Correlations between metabolic function and community temperateness across the breadth of cellular metabolism.

Spearman’s rho estimates (ρs) are shown with 95 % confidence intervals (CIs) in parentheses from 1,000 bootstrapped simulations. Panels are ordered by mean gene frequency. The y-axis shows the gene frequency as a percentage of known genes, an internally normalized and relative metric where all of the frequencies for a given sample across all panels sums to 100 %. Gene categories with average frequencies below 0.5 % were omitted. Viral community temperateness was calculated as the percentage of integrase and excision genes in the known genes in each virome. Genes where the 95 % confidence intervals exclude zero are significant at the p < 0.05 level. Bootstrapping code can be found at https://github.com/hopefulmonstersucla/viral-metabolism/tree/main/code.

Central Carbon Metabolism genes shown in Figures 2 and 3.

Genes are categorized by pathway and gene function, identified by EC number and gene name, and annotated with prior references if they have been observed in viruses before. The Spearman’s correlation between the frequency of each gene and viromes temperateness is indicated by bootstrapped mean ρs values and 95 % confidence intervals from 1,000 iterations are shown. Please refer to Supplementary Figure 2 for bootstrap ρs distributions. Genes where the 95 % confidence intervals exclude zero are significant at the p < 0.05 level. Genes shared between pathways are shown with a slash. Note, although anaplerotic and cataplerotic reactions often refer to steps in the TCA, we are using the term more generally here to refer respectively to the generation of central intermediate metabolites (‘filling up’; anaplerotic) and the consumption of central metabolites to generate specialized pre-cursor or cellular energy pools (‘drawing down’; cataplerotic) more generally31. Bootstrapping code can be found at https://github.com/hopefulmonstersucla/viral-metabolism/tree/main/code.

Correlation between central carbon metabolism genes and viral community temperateness.

Spearman’s rho correlation coefficients (ρs) are shown from 1,000 bootstrapped simulations correlating central carbon metabolism gene frequency as a percentage of known genes vs viral community temperateness. Panels are ordered by pathway, including Embden-Meyerhof-Parnas glycolysis (EMP), Entner-Doudoroff glycolysis (ED), pentose phosphate (PP), and tricarboxylic acid cycle pathways. Viral community temperateness was calculated as the percentage of integrase and excision genes in the known genes in each virome. Enzymes with rho values significantly different to zero are indicated by asterisks (i.e., p< 0.05, 95 % confidence intervals exclude zero). Genes are labeled by EC numbers; see Table 1 for gene names and summary statistics.

Pathway map of changes in the frequency of central carbon metabolism genes across the lytic to temperate spectrum.

Map of central carbon metabolism with each enzymatic step colored by the change in enzymatic metagenomic gene frequency across lytic to temperate viral communities. Red arrows show higher frequency in temperate communities while blue arrows are lytic associated, reflecting ρs values for each enzyme as seen in Table 1 and Figure 2. Enzymes are labeled by EC numbers (see Table 1 for gene names). Single-headed and double-headed arrows represent irreversible and reversible reactions, respectively. Black arrows show genes that were absent from all viromes. Central intermediate metabolites glyceraldehyde-3-phosphate (G3P) and pyryvate are shown in bold. *Note that fructose-6-phosphate is found in the EMP/gluconeogenesis pathways as well as the PPP, ** that oxaloacetate is produced in both the TCA and gluconeogenesis, and that EC1.1.1.42 in the TCA has an oxalosuccinate intermediate. Code can be found at https://github.com/hopefulmonstersucla/viral-metabolism/tree/main/code.

Correlations between viral community temperateness and central carbon gene frequency across pathways and gene function.

(a) Spearman’s rho correlation coefficients (ρs) for genes in the Embden-Meyerhof-Parnas glycolysis (EMP), gluconeogenesis (GNG), pentose phosphate (PP), Entner-Doudoroff glycolysis (ED, and tricarboxylic acid cycle (TCA) pathways are shown with median values marked by horizontal bars in each pathway. (b) Coefficients are also shown for genes involved in building up central intermediate metabolite pools like pyruvate and glyceraldehyde-3-phosphate (anaplerotic reactions) or consuming general metabolites to make specialized pools of precursors and cellular energy (cataplerotic reactions) in central carbon metabolism, compared with genes involved in neither anaplerosis nor cataplerosis, also with median values shown. Oxidoreductase genes are colored by their electron acceptor: blue for NADP+ electron acceptors involved in biomass generation, and red for NAD+ electron acceptors involved in cellular energy generation. Dashed horizontal lines indicate ρs = 0. Note, although anaplerotic and cataplerotic reactions often refer to steps in the TCA, we are using the term more generally here to refer respectively to the generation (‘filling up’; anaplerotic) of flexible general intermediate metabolites and the consumption of general metabolites towards precursor and cellular energy production (‘drawing down’; cataplerotic) more generally31.

Schematic model incorporating viral metabolic effects on resilience of healthy coral reefs and exacerbation of reef degradation.

The predicted effects of viral lysis and metabolic capability on algal (A; green), coral (C; coral), bacteria (B; blue) and viral (V; violet) population sizes in healthy and degraded reef scenarios with no viruses (a and d, respectively), with viral lysis (b and e, respectively), and with viral lysis and metabolism (c and f, respectively). Compartment models showing how the model was implemented for each panel are shown (left) as well as resulting dynamics (right). Dynamics show changes in populations over time, in arbitrary units (a.u.), reflecting the qualitative nature of the model. In each iteration, algal population size A was solved as the 1 – C to reflect the finite space available on the benthos. Parameter values, shown as line weights, qualitatively reflect known processes on healthy vs degraded reefs, and can be found in Supplementary Table 2. Rates include: (i) coral-algal inhibition, (c) coral proliferation, (a) algal-mediated DOC supply to bacteria, (m) bacterial anabolic metabolism, (d) bacterial-mediate coral disease, (k) viral lysis of bacteria, (l) lytic viral metabolism, and (t) temperate viral metabolism. Note that panels (b) and (e) depict the Kill-the-Winner and Piggyback-the-Winner models, respectively. Algal, coral, bacteria, and viral pools were initialized at 0.5 a.u. for all simulations where present. To make the model as conservative as possible, rates of algal carbon delivery (a), and direct coral-algal interference (i) were not changed across scenarios; doing so would lead to enhanced loss of coral. Note that flat lines at minimal values reflect functional population extinction.

Summary of the 19 samples analyzed, including sample accession numbers, the total number of base pairs, reads sequenced, and locations of viral metagenomes.

Note that all metagenomes have been previously published except Pearl and Hermes Reef Wreck and Caroline Wreck Site samples indicated by an asterisk. Caroline Island is now named Millenium Atoll. Fasta files for all metagenomes can be found at https://github.com/hopefulmonstersucla/viral-metabolism/tree/main/fastas.

Parameters and values used in the qualitative ecosystem model.

To isolate the effect of bacterial and viral proliferative genes on algal, coral, bacteria, and viral populations, as many parameters as possible were held constant, and only those relating directly to bacterial and viral proliferation and pathogenicity changed (listed in bold). All populations, shown with arbitrary units to show relative change in Figure 5, were initialized at 0.5. Note that, to account for finite space on the benthos, algal population size A is solved from coral population size C as A = 1 – C each iteration. Code can be found at https://github.com/hopefulmonstersucla/viral-metabolism/tree/main/code.

The frequency of photosynthesis genes across the lytic-temperate spectrum.

Genes were identified using SEED Level 2 categorization and are organized according to the KEGG photosynthesis map. Spearman’s Rho values indicate whether genes are over-represented (value < 0) or under-represented (value > 0) in lytic versus temperate viral communities.

Temperate viral communities are more cell-like and are more genetically diverse in known genes.

(a) The percent of genes with hits against the SEED database (i.e., percent known genes), and (b) the diversity of those known genes in the viral communities across the lytic to temperate spectrum. Note that the percent known genes is indicative of how metabolically similar the viruses are to cells given that known genes are commonly of identified in the study of cells. Viral community temperateness was calculated as the percentage of integrase and excision genes in the known genes in each virome.

Bootstrapped distribution of ρs from 1,000 iterations correlating the frequency of central carbon metabolism genes with viral community temperateness (number of observations vs ρs value).

Genes are grouped by pathway. Histogram shows the distribution of bootstrapped ρs values, with pink lines showing the median bootstrapped value and shaded grey areas showing the 95 % confidence interval of the median ρs. Dashed vertical lines show ρs = 0. Bootstrapping code can be found at https://github.com/hopefulmonstersucla/viral-metabolism/tree/main/code.