Bates KA et al. (2022), Microbiome function predicts amphibian chytridi...

XB-ART-60816

Microbiome 2022 Mar 10;101:44. doi: 10.1186/s40168-021-01215-6.

Show Gene links Show Anatomy links

Microbiome function predicts amphibian chytridiomycosis disease dynamics.

Bates KA , Sommer U , Hopkins KP , Shelton JMG , Wierzbicki C , Sergeant C , Tapley B , Michaels CJ , Schmeller DS , Loyau A , Bosch J , Viant MR , Harrison XA , Garner TWJ , Fisher MC .

???displayArticle.abstract???
BACKGROUND: The fungal pathogen Batrachochytrium dendrobatidis (Bd) threatens amphibian biodiversity and ecosystem stability worldwide. Amphibian skin microbial community structure has been linked to the clinical outcome of Bd infections, yet its overall functional importance is poorly understood. METHODS: Microbiome taxonomic and functional profiles were assessed using high-throughput bacterial 16S rRNA and fungal ITS2 gene sequencing, bacterial shotgun metagenomics and skin mucosal metabolomics. We sampled 56 wild midwife toads (Alytes obstetricans) from montane populations exhibiting Bd epizootic or enzootic disease dynamics. In addition, to assess whether disease-specific microbiome profiles were linked to microbe-mediated protection or Bd-induced perturbation, we performed a laboratory Bd challenge experiment whereby 40 young adult A. obstetricans were exposed to Bd or a control sham infection. We measured temporal changes in the microbiome as well as functional profiles of Bd-exposed and control animals at peak infection. RESULTS: Microbiome community structure and function differed in wild populations based on infection history and in experimental control versus Bd-exposed animals. Bd exposure in the laboratory resulted in dynamic changes in microbiome community structure and functional differences, with infection clearance in all but one infected animal. Sphingobacterium, Stenotrophomonas and an unclassified Commamonadaceae were associated with wild epizootic dynamics and also had reduced abundance in laboratory Bd-exposed animals that cleared infection, indicating a negative association with Bd resistance. This was further supported by microbe-metabolite integration which identified functionally relevant taxa driving disease outcome, of which Sphingobacterium and Bd were most influential in wild epizootic dynamics. The strong correlation between microbial taxonomic community composition and skin metabolome in the laboratory and field is inconsistent with microbial functional redundancy, indicating that differences in microbial taxonomy drive functional variation. Shotgun metagenomic analyses support these findings, with similar disease-associated patterns in beta diversity. Analysis of differentially abundant bacterial genes and pathways indicated that bacterial environmental sensing and Bd resource competition are likely to be important in driving infection outcomes. CONCLUSIONS: Bd infection drives altered microbiome taxonomic and functional profiles across laboratory and field environments. Our application of multi-omics analyses in experimental and field settings robustly predicts Bd disease dynamics and identifies novel candidate biomarkers of infection. Video Abstract.

???displayArticle.pubmedLink??? 35272699
???displayArticle.link??? Microbiome
???displayArticle.grants??? [+]

Genes referenced: diablo

???attribute.lit??? ???displayArticles.show???

	Fig. 1 Metagenomic sequencing-based exploration of Bd disease dynamics supports functional differences in skin bacterial communities from epizootic and enzootic populations. PCA and PERMANOVA of bacterial KO beta diversity for a) all KOs b) Metabolism (KEGG level 1) c) Environmental Information Processing (KEGG level 1) d) Cellular Processes (KEGG level 1). e Clustered image map of bacterial KOs (annotated by functional pathway) contributing to separation along sPLS-DA component 1. Samples are clustered using complete linkage and Euclidean distances. Sample sizes: Acherito n = 12, Lhurs n = 11, Puits n = 10, Arlet n = 14
	Fig. 2 Bd infection alters functional profile of the amphibian skin bacterial microbiome. a PCA of bacterial KO abundance on day 30 of the Bd exposure experiment b) PCA of Metabolism (KEGG level 1) c) PCA of environmental processes (KEGG level 1) d) Clustered image map of bacterial KOs (annotated by functional pathay) associated with Bd or control exposure as identified by sPLS-DA. Sample sizes: control group n = 11, Bd-exposed group n = 9
	Fig. 3 Multi-omics integration selects predictive targets of wild disease dynamics. a) PCA of skin metabolite profile of wild A. obstetricans populations b) volcano plot displaying differentially abundant metabolite features identified by univariate Wilcoxon’s test. c) Relevance networks produced by integration of microbiome and metabolome datasets using DIABLO for bacteria-metabolite interactions. A single network was identified that was indicative of epizootic dynamics based on the presence of taxa that were identified as enriched in the epizootic population from single omics analyses and their positive associations with epizootic metabolites. Bacteria are shown as diamonds and metabolites as circles. A positive correlation between nodes is indicated by red connecting lines, a negative correlation is shown by blue. Enzootic and epizootic enriched metabolites/bacteria have blue and red borders respectively. Sample sizes: Acherito n = 14, Lhurs n = 14, Puits n = 14, Arlet n = 14
	Fig. 4 Experimental Bd infection perturbs host skin bacterial and fungal communities. Beta diversity of a) bacteria and b) fungi during experimental Bd infection. Sample sizes bacteria: control = 20 (each sample day), Bd exposed = 20 (each sample day). Sample sizes fungi: day 1 control = 9, day 1 exposed = 12, day 30 exposed = 16, day 30 control = 8, day 60 exposed = 19, day 60 control = 20
	Fig. 5 Bd infection alters functional profile of the amphibian skin bacterial microbiome. a PCA of bacterial KO gene abundance on day 30 of the Bd exposure experiment b) PCA of Metabolism (KEGG level 1) c) PCA of environmental processes (KEGG level 1) d) Clustered image map of bacterial KO genes associated with Bd or control exposure as identified by sPLS-DA. Sample sizes: control group n = 11, Bd-exposed group n = 9
	Fig. 6 Integration of skin bacterial microbiome and metabolome identifies a Bd infection-associated multi-omics signature. DIABLO sample plots demonstrating discrimination of Bd-exposed and un-exposed midwife toads based on a) skin bacterial microbiome and b) skin metabolome c) bacterial taxa contributing separation along component 1 in (a). Bar length indicates loading coefficient weight of selected bacterial ASVs. Bar colour indicates the group in which the bacterial ASV has the highest median abundance, blue = control, red = Bd exposed. d Clustered image map (Euclidean distance, complete linkage) of the multi-omics signature. Samples are represented in rows, selected features of the first component are shown in columns. Sample sizes: Control = 20, Bd exposed = 20
	SI Figure 1. Map of study sites. Map generated using ArcGIS version 10.0 (http://www.esri.com/software/arcgis) with the World Imagery Basemap. Source: Esri, DigitalGlobe, GeoEye, Earthstar, Geographics, CNES/Airbus DS, USDA, USGS, AeroGRID, IGN, the GIS User Community.
	SI Figure 2. Boxplot of intensity for second ion of putative indole-3- carboxaldeyde (m/z 144.04606, RT 8.73 min). Sample sizes: Acherito n=14, Lhurs n=14, Puits n=14, Arlet n=14.
	SI Figure 3. Skin bacteria-metabolome interactions distinguishes wild Bd disease dynamics. Relevance networks produced by sPLS regression. a) bacterial subnetwork A, indicative of epizootic dynamics showing associations between bacterial ASVs and metabolite features b) bacterial subnetwork B. Metabolite features and ASVs are coloured according to identification from univariate analyses (metabolites: Wilcoxon test q < 0.05, log2 fold change > 1.5, ASVs: sPLS-DA analysis or ALDEx2). Sample sizes: Acherito n=14, Lhurs n=14, Puits n=14, Arlet n=14.
	SI Figure 4. Skin fungi-metabolome interactions distinguishes wild Bd disease dynamics. Relevance networks produced by sPLS regression. a) fungal subnetwork A showing positive associations between epizootic OTUs and epizootic metabolite features. b) subnetwork B. Metabolite features and OTUs are coloured according to identification from univariate analyses (metabolites: Wilcoxon test q < 0.05, log2 fold change > 1.5, OTUs: sPLS-DA analysis or ALDEx2). Sample sizes: Acherito n=11, Lhurs n=6, Puits n=8, Arlet n=10.
	SI Figure 5. Bd infection intensity in the Bd exposed treatment group over the course of the experiment. Sample sizes: Control=20, Bd exposed=20.
	SI Figure 6. Trends in ASVs abundance for taxa that were discriminatory in both the laboratory and field studies. Column 1: CLR transformed abundance of ASVs for each time point of the experiment. Column 2: Spearman’s correlation between CLR transformed ASV abundance and log 10 GE +1 on day 30 of the experiment. Column 3: CLR transformed abundance in wild populations. ASVs plotted are a) ASV17_Stenotrophomonas b) ASV45_Comamonadaceae and c) ASV6_Sphingobacterium.