Tryptophan metabolism and bacterial commensals prevent fungal dysbiosis in Arabidopsis roots

Abstract

In nature, roots of healthy plants are colonized by multikingdom microbial communities that include bacteria, fungi, and oomycetes. A key question is how plants control the assembly of these diverse microbes in roots to maintain host-microbe homeostasis and health. Using microbiota reconstitution experiments with a set of immunocompromised Arabidopsis thaliana mutants and a multikingdom synthetic microbial community (SynCom) representative of the natural A. thaliana root microbiota, we observed that microbiota-mediated plant growth promotion was abolished in most of the tested immunocompromised mutants. Notably, more than 40% of between-genotype variation in these microbiota-induced growth differences was explained by fungal but not bacterial or oomycete load in roots. Extensive fungal overgrowth in roots and altered plant growth was evident at both vegetative and reproductive stages for a mutant impaired in the production of tryptophan-derived, specialized metabolites (cyp79b2/b3). Microbiota manipulation experiments with single- and multikingdom microbial SynComs further demonstrated that 1) the presence of fungi in the multikingdom SynCom was the direct cause of the dysbiotic phenotype in the cyp79b2/b3 mutant and 2) bacterial commensals and host tryptophan metabolism are both necessary to control fungal load, thereby promoting A. thaliana growth and survival. Our results indicate that protective activities of bacterial root commensals are as critical as the host tryptophan metabolic pathway in preventing fungal dysbiosis in the A. thaliana root endosphere.

Keywords: microbial homeostasis; microbial interactions; plant holobiont; plant innate immunity; root microbiome.

Fig. 1.

A link between innate immunity and BFO-mediated plant growth promotion. (A) Schematic representation of investigated genes. (B) FW comparison between sterile and BFO-inoculated WT plants. t test, *P < 0.05, n = 132 plants for the BFO condition. (C) The relative growth promotion index was calculated by first subtracting the average sterile FW of each mutant from corresponding FW of BFO-treated plants and then by dividing this value by the average difference between BFO-treated and sterile WT (respective WT for each mutant). n = 48 to 132 plants per condition. Data comes from three independent biological replicates, with an exception for WT and cyp79b2/b3 mutant, in which, in total, six biological replicates were performed. Significant differences were calculated using Kruskal–Wallis and Dunn control test with Bonferroni correction (α = 0.05) based on transformed FW data. A total of 22 outliers with values above five are not shown on the graph for clarity reasons but were retained in the statistical analysis (SI Appendix, Fig. S2C).

Fig. 2.

Subtle shifts in root microbiota composition across immunocompromised mutants. (A–C) Alpha diversity Shannon index for bacterial (A), fungal (B), and oomycetes (C) communities. “Input” refers to initial microbial pool at T0, whereas “peat” corresponds to an unplanted pot containing peat and the BFO SynCom only. Significant differences were calculated using Kruskal–Wallis and Dunn test (α = 0.05). The number of samples per condition are the following: bacteria: n = 11 to 28, fungi: n = 11 to 28, and oomycetes: n = 1 to 18. (D–E) cPCoA, based on Bray–Curtis distances, constrained by “genotype” for bacterial (D), fungal (E), and oomycetes (F) communities. Each cross covers the minimal and maximal values per axis of the respective genotype. The number of samples per condition are the following: bacteria: bacteria: n = 11 to 28 (cut off 1,000 reads), fungi: n = 9 to 28 (cut off 1,000 reads), and oomycetes: n = 5 to 27 (cut off 100 reads). Percentage value above the graph represents the variance explained by the genotype effect. Genotypes significantly different from WT (ANOVA, P < 0.05, see Dataset S4) are highlighted with thick lines and their respective genotypes names are highlighted on the graphs.

Fig. 3.

Fungal load in roots explains BFO-mediated plant growth phenotypes. (A–C) Bacterial (A), fungal (B), and oomycetes (C) load in plant root samples, calculated based on qPCR data relative to plant UBQ10 reads. Asterisks indicate genotypes that were significantly different from WT. Significant differences were calculated using Kruskal–Wallis and Dunn control test with Bonferroni correction (α = 0.05) and WT as a control. The number of samples per condition are the following: bacteria: n = 10 to 23, fungi: n = 11 to 24, and oomycetes: n = 10 to 23. (D–F) Linear regression between mean bacterial (D), fungal (E), and oomycetes (F) load and mean plant relative FW (i.e., mean relative plant growth promotion index), P value, and R² were obtained from ANOVA (n = 15 genotypes).

Fig. 4.

Trp metabolism and bacterial commensals prevent fungal dysbiosis in roots. (A) Statistical differences of rosette’s dry weight (DW) were calculated with ANOVA and Tukey’s post hoc test (α = 0.05). (B) Days until bolting significant differences were calculated using Kruskal–Wallis and Dunn test with Bonferroni correction (α = 0.05). (C) Statistical differences in siliques numbers were calculated using Kruskal–Wallis and Dunn test with Bonferroni correction (P < 0.05). (A–C) n = 0 to 10 samples per condition. (D–F) Total bacterial (D), fungal (E), and oomycetes (F) abundance in the roots of 9-wk-old plants. Statistical differences for total microbial abundance were calculated using Kruskal–Wallis and Dunn test with Bonferroni correction (α = 0.05). The number of samples per condition are the following: bacteria: n = 11 to 15, fungi: n = 0 to 15, and oomycetes: n = 0 to 15. (G and H) PCoA based on Bray–Curtis distances between samples for bacterial (G) and fungal (H) community. The number of samples per condition are the following: bacteria: n = 8 to 15 and fungi: n = 6 to 15.