5B andC).

Phylogenetic analyses of the assembled sequences that we recovered from deeply sequenced TARA samples and other metagenomic studies showed 3 statistically supported clades within Group-I VirR proteins: one harboring only DTT (64%) and DTS (36%) motifs, another dominated by DTS (72%) and DTT (17%) and having 4 other motifs represented, and a small clade composed of 5 different motifs (SI Appendix, Fig. S13). Likewise, the Group-II VirR proteins delineated into 3 clades, 2 of which are dominated by the DTV motif, with 100 and 73% DTV, respectively. The latter had 4 additional motifs, including 18% DSV. The smallest Group-II VirR clade had just 4 sequences and 3 motifs. In total, our survey of VirR revealed 8 previously unreported motifs, in addition to the observed DTS, DTV, and DTT, and indicated that motifs generally grouped in a manner connecting to the evolutionary history of these proteins. Functional characterization will be important for understanding the cell biological implications during infection as will identification of the corresponding natural hosts.

Finally, more than 99% of environmental VirR had 1 of 2 amino acids (M, L; Met89 in VirR_DTS) that confer increased green light absorption relative to blue, in contrast to the multiple wavelengths used by marine prokaryotic rhodopsins (78). With VirR proteins being in only 2 genome-sequenced viruses with known hosts, we could not parse the metagenomic sequences recovered into percentages coming from viruses of phytoplankton vs. heterotrophs. However, our recruitment of assembled PolB genes from TARA (12,684 in total) recovered 6 times more than vintage TARA 454-metagenome analyses (13,79), and protein similarity networks identified 1,026 PolB as being fromMimiviridae (SI Appendix, Fig. S14). The computed ratio of VirR toMimiviridae PolB was 0.7, suggesting that VirR is a common component of many giant viruses in sunlit ocean environments (Fig. 5D andE andDataset S6).

• 1.Zimmerman A. E., et al. , Closely related viruses of the marine picoeukaryotic algaOstreococcus lucimarinus exhibit different ecological strategies. Environ. Microbiol.21, 2148–2170 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 2.Weitz J. S., Wilhelm S. W., Ocean viruses and their effects on microbial communities and biogeochemical cycles. F1000 Biol. Rep.4, 17 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 3.Ankrah N. Y., et al. , Phage infection of an environmentally relevant marine bacterium alters host metabolism and lysate composition. ISME J.8, 1089–1100 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 4.Lindell D., Jaffe J. D., Johnson Z. I., Church G. M., Chisholm S. W., Photosynthesis genes in marine viruses yield proteins during host infection. Nature438, 86–89 (2005). [DOI] [PubMed] [Google Scholar]
• 5.Breitbart M., Bonnain C., Malki K., Sawaya N. A., Phage puppet masters of the marine microbial realm. Nat. Microbiol.3, 754–766 (2018). [DOI] [PubMed] [Google Scholar]
• 6.Rosenwasser S., Ziv C., Creveld S. G. V., Vardi A., Virocell metabolism: Metabolic innovations during host-virus interactions in the ocean. Trends Microbiol.24, 821–832 (2016). [DOI] [PubMed] [Google Scholar]
• 7.Monier A., et al. , Phosphate transporters in marine phytoplankton and their viruses: Cross-domain commonalities in viral-host gene exchanges. Environ. Microbiol.14, 162–176 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 8.Monier A., et al. , Host-derived viral transporter protein for nitrogen uptake in infected marine phytoplankton. Proc. Natl. Acad. Sci. U.S.A.114, E7489–E7498 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 9.Raoult D., et al. , The 1.2-megabase genome sequence of Mimivirus. Science306, 1344–1350 (2004). [DOI] [PubMed] [Google Scholar]
• 10.Abergel C., Legendre M., Claverie J. M., The rapidly expanding universe of giant viruses: Mimivirus, Pandoravirus, Pithovirus and Mollivirus. FEMS Microbiol. Rev.39, 779–796 (2015). [DOI] [PubMed] [Google Scholar]
• 11.Colson P., La Scola B., Levasseur A., Caetano-Anollés G., Raoult D., Mimivirus: Leading the way in the discovery of giant viruses of amoebae. Nat. Rev. Microbiol.15, 243–254 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 12.Schulz F., et al. , Giant viruses with an expanded complement of translation system components. Science356, 82–85 (2017). [DOI] [PubMed] [Google Scholar]
• 13.Deeg C. M., Chow C. T., Suttle C. A., The kinetoplastid-infecting Bodo saltans virus (BsV), a window into the most abundant giant viruses in the sea. eLife7, e33014 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 14.Suttle C. A., Marine viruses–Major players in the global ecosystem. Nat. Rev. Microbiol.5, 801–812 (2007). [DOI] [PubMed] [Google Scholar]
• 15.Worden A. Z., et al. , Environmental science. Rethinking the marine carbon cycle: Factoring in the multifarious lifestyles of microbes. Science347, 1257594 (2015). [DOI] [PubMed] [Google Scholar]
• 16.Laber C. P., et al. , Coccolithovirus facilitation of carbon export in the North Atlantic. Nat. Microbiol.3, 537–547 (2018). [DOI] [PubMed] [Google Scholar]
• 17.Coutinho F. H., et al. , Marine viruses discovered via metagenomics shed light on viral strategies throughout the oceans. Nat. Commun.8, 15955 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 18.Santini S., et al. , Genome ofPhaeocystis globosa virus PgV-16T highlights the common ancestry of the largest known DNA viruses infecting eukaryotes. Proc. Natl. Acad. Sci. U.S.A.110, 10800–10805 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 19.Moniruzzaman M., et al. , Genome of brown tide virus (AaV), the little giant of the Megaviridae, elucidates NCLDV genome expansion and host-virus coevolution. Virology466–467, 60–70 (2014). [DOI] [PubMed] [Google Scholar]
• 20.Schvarcz C. R., Steward G. F., A giant virus infecting green algae encodes key fermentation genes. Virology518, 423–433 (2018). [DOI] [PubMed] [Google Scholar]
• 21.Fischer M. G., Allen M. J., Wilson W. H., Suttle C. A., Giant virus with a remarkable complement of genes infects marine zooplankton. Proc. Natl. Acad. Sci. U.S.A.107, 19508–19513 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 22.Gallot-Lavallée L., Blanc G., Claverie J. M., Comparative genomics of Chrysochromulina ericina virus and other microalga-infecting large DNA viruses highlights their intricate evolutionary relationship with the established Mimiviridae family. J. Virol.91, e00230-17 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 23.Wilson W. H., et al. , Complete genome sequence and lytic phase transcription profile of a Coccolithovirus. Science309, 1090–1092 (2005). [DOI] [PubMed] [Google Scholar]
• 24.Wilhelm S. W., et al. , A student’s guide to giant viruses infecting small eukaryotes: From acanthamoeba to zooxanthellae. Viruses9, E46 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 25.Keeling P. J., et al. , The marine microbial eukaryote transcriptome sequencing project (MMETSP): Illuminating the functional diversity of eukaryotic life in the oceans through transcriptome sequencing. PLoS Biol.12, e1001889 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 26.Yau S., et al. , Virophage control of antarctic algal host-virus dynamics. Proc. Natl. Acad. Sci. U.S.A.108, 6163–6168 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 27.Richter D. J., Fozouni P., Eisen M. B., King N., Gene family innovation, conservation and loss on the animal stem lineage. eLife7, e34226 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 28.Deng L., et al. , Grazing of heterotrophic flagellates on viruses is driven by feeding behaviour. Environ. Microbiol. Rep.6, 325–330 (2014). [DOI] [PubMed] [Google Scholar]
• 29.Gonzalez J. M., Suttle C., Grazing by marine nanoflagellates on viruses and virus-sized particles: Ingestion and digestion. Mar. Ecol. Prog. Ser.94, 1–10 (1993). [Google Scholar]
• 30.Labonté J. M., et al. , Single-cell genomics-based analysis of virus-host interactions in marine surface bacterioplankton. ISME J.9, 2386–2399 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 31.Frank N., Helge Abuldhauge T., Daniel J R., Bridging the gap between morphological species and molecular barcodes–Exemplified by loricate choanoflagellates. Eur. J. Protistol.57, 26–37 (2017). [DOI] [PubMed] [Google Scholar]
• 32.Limardo A. J., et al. , Quantitative biogeography of picoprasinophytes establishes ecotype distributions and significant contributions to marine phytoplankton. Environ. Microbiol.19, 3219–3234 (2017). [DOI] [PubMed] [Google Scholar]
• 33.Simmons M. P., et al. , Abundance and biogeography of picoprasinophyte ecotypes and other phytoplankton in the Eastern North Pacific Ocean. Appl. Environ. Microbiol.82, 1693–1705 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 34.Pennington J. T., Chavez F. P., Seasonal fluctuations of temperature, salinity, nitrate, chlorophyll and primary production at station H3/M1 over 1989-1996 in Monterey Bay, California. Deep Sea Res. Part II Top. Stud. Oceanogr.47, 947–974 (2000). [Google Scholar]
• 35.Derelle E., et al. , Diversity of viruses infecting the green microalgaOstreococcus lucimarinus. J. Virol.89, 5812–5821 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 36.Bachy C., et al. , Transcriptional responses of the marine green algaMicromonas pusilla and an infecting prasinovirus under different phosphate conditions. Environ. Microbiol.20, 2898–2912 (2018). [DOI] [PubMed] [Google Scholar]
• 37.Tomaru Y., Nagasaki K., Diatom Viruses (Springer, 2011). [Google Scholar]
• 38.Tulman E. R., et al. , The genome of canarypox virus. J. Virol.78, 353–366 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 39.Forterre P., The virocell concept and environmental microbiology. ISME J.7, 233–236 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 40.Yutin N., Wolf Y. I., Raoult D., Koonin E. V., Eukaryotic large nucleo-cytoplasmic DNA viruses: Clusters of orthologous genes and reconstruction of viral genome evolution. Vriol. J.6, 223 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 41.Sunagawa S., et al. ; Tara Oceans coordinators , Structure and function of the global ocean microbiome. Science348, 1261359 (2015). [DOI] [PubMed] [Google Scholar]
• 42.Yooseph S., et al. , The Sorcerer II Global Ocean sampling expedition: Expanding the universe of protein families. PLoS Biol.5, e16 (2007). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 43.Abrahão J., et al. , Tailed giant Tupanvirus possesses the most complete translational apparatus of the known virosphere. Nat. Commun.9, 749 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 44.De Corte D., et al. , Metagenomic insights into zooplankton-associated bacterial communities. Environ. Microbiol.20, 492–505 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 45.Blanc G., et al. , TheChlorella variabilis NC64A genome reveals adaptation to photosymbiosis, coevolution with viruses, and cryptic sex. Plant Cell22, 2943–2955 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 46.Thomas C. J., et al. , Localization of a baculovirus-induced chitinase in the insect cell endoplasmic reticulum. J. Virol.72, 10207–10212 (1998). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 47.Torruella G., et al. , Phylogenomics reveals convergent evolution of lifestyles in close relatives of animals and fungi. Curr. Biol.25, 2404–2410 (2015). [DOI] [PubMed] [Google Scholar]
• 48.Azam F., et al. , The ecological role of water-column microbes in the sea. Mar. Ecol. Prog. Ser.10, 257–263 (1983). [Google Scholar]
• 49.Ernst O. P., et al. , Microbial and animal rhodopsins: Structures, functions, and molecular mechanisms. Chem. Rev.114, 126–163 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 50.Gómez-Consarnau L., et al. , Proteorhodopsin phototrophy promotes survival of marine bacteria during starvation. PLoS Biol.8, e1000358 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 51.Martinez A., Bradley A. S., Waldbauer J. R., Summons R. E., DeLong E. F., Proteorhodopsin photosystem gene expression enables photophosphorylation in a heterologous host. Proc. Natl. Acad. Sci. U.S.A.104, 5590–5595 (2007). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 52.Govorunova E. G., Sineshchekov O. A., Li H., Spudich J. L., Microbial rhodopsins: Diversity, mechanisms, and optogenetic applications. Annu. Rev. Biochem.86, 845–872 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 53.Pushkarev A., et al. , A distinct abundant group of microbial rhodopsins discovered using functional metagenomics. Nature558, 595–599 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 54.Béjà O., Spudich E. N., Spudich J. L., Leclerc M., DeLong E. F., Proteorhodopsin phototrophy in the ocean. Nature411, 786–789 (2001). [DOI] [PubMed] [Google Scholar]
• 55.Gómez-Consarnau L., et al. , Microbial rhodopsins are major contributors to the solar energy captured in the sea. Sci. Adv. 5, eaaw8855 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 56.Marchetti A., Catlett D., Hopkinson B. M., Ellis K., Cassar N., Marine diatom proteorhodopsins and their potential role in coping with low iron availability. ISME J.9, 2745–2748 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 57.Yutin N., Koonin E. V., Proteorhodopsin genes in giant viruses. Biol. Direct7, 34 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 58.Lamarche L. B., et al. , Purification and characterization of RhoPDE, a retinylidene/phosphodiesterase fusion protein and potential optogenetic tool from the choanoflagellate Salpingoeca rosetta. Biochemistry56, 5812–5822 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 59.Yoshida K., Tsunoda S. P., Brown L. S., Kandori H., A unique choanoflagellate enzyme rhodopsin exhibits light-dependent cyclic nucleotide phosphodiesterase activity. J. Biol. Chem.292, 7531–7541 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 60.Olson D. K., Yoshizawa S., Boeuf D., Iwasaki W., DeLong E. F., Proteorhodopsin variability and distribution in the North Pacific Subtropical Gyre. ISME J.12, 1047–1060 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 61.Philosof A., Béjà O., Bacterial, archaeal and viral-like rhodopsins from the Red Sea. Environ. Microbiol. Rep.5, 475–482 (2013). [DOI] [PubMed] [Google Scholar]
• 62.López J. L., et al. , Microbial and viral-like rhodopsins present in coastal marine sediments from four polar and subpolar regions. FEMS Microbiol. Ecol.93, (2017). [DOI] [PubMed] [Google Scholar]
• 63.Spudich J. L., Jung K.-H., “Microbial rhodopsins: Phylogenetic and functional diversity” in Handbook of Photosensory Receptors, Briggs W. R., Spudich J. L., Eds. (Wiley, 2005), pp. 1–23. [Google Scholar]
• 64.Harris A., et al. , A new group of eubacterial light-driven retinal-binding proton pumps with an unusual cytoplasmic proton donor. Biochim. Biophys. Acta1847, 1518–1529 (2015). [DOI] [PubMed] [Google Scholar]
• 65.Bulzu P. A., et al. , Casting light on Asgardarchaeota metabolism in a sunlit microoxic niche. Nat. Microbiol.4, 1129–1137 (2019). [DOI] [PubMed] [Google Scholar]
• 66.Vogeley L., et al. , Anabaena sensory rhodopsin: A photochromic color sensor at 2.0 A. Science306, 1390–1393 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 67.Kyndt J. A., Meyer T. E., Cusanovich M. A., Photoactive yellow protein, bacteriophytochrome, and sensory rhodopsin in purple phototrophic bacteria. Photochem. Photobiol. Sci.3, 519–530 (2004). [DOI] [PubMed] [Google Scholar]
• 68.Mehler M., et al. , The EF loop in green proteorhodopsin affects conformation and photocycle dynamics. Biophys. J.105, 385–397 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 69.Sudo Y., et al. , A microbial rhodopsin with a unique retinal composition shows both sensory rhodopsin II and bacteriorhodopsin-like properties. J. Biol. Chem.286, 5967–5976 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 70.Tsukamoto T., Inoue K., Kandori H., Sudo Y., Thermal and spectroscopic characterization of a proton pumping rhodopsin from an extreme thermophile. J. Biol. Chem.288, 21581–21592 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 71.Luecke H., Schobert B., Richter H. T., Cartailler J. P., Lanyi J. K., Structure of bacteriorhodopsin at 1.55 A resolution. J. Mol. Biol.291, 899–911 (1999). [DOI] [PubMed] [Google Scholar]
• 72.Luecke H., Schobert B., Lanyi J. K., Spudich E. N., Spudich J. L., Crystal structure of sensory rhodopsin II at 2.4 angstroms: Insights into color tuning and transducer interaction. Science293, 1499–1503 (2001). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 73.Spudich J. L., The multitalented microbial sensory rhodopsins. Trends Microbiol.14, 480–487 (2006). [DOI] [PubMed] [Google Scholar]
• 74.Avelar G. M., et al. , A rhodopsin-guanylyl cyclase gene fusion functions in visual perception in a fungus. Curr. Biol.24, 1234–1240 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 75.Laundon D., Larson B. T., McDonald K., King N., Burkhardt P., The architecture of cell differentiation in choanoflagellates and sponge choanocytes. PLoS Biol.17, e3000226 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 76.Sabehi G., et al. , New insights into metabolic properties of marine bacteria encoding proteorhodopsins. PLoS Biol.3, e273 (2005). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 77.McCarren J., DeLong E. F., Proteorhodopsin photosystem gene clusters exhibit co-evolutionary trends and shared ancestry among diverse marine microbial phyla. Environ. Microbiol.9, 846–858 (2007). [DOI] [PubMed] [Google Scholar]
• 78.Pinhassi J., DeLong E. F., Béjà O., González J. M., Pedrós-Alió C., Marine bacterial and archaeal ion-pumping rhodopsins: Genetic diversity, physiology, and ecology. Microbiol. Mol. Biol. Rev.80, 929–954 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
• 79.Hingamp P., et al. , Exploring nucleo-cytoplasmic large DNA viruses in Tara Oceans microbial metagenomes. ISME J.7, 1678–1695 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

Supplementary Materials