Viruses infecting a warm water picoeukaryote shed light on spatial co-occurrence dynamics of marine viruses and their hosts
Isolation and characterisation of viruses infecting the picoeukaryote Bathycoccus Clade BIIBathycoccus BII isolates RCC716 and RCC715 used in our experiments were originally cultured from a nutrient-limited region in the Indian Ocean. Clade BII as a whole has been reported extensively in warm oligotrophic ocean gyres based on metagenome analyses [22,23,24]. Peak abundances occurr when well-developed deep chlorophyll maxima are present, or throughout the photic zone during mixing periods at Station ALOHA of the Hawaii Ocean Time-series [12]. We targeted BATS for viral isolation in springtime because Bathycoccus has been observed at relatively high abundance during this period using qPCR [74]. Here, three viruses were isolated against RCC716 [12] using seawater flown from BATS/Bermuda to the laboratory, obviating bringing this finicky strain into the field for use as a viral host. We then purified the viruses by serial dilutions and sequenced the partial PolB gene to determine whether they were evolutionarily different from other cultured viruses. BLASTn and preliminary phylogenetic analysis using GenBank nr sequences indicated they were distinct from described viruses with deposited sequences, with best BLASTn hits to Bathycoccus prasinos viruses (62–74% nucleotide identity). Transmission electron microscopy (TEM) revealed that all three viruses have similar morphology to other characterised prasinoviruses [75], with icosahedral capsids diameter ranging between 120 and 140 nm (Fig. 1A).Fig. 1: Morphology and evolutionary relationships of newly discovered Bathycoccus viruses.A Transmission electron micrographs of BII-V1, BII-V2 and BII-V3 (scale bar, 50 nm). The capsid diameters (n = 6 virions) measured 138 ± 2 nm (BII-V1), 150 ± 5 nm (BII-V2) and 152 ± 11 nm (BII-V3). B Maximum Likelihood (ML) phylogenetic reconstruction of green algal viruses inferred from a concatenated alignment of 22 core proteins shared among the viruses (7,001 positions) under the LG + G + F model. Node support was calculated from 1000 bootstrap (BS) replicates, with all branches acquired support values of 100% (white dots). Viruses infecting Chlorella were used as an outgroup and the branch connecting the prasinoviruses to the outgroup was truncated for display purpose. The new Bathycoccus viruses isolated against Bathycoccus Clade II (sensu [12]) isolate RCC716 (named as species Bathycoccus calidus herein, see below) are in bold. Colours reflect different host species within each genus. Letters alongside vertical lines (a and b) correspond to Bathycoccus viral clades. C Venn diagram of the shared and unique protein-encoding genes in the genome sequences of the new Bathycoccus viruses.Full size imageGenomic sequencing and multi-gene evolutionary analysesAssembly of DNA sequences from the viral isolates after deep sequencing by Illumina rendered one complete dsDNA genome sequence (BII-V3), and two others may still be partial (Table 1). The BII-V2 genome, which was in one contig, was similar in size (~208 kb) to that of BII-V3 (~212 kb). The BII-V1 genome assembly was ~174 kb and comprised of four linear dsDNA scaffolds. The viral concentrate was deeply sequenced ( >50x coverage) and minor fragmentation of the genome was partially related to repeats that were not resolved during assembly. The total number of putative open reading frames (ORFs) varied from 220 in BII-V1 to 235 in BII-V2 (Table 1). Gene synteny was globally well-conserved across the BII-Vs and the BpV1 and BpV2 viruses of B. prasinos (Fig. S1), with limited genomic rearrangements. Other genome characteristics such as the coding proportion (~90%) and G + C % (~36%) were similar to other described prasinoviruses infecting Mamiellophyceae [64, 75], for which the reported number of proteins range from 203 to 268 and G + C % from 37 to 45%. However, the full-length PolB gene from the genome assemblies differed for BII-V3 from the other two, in having a 329 amino acid intein (Supplementary information table S3). Likewise, inteins have been reported at the same PolB position in uncultivated prasinoviruses from the subtropical Pacific Ocean [76], where Bathycoccus BII is abundant [12].Table 1 Genomic characteristics of the three Bathycoccus viruses (BII-Vs) isolated against Clade BII isolate RCC716.Full size tableTo reconstruct a robust phylogeny for the new viruses, we employed 22 proteins previously identified as being shared across all available green algal virus genomes, including both prasinoviruses and chloroviruses [65]. We found all 22 in the predicted coding sequences of BII-V1; however, DNA helicase (SNF2) was not found in BII-V2 or -V3, FAD-dependent thymidylate synthase (thy1) and the topoisomerase IV were not found in BII-V2, nor was the prolyl 4-hydroxylase in the BII-V3 genome. Additional searches with tBLASTn did not recover these genes or fragments of them, suggesting they have been lost. Phylogenomic reconstruction grouped the three BII-Vs with the two BpVs [32], in a fully supported clade that branched adjacent to a large group of viruses that infect various species of Ostreococcus and Micromonas (Fig. 1B). The clade of Bathycoccus viruses was segregated in two subclades with BII-V2 and BII-V3 clustering together adjacent to BII-V1 and BpVs (Fig. 1B). While better resolution of the position of BII-V1 awaits greater taxonomic sampling, our results demonstrated that the three new viruses branch adjacent or basally to BpVs.Variation in prasinovirus gene content and functions encodedThe three Bathycoccus Clade BII viruses had 72–77% of their proteins held in common, and ~30 unique proteins as well as a few proteins shared by just two of the three viruses (Fig. 1C). The 170 shared proteins had higher amino acid identities between BII-V2 and BII-V3 (73% aa identity) than to BII-V1 (69% and 68%, respectively). Generally, only 19–21% of Bathycoccus viral genes could be assigned a functional category, based on EggNOG classification. Similar functional category distributions were observed across both prasinoviruses and chloroviruses, including lipid metabolism, RNA processing and modification, and nucleotide metabolism and transport (Fig. 2A). Other functional categories were more variable, such as cell wall/membrane/envelope biogenesis genes prevalent in chloroviruses (potentially related to their enveloped nature), as well as genes involved in modification of the capsid with compounds such as with chitin and hyaluronan [77, 78] that are absent from prasinoviruses sequenced to date (Fig. 2A). Within prasinoviruses, most of the unique proteins in the Bathycoccus viruses lack defined functional categories. Among those with functional assignments, all five Bathycoccus viruses encoded a P2X receptor in the intracellular trafficking and secretion category, and both BII-V2 and -V3 encode two proteins putatively involved in degrading the aromatic compound 4-hydroxy-2-oxopentanoate to acetyl-CoA (secondary metabolite category), that otherwise are only encoded by one other prasinovirus, MpV1 [32]. Similar to the phylogenetic relationships, the functional category distributions of BII-V1 were closer to those of BpVs than to BII-Vs. The primary difference was in carbohydrate metabolism, where BII-V2 and -V3 each encodes ribulose-phosphate 3-epimerase (involved in the pentose phosphate pathway and carbon fixation; not found in any other available virus genomes, but encoded by B. prasinos) and TDP-glucose 4,6-dehydratase (involved in biosynthesis of rhamnose and encoded by most other chloroviruses and prasinoviruses [79]). Notably, the putative high-affinity phosphate transporter (PHO4, also termed HAPT) was only present in BII-V1 and BpV1, as well as OtV2 (isolated against the Ostreococcus Clade OII ecotype), and most sequenced viruses of O. lucimarinus (Supplementary information table S3). This gene is hypothesised to enhance phosphate uptake during infection under phosphorus‐limited host growth [25], as observed for the PstS phosphate transport system expressed by cyanophages [80], mitigating limitation of this key component of viral genomes. However, most isolated prasinovirus genomes come from waters that are not considered phosphate-limited, hence presence of this gene may connect to poising the host for responding to sudden availability of other nutrients, such as nitrogen, which is often limiting in the ecosystems from which these viruses were isolated. Studies of virus-cell responses under various limiting nutrients are required to understand the retention of this host-derived HGT.Fig. 2: Distribution of functions and orthologous protein families across genome-sequenced prasinoviruses.A Functional category distributions across 21 genome-sequenced prasinoviruses and chloroviruses based on EggNOG categorisation. Viruses are clustered by similarity in their distribution of the functional categories on the y-axis and the frequency of each category across the viral genomes determines clustering along the x-axis ordering. Genes with homology to proteins in the EggNOG database but could not be assigned a function are in the “function unknown” category. B Orthogroups presence/absence patterns ordered along the x-axis by ranking according to the total number of genes in the orthogroup. For inclusion, the orthogroup was required to include protein sequences from at least two different viral genomes. Viruses are ordered along the vertical by their presence/absence pattern reconstructed by hierarchical clustering (topology on the left). Top histogram: frequency of each orthogroup in sequenced prasinoviruses. C Genes in each virus (number) not assigned to any orthogroup, with viruses in the same vertical order as B.Full size imageHierarchical clustering of orthologous proteins revealed patterns across prasinoviruses that generally corresponded with phylogenetic relationships. The BII- and Bp-viruses shared 130 orthologous proteins and hierarchical clustering (Fig. 2B) followed the clade structure of the phylogenomic reconstruction (Fig. 1B) with the exception of BII-V1 that grouped with BII-Vs, as well as OtV6, which grouped with Micromonas viruses. These orthologous proteins had on average 72% amino acid identity between BII-V2 and BII-V3, and 88% between the two B. prasinos viruses, but between 65 to 67% when comparing members of these two groups (Table 2). BII-V1 orthologs also had 67% and 66% amino acid identity to BII-V2 and BII-V3, respectively, while they had 83% and 80% identity to BpV1 and BpV2, respectively. Collectively, these results indicate that BII-V2 and -V3 diverged from BpVs prior to the divergence of BII-V1.Table 2 Average percent amino acid identity of the orthologous proteins between the five Bathycoccus viruses.Full size tableOf the 130 orthologous Bathycoccus virus proteins, 37% were assigned putative functions revealing core components of this viral group (Supplementary information table S3). These included genes involved in DNA replication and transcription, including PolB (type II), a DNA topoisomerase, a transcription factor S-II, mRNA capping enzymes, ribonucleases, a ribonucleotide reductase, and a dUTPase. Several others are necessary for viral particle synthesis, such as genes encoding structural elements for assembling the virion, including capsid proteins (5–6 copies per genome), as well as transcriptional regulators connected to the replication cycle. The BII viruses showed a number of differences among orthologous protein families. In addition to each having “unique” protein sets, there was a set of BII-V specific orthogroups, as well as some shared with BpVs, and/or other prasinoviruses (Fig. 1C and Supplementary information table S3). First, six predicted proteins showed orthologs across the three BII-Vs, but were not present in other prasinoviruses sequenced to date. Only one of these six was assigned putative function, belonging to the XRE family of transcriptional regulators. Additionally, all BII viruses harboured a tandem duplication of the FstH gene, while other sequenced prasinoviruses (including the two Clade BI viruses) have one copy (Supplementary information table S3). This ATP-dependent metalloprotease has been shown to be involved in photosystem II repair in cyanobacteria [81], and is present in genomes of photosynthetic eukaryotes, including all Mamiellophyceae [15, 16]. In Arabidopsis and Chlamydomonas it has been shown to be involved in protein quality control in the thylakoid membranes [82]. A gene of unknown function was also duplicated in the BII-virus genomes, that is a single copy in BpVs and absent from other sequenced prasinoviruses. Genes putatively encoding a glucose-1-phosphate adenylyltransferase, a glycosyltransferase and a thiamine pyrophosphate-requiring enzyme involved in amino acid biosynthesis were sporadically found in BII-viruses.Considering the two Bathycoccus virus subclades (Fig. 1B), there is one predicted protein of unknown function exclusive to BpV1, BpV2 and BII-V1 and six predicted proteins shared only by BII-V2 and BII-V3. Among the latter, one belonged to the Ribulose-5-Phosphate-3-Epimerase (RPE) family, which catalyses the interconversion of D-ribulose 5-phosphate (Ru5P) into d-xylulose 5-phosphate, as part of the Calvin cycle (although no transit peptide was detected using TargetP) and in the oxidative pentose phosphate pathway. The ortholog analyses further showed that among prasinoviruses, 9, 17 and 18 genes were unique to BII-V1, BII-V2 and BII-V3, respectively (Fig. 2B). Apart from one nucleotidyltransferase and one glycosyltransferase (group 1) in BII-V1, none of these unique genes had known functions.To study the evolutionary aspects of the shared prasinovirus proteins, we constructed and examined 130 phylogenies of orthogroups shared between Bathycoccus viruses. Nine showed a topology where all three BII-Vs grouped together with full support (100% bootstrap support), separate from the BpV orthologs, and in contrast to the multi-gene phylogeny where BII-V1 grouped with BpVs (Fig. 1B). The average amino acid similarities within each of these nine protein ortholog groups ranged from 85 to 88% between BII-Vs proteins, while they were 77 to 81% between BII-Vs and BpVs, different from overall amino acid similarity averages (Table 2). Interestingly, proteins from three of these nine ortholog groups, all lacking known functions, were adjacent to each other in the genome, or separated by only one gene. This synteny and co-location likely reflects the acquisition of these genes before co-infecting viruses diverged via viral recombination [83].Infection dynamics of Bathycoccus virusesGeneral host specificity of BII-viruses was assessed using two B. prasinos isolates (CCMP1898 and RCC4222, Clade BI), the two available Clade BII isolates (RCC715 and RCC716), four Ostreococcus species and one Micromonas species (Table 3). None were able to infect the B. prasinos, Ostreococcus or Micromonas isolates tested, suggesting BII-V specificity for Bathycoccus Clade BII. Similar host specificity has been observed in O. lucimarinus viruses, none of which infect O. tauri [64], and other viruses of eukaryotic and prokaryotic algae [84, 85]. Some other prasinoviruses appear to have broader host ranges [85,86,87], or their host species are less divergent than the two known Bathycoccus clades. For example, generalist viruses isolated against Micromonas commoda can infect M. bravo [85]. Further investigations are necessary to determine the extent to which the six shared proteins in BII-Vs (absent from BpVs), are responsible for the differences in host and virus specificity of interactions, versus variations in the shared Bathycoccus virus proteins (65–83% similarity). Importantly, host specificity tests for the new viruses described herein were limited by weak sampling of Bathycoccus diversity (in culture; all that we could acquire were tested).Table 3 Results of cross infectivity tests of BII-V1, BII-V2 and BII-V3 against isolates representing various picoprasinophyte species within the Class Mamiellophyceae.Full size tableAlthough specific for the BII clade, the three BII-Vs exhibited variations in infectivity of the two cultured BII strains available, despite their isolation from the same sample and having identical ITS1 and ITS2 sequences. BII-V1 lysed and cleared RCC715 and RCC716 cultures after four days (Table 3). The same was true for BII-V2 and BII-V3, when incubated with RCC716. Different from results for BII-V1, we found that while BII-V2 and -V3 initially lysed RCC715 cultures, resistant populations became evident at day 7 of infectivity tests, and measureable lysis of RCC715 could not be achieved thereafter. These results underscored the need to further examine host-virus interactions for the three new viruses.Infection dynamics over time course experiments further illuminated differences in BII-V impacts on hosts. In these experiments, growth rates of the uninfected (control) RCC715 and RCC716 cultures were 0.45 ± 0.04 day−1 and 0.49 ± 0.06 per day, respectively, similar to rates during the pre-experiment acclimation period (T-test, p > 0.05). Host and virus dynamics were similar for RCC715 and RCC716 infected with BII-V1 (Fig. S2 and Fig. 3), with cell numbers starting to diverge from control abundances 10 h after inoculation (T-test, p More
