Bacterial networks in Atlantic salmon with Piscirickettsiosis

Abstract

An unbalanced composition of gut microbiota in fish is hypothesized to play a role in promoting bacterial infections, but the synergistic or antagonistic interactions between bacterial groups in relation to fish health are not well understood. We report that pathogenic species in the Piscirickettsia, Aeromonas, Renibacterium and Tenacibaculum genera were all detected in the digesta and gut mucosa of healthy Atlantic salmon without clinical signs of disease. Although Piscirickettsia salmonis (and other pathogens) occurred in greater frequencies of fish with clinical Salmonid Rickettsial Septicemia (SRS), the relative abundance was about the same as that observed in healthy fish. Remarkably, the SRS-positive fish presented with a generalized mid-gut dysbiosis and positive growth associations between Piscirickettsiaceae and members of other taxonomic families containing known pathogens. The reconstruction of metabolic phenotypes based on the bacterial networks detected in the gut and mucosa indicated the synthesis of Gram-negative virulence factors such as colanic acid and O-antigen were over-represented in SRS positive fish. This evidence indicates that cooperative interactions between organisms of different taxonomic families within localized bacterial networks might promote an opportunity for P. salmonis to cause clinical SRS in the farm environment.

Figure 1

Illustration of the relative abundance of taxonomic profiles. (A) Relative proportion of phyla detected in healthy and SRS-positive salmon according the tissue sampled. (B) Relative abundance of taxonomic orders present in healthy and SRS-positive Atlantic salmon according to specific farm sites. Healthy fish were selected from sites C5 and C6 correspond to order shown in (A) while SRS positive fish were selected from farm sites C1, C2, C3 and C4 with corresponding order found in (A). The contents of the digesta are shown on top and the contents of the mucosa are shown on the bottom. The phylogeny are color coded to indicate individual orders.

Figure 2

Differential abundance of amplicon sequencing variants (ASVs) in the digesta and intestinal mucosa of healthy and SRS-positive Atlantic salmon. (A) Venn diagram of differentially abundant ASVs in digesta and intestinal mucosa of healthy and SRS samples. The top number indicates the total number of ASVs in that group. The number of ASVs derived from healthy and SRS positive fish are shown in parentheses. Contents of the digesta and mucosa are shown in green and yellow, respectively. (B) Venn diagram of differentially abundant ASVs in which we obtained a taxonomic assignment. The number of ASVs derived from healthy and SRS positive fish are shown in parentheses. Contents of the digesta and mucosa are shown in green and yellow, respectively. (C) The assigned taxa of differentially abundant ASVs comprising the 35 genera detected in (B) above. Each circle is a particular Amplicon Sequence Variant (ASV) identified and assigned to a particular species. There can be more than one ASV for a particular species, because each ASV represents a cluster of a group of reads representing a single DNA sequence recovered from the sequencing data. In this figure, each of the dots represents one of these ASVs within the particular taxonomic assignment. Fold-change (log₂FC) is represented in the logarithmic scale of base 2, red color means overabundance in infected samples, and blue represents overabundance in healthy samples. Adjusted p-values are represented in the negative logarithmic scale of base 10. Bigger circles mean higher p-value statistical significance.

Figure 3

Community shuffling plot of the changes in bacterial network structures of SRS-positive compared to healthy Atlantic salmon. This plot illustrates the relationships between members of the bacterial families found in the fish. This graphic consists of nodes arranged in a circle. Each node represents a member of the bacterial network. The co-occurrence between bacterial network members of the infected fish are indicated by a line connecting the two specific nodes. This line is called an ‘edge’ and red, green and blue indicate associations enriched in the SRS-positive, healthy and shared bacterial networks, respectively. This graphical output highlights the sub-networks and ‘driver’ nodes. All nodes belonging to the same community are randomly assigned a similar color. Grayed-out nodes represent the ones present in both but directly interacting with the common sub-network in either infected or healthy conditions. Node sizes are proportional to their scaled NESH score, as calculated using NetShift tool. A node is colored red if its proximity increases from healthy to infected samples. Nodes that are big and red indicate that these organisms are significant ‘drivers’ in the shift from healthy to a SRS-diseased state in the localized bacterial communities.

Figure 4

Illustration of co-occurrence network in SRS-positive salmon. This plot illustrates the statistically-significant (p ≤ 0. 005) associations detected between Piscirickettsiaceae and other microbial families. Line color indicates the type of interaction with red lines representing negative associations (occurring less than expected), and green lines representing positive associations (occurring more than expected). The density of a node relates to the number of other members the organism communicates with. The length of the line represents the closeness of the association between bacterial members. We highlighted the Piscirickettsiaceae for easy identification.

Figure 5

Metabolic phenotypes predicted to exist in health and SRS-positive salmon based on unique bacterial network compositions. Shown on the left are the metabolic pathways detected with statistical significance (q-value). The mean proportion of organisms expressing specific metabolic pathways in the digesta of healthy and sick fish are shown as horizontal bar graphs in green and red, respectively. The percentage differences in mean proportion of each metabolic phenotype is shown as a horizontal spread for healthy and sick salmon, where the error bars indicate the 95th percentile. This analysis was performed using the PICRUSt2 tool.