Strain matters: host responses reflect symbiont origin in the squid-vibrio symbiosis

Abstract

Understanding the cause and consequences of bacterial strain variation remains a challenge in the study of symbioses. While the diverse reactions of the host immune system to strain variants have been well studied in pathogenesis, much less is known about how strain variation influences beneficial associations. From the complex vertebrate gut microbiome to the more tractable invertebrate models of symbiosis, the host's cellular and molecular responses to this diversity remain largely a mystery. Here, we explore strain diversity in Vibrio fischeri, the bioluminescent bacterial symbiont of the Hawaiian bobtail squid, Euprymna scolopes. Phylogenetic analyses of the genomes of 62 V. fischeri strains, including 50 light organ-associated and 12 planktonic isolates, revealed several genes that were absent in planktonic strains, but uniformly present in symbiotic ones. To better understand the consequences of this diversity to the host, we selected five light-organ associated strains: three from E. scolopes but having different combinations of colonization factors, one from a congeneric squid host, and one from a marine fish. We colonized juvenile E. scolopes with these strains and, using RNAseq, found that (i) the most similar host transcriptomic responses occurred among the native E. scolopes strains, (ii) intermediate was the strain from the related squid, and (iii) least similar was the fish strain. Importantly, native strains downregulated immune-related genes more than non-native ones. Finally, host development was atypical or delayed when colonized by non-native strains. These experiments point the way to more targeted studies of the mechanisms underlying host responses to symbiont strain diversity.

Importance: Variation among strains of a bacterial species is a powerful factor underlying the intensity of host responses during pathogenic infections. Less is known about the cellular and molecular responses of host tissues to differences between the strains present in an animal's normal microbiome. We use a natural, species-specific, symbiosis to explore the influence of strain-level differences on host gene expression and morphogenesis. Analysis of symbiotic strains from squids and fishes, as well as free-living strains, shows that the carriage of colonization determinants, while critical to competitive success among strains of a species, has a minimal effect on the transcriptional response of the host. We provide evidence that a more important driver of normal gene expression during the development of symbiosis is the history of a strain's co-diversification with its host species. Such studies, using simple invertebrate models, allow the recognition of otherwise obscured interactions underlying the more complex microbiomes of vertebrates.

Keywords: Vibrio; development; diversification; genomics; specificity; squid; transcriptomics.

Fig 1

Patterns of gene content in Vibrio fischeri isolates diverge across niches and hosts. (A) Cartoon of E. scolopes nascent light-emitting organ (LO). Left side: external view showing ciliated appendages and three surface pores leading to the internal crypts in which symbionts will be housed. Right side: cut-away view of LO interior showing three of the organ’s six crypts. (B) Co-phylogenetic analysis. Left tree: Core genome comparison (nucleotide level) based on orthologous genes. Right tree: Accessory genome clustering based on presence/absence of gene content. Analysis includes 50 light organ-associated strains (from squids or fishes) and 12 planktonic (i.e., seawater) isolates. Trees are rooted on the outgroup species Pseudomonas syringae. Branch lengths are log-transformed for visualization. The branch length of P. syringae was shortened to fit within the figure. (C) Functional enrichment analysis (percent of strains encoding these symbiosis-associated proteins). Left: Proteins enriched in V. fischeri LO isolates (50 strains) compared to V. fischeri isolates not associated with a LO (12 strains). Right: Proteins enriched in strains associated with squid LOs (47 strains) compared to fish LOs (3 strains).

Fig 2

Colonization by non-native V. fischeri strains differentially impacts host gene expression at 72 hpi. (A) Principal component analysis (PCA) plots representing expression of the top 3,000 most variable host genes 72 h after colonization by one of 5 strains (color-coded). The top-left PCA plot compares all strains to show global variation. The additional three plots highlight differences between specific strain combinations. (B) UpSet plot, capturing both shared and strain-specific transcriptional responses, and showing overlap in differentially expressed genes (DEGs) of the host from pairwise comparisons between strains. There were no DEGs between ES114 and the other two E. scolopes-derived strains (bottom two rows). DEGs were defined as those with adjusted P-value (padj) < 0.05 and absolute log2-fold change (|log2FC|) ≥0.58, with log2FC values shrunken using the ashr method. In the MJ11 vs. ES114 comparison, 18 genes had |log2FC| ≥ 1 and 10 had |log2FC| ≥ 2. The 0.58 threshold was selected to capture biologically meaningful changes influenced by bacterial colonization and bioluminescence. Color legend: Blue bar: DEGs unique to MJ11 vs. ES114; Purple bar: DEGs unique to MJ11 vs. MB15A5; Green bar: DEGs unique to MJ11 vs. MB11B1; Gray bar: DEGs in MJ11 vs. one or two other strains (but not all four); Black bar: DEGs shared in MJ11 vs. all E. scolopes strains (used in heatmap); White bar: DEGs not included in the heatmap. Color-group memberships of individual DEGs are identified on left column of heat map. (C) Heatmap of selected DEGs. Gene annotations are based on BLAST nucleotide hits; annotated gene IDs abbreviated (Table S4); (–) =no BLAST match found; (HPP) = hypothetical/predicted protein; (UPP) = uncharacterized/predicted protein. Functional groupings of GO-term classes are indicated by a colored line on the far right; Black = Transpost and oxidoreductase activity; Teal = Membrane associated proteolysis; Orange = Cytoskeleton, transport and stress response. Gene names with immune-related annotations are in dark green. The five genes examined in Fig. 3 are in bold dark green. Expression values are variance-stabilizing transformed (VST) and row-scaled for visualization. Rows (genes) are clustered by a tree, which is not represented.

Fig 3

Strain colonization impacts host genes associated with symbiotic interactions. Patterns of expression at 72 hpi of 5 host genes (A–E) of interest from Fig. 2C (names in bold). Counts are VST transformed. Expression levels that were significantly different between pairs of strains are indicated by the line below them; padj < 0.05, |log2FC| ± 0.58. For simplicity, SR5 was only compared with ES114 or MJ11. Data are also present in heatmap, Fig. 2C.

Fig 4

Colonization by non-native strains delays host development. (A) Cartoon illustrating some of the developmental events typically occurring in the juvenile LO during the first 72 hpi. The leftmost image shows the entire organ of a hatchling, while the other three depict an interior view of only one side at three times post inoculation. During the first 3 days after colonization, the ciliated appendages begin their trajectory toward full regression, and the bottleneck (arrows) connecting the migration path to the crypts reduces in diameter. aa, anterior appendage; pa, posterior appendage. (B) Colony-forming units per light organ (CFU/LO) recovered from individual squids at 24 hpi. Statistical comparisons were performed using a Kruskal–Wallis test followed by Dunn’s post hoc tests with Bonferroni correction. Asterisks indicate significance: **P < 0.01. (C) Image of the LO of a juvenile squid at (72 hpi) colonized on only one side with strain MJ11. This condition was noted in about 25% of MJ11-colonized light organs but was never observed in the several hundred animals colonized by the other four strains. (C’) Average length of the ciliated appendages of the LO at 72 hpi with one of five strains. Measurements were made only on animals in which both sides of the organ were colonized. Data shown are for the length of the left anterior appendage. Statistical comparisons were performed using a Kruskal–Wallis test followed by Dunn’s post hoc tests with Bonferroni correction. Asterisks indicate significance: ***P < 0.001. **P < 0.01, * P < 0.05. APO (aposymbiotic, not colonized). (D) LO bottleneck width at 24 hpi (left) and 72 hpi (right) of colonized LOs compared to uncolonized LOs (APO). Statistical comparisons were performed at each time point, using a Kruskal–Wallis test followed by Dunn’s post hoc tests with Bonferroni correction. Asterisks indicate significance: ***P < 0.001. (E) Bottleneck closure response to the carriage of the rscS gene. At 24 hpi, bottleneck closure induced by ES114 and ES114 ∆rscS was not statistically different; however, MJ11 closed less than both of them. MJ11 rscS+ has rscS present on a plasmid. MJ11 ctrl carries only the backbone plasmid. Statistical comparisons were performed using a Kruskal–Wallis test followed by Dunn’s post hoc tests with Bonferroni correction. Asterisks indicate significance: ***P < 0.001. **P < 0.01, *P < 0.05. Only some statistical significances are shown for plot clarity.