Systematic and quantitative view of the antiviral arsenal of prokaryotes

Abstract

Bacteria and archaea have developed multiple antiviral mechanisms, and genomic evidence indicates that several of these antiviral systems co-occur in the same strain. Here, we introduce DefenseFinder, a tool that automatically detects known antiviral systems in prokaryotic genomes. We use DefenseFinder to analyse 21000 fully sequenced prokaryotic genomes, and find that antiviral strategies vary drastically between phyla, species and strains. Variations in composition of antiviral systems correlate with genome size, viral threat, and lifestyle traits. DefenseFinder will facilitate large-scale genomic analysis of antiviral defense systems and the study of host-virus interactions in prokaryotes.

Fig. 1. DefenseFinder, a tool to detect all known prokaryotic antiviral systems.

a List of systems included in DefenseFinder. Systems are ordered alphabetically. For systems with several types, each system is represented by one type and other types are indicated in parentheses (Full list, Supplementary Data Table 1). b Workflow for the creation of DefenseFinder. c Example of a DefenseFinder rule (in the MacsyFinder syntax) for the detection of system CBASS type I. Cyclase and effector proteins are mandatory while the sensing protein is only accessory. This means the system allows for it to be missing in a detected CBASS type I system. Different profiles are recognized for a protein. ex cyclase (SMODS, AGC_C, Cycl100). d DefenseFinder function layout. DefenseFinder takes an ordered multifasta protein file. A search for specific HMM profiles is conducted, the MacSyfinder rules specific for antiviral systems are applied on the search results, generating three results files.

Fig. 2. Distribution of antiviral systems in prokaryotic genomes.

a Distribution of the total number of antiviral systems per genome. The x-axis was cut at 20 for data visualization purposes. Max number is 57 (Desulfonema limicola). b Average number of antiviral systems (blue) and number of families of antiviral systems (gray) are correlated with genome size (two-sided Spearman, ρ = 0.25 and ρ = 0.44, both p-values < 0.0001 on all sizes; ρ = 0.27 and ρ = 0.30, p-value < 0.0001 for the 2 Mbp–5 Mbp interval) (CI = 95%). Bottom plot, distribution of the genome size. c Average number of antiviral systems (blue) and number of families of antiviral systems (gray) are correlated with number of prophages (two-sided Spearman, ρ = 0.16 and ρ = 0.27, both p-values < 0.0001) (CI = 95%). Bottom plot, distribution of the number of prophages in genomes.

Fig. 3. Families of antiviral systems have a heterogenous distribution.

a Frequency of systems in genomes. b Classification of antiviral mechanisms and detected operons according to their molecular functioning. c System length according to genomic location (***, two sided Wilcoxon test p-value = 6.5e−273)). The box extends from the first quartile (Q1) to the third quartile (Q3) of the data, with a line at the median. The whiskers extend from the box by 1.5× the inter-quartile range (IQR). n chromosome = 106,104, n prophage = 8088. Plasmids were excluded from this analysis. A system was deemed to be part of a prophage when the first and last protein of the system was inside the prophage boundary. The size of a system was computed as the difference between the end of the last protein and the beginning of first protein. d Frequency of genomic location for each system. e Number of systems per prokaryotic phylum. Only phyla with more than 50 genomes are represented. The number of genomes in the dataset is represented under each phylum’s name. The heatmap represents the frequency of each system in a given phylum (per line), color legend on the right. Absolute numbers of genomes encoding a given system is indicated in each cell.

Fig. 4. The antiviral arsenal is species specific.

a Number of systems and number of families of systems per species. Species with more than a hundred genomes were selected for this analysis. b, c Antiviral arsenal of two different species: Pseudomonas aeruginosa and Neisseria meningitidis. Each panel shows the distribution of the total number of systems in the species (top panel), the frequency of the 20 most common antiviral systems in prokaryotes in this species (bottom panel) and a phylogenetic tree of the species with the presence/absence of the 10 most common antiviral systems in prokaryotes. d Correlation at the species level between the number of prophages and the families of antiviral systems (linear regression, two sided Pearson r = 0.56, p-value = 0.008, light pink represents the confidence interval at 95%).