Genomic differentiation among wild cyanophages despite widespread horizontal gene transfer

Abstract

Background: Genetic recombination is a driving force in genome evolution. Among viruses it has a dual role. For genomes with higher fitness, it maintains genome integrity in the face of high mutation rates. Conversely, for genomes with lower fitness, it provides immediate access to sequence space that cannot be reached by mutation alone. Understanding how recombination impacts the cohesion and dissolution of individual whole genomes within viral sequence space is poorly understood across double-stranded DNA bacteriophages (a.k.a phages) due to the challenges of obtaining appropriately scaled genomic datasets.

Results: Here we explore the role of recombination in both maintaining and differentiating whole genomes of 142 wild double-stranded DNA marine cyanophages. Phylogenomic analysis across the 51 core genes revealed ten lineages, six of which were well represented. These phylogenomic lineages represent discrete genotypic populations based on comparisons of intra- and inter- lineage shared gene content, genome-wide average nucleotide identity, as well as detected gaps in the distribution of pairwise differences between genomes. McDonald-Kreitman selection tests identified putative niche-differentiating genes under positive selection that differed across the six well-represented genotypic populations and that may have driven initial divergence. Concurrent with patterns of recombination of discrete populations, recombination analyses of both genic and intergenic regions largely revealed decreased genetic exchange across individual genomes between relative to within populations.

Conclusions: These findings suggest that discrete double-stranded DNA marine cyanophage populations occur in nature and are maintained by patterns of recombination akin to those observed in bacteria, archaea and in sexual eukaryotes.

Keywords: Bacteriophage; Cyanophage; Double-stranded DNA; Evolution; Phage; Species; Virus.

Fig. 1

Phylogenomic analyses of 142 cyanophages. a Unrooted phylogenomic maximum likelihood tree of 51 concatenated core genes (see Additional file 2: Table S2) in 142 genome-sequenced isolates reveals 10 distinct cyanophage genomic lineages. Six lineages (designated I-VI) contain three or more representatives, while the remaining four are less well-represented and indicated by colored hexagons. Isolate origin (coastal = white or offshore = black) is designated in the outer ring. b Pairwise comparisons of average nucleotide identity (ANI) of shared genes between genomes in the well-represented lineages reveals six with ANI >98% that correspond to phylogenomic clusters I-VI. (C) The pairwise fraction of shared genes within clusters are high (>96%). Clustering of ANI and shared gene content are statistically not random (Additional file 2: Table S6) and correspond to bootstrapped phylogenomic lineages

Fig. 2

Alternative sequence-based clustering of genomes match phylogenomic lineages. a Principal component (PC) projection of the relationship between the 142 cyanophage genomes, previously sequences T4-like cyanophage genomes, and viral-tagged T4-like phages from the same coastal waters revealed tight clustering of genomes within the same phylogenetic lineage. Genomes were clustered based on ANI of genes within shared protein clusters. b Automatic barcode gap discovery method which sorts individuals in to the same population when divergence is smaller within than between also revealed identical clustering to the phylogenomic lineages

Fig. 3

Comparative Genomics of lineages. a Synteny plot (blastn) within and between clusters shows high conservation of synteny within a cluster and between phylogenetically close lineages. Lineages V and VI shows an erosion of synteny. b Codon usage within each cluster reveals similar codon usage between lineages I-V with lineage VI as an outgroup

Fig. 4

Host range analyses of 15 Synechococcus host strains against 138 cyanophage isolates. Four of the genome-sequenced isolates (S-MbCM6, 7, 25, and 100) were not examined. Cyanophage standard host ranges, i.e. infection or no infection host range, exhibit little correspondence with phylogenomic lineage or environmental origin with at least one member of each lineage sharing the ability to infect CC9311, WH7803, MITS9220, UW140, and WH8101, indicating a lack of physical boundaries between cyanophage populations. There is little correspondence between host range and phylogenomic lineages, even if the structure of the host range is unlikely to occur as a result of chance (see Additional file 3: Table S8)

Fig. 5

Selection and recombination results across lineages. a Non-polarized McDonald-Kreitman analyses reveal different selective signatures across phylogenomic lineages (Fisher’s exact tests, p < 0.05) and effect size cut-offs (phi coefficient ≥ 0.1). b The number of recombination events detected within and between lineages as inferred using coalescent (turquoise) and substitutions (maroon). Detailed gene annotations, p-values, and effect sizes for panel A are available in Additional file 3: Table S9