Microbiome composition modulates secondary metabolism in a multispecies bacterial community

Abstract

Bacterial secondary metabolites are a major source of antibiotics and other bioactive compounds. In microbial communities, these molecules can mediate interspecies interactions and responses to environmental change. Despite the importance of secondary metabolites in human health and microbial ecology, little is known about their roles and regulation in the context of multispecies communities. In a simplified model of the rhizosphere composed of Bacillus cereus, Flavobacterium johnsoniae, and Pseudomonas koreensis, we show that the dynamics of secondary metabolism depend on community species composition and interspecies interactions. Comparative metatranscriptomics and metametabolomics reveal that the abundance of transcripts of biosynthetic gene clusters (BGCs) and metabolomic molecular features differ between monocultures or dual cultures and a tripartite community. In both two- and three-member cocultures, P. koreensis modified expression of BGCs for zwittermicin, petrobactin, and other secondary metabolites in B. cereus and F. johnsoniae, whereas the BGC transcriptional response to the community in P. koreensis itself was minimal. Pairwise and tripartite cocultures with P. koreensis displayed unique molecular features that appear to be derivatives of lokisin, suggesting metabolic handoffs between species. Deleting the BGC for koreenceine, another P. koreensis metabolite, altered transcript and metabolite profiles across the community, including substantial up-regulation of the petrobactin and bacillibactin BGCs in B. cereus, suggesting that koreenceine represses siderophore production. Results from this model community show that bacterial BGC expression and chemical output depend on the identity and biosynthetic capacity of coculture partners, suggesting community composition and microbiome interactions may shape the regulation of secondary metabolism in nature.

Keywords: community interactions; metametabolomics; metatranscriptomics; microbiome; secondary metabolism.

Fig. 1.

Interspecies coculture affects the expression of BGCs in a model community. (A) BGCs of Bc, Fj, and Pk are expressed at different levels when in pairwise or three-member coculture with each other. Psi log₂(fold change) is calculated as the fold change of the coculture condition compared to the BGC-containing strain in monoculture. BGCs of Bc, Fj, and Pk are shown in blue, tan, and red, respectively. (B–D) The 12 most differentially expressed BGCs of Bc (B), Fj (C), and Pk (D) as heatmaps, with green indicating increased expression in coculture, brown indicating decreased expression in coculture, and light gray indicating no change. Columns denote coculture conditions, and rows denote individual BGCs and their antiSMASH region numbers in parentheses. SI Appendix, Fig. S2 shows all BGCs in all three strains.

Fig. 2.

Several BGCs of Bc exhibit different expression patterns in coculture. (A) Gene maps of the BGCs of zwittermicin/kanosamine, petrobactin, and bacillibactin are colored based on antiSMASH annotations. (B–D) Gene expression for zwittermicin/kanosamine, petrobactin, and bacillibactin BGCs, with green indicating increased expression in coculture, brown indicating decreased expression in coculture, and light gray indicating no change. Rows indicate Bc coculture condition, and columns indicate individual Bc open reading frames.

Fig. 3.

Coculture shapes the metabolomes of a model multispecies community. (A) The t statistics for each LC-MS molecular feature are shown for the effect of each species (Bc, Fj, or Pk), or pairwise or three-way interaction terms. Individual molecular features are shown as dots (left), and distributions are shown as frequency curves and boxes (right). Box heights denote the interquartile range, and the center line corresponds to the median. Molecular features in the top and bottom 2.5 percentile of the three-way interaction term are shown as red and yellow, respectively. A “community metabolite” with the highest estimated effect size in the three-way interaction is labeled with a red arrow. (B) LC-MS/MS subnetwork of Pk-produced lokisin and related molecular features is shown to the left. Molecular features are shown as nodes with edges connecting those with similar spectra. Nodes are sized and colored corresponding to their abundance across all sample conditions. Calculated MWs are shown. The structure of lokisin is shown to the right. Regions of lokisin corresponding to a putative ring-open form (blue atoms and arrow) or putative amino acid cleavages (community metabolite; red atoms and arrow) are highlighted. (C) Peak area of selected molecular features across Pk-containing conditions (B, Bc; F, Fj; K, Pk; combinations denote cocultures). The molecular features for the putative ring-open form of lokisin and the community metabolite are highlighted in blue and red, respectively.

Fig. 4.

Koreenceine produced by Pk modifies BGC expression in Bc. (A) BGC expression (transcript CPM) is shown for Bc siderophores bacillibactin (Top) and petrobactin (Bottom) in either Bc–Pk pairwise (Left) or Bc–Fj–Pk three-member coculture (Right). Cocultures with wild-type Pk are shown in green, and cocultures with koreenceine-null mutant Pk are shown in orange. (B) At the zwittermicin/kanosamine BGC locus, genes involved in the biosynthesis of kanosamine (kabABCD) and β-Uda (zmaTUV) are up-regulated when Bc is cocultured with wild-type Pk (green). Up-regulation of kabABCD and zmaTUV is abolished when Bc is cocultured with koreenceine-null mutant Pk (orange).