Correlative metabologenomics of 110 fungi reveals metabolite-gene cluster pairs

Abstract

Natural products research increasingly applies -omics technologies to guide molecular discovery. While the combined analysis of genomic and metabolomic datasets has proved valuable for identifying natural products and their biosynthetic gene clusters (BGCs) in bacteria, this integrated approach lacks application to fungi. Because fungi are hyper-diverse and underexplored for new chemistry and bioactivities, we created a linked genomics-metabolomics dataset for 110 Ascomycetes, and optimized both gene cluster family (GCF) networking parameters and correlation-based scoring for pairing fungal natural products with their BGCs. Using a network of 3,007 GCFs (organized from 7,020 BGCs), we examined 25 known natural products originating from 16 known BGCs and observed statistically significant associations between 21 of these compounds and their validated BGCs. Furthermore, the scalable platform identified the BGC for the pestalamides, demystifying its biogenesis, and revealed more than 200 high-scoring natural product-GCF linkages to direct future discovery.

Fig. 1 |. Workflow for the metabologenomics approach for natural products discovery in fungi.

a, Using interpreted sequences of 110 assembled fungal genomes, we grouped BGCs into GCFs. b, In tandem, LC–MS/MS profiles (that is, both MS¹ and MS² datasets) were collected from extracts from all 110 strains, each grown on three conditions (oats, rice and Cheerios). c, Correlative analyses were completed using three scoring methods (pattern matching, correlation scoring and intensity ratio analysis). d, Gene cluster-metabolite linkages identified through correlative analysis were then confirmed through targeted biosynthetic studies. All panels were created with BioRender.com. Avg, average; w/, with; w/o, without.

Fig. 2 |. Comparison of GCF-natural product score distributions using different GCF grouping parameters.

For a, b, d and e, each point represents a unique metabolite–GCF pair, and the location on the plot reveals the strength of the associated weighted correlation scores (x axis), −log₁₀(P) values (y axis) and intensity ratios (point size). P values are the result of the chi-squared test with Bonferroni correction. Significant correlations (P ≤ 0.05 after multiple-hypothesis correction) are colored green, nonsignificant correlations are colored pink and known metabolite–GCF pairs are colored purple. a, Distributions at the 60% similarity threshold for NRPS-containing GCFs. b, Distributions at the 70% similarity threshold for NRPS-containing GCFs. c, Total number of significant (green) and nonsignificant (pink) correlations for knowns with validated BGCs belonging to NRPS-containing GCFs calculated with the nine different GCF similarity thresholds using the pattern-matching approach. The 60% similarity threshold maximizes the number and significance of validated matches. d, Distributions for NRPKS-containing GCFs at the 60% cutoff. e, The 70% cutoff for NRPKS-containing GCFs. f, Total number of correlations to validated knowns of the NRPKS biosynthetic type calculated with different GCF similarity thresholds using the pattern-matching approach, illustrating that the 70% threshold is optimal for this biosynthetic type.

Fig. 3 |. Compiled metabolite–GCF correlations using optimized GCF network.

a, Each point represents a unique metabolite–GCF pair and its location corresponds to the strength of the association. Weighted correlation scores are on the x axis and −log₁₀(P values) calculated using the pattern-matching approach on the y axis. P values are the result of a chi-squared test with a Bonferroni correction. Point size corresponds to the square root of the intensity ratio. Significant correlations (P ≤ 0.05 after multiple-hypothesis correction) and nonsignificant correlations are colored in green and pink, respectively. Correlations for validated metabolite–GCF pairs are in purple (with selected known linkages labeled with metabolite–GCF names) and correlations between the pestalamide B and three GCFs of interest are colored in orange. b–e, Cooccurrence plots for monacolin K to HRPKS_85 (3 of 3 strains with the BGC produce monacolin K) (b), notoamide A to NRPS_66 (3 of 7 BGC-containing strains produce notoamide A) (c), roquefortine C to its GCF; 7 of 11 strains with DMAT_31 BGCs produce roquefortine C (d) and pestalamide B-GCF linkages (e). Strains are on the x axis and log-transformed peak heights on the y axis. Presence/absence patterns for three candidate GCF linkages are highlighted along the gridlines. NRPKS_59 (dark blue) is missing in the top-producing strain (orange box, far right). TERPENE_288 (light blue) was the top-ranked linkage (due to smaller GCF size than other high-scoring GCFs, orange boxes, left and middle), but ruled out using MS² data. HYBRID_85 (neon green), the second-ranked linkage, was targeted for follow up studies (for b–e, only a subset of strains are shown for clarity). HRPKS, highly reducing polyketide synthase; DMAT, dimethylallyl tryptophan synthase.

Fig. 4 |. Heterologous expression of pestalamide B in Aspergillus nidulans.

a, The expression strain A. nidulans-pst produces pestalamide B at a high MS titer, but lower than the native producer A. brasiliensis CBS 101.740; the expression strain lacking the backbone synthase (A. nidulans-ΔpstD) does not produce this metabolite. b, MS¹ spectral shifts of pestalamide B following feeding with [¹³C₆]-leucine (green), phenylacetic acid (phenyl-d₅) (pink) or without heavy isotopes (purple). c, MS² spectral shifts of pestalamide B fragments following feeding with [¹³C₆]-leucine and phenylacetic acid (phenyl-d₅). Fragment ions are annotated with their unlabeled m/z values, but putative structures have been color-coded based on the proposed incorporation of isotopically labeled precursors. Notably, while fragment ions show either +0 or +5 Da shifts on phenylacetic acid feeding (pink circles on structures), fragment ions often show +1, +4, +5 or +6 Da shifts (green circles on structures) following leucine feeding, consistent with substantial rearrangement of leucine during pestalamide biosynthesis. All panels were created with BioRender.com.

Fig. 5 |. Proposed biosynthesis of pestalamide B.

All panels were created with BioRender.com.