Mining for a new class of fungal natural products: the evolution, diversity, and distribution of isocyanide synthase biosynthetic gene clusters

Abstract

The products of non-canonical isocyanide synthase (ICS) biosynthetic gene clusters (BGCs) mediate pathogenesis, microbial competition, and metal-homeostasis through metal-associated chemistry. We sought to enable research into this class of compounds by characterizing the biosynthetic potential and evolutionary history of these BGCs across the Fungal Kingdom. We amalgamated a pipeline of tools to predict BGCs based on shared promoter motifs and located 3800 ICS BGCs in 3300 genomes, making ICS BGCs the fifth largest class of specialized metabolites compared to canonical classes found by antiSMASH. ICS BGCs are not evenly distributed across fungi, with evidence of gene-family expansions in several Ascomycete families. We show that the ICS dit1/2 gene cluster family (GCF), which was prior only studied in yeast, is present in ∼30% of all Ascomycetes. The dit variety ICS exhibits greater similarity to bacterial ICS than other fungal ICS, suggesting a potential convergence of the ICS backbone domain. The evolutionary origins of the dit GCF in Ascomycota are ancient and these genes are diversifying in some lineages. Our results create a roadmap for future research into ICS BGCs. We developed a website (https://isocyanides.fungi.wisc.edu/) that facilitates the exploration and downloading of all identified fungal ICS BGCs and GCFs.

Graphical Abstract

Figure 1.

Full isocyanide synthase (ICS) biosynthetic gene cluster (BGC) genome-mining pipeline. First, clusters of co-regulated genes around ICSs are predicted. These BGCs are then grouped into gene cluster families (GCFs) based on sequence similarity and shared protein domains. To identify the core genes in GCFs, we searched for co-localized orthologous genes across every genome in the dataset. Finally, a subset of the GCFs were validated by confirming that the co-occurrence of genes found in the GCFs (i.e. conservation of genes across BGCs in a given GCF) exhibited significant linkage disequilibrium (LD) relative to whole-genome levels of LD (i.e. conservation of orthologous genes around BUSCOs).

Figure 2.

Distribution of isocyanide synthases (ICSs) and associated clusters biosynthetic gene clusters (BGCs) across the fungal kingdom. The species tree was estimated using a multi-species coalescent model, and is rooted at the early-divergent subphylum, Cryptomycota. Tips are colored based on phylum, and major classes within the Ascomycota and Basidiomycota are labelled. The overlay bar chart on the outer ring depicts the number of ICS genes found (dark blue), and number of ICS BGCs (light blue). Fully dark blue bars denote that all ICS genes act as backbones in ICS BGCs, while fully light blue bars indicate that none of the predicted ICS genes were found in co-regulated gene clusters.

Figure 3.

The relationships between isocyanide synthase (ICS) biosynthetic gene clusters (BGCs), canonical BGCs (predicted by antiSMASH), and genome sizes. (A) A scatterplot of the total number of BGCs (canonical + ICS) and the number of genes in a genome. Each point represents a genome and is colored and sized based on the number of ICS BGCs predicted within the genomes. Density plots are added to the margins of the scatterplot to show the distribution of the datapoints along the x and y axes. (B) The distribution of BGC sizes (in base pairs) for canonical BGC classes, and the ICSs elucidated in this study (n = BGCs). BGC classes that were significantly different (larger or smaller) in size (rank sums test; see Supplementary Methods) from ICS BGCs are marked on the plot.

Figure 4.

Architecture, evolution, and distribution of the 10 sampled isocyanide synthase (ICS) gene cluster families (GCFs) under significant linkage disequilibrium (LD). (A) A subset of the most common protein domains in the ICS GCFs. Domains frequently found within ICS backbones variations (e.g. TauD, NRPS-like) were filtered out, except for the ICS domain (top row) which is present in 100% of the clusters. GCFs are labeled with colors (see key), and domains implicated in specialized metabolism biosynthesis, transportation, or regulation are bolded. An un-subset version of the plot can be found in Supplementary Figure S12. (B) Midpoint rooted maximum likelihood phylogeny of ICS proteins from the analyzed GCFs. The tree is colored and labeled with the same key found in panel ‘a’. The outer black ring conveys the variation of the ICS enzyme based on its dashed pattern. (C) Presence (yellow) and absence (purple) heatmap of each GCF mapped onto a species tree of Ascomycota. Columns are sorted by the relatedness of each GCF’s ICS backbone using a collapsed ICS protein phylogeny.

Figure 5.

Reconstructed evolutionary history of the dit superfamily. (A) The biosynthetic gene cluster (BGC) architectures of the three dit clans that make up the dit superfamily. The conservation of orthologs encoding specific protein domains is indicated above and below the corresponding gene arrow, respectively. The primary taxonomic class(es) containing the most representatives of each clan is labeled. (B) Unrooted maximum likelihood ICS phylogeny of backbones found in the 10 sampled GCFs (the unfiltered ICS tree can be found in the Supplementary Repository). The ‘lower’ clade's ICS enzymes (marked in grey) are colored to show the diversity of GCFs. All ICSs within the dit superfamily (i.e. dit1 homologs) are found in the ‘upper’ clade (marked with light-blue) and colored according to their clan. (C) The p450 proteins found within the dit superfamily BGCs were mapped onto a phylogeny of other p450s found in 10 selected genomes to establish if they are all homologs of dit2. As in panel ‘B’, the clans are colored and labeled accordingly. (D) Topological comparison between the concatenated maximum likelihood phylogeny of the Dit1/2 protein sequences (left) and the coalescent-based species tree (right). Branch support values are displayed as pie charts, with the left tree indicating bootstrap support values (1000 bootstrap replicates) and the right tree posterior probabilities. Phylogenetic relationships are highlighted as follows: (i) Purple: incongruencies between gene tree and species tree, (ii) Light blue: when a species has two copies of the dit cluster, both are highlighted regardless of incongruence. The most common dit clan of each taxonomic class is indicated on the Dit1/2 Tree. The highlighted clade (tan) on the Dit1/2 tree represents the region associated with most of the observed incongruencies and long branch lengths. This region also includes all the sequences corresponding to the dimorphic order, Onygenales.