Introducing THOR, a Model Microbiome for Genetic Dissection of Community Behavior

Abstract

The quest to manipulate microbiomes has intensified, but many microbial communities have proven to be recalcitrant to sustained change. Developing model communities amenable to genetic dissection will underpin successful strategies for shaping microbiomes by advancing an understanding of community interactions. We developed a model community with representatives from three dominant rhizosphere taxa, the Firmicutes, Proteobacteria, and Bacteroidetes We chose Bacillus cereus as a model rhizosphere firmicute and characterized 20 other candidates, including "hitchhikers" that coisolated with B. cereus from the rhizosphere. Pairwise analysis produced a hierarchical interstrain-competition network. We chose two hitchhikers, Pseudomonas koreensis from the top tier of the competition network and Flavobacterium johnsoniae from the bottom of the network, to represent the Proteobacteria and Bacteroidetes, respectively. The model community has several emergent properties, induction of dendritic expansion of B. cereus colonies by either of the other members, and production of more robust biofilms by the three members together than individually. Moreover, P. koreensis produces a novel family of alkaloid antibiotics that inhibit growth of F. johnsoniae, and production is inhibited by B. cereus We designate this community THOR, because the members are the hitchhikers of the rhizosphere. The genetic, genomic, and biochemical tools available for dissection of THOR provide the means to achieve a new level of understanding of microbial community behavior.IMPORTANCE The manipulation and engineering of microbiomes could lead to improved human health, environmental sustainability, and agricultural productivity. However, microbiomes have proven difficult to alter in predictable ways, and their emergent properties are poorly understood. The history of biology has demonstrated the power of model systems to understand complex problems such as gene expression or development. Therefore, a defined and genetically tractable model community would be useful to dissect microbiome assembly, maintenance, and processes. We have developed a tractable model rhizosphere microbiome, designated THOR, containing Pseudomonas koreensis, Flavobacterium johnsoniae, and Bacillus cereus, which represent three dominant phyla in the rhizosphere, as well as in soil and the mammalian gut. The model community demonstrates emergent properties, and the members are amenable to genetic dissection. We propose that THOR will be a useful model for investigations of community-level interactions.

Keywords: Bacillus cereus; Flavobacterium johnsoniae; Pseudomonas koreensis; biofilm; colony expansion; emergent properties; inhibitory network; model community; rhizosphere.

FIG 1

Network analysis of inhibitory interactions among rhizosphere isolates. (A) Venn diagram of the inhibitory interactions identified between the isolates as determined by the presence of zones of inhibition in three media: Luria-Bertani agar (LBA), ½-strength tryptic soy agar (TSA), and 1/10-strength TSA. (B) Colors indicate phylogeny of isolates used in the inhibitory matrix and network. (C) Inhibitory interaction matrix between B. cereus UW85 and hitchhiker isolates in three media. Potential producers, which are isolates tested for the ability to inhibit others, are on the y axis, and receivers, which are isolates tested for inhibition by others, are on the x axis. There are two different color codes used in panel C. One indicates the phylogeny of isolates in the row and column title, and the second one for the matrix results using the colors scheme shown in the Venn diagram corresponding to the medium in which the interaction appears. (D) Hierarchy scoring scheme used to organize the isolates in the hierarchy interaction network, in which black is the focal point. (E and F) Inhibitory interaction network organized by hierarchy score. (E) Inhibitory interactions generated from the isolates with high hierarchy scores. (F) Inhibitory interactions generated from the isolates with medium and low hierarchy scores. (G) Reciprocal inhibitory interactions observed in the inhibitory network. Orange indicates interactions observed in only 1 medium, and black indicates interactions observed in 2 or 3 media.

FIG 2

Coculture of rhizosphere isolates with Pseudomonas spp. and B. cereus UW85. (A) Competition experiments between F. johnsoniae CI04 and either Pseudomonas sp. CI14, Pseudomonas sp. RI46, or P. koreensis CI12 in three media, Luria-Bertani broth (LB), ½-strength tryptic soy broth (TSB), and soybean root exudate. (B) Rhizosphere isolates were grown alone or in coculture with P. koreensis CI12 in soybean root exudate. Colored bars under x axis indicate phylogenetic groups as in Fig. 1. (C) F. johnsoniae CI04 grown alone, in coculture with P. koreensis CI12 or B. cereus UW85, and in triple culture with P. koreensis CI12 and B. cereus UW85. Gray dotted line, limit of detection.

FIG 3

Effect of Bacillus spp. on populations of F. johnsoniae in the presence of P. koreensis. Triple culture was of P. koreensis CI12, F. johnsoniae CI04, and either B. subtilis NCIB3160 or B. cereus UW85, or their mutants. Gray dotted line, limit of detection.

FIG 4

B. cereus UW85 colony expansion in the presence of the rhizosphere isolates. (A) Schematic representation of the spread-patch plates. (B) B. cereus UW85 grown alone. (C) B. cereus UW85 grown on a lawn of each member of the community. Photographs were taken after 4 days at 28°C. Arrows indicate the limits of the B. cereus colony.

FIG 5

Effect of community members on B. cereus UW85 colony expansion. (A) B. cereus UW85 plasmid-dependent GFP strain grown alone or on a lawn of P. koreensis CI12, F. johnsoniae CI04, or Paenibacillus sp. RI40. Bright-field (BF) and GFP imaging of colonies 5 days after inoculation. (B) B. cereus UW85 GFP strain grown in close proximity to a colony of P. koreensis CI12, F. johnsoniae CI04, or Paenibacillus sp. RI40. Arrows indicate B. cereus UW85 expansion over colonies of the other isolates. Bright-field, GFP channel, and overlay of the two channels of plates after 2 days of growth (F. johnsoniae CI04) and after 5 days of growth (P. koreensis CI12 and Paenibacillus sp. RI40).

FIG 6

Biofilm formation by rhizosphere isolates. Biofilm was quantified by measuring the optical density at 595 nm (OD₅₉₅) after staining with crystal violet. (A) Crystal violet quantification of biofilm formation for each of the 21 isolates at 12 and 24 h after inoculation. (B to D) Pseudomonas biofilm production when grown with a poor biofilm producer normalized against Pseudomonas biofilm in pure culture at 12 and 24 h, either Pseudomonas sp. RI46 (B), Pseudomonas sp. CI14 (C), or P. koreensis CI12 (D). (E) Biofilm formation by P. koreensis CI12 growing alone, in coculture with either F. johnsoniae CI04 or B. cereus UW85, and in triple culture at 12, 18, 24, and 36 h. (F) P. koreensis CI12 biofilm production when grown with two other isolates normalized against P. koreensis growth in pure culture. *, P < 0.01. Colored bars under the x axis indicate phylogenetic groups as in Fig. 1. Gray dotted line, limit of detection.