Use of metabolomics for the chemotaxonomy of legume-associated Ascochyta and allied genera

Abstract

Chemotaxonomy and the comparative analysis of metabolic features of fungi have the potential to provide valuable information relating to ecology and evolution, but have not been fully explored in fungal biology. Here, we investigated the chemical diversity of legume-associated Ascochyta and Phoma species and the possible use of a metabolomics approach using liquid chromatography-mass spectrometry for their classification. The metabolic features of 45 strains including 11 known species isolated from various legumes were extracted, and the datasets were analyzed using chemometrics methods such as principal component and hierarchical clustering analyses. We found a high degree of intra-species consistency in metabolic profiles, but inter-species diversity was high. Molecular phylogenies of the legume-associated Ascochyta/Phoma species were estimated using sequence data from three protein-coding genes and the five major chemical groups that were detected in the hierarchical clustering analysis were mapped to the phylogeny. Clusters based on similarity of metabolic features were largely congruent with the species phylogeny. These results indicated that evolutionarily distinct fungal lineages have diversified their metabolic capacities as they have evolved independently. This whole metabolomics approach may be an effective tool for chemotaxonomy of fungal taxa lacking information on their metabolic content.

Figure 1. Colony morphology of 45 Ascochyta, Phoma, and Alternaria strains.

Strains were indicated by their respective codes. Detailed information of the strains is in Supplementary Table S1. Two Alternaria solani strains (ALS1 and ALS2) were also included in this study as an outgroup for the Ascochyta and Phoma taxonomy. Photos were taken 1 week after incubation on PDA.

Figure 2. Metabolic profiles among 21 Ascochyta and Phoma strains.

(a) PCA scores plot (upper), PC 2 versus PC 3 showing the variation in the metabolic profiles from 21 fungal strains: AP1, AP4, and AP5 (closed circles: A. pisi); ID4A (cross: an Ascochyta-like strain); PK4 (asterisk: P. koolunga); ID1A, ID3A, and G11 (open circles: Ascochyta-like strains); AS1 and AS4 (diamonds: P. medicaginis); MP1, MP2, and MP19 (open squares: A. pinodes); PMP1, PMP3, and PMP4 (closed squares: A. pinodella); AR628, AR21, AR738, G10, and M305 (triangles: A. rabiei). PLS scores plot (lower), showing the supervised separation of the strains. (b) Dendrogram of the 21 strains based on chemical similarity, based on UPGMA clustering method. The heatmap of respective metabolites corresponding to each strain is presented. (c) Overlaid total ion chromatograms of A. pisi strains in upper panel, A. rabiei strains in middle panel, and A. pinodes strains in lower panel. Chemical structure of ascochitine (upper), solanapyrone A (middle), or pinolidoxin (lower) was shown within each panel.

Figure 3. Chromatograms relating to the detection of ascochitine and pinolidoxin.

(a) Base peak ion chromatograms (BPIs) of AP5, PK4, and ID4A strains. Peaks corresponding to ascochitine were indicated by observed m/z values and retention times (in parenthesis). (b) BPIs of MP1 and PMP3 strains. Peaks corresponding to pinolidoxin were indicated by observed m/z values and retention times (in parenthesis).

Figure 4. Chemical diversity and grouping observed in 40 Ascochyta, Phoma and Alternaria strains.

(a) Clustering of fungal strains based on chemical similarity. Numbers and color coding indicate the five main chemical groups according to UPGMA clustering method at a 0.75 distance threshold. (b) PLS scores plot, PC1 versus PC 2, showing the supervised separation of the major chemical groups and singletons (grey open circles).

Figure 5. Bayesian phylogeny of Ascochyta and Phoma species.

The phylogeny was estimated from the combined dataset of CHS, EF and G3PD for Ascochyta and Phoma spp. isolated from various legumes and rooted by Alternaria solani strains. The upper numbers at major nodes indicate Bayesian posterior probabilities (PP), and the lower numbers represent bootstrap values (BS) from 1,000 bootstrapped samples in a maximum likelihood (ML) phylogeny. Clades were inferred based on PP greater than or equal to 95% and BS greater than or equal to 70%. (clade A: A. pisi [APs] and A. fabae [AFs]; clade B: A. lentis [ALs] and A. viciae-villosae [AV1]; clade C: A. rabiei [ARs] and P. medicaginis [ASs]; clade D: A. pinodes [MPs] and A. pinodella [PMPs]). The five major chemical groups detected in the hierarchical clustering analysis are mapped on the Bayesian phylogeny, showing distinct metabolic features in evolutionary lineages. Gray lines at the terminal clades indicate ungrouped strains (singletons). Branch lengths are proportional to the inferred amount of evolutionary change and the scale represents 0.05 nucleotide substitutions per site. The asterisks indicate the PP of the corresponding nodes in the Bayesian phylogeny. Placement of these nodes differed under the ML analysis where P. koolunga (PK4) and an Ascochyta-like strain (ID4A) were placed external to clade C.