Abstract
Climate change-driven ocean warming is increasing the frequency and severity of bleaching events, in which corals appear whitened after losing their dinoflagellate endosymbionts (family Symbiodiniaceae). Viral infections of Symbiodiniaceae may contribute to some bleaching signs, but little empirical evidence exists to support this hypothesis. We present the first temporal analysis of a viral lineage—the Symbiodiniaceae-infecting ‘dinoRNAVs’—in coral colonies exposed to a 5-day heat treatment. Throughout the experiment, all colonies were dominated by Symbiodiniaceae in the genus Cladocopium, but 124 dinoRNAV major capsid protein ‘aminotypes’ (unique amino acid sequences) were detected across coral genets and treatments. Seventeen dinoRNAV aminotypes were found only in heat-treated fragments, and 22 aminotypes were detected at higher relative abundances in heat-treated fragments. DinoRNAVs also exhibited higher alpha diversity and dispersion under heat stress. Together, these findings provide the first empirical evidence that exposure to high temperatures triggers some dinoRNAVs to switch from a persistent to a productive infection mode within heat-stressed corals. Over extended time frames, we hypothesize that cumulative dinoRNAV lysis of Symbiodiniaceae cells during productive infections could decrease Symbiodiniaceae densities within corals, observable as bleaching signs. This study sets the stage for reef-scale investigations of dinoRNAV dynamics during bleaching events.
Introduction
Warming seas, driven by climate change, are increasingly causing bleaching events: mass losses of endosymbiotic dinoflagellates (family Symbiodiniaceae) from corals and other invertebrate hosts. Bleaching events often result in coral mortality and are contributing to the degradation of reef ecosystems globally (1, 2). Viruses, which are diverse and abundant on coral colonies (3–7), are hypothesized to contribute to some coral bleaching signs by lysing Symbiodiniaceae cells (e.g. 8–10). Alternatively, viral shifts in conjunction with bleaching-associated stressors (11,e.g. 12–14) could merely be correlated with bleaching signs or constitute opportunistic secondary infections (15, 16). Beyond bleaching, viruses may influence colony health by altering the function of resident microbial symbionts or coral tissues (e.g. 17–23). Although various roles for viruses in coral bleaching, disease and function have been hypothesized, thus far, these roles have been difficult to test empirically.
Symbiodiniaceae are putative target hosts of DNA and RNA viruses (reviewed in 6,7), including the ‘dinoRNAVs’, a group of dinoflagellate-infecting positive-sense single stranded RNA viruses. Although dinoRNAVs have yet to be isolated, stably propagated, and fully characterized, they have been detected from Symbiodiniaceae cultures, as well as Atlantic and Pacific corals spanning 6 genera (Table 1). These associations suggest that dinoRNAVs are prevalent as persistent infections in Symbiodiniaceae cells (8–10,24,25). Furthermore, increased dinoRNAV detection (24) and enhanced host anti-viral response (26) suggest that under stressful conditions, dinoRNAVs may switch to a more productive replication mode in which they directly lyse their hosts. This infection strategy has recently been identified in other algal host- virus systems (e.g., coccolithophore Emiliana huxleyi-EhV system, 27).
Heterocapsa circularisquama RNA virus (HcRNAV), which infects free-living dinoflagellates, is among the closest known relatives to Symbiodiniaceae-infecting dinoRNAVs (24, 28). HcRNAV undergoes a strictly lytic replication cycle following a latent period of 24-48 hours, during which the host is infected but not lysed and viruses have not been released (29). We therefore posited that shifts by Symbiodiniaceae-infecting viruses into a more productive replication mode might also be detectable within the first few days of exposure to stress (13,24,26). Examining viral dynamics within individual colonies at the onset of thermal stress should clarify the relationship between viral infection of Symbiodiniaceae and coral bleaching signs.
To investigate the role of Symbiodiniaceae-infecting viruses in coral bleaching, we quantified, for the first time, the temporal dynamics of dinoRNAVs within individual coral colonies. To accomplish this, colonies of the stony coral Pocillopora species complex were exposed to acute (+∼2°C) thermal stress, and dinoRNAV diversity was quantified over a 5-day period. We hypothesized that: (1) dinoRNAVs are common in coral colonies; (2) dinoRNAV richness increases and compositions shift under thermal stress; and (3) changes to dinoRNAVs occur within 72 hours. By analyzing dinoRNAV diversity at the amino acid level, this study partially circumvented methodological challenges arising from the high mutation rates and genetic diversity of single-stranded RNA viruses (30–32), which have previously made it difficult to compare RNA viral dynamics across ecologically relevant scales (33). As dinoRNAV work on reefs continues to progress, the aminotypes presented here may eventually merit further collapse into ‘quasispecies’—heterogeneous mixtures of related genomes (33–36)—a common approach for conceptualizing diversity within RNA virus populations.
Materials and methods
Experimental design
We conducted a replicated aquarium experiment (two treatments; four aquariums per treatment) in which fragments from five colonies of Pocillopora species complex (37) were exposed to control conditions (ambient reef water; 28.2°C) or a +2.1°C heat treatment (summer bleaching temperatures; 30.3°C) for 5 days (See Figure S1 and Supplementary Methods, 38,39). At the start of the experiment (t(h) = 0), all fragments were photographed with a Coral-Watch Health Monitoring Chart (40) in the frame, and one fragment per colony in the control aquaria was preserved as an initial control. At five time points (t(h) = 4, 12, 24, 72 and 108 h), all fragments were photographed again and visually inspected for signs of stress (e.g., excessive mucus production), lesions and/or paling. A control and a heat-stressed fragment per colony were also preserved at each time point (generating 10 fragments per time point). DNA and RNA were extracted from each sample (which included coral animal tissue, Symbiodiniaceae cells and viruses) using a ZymoBIOMICS DNA/RNA Miniprep Kit (Zymo Research, Irvine, CA, USA) with an additional enzyme digestion step to improve viral RNA yields; cDNA was then synthesized from eluted RNA (see Supplementary Methods). At the 24 and 72 h timepoints, samples were also fixed for transmission election microscopy (TEM) imaging in 3X PBS with 2% paraformaldehyde (see Supplementary Methods).
To characterize the effect of the heat treatment on Symbiodiniaceae cell densities (a metric of coral health and bleaching status), we compared brightness values from standardized photographs of each coral fragment—a proxy for Symbiodiniaceae chlorophyll concentrations— at the start of the experiment and at the time of preservation (see Supplementary Methods, 40,41). Symbiont diversity was characterized by sequencing the internal transcribed spacer-2 (ITS-2) region of Symbiodiniaceae rDNA from fragments of each coral colony using primers from Hume et al (42) following methods in Howe-Kerr et al (43; see Supplementary Methods). Sequences were processed using Symportal (44).
Sequencing of dinoRNAV major capsid protein gene amplicons, bioinformatics processing, and phylogenetic analysis
The dinoRNAV major capsid protein (mcp) gene was amplified from cDNA libraries using a nested PCR protocol with degenerate primers (28); cleaned and normalized libraries were sequenced on the Illumina MiSeq platform using PE300 v3 chemistry. Processing and analysis of dinoRNAV mcp gene reads was conducted using the program vAMPirus (See Supplementary Methods for sequencing and read processing details; Veglia et al., 2021). Briefly, ASVs were generated using the UNOISE (46) algorithm with vsearch (47), and all ASVs were then translated and collapsed into unique amino acid sequences, ‘aminotypes’. Any sequences containing stop codons were removed prior to further analysis.
An alignment was made in MUSCLE using aminotypes from this study, a dataset from the Great Barrier Reef (28) that was reprocessed using the methods above, and a set of reference best BLASTx hits to the NCBI database (48). The alignment (Supplementary Data 1) was trimmed to the first column on either side which contained no gaps, and then used to determine the best model for evolution (LG+G4+I) according to ModelTest-NG (49). A maximum- likelihood phylogeny was inferred with RAxML-NG (1000 bootstrap iterations) and rooted with HcRNAV as an outgroup.
DinoRNAV diversity metrics based on mcp aminotypes
All data processing, visualization, analysis and statistical tests were conducted in R version 4.0.2 and Vegan 2.5-6 (50) on an aminotype counts table (Supplementary Data 2). For some analyses, the dataset was rarefied to 59 837 amino acid sequences. Shannon’s diversity index (H) value were calculated based on rarefied data; expected aminotype richness values were calculated using repeated random subsampling of non-rarefied data (sample size=59 837 amino acid sequences). Venn diagrams were made from non-rarified data using the online tool http://bioinformatics.psb.ugent.be/webtools/Venn/ (accessed August 17th, 2020). For comparison, Venn diagrams were also made based on rarified data (results not shown); these Venn diagrams exhibited similar patterns but were less conservative and were therefore not included.
To quantify the dispersion of dinoRNAVs between treatments and over time, a non- metric multidimensional scaling (NMDS) plot was constructed based on Bray-Curtis distances from square-root-transformed rarefied data (k=2, 999 iterations), and the distance to centroid for each sample was calculated. Since dinoRNAVs differed among coral colonies, we calculated centroids for each individual colony in the control and heat treatments separately (5 colonies x 2 treatments = 10 centroids) in order to examine the effect of heat treatment on dinoRNAV dispersion in a given coral colony. For this analysis, different timepoints were used as replicates.
Statistical analyses
We tested for differences in brightness values from the coral fragment colorimetric analysis, as well as differences in dinoRNAV aminotype alpha diversity (Shannon’s index and aminotype richness) and dispersion (distance to centroid) using linear mixed effects models (LMM) with the package LME4 v1.1-23 (40, 51). We tested for an interaction between treatment and timepoint; by incorporating colony ID as a random effect, this analysis approaches a repeated measures test. F-tests were used for model selection with car v3.0-8. Assumptions of normality of the residuals were assessed visually with quantile-quantile plots and Shapiro-Wilk tests; the assumption of homogeneity of variance was visually assessed using plots with residuals versus fitted values. Dispersion values did not follow the assumption of normality of the residuals and were square root-transformed. We also tested for differences between control and heat-treated fragments at each timepoint (using colonies as replicates), as well as per colony (using time points as replicates), on dinoRNAV aminotype diversity and dispersion with ANOVAs and controlled for type 1-errors using a Bonferroni correction.
The effect of heat treatment on the overall composition of dinoRNAVs was tested with a PERMANOVA using adonis() in vegan based on Bray-Curtis distances from square-root- transformed rarefied data. We tested for an interaction between treatment and colony and added time as an additional factor. Although dispersion differed significantly among groups (betadisper test), PERMANOVA is robust to heterogeneous dispersion if the design is balanced (52).
Lastly, we conducted a differential abundance analysis using the non-rarefied amino acid counts table with DESeq2 v1.26.0 (53). We fitted a negative binomial model and Benjamini-Hochberg FDR-corrected Wald tests (=0.05) were used to test for differences in taxon abundance between treatment within each colony and at each timepoint after the start of the experiment (t(h)= 4, 12, 24, 72, 108). We excluded all fragments sampled at timepoint 0 and any fragments with <10 000 reads, as well as the fragment from the other treatment corresponding to the fragment with insufficient reads at the same time (colonies 3 and 5 at timepoint t(h)=4, colony 3 at t(h)=72).
Results
Coral holobiont traits
Coral fragments in the control and heat aquariums remained apparently healthy throughout the experiment; no signs of stress such as paling, mucus production, or tissue sloughing were observed. Linear mixed effects models of color values did not reveal significant paling in heat-treated coral fragments (treatment F=0.10, p=0.75; timepoint F= 0.30, p=0.59; treatment*timepoint F= 0.09, p= 0.77). Lack of color change was expected; bleaching signs are generally only detectable after weeks of ecologically relevant temperature stress—even though vital molecular processes are affected after several days (54). All colonies contained what we interpret as a species of Symbiodiniaceae closely related to Cladocopium goreaui (44, 55); Symbiodiniaceae diversity did not change over time (Figure S2).
Transmission electron microscopy imaging of Symbiodiniaceae and associated virus-like particles
Symbiodiniaceae cells in coral fragments exposed to 24 hours of heat treatment showed a pattern of vacuolization in the chloroplasts (Figure 1a) compared to cells in control fragments (Figure 1g). Many vacuoles in Symbiodiniaceae cells contained one to several VLPs with icosahedral, electron-dense capsids that were 110-170 nm in size and lacking envelopes (Figure 1a-d). Similar VLPs were observed immediately outside Symbiodiniaceae cells; some also appeared to be endocytosed by—or exiting out of—Symbiodiniaceae cells (Figure 1e-h).
DinoRNAV major capsid protein (mcp) gene sequencing overview
Amplicon sequencing of the dinoRNAV major capsid protein (mcp) gene resulted in 10 222 055 paired raw reads from all samples and one negative control. A total of 7 593 537 reads with a mean length of 423 bases (before trimming to 422 bases) remained after merging and quality control. Denoising resulted in 273 unique amplicon sequence variants (ASVs) across all samples, and translation revealed that 11 ASVs contained stop codons; these ASVs were removed. Translated ASVs collapsed into 124 ‘aminotypes’ (unique amino acid sequences) at a length of 140 amino acids.
DinoRNAV mcp genes were detected in fragments from all 5 coral colonies, but 5 control samples and the negative control were not retained for further analysis because they contained few (<10 000) amino acid sequences (the negative control had 20 amino acid sequences). All remaining samples had between 50 000 and 210 000 amino acid sequences with a mean read depth of 142 810 ± 36 163 (SD).
DinoRNA viruses in Pocillopora species complex colonies are closely related to dinoRNAVs in other coral species
The 124 unique dinoRNAV aminotypes (Figure 2; Table S1) identified in this study were most similar to 13 amino acid sequences from a previously published dinoRNAV library (28); sequence similarities to aminotypes from published dinoRNAVs ranged from 52.1-99.3% (e- values ranged between 9.6·10-73-2.7·10-35, Table S1). All dinoRNAV mcp aminotypes in this study form a clade that is more recently derived than reference sequences such as HcRNAV, Beihai sobemo-like virus and sponge weivirus-like virus (Figure 2). The dinoRNAV aminotypes in this study may form (at least) three quasispecies (e.g. red rectangles in Figure S3, sensu 56). Strikingly, the majority of dinoRNAV mcp genes identified in our study occupy one clade; on average, sequences in this clade vary 9.8% ± 5.5 (SD) from each other (Figure 2; Figure S3). Ultra-deep sequencing may be necessary to further clarify the drivers of these potential ‘mutant cloud dynamics’ in Symbiodiniaceae hosts (57). In several branches of the tree, aminotypes from our study are most closely related to aminotypes detected from Symbiodiniaceae in Acropora tenuis, Favia fungites, Galaxea fascicularis, Pocillopora damicornis, Porites cylindrica, and Porites lutea corals sampled from the Great Barrier Reef (28). Dominant aminotypes (>1% abundance across total dataset) are present in all major branches of the tree but differ by as much as ∼40% in amino acid sequence (Supplementary Figure 3). Aminotypes detections varied across treatment and time in the experiment (shading of red and black squares in Figure 2).
DinoRNAV aminotypes differed among coral colonies and treatments
Approximately 34% (42 of 124 aminotypes) of aminotypes in this study were present in all colonies, 48% (60) were shared between 2-4 colonies and 18% (22) were unique to individual coral colonies (with individual colonies having up to 6 unique colony-specific aminotypes, Figure 3a). All 14 aminotypes that each comprised >1% reads in the total dataset (listed in Figure 3b) were detected in all five coral colonies. DinoRNAV aminotypes varied among individual colonies (Figure 3b); colony ID was the most powerful predictor of viral composition (PERMANOVA: R2 = 0.76, p<0.001). DinoRNAVs responded to elevated temperatures in colony-specific ways, as indicated by a significant interaction effect between treatment and colony (PERMANOVA: R2=0.068, p<0.001). Treatment was also significant by itself (PERMANOVA: R2= 0.02, p<0.001), demonstrating a subtle but consistent response to elevated temperatures across all colonies. Time was not a significant predictor (PERMANOVA: R2= 0.01, p=0.77; Figure 3b).
Heat treatment rapidly increased the diversity of dinoRNAV aminotypes
A total of 17 aminotypes were unique to the heat treatment; 1 aminotype was unique to the controls (Figure 4a; Table 2). Heat-specific aminotypes were observed in all colonies and ranged from 1-6 unique aminotypes per colony. Three heat-specific aminotypes were shared among multiple (2–3) coral colonies; the remaining 14 heat-specific aminotypes were not shared between colonies. These findings demonstrate that heat-specific aminotypes were not restricted to individual fragments but were shared between fragments of the same colony or among colonies. However, most heat-specific aminotypes were relatively rare, with only two aminotypes comprising >1% of reads in each fragment (aminotypes 25 with 1.8%; aminotype 114 with 1.4%; other aminotypes with means of 0.01-0.7% of reads in individual fragments). The single aminotype that was unique to control fragments was identified in one fragment after 72 hours (0.02% of reads in that fragment).
Alpha diversity (Shannon’s index, H) of aminotypes was positively associated with the heat treatment (F=4.92, p=0.03) and time (F=4.86, p=0.03) in our linear mixed effects model (LMM; Figure 4b, d, h). There was no significant interaction between heat and time (F=0.54, p=0.47). On average (± SE), Shannon’s index was 27% higher in heat-treated fragments (1.1 ± 0.6) than controls (0.8 ± 0.6; Figure 4b). Within individual timepoints, heated fragments had 4- 93% higher mean H values (0.9-1.3) than controls (0.7-1.0; Figure 4d), but individual comparisons were not significant. Means of Shannon’s index per colony ranged between 0.1-1.6 for control fragments and 0.5-1.8 for heat-treated fragments (Figure 4f).
There was a significant positive association between viral aminotype richness and heat treatment (F=5.71, p=0.02), but not aminotype richness and time (F=0.36, p=0.55) in our LMM (Figure 4c, e, i). There was no significant interaction between treatment and time (F=0.10, p=0.75). On average, aminotype richness was 16% higher in heat-treated (37.2 ± 2.0) than control fragments (32.0 ± 2.1; Figure 4c). At individual timepoints, heat-treated fragments had 7- 23% higher mean aminotype richness (34.8-39.1) than controls (29.2-35.4; Figure 4e), but these differences were not significant. Mean aminotype richness for individual colonies (using timepoints as replicates) ranged between 19.5-44.6 for control fragments and 28.0-44.9 for heat- treated fragments (Figure 4g).
Twenty-two aminotypes had higher relative abundances in heat-treated fragments
DESeq2 analysis revealed 28 aminotypes had significantly altered relative abundances in heat- treated or control fragments (Figure 5). Twenty-two aminotypes had higher relative abundances in heated fragments, whereas 6 aminotypes had higher relative abundances in controls. Ten aminotypes were differentially abundant at multiple (2-4 out of 5) timepoints throughout the experiment. Aminotypes 29, 38 and 98 were significantly more abundant in the heat treatment in 4 out of 5 sampled timepoints (from 12 to 108 h); Aminotype 93 was more abundant in the final 3 timepoints (from 24 to 108 h).
Dispersion of dinoRNAV mcp amino acid sequences increased in heat-treated fragments
Dispersion (measured as distance to centroid) of dinoRNAVs was positively associated with heat treatment (Figure 6a, b, e; F=11.00, p=0.002) in our LMM. There was no effect of time (F=1.07, p=0.31) and no interaction between treatment and time (F=0.64, p=0.43). Overall, mean (±SE) dispersion was 62% higher in the heat treatment when all colonies and timepoints were pooled (control: 0.14 ± 0.09; heat: 0.23 ± 0.13). A trend of increasing dispersion (32-100% higher) of dinoRNAVs was observed in heat-treated samples at individual timepoints (Figure 6c; controls: 0.12-0.17; heat: 0.2-0.29), but individual comparisons were not significant. Mean dispersion of individual colonies (across timepoints) ranged between 0.1-0.24 for controls and 0.13-0.37 for heat-treated fragments, resulting in an increase of 11-93% per colony, but individual comparisons were not significant (Figure 6d).
Discussion
Viruses can have diverse impacts on hosts, ranging from antagonistic to beneficial (58–61). Efforts to understand how viral infections impact coral colonies have been stymied by the lack of (1) a high-throughput approach to track a viral lineage in colonies across an acute stress event; and (2) established cultures of viruses associated with corals (for use in viral addition experiments). This study tracks a group of Symbiodiniaceae-infecting viruses, the dinoRNAVs, in a controlled experiment to interrogate how infection responds to temperatures associated with bleaching. DinoRNAVs were detected from all five coral colonies examined, and from every heat-treated coral fragment, but only some controls. These observations, as well as detections of higher alpha diversity of mcp aminotypes, increased dinoRNAV dispersion, the identification of unique mcp aminotypes, and greater relative abundances of specific aminotypes in heat-treated fragments, together strongly indicate that dinoRNAV infections were more active in heat-treated fragments. DinoRNAV responses were detectable within a single day, more quickly than the several weeks over which bleaching signs typically manifest. If viral infection of Symbiodiniaceae cells is enhanced (e.g., increased production, accumulation of viral diversity) during thermal anomalies in situ, then this cumulative viral activity (if maintained over weeks) could contribute to bleaching onset on reefs, or otherwise disrupt of coral-Symbiodiniaceae partnerships.
DinoRNAVs as a common, persistent virus of Symbiodiniaceae
DinoRNAV mcp genes were detected in all colonies (N=5), and in most to all fragments per colony (73-100% of fragments, N=8/11 – 11/11 fragments per colony). DinoRNAV genes have additionally been reported from colonies of seven other stony coral species across the Atlantic and Pacific Oceans (Table 1, Figure 2), suggesting that these viruses are commonly associated with coral microbiota. In this experiment, viral aminotype compositions (and potential viral quasispecies) differed among coral colonies, but were similar within the fragments of a given colony, despite maintenance in separate aquaria (Figure 3). These results suggest that, under ambient conditions, dinoRNAV populations are driven strongly by within-colony factors and are relatively homogeneously distributed throughout entire coral colonies. RNA viruses rely on RNA-dependent RNA polymerase (RdRp) for replication, which is relatively error-prone. Therefore, individual viral progenitors infecting Symbiodiniaceae cells in a given colony may each produce a variety of genetically distinct viral “progeny” during a single replication cycle (34, 62); hence, the observed pattern of colony-specificity in dinoRNAV aminotypes likely results from diversification within ‘quasispecies’ and (potentially) subsequent purifying selection (33,63–65). This production of a ‘mutant cloud’ of dinoRNAV diversity may even help ensure these viruses are successful (at the population level) at infecting Symbiodiniaceae under changing environmental conditions (35).
Healthy corals contain millions of Symbiodiniaceae cells per cm2 of coral tissue; our results suggest that Symbiodiniaceae cells in hospite may be infected by one or perhaps several dinoRNAV quasispecies at any given time (e.g., Symbiodiniaceae in colonies 2 and 3, Figures 2 and 3). Recent surveys of free-living marine microbial communities reported that viral infections may occur in ∼33% (66, 67) to over 60% (68) of marine microorganisms, and many individual cells may be infected by several viruses at a given time (66, 69). Considering that our sequencing data are based on mcp gene amplicons from RNA libraries generated from unfractionated coral tissue, dinoRNAV mcp gene detections could potentially arise from several sources, including RNA genomes within intact dinoRNAV capsids (28), mcp genes that are being expressed in host cells during the dinoRNAV replication cycle, free dinoRNAV genomes that may occur as extrachromosomal RNA in a latent (25) or carrier state similar to pseudolysogeny (70, 71), and/or expressed endogenized viral elements in host genomes (72, 73). The amplicon sequencing approach employed here limits our ability to discern amongst these potential sources of viral mcp gene detections, as does the dearth of information available on dinoRNAV replication cycles. Approaches such as single cell RNA-seq (e.g. 74) of Symbiodiniaceae cells, as well as sequencing methods designed to identify and characterize infective viruses (similar to viral tagging, 75,adsorption sequencing, 76) represent critical next steps in assessing the prevalence of dinoRNAV infections within individual Symbiodiniaceae cells, and the provenance of dinoRNAV mcp gene detections (e.g RNA genomes from intact viruses vs mcp genes expressed during the replication cycle), respectively.
Switching from persistent to more productive infection under stress
Marine viruses have diverse infection strategies; many viruses switch between strategies upon environmental changes (58,77–81). Some marine viruses, for example, may switch between latent and productive cycles upon a variety of host-related or environmental triggers (8,82,83). RNA virus infections can range from exclusively lytic (29,84,85), to infections that are ‘persistent’ or ‘chronic’ and do not immediately kill the host (86, 87). Persistent RNA viruses of plants, for example, may show altered activity based on seasonality (88), temperature (89, 90), or other environmental factors (90). For persistent viruses of Symbiodiniaceae, a variety of mechanisms may modulate viral infection strategies; the role of temperature in triggering persistent infections to become more productive has received particular attention due to the link between high temperature and coral bleaching (8,24,26,83,91).
We identified 22 aminotypes that were present at higher relative abundances in heat- treated coral fragments (Figure 5), and 17 aminotypes were unique to these fragments (Figure 4a, c). We hypothesize that these findings indicate a switch from persistent to more productive infections by some dinoRNAV strains (or quasispecies). In this study, aminotypes 29, 38, 93 and 98 are particular candidates for this hypothesis, given their sustained relative increase in abundance in heat-treated corals starting at 12-24 hours (Figure 5). We encourage future studies to screen corals and Symbiodiniaceae communities (and existing sequencing libraries) for these aminotypes.
The observation of a rapid increase in dinoRNAV aminotypes diversity during the onset of thermal stress is consistent with a previous report of increased (but overall low) abundance of dinoRNAV transcripts in the Caribbean coral Montastrea cavernosa following exposure to elevated temperatures for 12 hours (24). Further, a thermosensitive culture of Symbiodiniaceae C1 had high expression of dinoRNAVs under ambient temperatures, suggesting that dinoRNAV may modulate resistance to bleaching in some species or populations of Symbiodiniaceae (26). Similarly, viral metagenomes from bleached pocilloporid colonies in situ contained significantly more eukaryotic virus sequences than unbleached, apparently healthy colonies (12). Experiments with the cricket paralysis virus—a +ssRNAV with similar genome architecture to dinoRNAVs in Symbiodiniaceae cultures—also showed increased viral replication after two hours when temperatures were raised by 5 °C (92). Taken together, our findings and those from other studies indicate that dinoRNAVs may play key roles in altering the thermal sensitivity of Symbiodiniaceae, and perhaps, coral colonies.
Virus-like particles associated with heat-stressed Symbiodiniaceae cells
After 24 hours of heat treatment, virus-like particles (VLPs) 110-170 nm in diameter were observed in vacuoles within Symbiodiniaceae cells (Figure 1). These VLPs were similar in size members of the Phycodnaviridae, such as Chloroviruses, Coccolithoviruses, and Prasinoviruses, within the nucleocytoplasmic large DNA viruses (NCLDVs). These viral groups have been previously reported from cultured Symbiodiniaceae cells via metagenomic approaches (26) and from Symbiodiniaceae in hospite through a combination of metagenomic analysis and TEM imaging (12, 13). We observed VLPs that may have been entering or exiting Symbiodiniaceae cells (Figure 1), suggesting active infections were occurring. Such observations—concurrent with genetic shifts in dinoRNAVs—corroborate that diverse viral groups likely respond to increased temperatures (8,10,91,93,94).
RNA viruses known to infect free-living dinoflagellates, such as HcRNAV, have icosahedral capsids ∼30 nm in diameter (29). Similarly sized VLPs were not observed in this study, but the capsid morphology of Symbiodiniaceae-infecting dinoRNAVs remains unconfirmed. Isolation of Symbiodiniaceae-infecting dinoRNAVs and characterization of their replication cycle will further clarify which VLPs in Symbiodiniaceae TEM images are potential dinoRNAVs.
Colony-specific dinoRNAV responses to elevated temperatures
Since higher temperatures increase host cell enzymatic activity, increased viral production and accumulation of mutations within quasispecies exposed to heat stress might be expected based on thermodynamics alone (95). However, dinoRNAV responses to elevated water temperatures differed among individual coral colonies (Figures 3b, 6a), even in two colonies (3 and 4) that were both dominated by aminotype 1 in control aquaria. While dinoRNAVs in heat-treated fragments from colonies 1 and 3 exhibited strong shifts in putative quasispecies compositions, such changes were less pronounced in the other colonies. These findings suggest colonies 1 and 3 and/or their dinoRNAV quasispecies were more sensitive to heat stress. Coral colonies generally exhibit heterogeneous resistance to coral bleaching (e.g. 96,97), and bleaching susceptibility in Pocillopora species complex has been correlated to differential communities of eukaryotic viruses (12). Taken together, we hypothesize that colonies 1 and 3 in our study were more sensitive to temperature stress and likely would have shown signs of bleaching had the experiment run longer. Subsequent experiments that extend multiple weeks to sample both the onset of thermal stress and the onset of bleaching signs are a critical next step in understanding how dinoRNAV dynamics relate to colony health trajectories and relative bleaching resistance.
Conclusions
This is the first study to characterize the dynamics of Symbiodiniaceae-infecting dinoRNAVs in coral colonies exposed to ecologically relevant bleaching temperatures. We identified dinoRNAVs in each sampled coral; temperature stress elicited rapid changes to potential dinoRNAV quasispecies in terms of diversity and composition, and a subset of viral aminotypes were significantly associated with heat-treated corals. Together, these multiple lines of evidence suggest that dinoRNAVs are common as persistent infections of Symbiodiniaceae. Environmental stress may increase the productivity of these viruses, contributing to bleaching signs or other impacts on coral-Symbiodiniaceae partnerships (if stress is prolonged). Overall, these findings add to the growing body of literature demonstrating that viruses of microorganisms affect emergent phenotypes of animal and plant holobionts, and may modulate holobiont responses to changing environmental conditions.
Author contributions
C.G., L.H.K. and A.C. conceived of the experiment; C.G., L.H.K., and A.J.V. developed the methods with support from A.C.; C.G., L.I.H., R.B., and A.C. conducted the experiments and processed samples; C.G. led data analysis, with contributions by all authors; C.G. wrote the first draft of the manuscript, with contributions by all authors.
Acknowledgements
The authors extend their sincere appreciation to Drs. Rebecca L. Vega Thurber, Andrew R. Thurber, and Craig E. Nelson for input on, and help with setting up, the experiment. Many thanks to Rebecca L. Maher and J. Grace Klinges for help with sampling, and to Dennis Conetta for assistance with DNA and RNA extractions. We additionally thank Mark Dasenko at Oregon State University’s Center for Genome Research & Biocomputing (Corvallis, OR) for his support in designing the sequencing methods. Financial support was provided by a Sigma-Xi Grant-in- aid of Research to C.G., a U.S. National Science Foundation award (OCE #1635798) to A.M.S.C. and an Early-Career Research Fellowship (#2000009651) from the Gulf Research Program of the National Academies of Sciences to A.M.S.C.
Footnotes
Competing Interests statement: The authors declare that they have no competing interests.
References
- 1.↵
- 2.↵
- 3.↵
- 4.
- 5.
- 6.↵
- 7.↵
- 8.↵
- 9.
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.
- 19.
- 20.
- 21.
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.
- 39.
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.
- 57.↵
- 58.↵
- 59.
- 60.↵
- 61.↵
- 62.↵
- 63.
- 64.
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.
- 78.
- 79.
- 80.
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵