Metabolomics insights into the interaction between Pseudomonas plecoglossicida and Epinephelus coioides

Abstract

As a highly infectious epidemic in aquaculture, Pseudomonas plecoglossicida infection results in high mortality of teleosts and serious economic losses. Host-pathogen interactions shape the outcome of an infection, yet we still understand little about the molecular mechanism of these pathogen-mediated processes. Here, a P. plecoglossicida strain (NZBD9) and Epinephelus coioides were investigated as a model system to characterize pathogen-induced host metabolic remodeling over the course of infection. We present a non-targeted metabolomics profiling of E. coioides spleens from uninfected E. coioides and those infected with wild-type and clpV-RNA interference (RNAi) strains. The most significant changes of E. coioides upon infection were associated with amino acids, lysophospatidylcholines, and unsaturated fatty acids, involving disturbances in host nutritional utilization and immune responses. Dihydrosphingosine and fatty acid 16:2 were screened as potential biomarkers for assessing P. plecoglossicida infection. The silencing of the P. plecoglossicida clpV gene significantly recovered the lipid metabolism of infected E. coioides. This comprehensive metabolomics study provides novel insights into how P. plecoglossicida shape host metabolism to support their survival and replication and highlights the potential of the virulence gene clpV in the treatment of P. plecoglossicida infection in aquaculture.

Figure 1

Characteristics of E. coioides upon infection. (A) Scheme of the experimental design. (B,C) Typical metabolite features of pooled tissue extract acquired using LC–MS in ESI positive mode and negative mode, respectively. (D) PCA score plot of the metabolic profile composed of metabolites after UV scaling pretreatment. (E) Metabolic trajectories of the control and different infection groups based on the PCA model. Each point represents the average score values of samples with SEM. (F) Euclidean distance between WT and time-matched PBS (or clpV-RNAi). (G) Venn diagram for overview of univariate statistical analysis and multivariate VIP values. PBS denotes the PBS control group, WT denotes the wild-type strain infection group, and clpV-RNAi denotes the clpV-RNAi strain infection group.

Figure 2

Comparison between infection-induced and gene silencing-related changes. (A,B) Overview of enriched metabolite sets. (C,D) Summary of enriched metabolic pathways. Enriched analyses of all significant differential metabolites were performed based on comparisons between PBS and WT (i.e., infection-induced changes, left panel), and clpV-RNAi and WT (i.e., silencing-related changes, right panel). Analyses of those significant differential metabolites resulted in the top 3 significant metabolite sets (red asterisk*), and revealed the global metabolic disorders of the most relevant pathways.

Figure 3

Response patterns of the metabolome at the terminal stage of 96 hpi. (A) Heatmap. The intersection of representative differential metabolites (p < 0.05, FDR < 0.05 and VIP > 1) from both comparisons (i.e., PBS vs. WT and clpV-RNAi vs. WT) was UV-scaled and subjected to hierarchical clustering. The union of all representative differential metabolites was further specialized by correlation networks. (B) The number of lines in each correlation network. (C) Correlation networks. In correlation networks, each point represents one metabolite with the relative content of UV scaled. Each black dotted (or gray solid) line represents a positive (or negative) correlation with |Cij|> 0.9.

Figure 4

Correlation between response increase in WT to PBS ratio (or WT to clpV-RNAi ratio) and the number of double bonds. (A,B) Depict the changes in FAs and LPCs, respectively, at the terminal stage of 96 hpi.

Figure 5

Evaluation of potential biomarkers. (A,B) Are the response trajectories of dihydrosphingosine and FA 16:2, respectively. Each point in the trajectory is presented as the mean ± SE. (C,D) ROC curves for the combination of biomarkers of dihydrosphingosine and FA 16:2. Diagnostic potential was evaluated based on binary logistic regression.

Figure 6

Profile of metabolite response at the terminal stage of 96 hpi. (A,B) Present the mapping of key pathways of amino acid metabolism. (C) Metabolism of lipids. All data are presented as the mean ± SE. The black asterisk indicates statistical significance (p < 0.05).