The lactonase BxdA mediates metabolic specialisation of maize root bacteria to benzoxazinoids

Abstract

Root exudates contain specialised metabolites that shape the plant's root microbiome. How host-specific microbes cope with these bioactive compounds, and how this ability affects root microbiomes, remains largely unknown. We investigated how maize root bacteria metabolise benzoxazinoids, the main specialised metabolites of maize. Diverse and abundant bacteria metabolised the major compound in the maize rhizosphere MBOA (6-methoxybenzoxazolin-2(3H)-one) and formed AMPO (2-amino-7-methoxy-phenoxazin-3-one). AMPO forming bacteria were enriched in the rhizosphere of benzoxazinoid-producing maize and could use MBOA as carbon source. We identified a gene cluster associated with AMPO formation in microbacteria. The first gene in this cluster, bxdA encodes a lactonase that converts MBOA to AMPO in vitro. A deletion mutant of the homologous bxdA genes in the genus Sphingobium, did not form AMPO nor was it able to use MBOA as a carbon source. BxdA was identified in different genera of maize root bacteria. Here we show that plant-specialised metabolites select for metabolisation-competent root bacteria. BxdA represents a benzoxazinoid metabolisation gene whose carriers successfully colonize the maize rhizosphere and thereby shape the plant's chemical environmental footprint.

Fig. 1. AMPO-forming bacteria are abundant on benzoxazinoid-exuding maize roots.

a–c AMPO-forming colonies in extracts from different plant species. Root, rhizosphere or soil extracts were plated on 10% TSA supplemented with DMSO or MBOA (200 µg/mL) and cycloheximide to suppress fungal growth. Colonies were scored for AMPO formation after 10 days of incubation. a AMPO-forming colonies in maize root extracts. AMPO-forming colonies appear red on the MBOA-supplemented medium. b Percentage of total colony-forming units (CFU) that form AMPO on wild-type maize (WT) or benzoxazinoid-deficient bx1 mutant roots, in the rhizosphere and soil. Means ± standard error and individual data points are shown (WT n = 8, bx1 n = 9). Results of two-tailed t-tests are shown inside the panels. c Percentage of AMPO-forming CFUs in root extracts of benzoxazinoid-producing Zea mays (maize), Triticum aestivum (wheat) and non-benzoxazinoid-producing Medicago sativa (lucerne), Brassica napus (oilseed rape) and Arabidopsis thaliana. Means ± SE and individual data points are shown (n = 10, except maize n = 8, see b) ANOVA and compact letter display of all pairwise comparisons (Significance-level: FDR-corrected p < 0.05) of estimated marginal means are shown. d Phylogenetic tree of the MRB strain collection. The inner ring is coloured according to the relative abundance (%) of the corresponding partial 16S rRNA gene sequence in the microbiome profile of maize roots, from which most of the isolates were obtained. The outer ring displays the phenotype of the strains on 100% TSA plates supplemented with MBOA (200 µg/mL). Strains were classified as strong AMPO-former based on a dark red colouring, weak AMPO-former for strains with a light red colour change or non-AMPO-former for strains not showing a colour change compared to the control. Tree tips are coloured by family taxonomy and the ring next to the strain IDs reports phylum taxonomy. The maximum likelihood phylogeny is based on partial 16S rRNA gene sequences.

Fig. 2. Metabolisation of benzoxazinoids and use as sole carbon source by maize root bacteria.

a, b Heatmaps displaying qualitative detection of MBOA, DIMBOA-Glc and their metabolisation products for 46 maize root bacteria. Detected compounds and classifications are indicated in the legend. The corresponding metabolite analyses and classification cut-offs are shown in Supplementary Fig. 5. a–c Strains were grown in liquid 50% TSB supplemented with 500 µM of the respective chemical. “No bacteria control” NBC only contains the medium supplemented with the respective chemicals. a Detection of MBOA and its metabolisation products AMPO and AAMPO and b DIMBOA-Glc and its metabolisation products MBOA, AMPO and AAMPO. c Time course of MBOA metabolisation to AMPO and AAMPO for selected single strains: strong AMPO-formers Sphingobium LSP13, Pseudoarthrobacter LMD1, Microbacterium LMB2, Enterobacter LME3; weak AMPO-formers Acinetobacter LAC11 and Rhizobium LRC7.O; non-AMPO-formers Pseudomonas LMX9, Bacillus LBA112 and Microbacterium LMI1x. Metabolite measurements (n = 1) were made on pools of three independently grown cultures (#: sample with failed pooling). d Bacterial growth within 68 h (reported as the area under the growth curve, AUC) for selected strains in minimal medium supplemented with DMSO (negative control), glucose (positive control), MBOA and DIMBOA-Glc each in two concentrations (500 µM or 2500 µM). Means ± standard error and individual data points are shown (n = 5). Growth curves are available in Supplementary Fig. 6. Asterisks indicate significant differences between treatment and DMSO control (pairwise t-test, P < 0.05).

Fig. 3. Phenotypic diversity of AMPO-formation in microbacteria.

Phylogenetic tree constructed from whole genome alignment of 39 microbacteria. Tips are coloured by the host plant from which the strains were isolated. The column “plate assay” shows the AMPO classification (strong AMPO-former or non-AMPO-former) of the strains based on red colour formation on 100% TSA plates supplemented with MBOA (200 µg/mL). The adjacent columns display the classifications of metabolite analyses (MBOA, AMPO, HMPAA and DIMBOA-Glc) of liquid 50% TSB cultures supplemented with 500 µM MBOA (“MBOA metabolization”) or DIMBOA-Glc (“DIMBOA-Glc metabolization”) after 68 h. The column “MBOA C-source” refers to the assay where the strains were grown in minimal medium supplemented with 500 µM MBOA as a sole carbon source (based on mean results of 12 independent replicates grown in two independent runs). Columns “OG000xxxx” report copy numbers of gene orthogroups (Supplementary Data 2) that are unique and specific to AMPO-formers.

Fig. 4. Bxd gene cluster in microbacteria.

a Overlap of candidate genes for AMPO formation identified by three approaches: orthogroups, kmers and RNAseq. b Position of identified candidate genes in the genome of Microbacterium LMB2. A zoom-in of the benzoxazinoid degradation (bxd) gene cluster, annotated with its gene architecture including all genes named bxdA to bxdO is shown. c Synteny plot of the genomic region across microbacteria samples where the bxd gene cluster is found. The gene cluster can be categorized into four types (type I, type II, type III and type IV) based on their gene order, gene content and chemical phenotype.

Fig. 5. BxdA converts MBOA to AMPO.

a In vitro activity of purified recombinant BxdA (from LMB2) with the substrate MBOA. Extracted ion chromatograms (EIC, HPLC-MS analysis in positive mode) for MBOA and AMPO are shown. An empty vector (EV) control and MBOA and AMPO standards were included. b–e Characterization of Sphingobium LSP13 Δ3bxdA mutant lacking the three bxdA homologs. b Photograph of LSP13 wild-type (WT) and the Δ3bxdA mutant cultivated for 10 days on 100% TSA supplemented with 200 µg/mL MBOA or DMSO. c MBOA metabolisation to AMPO over time in 50% TSB supplemented with 500 µM MBOA. The 8 and 24 h timepoints (measurements (n = 1) were pools of three independently grown cultures) and the 48 h timepoints (measurements (n = 3) were biological replicates) come from different experiments. d MBOA metabolisation to AMPO in minimal medium supplemented with 500 µM MBOA. Means and individual data points are shown (n = 3). e Bacterial growth within 94 h (reported as the area under the growth curve, AUC) in minimal medium supplemented with DMSO (negative control), glucose (positive control) and MBOA (500 µM or 1250 µM). Means and individual data points of three biological replicates are shown (n = 3). Growth curves are available in Supplementary Fig. 11. f Proposed reaction sequence from MBOA to AMPO catalysed by BxdA. The dashed arrows refer to possible alternative MBOA degradation pathways. The potential intermediate HMPCA ((2-hydroxy-4-methoxyphenyl)carbamic acid) is proposed but was not confirmed experimentally.