An effluent pump family distributed across plant commensal bacteria conditions host- and organ-specific glucosinolate detoxification

Russ, Dor; Fitzpatrick, Connor R.; Saha, Chinmay; Law, Theresa F.; Jones, Corbin D.; Kliebenstein, Daniel J.; Dangl, Jeffery L.

doi:10.1038/s41467-025-61266-3

Download PDF

Article
Open access
Published: 01 July 2025

An effluent pump family distributed across plant commensal bacteria conditions host- and organ-specific glucosinolate detoxification

Nature Communications volume 16, Article number: 5699 (2025) Cite this article

3769 Accesses
22 Altmetric
Metrics details

Subjects

This article has been updated

Abstract

In nature, plants recruit a diverse microbial community, the plant microbiome, that is distinct from the surrounding soil community. To understand the forces that shape the plant microbiome we need to characterize the microbial traits that contribute to plant colonization. We used barcoded mutant libraries to identify bacterial genes that contribute to the colonization of a monocot and a eudicot host. We show that plant colonization is influenced by dozens of genes. While some of these colonization genes were shared between the two host plant species, most were highly specific, benefiting the colonization of a single host and organ. We characterized an efflux pump that specifically contributes to Arabidopsis shoot colonization. This efflux pump is prevalent across Pseudomonadota genomes yet benefits the bacterial association with only a small subset of Arabidopsis thaliana accessions. Leveraging genomic diversity within Arabidopsis thaliana, we confirmed that specific glucosinolate breakdown products are detoxified by this family of efflux pumps. The broad prevalence of this efflux pump family suggests that its members contribute to protection of commensal bacteria from collateral damage of plant glucosinolate-based defense responses to herbivores and necrotrophic pathogens.

Network mapping of root–microbe interactions in Arabidopsis thaliana

Article Open access 07 September 2021

Synthetic microbe-to-plant communication channels

Article Open access 28 February 2024

Genome-resolved metatranscriptomics reveals conserved root colonization determinants in a synthetic microbiota

Article Open access 13 December 2023

Introduction

Plants host diverse microbial communities that play critical roles in plant development, nutrition, and resistance to disease^1,2,3,4. Plants can contribute to the selection of those communities from the surrounding environment through a myriad of chemical and physical means: production of antimicrobials; deposition of root exudates into the soil; competition for nutrients; physical barriers, and more^5,6,7,8,9,10. While the microbial communities associated with different plants share similarities at high taxonomic levels, the community composition diverges at lower taxonomic ranks^11,12,13,14. Moreover, plants recruit distinct communities to different organs¹⁵. Our knowledge of processes and proteins dictating plant or species-specific microbiome assembly remains limited. To fill this gap, we must define the microbial genes determining colonization and persistence in the plant microbiome¹⁶. Understanding the forces that shape plant microbiomes will enhance our ability to predict the behavior of microbial communities and to harness those communities for sustainable agriculture.

Different approaches can be taken to study microbial traits that help microbes colonize plants. Studies comparing the genomes of plant-root-associated bacteria versus those found in other habitats found enrichment of genes related to carbohydrate utilization and higher predicted growth rate potential^17,18. Furthermore, in planta meta-transcriptome profiling of root-inhabiting bacteria revealed an upregulation of genes involved in energy production relative to free-living bacteria¹⁹. Assimilation of specific glucosinolates can drive differences in bacterial communities between closely related hosts²⁰. However, direct interrogation of microbial genes required for persistence in plant habitats remain rare. In recent years, whole genome, high-resolution bacterial mutant libraries have enabled genetic screens to infer causality in bacterial–host microbiome interactions. These were deployed to understand antibiotic resistance, phage resistance, metabolite assimilation, gut colonization and more^21,22,23. Due to the complexity of mutant library application on plants, colonization screens to date have been typically performed in simplified, artificial systems, often limited to a single host and/or organ^24,25,26,27. These experiments identified genes important for plant colonization, including genes that help bacteria evade plant immunity, control bacterial motility, and regulate different metabolic processes^24,25,26. Yet, the diversity of bacterial genes important for long-term colonization is unknown as are the diversity of mechanisms by which they encode colonization traits are for individual plant hosts and organs. To address this knowledge gap, we deployed a high-resolution screen to reveal genes defining host and organ specificity in the plant microbiome.

We identified ~250 plant-association genes through a genome-wide approach using bacterial mutant libraries of two different robust plant-colonizing bacterial strains growing on either the monocot Brachypodium distachyon or the eudicot Arabidopsis thaliana. We demonstrate that the majority of colonization genes affect the association with specific organs or specific hosts with only a few genes benefiting the colonization of different organs in both plant species. We focused on genes inferred to function in bacterial adaptation to plant defense and immunity and identified two types of organ-specific efflux pumps, one that protects the bacteria against an Arabidopsis-shoot-derived glucosinolate and another that likely protects the bacteria against an as yet unidentified Arabidopsis-root-derived compound. Searching for functional orthologs of the efflux systems in the genomes of a large bacterial isolate collection, we found that the shoot-specific efflux pump (ef90) is prevalent among Arabidopsis colonizing commensals. The ef90 pump protects the bacteria from a toxic compound expressed in the leaves of small subset of A. thaliana accessions. We leveraged recombinant inbred lines (RIL) and genome-wide association (GWA) analysis to identify the glucosinolate break-down product sulforaphane as the antibacterial compound detoxified by ef90 encoded pumps in A. thaliana Col-0. The ef90 pump was previously defined as saxF in hemi-biotrophic leaf pathogenic strains of Pseudomonas syringae, where it was proposed to protect the pathogen from glucosinolates produced during necrotrophic growth²⁸. Our data prompt re-evaluation of this hypothesis.

Results

A. thaliana and B. distachyon assemble distinct microbial communities

Plant hosts actively recruit their associated microbiota. General colonization requirements would select for traits that enable generalized colonization across species and organs. Conversely, distinct species and strain-level microbiome assembly associated with different plant organs and hosts would suggest specific interactions selecting for specialized bacterial traits. We used two highly diverged model plants, the eudicot A. thaliana Col-0 and the monocot B. distachyon BD21 grown in a single wild soil that we have sampled extensively over the last 15 years (Mason Farm, Chapel Hill, NC) (“Methods”). After 6 weeks of growth in this soil, we harvested roots and shoots and characterized community assembly using 16S rRNA sequencing (“Methods”). We found that the two organs of the two plants assembled bacterial communities distinct from the soil community and from each other (Fig. 1a). We noted that the betaproteobacteria genus Paraburkholderia exhibited robust colonization of both organs of both host species and was readily detectable in soil (Supplementary Fig. 1). This genus had high and stable relative abundance in all five habitats. Genetic tools are available for Paraburkholderia bryophila MF376²⁹ (hereafter MF376), a genome-sequenced isolate previously cultured from A. thaliana Col-0 roots from plants growing in the same wild soil¹⁷. The evolutionary distance between the two hosts, their distinct microbiomes, and the ability to define the molecular colonization requirements of MF376 on both hosts represented an ideal opportunity to compare and contrast the bacterial genes that affect association with divergent plant species and different organs.

**Fig. 1: Identification of bacterial genes required for full colonization of two divergent plant host species.**

Dozens of bacterial genes affect association with plants

We deployed a TnBarSeq screen²⁹ to identify MF376 genes that regulate host association across different organs of Arabidopsis or Brachypodium. Each bacterial clone in the library has one gene disrupted by a randomly barcoded mariner transposon. We inoculated the MF376 TnBarSeq mutant library onto A. thaliana Col-0 and B. distachyon BD21 seeds after sowing in a soil-like gnotobiotic system³⁰ to mimic the natural colonization process (Fig. 1b; “Methods”). We harvested soil and plants after 6 weeks of growth, separated the above-ground (shoots) from below-ground (roots) organs, and sequenced the population of barcodes (“Methods”). We calculated the relative abundance of mutations in each gene²¹. We identified barcodes in 5962 genes (90% of the genome) and a total of 80,047 unique barcodes with an average of 50,544 barcodes per sample and ~8.5 unique barcodes (insertions) per gene (“Methods”; Supplementary Fig. 2a).

Mutations could, in principle, affect bacterial colonization of the different habitats in different ways (Fig. 1b): no effect (indifferent genes, gray); reduction of barcode abundance in all habitats (general plant association genes, blue), or in only some plant tissues compared to soil (organ-specific, green, or host-specific, turquoise); or reduction of the barcode abundance in soil compared to plant tissues (negative plant-association genes, red). We defined from these data plant-association genes as those genes where clones with mutations in them were depleted from plant tissue compared to the soil (Fig. 1c, blue dots and left inset) and negative plant association genes as those where clones with mutations in them were enriched in plant tissue (Fig. 1c, red dots and right inset). Characterizing the genes that affected association with different plant tissues (Fig. 1c, Supplementary Fig. 2b, and Supplementary Data 1) provided mechanistic insight into the principles and mechanisms of MF376 colonization.

Most MF376 colonization genes are host- and organ-specific

Surprisingly, we found that most colonization genes benefited the association of MF376 with a specific host and/or organ. Thus, while we identified 244 genes as context-specific colonization genes, only seven were identified as general colonization genes that benefited the association with all four tested plant tissues. We showed that 210 of the 251 plant-association genes were host- and organ-specific (first four columns of Fig. 2a). Only a few genes were host-specific and required for colonization of both organs (Fig. 2a, columns 5 and 10). Likewise, a few genes were required for colonization of the same specific organ in both species (Fig. 2a, columns 6 and 9). Few to no genes benefited different organs in the divergent species, i.e. benefit in Arabidopsis root and Brachypodium shoot, but not elsewhere (Fig. 2a, columns 7 and 8). More genes specifically benefited the association with plant shoots (Fig. 2a, columns 1, 3, and 6; 179 genes) than roots (Fig. 2a, columns 2, 4, and 9; 48 genes). This result implies that the selection on bacterial genes in plant shoots is more stringent than that in root (versus selection in the soil). This inference is consistent with the observation that shoot communities exhibit lower similarity with soil than root communities (Fig. 1a). Negative plant-association genes were usually shared among multiple organ–host combinations, 48 of the 104 negative plant association genes were shared among three organ-host combinations (Fig. 2a, red and column 11; the Brachypodium root was the outlier because only a few bacterial genes significantly affected association with it, either positively or negatively). The high number of small-effect association genes (Figs. 1c and 2a), together with the robust colonization by Paraburkholderia of both host plants (Supplementary Fig. 1) and the high species and/or organ context-dependent specificity of colonization genes (Fig. 2a) emphasizes the notion that plant association is a complex, multigenic, phenotype driven by evolution to maximize context-dependent adaptive flexibility.

**Fig. 2: Most MF376 colonization genes affect specific host and organ association.**

We analyzed the functional categories of these genes (Fig. 2b and Supplementary Data 1) using Clusters of Orthologous Genes (COG). GO processes such as carbohydrate and amino-acid metabolism, energy production, transcription, and cell wall integrity (Fig. 2b) are enriched across plant association genes for all plant organs. These results recapitulated, in general, findings from previous studies^19,25,26. We found that genes related to cell motility and to intracellular trafficking and secretion were enriched among negative association genes (see full list of enriched categories in Supplementary Data 2). The genes driving the signal in the intracellular trafficking and secretion category were, in fact, structural genes in the bacterial flagellum apparatus essential for cell motility. These results suggest that the loss of MF376 motility enhances colonization. While this result differs from a previous study²⁵, we note that the two studies surveyed colonization requirements of different focal strains in very different experimental systems.

Two efflux pump systems promote association with either A. thaliana Col-0 roots or shoots

We identified two efflux systems that confer organ and host specificity and likely interface with plant defenses. First, we identified three genes in a predicted operon at MF376 locus IDs H281DRAFT_03189, 03190, and 03191 (with 5, 16, and 13 insertions per gene respectively) that encode proteins with high homology to the three subunits of an efflux pump from the resistance, nodulation, and cell division (RND) superfamily: an inner membrane protein (Locus ID 03190), a periplasmic protein (03189), and an outer membrane protein (03191) (hereafter, the ef90 operon). Insertions into the three genes comprising this operon had similar effects on association. RND efflux pumps are membrane-bound complexes that protect bacteria from antimicrobials by specifically binding deleterious compounds and transporting them out of the bacterial cell³¹. RND systems are prevalent among Gram-negative bacteria and each such system has a specific repertoire of ligands it can bind and transport³². All three efflux pump genes (Fig. 3a, green, top panel) specifically benefited the association of MF376 with A. thaliana Col-0 shoots. Second, we identified a similar efflux system at MF376 loci IDs H281DRAFT_05196, 05197, and 05198 (with 13, 26 and 19 insertions per gene, respectively) that specifically benefited the association with A. thaliana Col-0 roots (Fig. 3a, purple, second panel). Neither of these two putative efflux systems benefited the association of MF376 with either B. distachyon BD21 shoots or roots (Fig. 3a).

**Fig. 3: Organ-specific efflux pumps are required for full colonization of *Arabidopsis thaliana* Col-0 and fitness in leaves.**

RND efflux pumps often remove antimicrobials to confer antimicrobial resistance. Logically, our findings suggested that A. thaliana Col-0 produces at least two different antimicrobials, one in the shoot and one in the root (Fig. 3b), that affect MF376 fitness. We hypothesized that MF376 encodes the two organ-specific efflux pumps to remove putative growth-inhibiting compounds to retain full fitness during plant colonization (Fig. 3b, left). To expand our understanding of commensal plant-association genes in general, and the role of RND efflux pumps in colonization in particular, we repeated the mutant library screen in Arabidopsis and Brachypodium across both organs with a second TnBarSeq library of Pseudomonas simiae WCS417 (hereafter, WCS417), a model gammaproteobacteria commensal³³ (“Methods”; Supplementary Fig. 3a and Supplementary Data 3). The WCS417 results qualitatively recapitulated our findings using MF376 (Supplementary Fig. 3b). We again identified two RND organ-specific efflux systems that promoted the association with either A. thaliana Col-0 shoots (Loci IDs PS417_13190, and 13195 with 53 and 7 insertions per gene respectively) or roots (Loci IDs PS417_01045, and 01050 with 7 and 53 insertions per gene respectively) with (Fig. 3b, right; Supplementary Fig. 3c).

We tested the requirement of the ef90 operon for fitness on A. thaliana Col-0 shoots using a competition assay (Fig. 3c; “Methods”). We first engineered parental MF376 derivatives by knocking resistance to either chloramphenicol (Chl^R wild-type (WT)) or tetracycline (Tet^R WT) into an intergenic region (see “Methods”). We next deleted the entire ef90 operon from each of the parental MF376 derivatives to create scarless derivatives called Chl^R d90 and Tet^R d90, respectively (see “Methods”). We then competed parental WT strains against reciprocally antibiotic-resistant mutant d90 strains in macerated A. thaliana Col-0 leaf extract^34,35 by mixing equal cell numbers of Chl^R WT with Tet^R d90 or Chl^R d90 with Tet^R WT. We chose leaf extracts because we hypothesized that pre-formed antimicrobials would be at higher concentration than in apoplastic wash fluids. We quantified the growth of each strain after plating the overnight mixed culture on Luria Broth (LB) agar plates supplemented with either tetracycline or chloramphenicol (see “Methods”). We found that while the WT strains were unaffected by the presence of Col-0 macerated leaf extract, the ef90 deletion, MF376 d90, resulted in complete growth inhibition when in competition with WT MF376 (Fig. 3c). The parental WT strains and the d90 mutants grew to similar levels when competed in non-extract-supplemented medium or in root extract (Supplementary Fig. 4). We also tested an independent WCS417 mutant allele with a disrupted inner-membrane subunit (ef90 ortholog) isolated from a gridded, sequence-indexed copy of the WCS417 TnBarSeq library²⁵ (Locus ID PS417_13190) (hereafter, WCS417-d90). We measured the bacterial growth of the WT WCS417-WT strain and the mutant WCS417-d90 separately in macerated A. thaliana Col-0 leaf extract diluted in medium (Methods). We found that macerated leaf extract inhibited the mutant WCS417-d90 growth when the extract was diluted 1:8 into medium. By contrast, the growth inhibition of the WCS417-WT strain was only evident at a 1:2 dilution (Fig. 3d). Overall, these results demonstrate that the genes that encode the ef90 efflux pumps in both MF376 and in WCS417 confer fitness against compounds found in macerated A. thaliana Col-0 shoots. The analogous functions of ef90 efflux pumps in MF376 and WCS417 suggests a common mechanism determining their requirement for association with a particular organ, leaves, of a specific host, Arabidopsis, is defined.

Bacterial RND efflux pump gene expression is often induced when the bacteria are exposed to a cognate compound^36,37. We leveraged that regulatory mode to test bacterial transcriptional response to A. thaliana Col-0 shoots in planta. We infiltrated MF376 and WCS417 into leaf apoplasts of 6-week-old plants and quantified the expression levels of each strain’s ef90 inner membrane subunit using reverse transcriptase quantitative polymerase chain reaction (RT-qPCR; Fig. 3e; “Methods”). We found that within 15 min of infiltration ef90 expression was induced fourfold in both MF376 and WCS417. That induction implies that upon leaf infiltration the bacteria are exposed to the A. thaliana Col-0-derived compound that inhibits growth of the d90 mutant.

Homologs of the A. thaliana Col-0-associated, shoot-specific ef90 efflux pump are prevalent among Arabidopsis-associated bacteria

Given the specificity of the MF376 ef90 fitness benefit for growth on macerated leaf extracts of A. thaliana Col-0, and the presence of a similar specific efflux pump in WCS417, we asked if these efflux pumps were a common feature of bacteria strongly associated with Arabidopsis. We used a 185-member collection of genome-sequenced bacteria isolated from Arabidopsis roots^4,13,17. The RND gene family of efflux pumps is widespread among Gram-negative bacteria. We identified the genes that encode all RND efflux pump inner-membrane subunits (also called RND transporter, and hereafter, RND pump) in Gram-negative bacteria in our isolate collection (Fig. 4a; “Methods”) because this subunit determines the compound specificity of RND efflux pumps³⁸. RND efflux pumps are highly abundant in all Gram-negative bacteria in our collection, regardless of taxonomic class, with an average number of 11 genes per isolate. We generated a neighbor-joining tree of all the genes based on sequence similarity (Fig. 4b; “Methods”). The tree is diverse; genes from bacteria of all RND efflux pump classes are distributed along it. Based on the most recent common ancestor of the functionally tested RND efflux pumps from MF376 and WCS417 (orange and cyan dots), we identified clades of genes potentially associated with A. thaliana Col-0 shoots (green, hereafter shoot clade) or roots (purple, hereafter root clade). While the root clade is deep and diverse, the shoot clade is shallow and monophyletic, implying that genes in that ef90 ortholog group perform similar functions and are diverging.

**Fig. 4: Homologs of *ef90* are prevalent among *Arabidopsis thaliana* colonizing Pseudomonadota.**

We found that most of the Pseudomonadota in our collection encode an ef90 ortholog in the shoot clade (Fig. 4c; 78% of Alphaproteobacteria, 100% of Betaproteobacteria, and 91% of Gammaproteobacteria; Supplementary Data 4). Most isolates encode only one gene in that clade (Fig. 4d). Five of the seven bacteria that encode two genes in the shoot clade belong to Variovorax, where a second copy of the gene has diverged from the rest of the tree after gene duplication (Fig. 4e, long diverging orange Comamonadaceae branch). Because the shoot clade genes are clustered with orthologs from bacteria of the same family, we tested whether gene expression increased in A. thaliana Col-0 leaf extracts from a single representative strain of each major family (Fig. 4e; “Methods”; Supplementary Data 5). We found that ef90 ortholog expression is induced following exposure to macerated leaf extract from A. thaliana Col-0 in five of seven tested, clade representative strains (Fig. 4f).

The prevalence of ef90 orthologs in our collection also led us to a wider search for ef90 orthologs in bacterial genomes¹⁷. We found that ef90 is also widespread among bacteria isolated from other plants (Supplementary Fig. 5 and Supplementary Data 6). Importantly, we noticed that many plant pathogens, including P. syringae pv. tomato DC3000, carry ef90 orthologs (Supplementary Fig. 6). That ortholog in DC3000, known as saxF, is a clade A RND efflux pump that protects the bacteria against the antibacterial aliphatic glucosinolate breakdown product sulforaphane^28,39. The existence of this third example of an ef90 pump playing a role in colonization, the shallow monophyletic shoot clade, the high prevalence of a single copy of the shoot pump gene among Arabidopsis-associated bacteria and the transcriptional induction by A. thaliana Col-0 macerated leaf extract suggest that many saxF/ef90 orthologs play roles in removing shoot-derived defense toxins, likely aliphatic glucosinolate breakdown products²⁸, produced in A. thaliana Col-0 shoots.

The plant-derived antimicrobial detoxified by saxF/ef90 is rare among angiosperms and even across A. thaliana accessions

Our screen identified mutants that affected the association of MF376 and WCS417 with A. thaliana Col-0 shoots but had no effect on colonization of B. distachyon BD21. Given the large phylogenetic distance between these two species, it is not surprising that the bacteria associated with these diverged taxa would have efflux pumps tuned to different host antimicrobials like glucosinolates. Given the specificity of glucosinolate-based defenses mainly to the Brassicales⁴⁰, we asked how conserved the observed leaf-derived inhibitory function was across finer plant evolutionary scales. Using the competition assay described in Fig. 3c, we surveyed additional eudicot host plants from different families, the solanaceae Solanum lycopersicum (S.l.), the legume Medicago truncatula (M.t.). These extracts did not inhibit MF376 d90, consistent with expectations, because the aliphatic glucosinolates detoxified by SaxF are limited to the Brassicaceae and possibly the Capparaceae (Fig. 5a). Shoot extract from Capsella bursa-pastoris (C.b.), a close relative of Arabidopsis from the Brassicaceae that makes a completely different set of aliphatic glucosinolates⁴⁰, also did not inhibit MF376 d90 growth, nor induce expression of ef90 (Fig. 5a and Supplementary Figs. 7 and 8). To further refine specificity of the production of the compound detoxified by saxF/ef90, we assessed seven accessions that represent a broad swath of A. thaliana diversity¹³. The MF376 d90 mutant was only inhibited by A. thaliana Col-0 extract (Fig. 5b). These results suggested that the beneficial effect of the MF376 efflux pump is narrow and possibly limited to only a fraction of Arabidopsis accessions.

**Fig. 5: The shoot-derived antibacterial compound is host-genotype specific.**

The saxF/ef90 efflux pump protects commensal bacteria from shoot-derived aliphatic isothiocyanates

We leveraged population genomics tools to dissect the genetics of the production of the antimicrobial detoxified by saxF/ef90. First, we exploited an established panel of 98 RILs of a cross between Col-0 and the non-inhibiting accession Ler-1^41,42. We divided the phenotypes observed in the RIL population into three groups (Fig. 5b): Non-toxic (Ler-1-like behavior, magenta) did not inhibit the WT or the d90 mutant; d90-toxic (Col-0-like behavior, turquoise) inhibited only the d90 mutant; and a novel transgressive phenotype (yellow) that inhibited both the WT and the d90 mutant (WT-toxic; Supplementary Fig. 9a and Supplementary Data 7). Three broad loci exhibited strong linkage to the d90- and WT-toxic phenotypes (Supplementary Data 7). The homology between saxF and d90 (Supplementary Fig. 6) suggested that the Col-0-derived compound that inhibits the d90 mutant is an aliphatic glucosinolate breakdown product²⁸. Consistent with this observation, each of the three peaks in the RIL analysis include a locus that exhibits presence/absence variation between Ler-1 and Col-0 affecting the production of an enzyme in glucosinolate metabolism (Fig. 6a, gray lines; see below). The RIL loci encompass wide genomic regions. We therefore refined our genetic mapping to test if these glucosinolate loci may be the cause of the d90-toxic and WT-toxic phenotypes using 92 select accessions from the A. thaliana 1001 genomes project⁴³. Six out of the 92 accessions expressed the d90-toxic response; these accessions were all isolated in Germany and the Netherlands. Another five expressed the WT-toxic response from diverse sources across Eurasia, which suggests that the novel transgressive WT-toxicity observed in the RILs is found naturally in some wild accessions (Supplementary Fig. 9b and Supplementary Data 8). We performed a GWA analysis comparing the d90-toxic and non-toxic accessions (Supplementary Data 9) and observed a remarkable convergence with the RIL results. The single-nucleotide polymorphism (SNP) with strongest association colocalized with the first peak in the RILs analysis (Fig. 6a, b, ESP peak) and resides within the promoter of the gene epithiospecifier protein (ESP, AT1G54040). Whether or not this is the causal SNP remains to be determined, but we did correlate its presence to decreased expression of ESP (Supplementary Fig. 9c).

**Fig. 6: The saxF/ef90 efflux system protects a commensal bacterial strain against glucosinolate breakdown products from *Arabidopsis thaliana* Col-0 leaves.**

The ESP protein plays a key role in glucosinolate breakdown. Upon foliar damage following herbivory or necrotrophic pathogen infection in Brassicaceae, glucosinolates are broken-down by myrosinases to one of two possible bio-active products in an ESP-dependent manner. First, ESP can promote the formation of a nitrile moiety instead of the isothiocyanate moiety that is spontaneously formed when EPS levels are low or absent⁴⁴ (Fig. 6c). If the glucosinolate has a terminal alkene, ESP will create an epithionitrile. Second, natural genetic variation at the other two RIL peaks encode enzymes that epistatically interact with ESP to differentiate Col-0 and Ler-1 glucosinolate products: methylthioalkylmalate synthase (MAM also known as the GSL-Elong locus), where Col-0 contains a copy of MAM1 (AT5G23010) that produces longer chained glucosinolates compared to MAM2 in Ler-1; and 2-oxoglutarate-dependent dioxygenases (AOPs), where Ler-1 shoots, but not Col-0 shoots, express an active AOP3 (AT4G03050) that converts a methylsulfinyl (MSO) glucosinolate into a hydroxyl glucosinolate⁴⁵ (Fig. 6c).

The parental accessions, Col-0 and Ler-1, produce 4MSO-ITC and 3OHP-Nitrile, respectively, as their dominant chemotypes⁴⁶ (Fig. 6c, thick lines). The hybrid nature of the Col-0 x Ler-1 RILs population includes lines that by recombination shuffle the ESP, MAM and AOP loci to encode non-parental combinations of glucosinolate metabolic enzymes and diverse breakdown products. This leads to eight possible glucosinolate-derived active compounds (Fig. 6c). Of the breakdown products, 3OHP-ITC and 4OHB-ITC go through a rapid spontaneous cyclization to form 1,3-Oxazinane-2-thione and 1,3-Oxazepane-2-thione, respectively⁴⁷. Bacterial inhibition by leaf extracts from specific lines in the Col-0 × Ler-1 RIL population (Fig. 6a and Supplementary Fig. 9a) suggests that three of those products (4MSO-ITC, 3MSO-ITC, and 1,3-Oxazepane-2-thione) are toxic to MF376 and are detoxified by the saxF/ef90 efflux system. We tested commercially available isothiocyanates to validate that the saxF/ef90 efflux system protects MF376 against 4MSO-ITC (Fig. 6d; Sulforaphane) and 3MSO-ITC (Fig. 6e; Iberin) as predicted. A fourth product (1,3-Oxazinane-2-thione) is specifically found in the RIL genotypes and accessions that have WT toxic extracts and is predicted to be highly toxic. We speculate that this compound may bypass the saxF/ef90 bacterial resistance and may be responsible for the WT-toxic phenotype, but it is not commercially available. Note that all of the lines that express a functional ESP, and thus produce nitriles, yield non-toxic extracts.

Discussion

Plant microbiomes affect plants in natural ecological settings and in agriculture^2,48. Yet, despite a growing knowledge base about the distinct microbiomes that inhabit different plants and plant organs, our understanding of the forces that drive microbiome assembly and function remains limited. Here, we show, first, that species- and organ-specific genes dominate the functional landscape of colonization and, second, that diversity in a class of antimicrobial efflux pumps contributes significantly to colonization by commensals that must contend with a diversity of host-produced glucosinolate metabolites deployed upon tissue damage.

We functionally surveyed the genomes of two plant colonizing bacteria, P. bryophila MF376 and P. simiae WCS417, for genes that affect the association with the above- (shoot) or below- (root) ground tissues of the two distantly related host-plant species A. thaliana Col-0 and B. distachyon BD21, we found ~250 genes that promote association with these plants. We demonstrated that a super-majority of plant-association genes in these strains are organ- and host-specific. We focused on two bacterial efflux systems that exemplify such specificity, benefiting the association with either the roots or the shoots of A. thaliana Col-0, but not B. distachyon. Detailed analysis of one of these efflux systems, saxF/ef90, demonstrated that the plant-association benefit of the shoot-specific saxF/ef90 operon is host genotype-specific, even within A. thaliana, and that orthologs of saxF/ef90 are prevalent among Arabidopsis colonizing Pseudomonadota. Further genomic and functional analyses found that the saxF/ef90 pump detoxifies specific aliphatic glucosinolate isothiocyantate break-down products (and not other phytoalexins, Supplementary Fig. 10) whose presence is polymorphic across A. thaliana and the Brassicaceae more broadly. It is noteworthy that the roots of both Col-0 and Ler-1 produce nearly all nitrile glucosinolates, and no glucosinolate isothiocyanates⁴⁹ (see below), possibly explaining the shoot specificity of the ef90 operon mutants.

The aliphatic isothiocyanates 3MSO-ITC (Iberin) and 4MSO-ITC (Sulforaphane) that are detoxified by saxF/ef90 are present only in the shoots of a limited number of geographically restricted A. thaliana accessions, including Col-0. This observation is consistent with the diversity of glucosinolate chemotypes defined across this species⁴⁶. Glucosinolate diversity has been associated with specific activity against diverse herbivores and necrotrophic pathogens⁵⁰. Specific glucosinolates alter interactions with a diversity of biotic organisms and their structural diversity is likely being shaped by these combinatorial interactions; there net effects in any given leaf microhabitat including interactions affecting herbivores, pathogens and now commensals^51,52,53. Conversely, the plant-associated commensal bacterial genomes we surveyed carry an average of 11 RND transporter genes per genome. The widespread distribution of closely related saxF/ef90-homologous operons in the genomes of plant-associated bacteria indicates that commensal bacteria encounter 3MSO-ITC (Iberin) and 4MSO-ITC (Suforaphane), or a structurally related product often enough for the pump function to be widely distributed and retained across evolutionary time. This may arise from the widespread albeit scattered presence of these compounds across the Brassicaceae⁴⁰. Alternatively, the cost of maintaining the saxF/ef90 operons might be so low that diverse plant-associated bacteria retain them due to the likelihood that they may encounter Brassicaceae hosts that do produce these aliphatic glucosinolates.

While we observed enrichment of saxF/ef90 across the genomes of Arabidopsis-associated bacterial isolates compared to non-plant associated isolates, we did not observe enrichment across the genomes of Arabidopsis-colonizing bacteria compared to genomes of bacteria isolated from other plant species (Supplementary Fig. 5). Closely related plants can produce a variety of secondary metabolites that can function as antimicrobials with variation in tissue-specificity of expression and variation in toxin production within a species^54,55. We speculate that saxF/ef90 pumps are functionally important for removing these toxins from commensal bacteria across a wide range of plants. Plant-associated bacteria would be expected to benefit from carrying an array of efflux pumps to allow them to adapt to the unpredictable diversity of antimicrobial compounds they are likely to encounter when colonizing diverse plants.

Glucosinolates and their breakdown products differentially influence herbivory depending on whether the herbivores are specialists or generalists. But our understanding of the role of glucosinolates in shaping the plant microbiome is still limited. Some commensals can utilize specific glucosinolates as nutrient sources. Thus, utilization of allyl-glucosinolate derived isothiocyanates by bacteria^20,40, and plant ER body-associated myrosinases linked to non-host microbe resistance⁵⁶ can affect plant microbiome assembly. While the production of 3MSO-ITC (Iberin) and 4MSO-ITC (Sulforaphane), substrates of the saxF/ef90 pump (Fig. 6d), is not widespread among the tested A. thaliana accessions (Fig. 5b and Supplementary Fig. 9b), they are produced by numerous widespread Brassicaceae species that can be tested in the future, including significant crops like Broccoli^57,58. In addition, RND efflux pumps can have a repertoire of ligands^32,38, and the selection for retention and diversification of the saxF/ef90 efflux pumps may be augmented by additional selective agents, potentially unrelated plant defense metabolites.

Interestingly, 3MSO-ITC (Iberin) and 4MSO-ITC (Sulforaphane) are products of glucosinolate breakdown by myrosinase following tissue damage⁵⁹. The sampling of our in planta bacterial mutant library screen included plant tissue damage, known to lead to near complete conversion of the intact glucosinolates to the isothiocyanates within approximately 1 min⁶⁰. It remains unclear how commensal bacterial confront such glucosinolate breakdown products in natural settings. Some glucosinolates can be found on the leaf surface, though their function there is not clear and there is evidence for steady state endogenous isothiocyanates expressed in the Col-0 accession^61,62. Moreover, foliar microbial communities are likely under constant threat of collateral damage due to occasional bursts of a variety of glucosinolate breakdown products resulting from the many sources of mechanical damage plants face in nature. Indeed, herbivory and necrotrophic pathogen attack alter the plant microbiome by unknown mechanisms⁶³, potentially linked to glucosinolate breakdown²⁰. Importantly, the taxon with highest prevalence of saxF/ef90 pumps (Supplementary Fig. 5), Pseudomonas, is also the most enriched following herbivore attack of the glucosinolate producing Cardamine cardifolia⁶³. The high prevalence of saxF/ef90 among plant commensal bacteria and the enrichment of Pseudomonas upon herbivory suggest that saxF/ef90 protects bacterial microbiota strains from collateral damage of the plant response to such challenges.

More broadly, the efflux systems we defined that contribute to host and organ specificity (Fig. 3) represent essential new details in our understanding of how a commensal bacterium determines host and organ specificity. Antimicrobial resistance in general, and efflux systems specifically, are established mechanisms for niche adaptation⁶⁴. Antimicrobial resistance conferred, for example, by efflux systems, protects human pathogens from antibiotics applied in clinics, and can protect plant pathogens from plant-derived toxins^28,35,65,66. These examples reinforce the concept that microbiome assembly, while largely deterministic, is also largely context dependent. Moreover, the high abundance of efflux pumps among Arabidopsis colonizers implies that other accessions and other plant species produce different repertoires of antimicrobials that select for other pumps. In addition, efflux pumps can also have a role in homeostasis management, siderophore efflux and more. Overall, our work demonstrates a role for efflux pumps in the host- and organ-specificity of plant commensal bacteria colonization.

In our in planta experiments testing the relative fitness of the MF376 d90 mutants, we observed no growth inhibition compared to WT bacteria, even in wounded Arabidopsis leaves where glucosinolates are typically activated. Several factors may explain this outcome. First, glucosinolate activation is a dynamic process where exposure to herbivores or necrotrophic pathogens leads to continuous activation and breakdown. As such, our experimental wounding of A. thaliana leaves may have resulted in a transient response insufficient to reach inhibitory concentrations during our experimental timeframe. Second, and perhaps more significantly, bacteria often possess redundant defense mechanisms against plant antimicrobials. This aligns with the findings of Fan et al.²⁸ that individual mutations, including saxF, exhibit prominent loss of detoxification function phenotypes in vitro but have negligible in planta effects. In this example significant susceptibility only manifests in a quadruple mutant (saxA, saxD, saxF, saxG). Therefore, while our d90 mutant demonstrates increased sensitivity to purified glucosinolates like sulforaphane or shoot extract in vitro, complementary efflux systems or detoxification pathways might provide sufficient protection against physiological concentrations of structurally diverse glucosinolates encountered in the plant environment. Further research is required to dissect this complex response.

Finally, the high specificity of plant-association genes and the plethora of genes and mechanisms that affect plant-association should inform methods to tailor therapeutic bacteria developed for the field⁴⁸. Our results show that the presence of a particular bacteria trait, the saxF/ef90 efflux pump, can have profound effects on colonization success across even within a species. This high specificity potentially poses a challenge for application of microbial interventions across different cultivars and across different crop species, but this high specificity might also may allow opportunities for fine scale tailoring of microbial strains or synthetic communities to specific plant host genotypes.

Methods

Plant material

The following plant species were used in this study: A. thaliana Col-0, B. distachyon BD21, S. lycopersicum Microtom, M. truncatula R108-1, C. bursa-pastoris. In addition, we tested 98 RILs coming from a Col-0 × Ler-1 cross (Supplementary Data 7) and 92 other A. thaliana accessions (Supplementary Data 8).

Microbiome analysis in wild soil

Soil sample and plant growth

Soil was collected from Mason Farm in Chapel Hill, North Carolina, USA (coordinates 35.889114, −79.0145183), and sieved through a 2 mm sieve. Round 2-inch pots were filled with soil and sown with seeds of either A. thaliana Col-0 or B. distachyon BD21. Seeds were stratified for 2 days at 4 °C in the dark then placed into a growth chamber at 22 °C under short day conditions (8 h light 16 h dark) for 6 weeks.

Harvest and DNA extraction

After 6 weeks, plants were harvested, and shoots were separated from roots. Loosely attached soil was removed manually from the roots, and then the roots and the shoots were washed separately three times by vortexing 15 s in water. The washed tissues were dab dried, flash frozen and lyophilized. In addition, samples of 250 mg of soil were collected, flash frozen and lyophilized. DNA was extracted from the lyophilized samples using the Dneasy Powersoil Pro kit (QIAGEN).

Sequencing library prep

The V3V4 region of the 16S rRNA gene was amplified using a 2-step, dual-index approach to barcode the amplified DNA at the 3’ and 5’ ends⁶⁷. Briefly, the V3V4 locus was amplified using Kapa Hifi (Kapa Biosystems) with primers 16S_V3 F and 16S_V4 R for 30 cycles. The products of this first PCR were then diluted 100-fold and used as template for the barcoding PCR with primers illumina_inx_x F and illumina_inx_x R for 10 cycles. Individual PCR libraries were pooled in equimolar ratios and cleaned using 0.8X AMPure XP beads (Beckman Coulter). Pooled libraries were sequenced on an Illumina MiSeq using 2 × 300 bp paired-end reads. We recovered on average 105,485 (range: 46,579–254,091) high quality bacterial sequences from each sample.

Community assembly analysis

We used the R package “DADA2” to process our sequencing reads⁶⁸, which infers amplicon sequence variants (ASV), and the R package “phyloseq” to further process our samples⁶⁹. We constructed a phylogenetic tree of our ASVs with the phylogenetic placement method outlined in Janssen et al.⁷⁰ and implemented using the SILVA 128 reference alignment in QIIME2⁷¹. We simplified our dataset to include only common ASVs for downstream analyses of community composition. We defined common as occurring in at least 5 samples at an abundance of at least 25 sequences¹³, which yielded 702 ASVs and accounted for 94% of the total number of sequences in the dataset. We performed a proportional abundance normalization (sequencing reads for an ASV in a given sample divided by the total number of sequencing reads in that sample) on this common set of ASVs for our analysis of community composition⁶⁹. We also rarefied samples to the lowest read count in the dataset (46,579 reads) and found nearly identical results to the proportional abundance normalized analyses. We report results with the proportional abundance normalized dataset with the 5 × 25 threshold. We used permutational multivariate analysis of variance (R function “adonis2” from the “vegan” package v. 2.5-7, marginal significance of terms computed) to analyze bacterial composition using the common ASV dataset described above. We used three different measures of community distance, weighted UniFrac, unweighted UniFrac, and Bray–Curtis. We found highly significant effects of fraction for all samples, and host species for root and shoot samples regardless of the community distance measure used. For visualization purposes we performed principal coordinates analyses using unweighted UniFrac distance matrices.

Gnotobiotic system, bacterial inoculum, and plants’ growth

Approach

Bacterial mutant abundance on the plant and in the soil are quantified using amplicon sequencing of barcodes, Therefore, natural resident bacteria will not interfere with the identification of mutants. Yet, those resident bacteria can compete with members of the mutant library of a focal strain and reduce its colonization. Many colonizing bacteria are essential for successful analysis of differences in mutant fitness in the different environments. To enhance the colonization of members of the mutant library we followed techniques taken from gnotobiotic systems.

Pot preparation

Plants were grown in a Gnotopot system based on Kramer et al.³⁰. In short, peat pellets were placed in 1 inch round pots and soaked in 0.5xMS until the pellets were saturated. The saturated pellets were then autoclaved three times, with 48 h incubation at room temperature between the first and second autoclave. The pots were distributed into flats and covered with tall domes.

Seed preparation

Seeds were sterilized for 5 min in 50% bleach, then were washed 3 times in sterile water. The sterile seeds were stratified for 3 days at 4 °C in the dark before sowing.

Bacteria preparation

Mutant libraries of P. bryophila MF376 and P. simiae WCS417 were grown overnight in 2xYT broth then washed and diluted in sterile water. A diverse auxiliary community of four plant-associated bacterial strains were also prepared in order to test the effect of competing/promoting strains. These strains, (Actinobacteria CurtobacteriumCL20, Alphaproteobacteria Sphingomonas MF220A, Betaproteobacteria Ralstonia CL21, and Gammaproteobacteria Stenotrophomonas MF92) were grown individually in 2xYT broth overnight, mixed to equal amounts, and the mixture was diluted in sterile water.

Seed sowing and bacteria inoculation

Each pot was sown with five stratified sterile seeds of either A. thaliana Col-0 or B. distachyon BD21. Half of the pots were inoculated with 10⁸ bacteria of the focal, mutant library strain (either MF376 or WCS417), in isolation, while the other half was inoculated with 10⁸ bacteria from the focal strain accompanied with 10⁷ bacteria from the small auxiliary community. In each flat, two inoculated pots were left without plants as soil only controls.

Plant growth

The plants were transferred to a growth chamber in flats with plastic domes. They were grown for 6 weeks under short day conditions (8 h light, 16 h dark). After 2 weeks, the plants were thinned to three plants per pot. After 4 weeks, cracks in the domes were opened to reduce the humidity around the plants. The plants were grown for a total of 6 weeks before harvest.

Harvest, preparation of sequencing libraries, and sequencing

Samples harvest

Six-week-old plants were harvested. To increase the sample-to-sample reproducibility, the plants were pooled to groups of 4–5 pots per sample (12–15 plants total). The above (shoot) and below (root) the ground tissues were separated. To remove non-attached bacteria, the shoots and roots were washed three times as detailed above. The clean tissue was then ground using a mortar and pestle, then the ground tissue was emersed in 20 ml of antibiotic-supplemented 2xYT broth to enrich the bacteria within the plant, similar to previous work²⁵. For soil samples, 1 g of soil was harvested for each sample and placed into antibiotic-supplemented 2xYT broth in a similar manner. The 2xYT medium was supplemented by antibiotic combinations that inhibit the growth of members of the auxiliary community but not of the focal strain: 100 mg/l Kanamycin and 4 mg/l Carbenicillin for MF376, and 100 mg/l Kanamycin and 8 mg/l Rifampicin for WCS417. After a short overnight growth (8 h), the DNA was extracted from the culture using DNeasy Power soil 96 kit (QIAGEN). Due to the high replication and the high number of samples, we had to split the harvest to 2 consecutive days.

Sequencing library preparation

Preparation of the libraries for sequencing was done following a 2-step, dual-index approach to barcode amplified DNA at the 3’ and 5’ ends: firstly a PCR to amplify the barcode region and add the internal parts of Illumina adapters, followed by a second PCR that added multiplexing indexes and the external parts of Illumina adapters. All amplification steps were done with KAPA HIFI readymix (KAPA biosystems). The first PCR with primers BS_int F and BS_int R (Supplementary Data 10) for 25 cycles amplified the barcodes and added adapters. The first PCR product was diluted 1:50 in water and a second PCR was performed with primers Illumina_inx_x F and illumina_inx_x R for 15 cycles to add multiplexing indexes and Illumina adapters. The libraries were then quantified on an agarose gel and pooled equimolarly. The pooled libraries were run again on 1.5% agarose gel for size selection, the fragments around 150 bp were cut from gel and cleaned with QIAquick Gel Extraction Kit (QIAGEN). The libraries were sequenced using 50 bp single reads on an Illumina Hiseq 2500.

Sequencing analysis and identification of differentially abundant mutants

Basic analysis

The assignment of fitness score to each mutant in each treatment was done based on the BarSeqR pipeline established by Wetmore et al.²¹. In short, the 20 bases barcodes were identified based on the preceding six nucleotide. The barcodes were then mapped to the bacterial genome using a mapping table. The BarSeqR pipeline calculates the fitness by comparing the mutants’ abundance in each sample to its abundance in a T0 control. Since bacteria colonize the plant from the surrounding substrate, we defined the substrate samples (soil) as the control. The fitness of mutants in a gene is then computed as the log₂ transform of a weighted average of all mutants in that gene²¹. The calculated fitness score of mutants in each gene for each sample is used for further analysis. Overall, we identified an average of 50,544 unique barcodes per sample, which sums up to ~8.5 unique barcodes per gene per sample on average (Supplementary Fig. 2).

Identification of differentially abundant mutants

The BarSeqR pipeline also computes a P value based on T-like statistics from multiple competing mutants in the same gene. While that approach yields good results in culture with large inoculum²¹, the limited number of plant colonization events leads to incomplete sampling from the diverse library and hence different starting population on each plant sample. Those differences in starting population hinder the use of that approach. Instead, we build a regression model for the fitness score of each mutant, comparing its fitness in the sample to its fitness in the soil: Fitness ~ Sample_type + Auxiliary_community + Harvest_day + (1|flat), where the Sample type (plant sample or soil), Auxiliary_community (Focal strain inoculated in isolation or with an auxiliary community) and Harvest_day (first or second day of harvest) are fixed variable and the flat number (overall we had 16 flats) is a random variable. Following the fit to the model, the distribution of residuals was tested, and genes that did not follow normal distribution in any of the samples were omitted from the analysis (Shapiro–Wilk test, P value <0.05, a total of 486 genes). Genes that satisfy normal distribution and with a false discovery rate (FDR) corrected (Benjamini–Hochberg procedure) P value <0.05 and Log₂(Fold change) > |1| were defined as “Affecting association with plants.” The addition of the auxiliary community and the day of the harvest affected only a small number of genes and are not discussed in the paper.

Engineer antibiotic-resistant derivatives of P. bryophila MF376

Approach

Two derivatives of MF376 with differentiating antibiotic resistance were engineered to conduct a competition assay between WT and mutant derivatives. Antibiotic resistance cassettes were added to the genome in an intergenic region by a gene knock-in approach based on the SacB negative selection gene to create antibiotic-resistant parental derivatives Tet^R WT and Chl^R WT. All oligos sequences can be found in Supplementary Data 10.

Cloning plasmids pMo130_par_ TetR and pMo130_par_ ChlR

An intergenic region in the MF376 genome was identified at position 319,584–319,849 between genes H281DRAFT_00298 and H281DRAFT_00299. To clone antibiotics resistance genes into that region, we used the plasmid pMo130 that contains a Kanamycin resistance gene for positive selection and the SacB gene for negative selection. The plasmid was amplified by PCR using primers GD_pMo130 F and GD_pMo130 R. Fragments of 1000 bp upstream and downstream the intergenic point were picked from the MF376 genome by PCR using primers GD005_par_flkU F and GD006_par_flkU R (upstream) and GD007_par_flkD F and GD008_par_flkD R (downstream). The TetA gene that confers resistance to Tetracycline was picked from plasmid pBBR1-MCS3 with primers GD001_TetR R and GD002_TetR F. The Chloramphenicol acetyl transferase (CAT) gene that confers resistance to Chloramphenicol was picked from plasmid pBBR1-MCS1 with primers GD003_ChlR R and GD004_ChlR F. The primers that amplified the antibiotic resistance cassettes and the genomic fragments added complementary sequences on their 3’ end. Each of the antibiotic resistance genes were cloned into plasmid pMo130 between the upstream and the downstream genomic fragments using Gibson assembly to create plasmids pMo130_par_TetR and pMo130_par_ChlR. Each of the new two plasmids were transformed to E. coli Top10 cells by heat shock transformation for propagation in 2xYT supplemented by 50 mg/l Kanamycin and extracted using QIAprep Spin Miniprep Kit (QIAGEN).

Selecting antibiotic resistance derivatives of P. bryophila MF376

Each of the plasmids pMo130_par_TetR and pMo130_par_ChlR were transformed into the 2,6-diaminopimelic acid (DAP) auxotrophic biparental mating strain E. coli WM3064 by electroporation. The transformed mating strains were selected on LB supplemented by 0.3 mM DAP, 50 mg/l Erythromycin, and 50 mg/l Kanamycin. The transformed mating strains were each mixed with the parental strain MF376 and plated on LB agar plates supplemented with 0.3 mM DAP. After overnight growth, the transconjugated MF376 derivatives were selected on LB agar plates supplemented with 3 mg/l Tetracycline or 10 mg/l Chloramphenicol. The antibiotic-resistant mating strain derivatives were counter selected due to their DAP auxotrophy. Transconjugate MF376 clones were cleaned twice by streaking onto antibiotic LB agar plates, where the plasmid was able to integrate into the genome by homologous recombination and then plated on LB agar plates supplemented by 10% sucrose to select against the SacB gene. To survive the sucrose selection, bacteria must lose the integrated pMo130 plasmid and recombine either to the parental form (no antibiotic resistance cassette) or to the antibiotic resistance derivative (where the antibiotic resistance cassette resides between the genomic fragments). Next, colonies that were selected on 10% sucrose were streaked onto LB agar plates supplemented with 3 mg/l Tetracycline or 10 mg/l Chloramphenicol to select for the antibiotic-resistant derivatives Tet^R WT and Chl^R WT, respectively.

Gene deletion in P. bryophila MF376

Approach

The ef90 operon was deleted from both antibiotic-resistant MF376 derivatives by a similar approach. All oligos sequences can be found in Supplementary Data 10.

Cloning plasmid pMo130_d90

Flanking DNA fragments (1 kb long) upstream and downstream of the ef90 operon were amplified from the genome of MF376 using primers GD055_flkU-3190 F and GD056_flkU-3190 R (upstream gene loci ID H281DRAFT_03188) and GD057_flkD-3190 F and GD058_flkD-3190 R (downstream gene loci ID H281DRAFT_03191). Primers GD056_flkU-3190 R and GD057_flkD-3190 F have 5’ overhanging ends complementary to each other, and primers GD055_flkU-3190 F and GD058_flkD-3190 R have overhanging 5’ ends complementary to the amplified pMo130 plasmid. The upstream and downstream genomic fragments were then integrated into the pMo130 plasmid by Gibson assembly to create plasmid pMo130_d90 that was then transformed into E. coli Top10 cells by heat shock transformation. The transformed Top10 cells were plated on LB agar supplemented by 50 mg/l Kanamycin, and after 2 days of growth colonies were validated using primers GD061_vo_3190 F and GD062_vo_3190 R that sit within the upstream and downstream MF376 genomic fragments, respectively. The plasmid was propagated in 2xYT supplemented by 50 mg/l Kanamycin and extracted using the QIAprep Spin Miniprep Kit (QIAGEN).

Deleting operon ef90 from MF376

Plasmid pMo130_d90 was transformed into the biparental mating strain E. coli WM3064 by electroporation. The transformed derivative of the mating strain was selected onto LB agar plates supplemented by 0.3 mM DAP, 50 mg/l Erythromycin, and 50 mg/l Kanamycin. Next, the transformed derivative of the mating strain was mixed with each of the antibiotic-resistant parental derivatives Tet^R WT and Chl^R WT and plated onto LB agar plates supplemented with 0.3 mM DAP. After overnight growth, the transconjugated MF376 derivatives were selected on LB agar plates supplemented with 50 mg/l Kanamycin. Transconjugate Tet^R WT and Chl^R WT MF376 clones were cleaned twice by streaking on Kanamycin LB agar plates and then were plated onto LB agar plates supplemented with 10% sucrose to select against the SacB gene. Colonies growing on the sucrose plates were tested for the deletion of the ef90 operon by two PCRs. A reaction with primers GD061_vo_3190 F and GD062_vo_3190 R (within the upstream and downstream flanking fragments) yielded a band since the deleted operon left the flanking fragments in proximity to one another, and a reaction with primers GD059_vi_3190 F and GD060_vi_3190 R (within the ef90 operon) yielded no band indicating that the fragment was deleted. The new two derivatives, TetR d90 and ChlR d90, were cleaned by streaking on LB agar plates and revalidation by PCR.

Competition assay between WT and d90 mutant derivative of P. bryophila MF376

Macerated leaf extract preparation

Plants were grown in a potting mix for 6 weeks under short day conditions (8 h light 16 h dark 22 °C/18 °C, respectively). After 6 weeks, plants were harvested the shoots and roots separated, and then weighed. Next, the shoots were mixed with equal volume of 2xYT broth and were ground using a mortar and pestle. The macerated leaf extract was filtered through a 0.22 µm filter to remove bacteria, serial diluted in 2XYT broth and 90 µl was pipeted into 96-well flat bottom plates.

Bacterial preparation

Each of the four MF376 derivatives was cultured shaking overnight at 28 °C in 2xYT. All derivatives were diluted to OD₆₀₀ 0.01 and WT derivatives were mixed with mutant derivatives as follows: ChlR WT was mixed with mutant TetR d90 and reciprocally TetR WT was mixed with mutant ChlR d90.

Competition assay, plating, and imaging

The mixed bacterial cultures (ChlR WT versus TetR d90 and TetR WT versus ChlR d90) were diluted 1:10 into the macerated soot extract (10 µl into 90 µl) to yield an inoculum of 10⁵ bacteria. After overnight growth in a shaking incubator at 28 °C, the culture was serial diluted in water and 5 µl drops from each dilution of each culture were plated onto LB agar plates supplemented with either 8 mg/l Chloramphenicol or 3 mg/l Tetracycline. Colonies were grown for 2 days at 28 °C and the plates were imaged using a tabletop scanner. The binary nature of the growth (full growth inhibition vs full growth) negated the need for quantification.

Comparing the growth of WT P. simiae WCS17 and an ef90 homolog mutant in macerated leaf extract

Both WT WCS417 and the ef90 mutant were cultured shaking overnight at 28 °C in 2xYT. The bacterial cultures were diluted to OD₆₀₀ 0.01 and further diluted 1:10 into serial dilutions of macerated leaf extract (10 µl into 90 µl) to yield an inoculum of 10⁵ bacteria in each well of a flat bottom 96-well plate. Bacteria were grown overnight shaking at 28 °C in a temperature-controlled plate reader (Infinite 200 pro, Tecan). During growth, OD₆₀₀ was measured every 10 min. Plotting bacterial growth over time, we defined the time when the no-plant control culture reached 90% culture saturation as the endpoint of the experiment. For each sample, the final OD600 was defined as the OD600 at the endpoint. Background OD600 was defined as the fifth lowest OD600 measurement (absorbance of plant material diminishes over time so taking initial OD600 yields negative values). The reported OD600 is final OD600 minus background OD600.

Quantifying ef90 expression in A.t. leaves and in macerated leaf extract

Infiltration and harvest

Overnight culture of WT MF376 and WCS417 were diluted 1:20 in 2xYT and recovered for 2 h shaking at 28 °C. The recovered bacteria were diluted to OD600 0.01, washed twice in 10 mM MgCl₂, and hand infiltrated into leaves of 6-week-old A. thaliana Col-0. Leaves were harvested after 15 and 60 min. As a control, leaves infiltrated with MgCl₂ without bacteria were sampled. All samples were flash froze upon harvest.

Inoculation in macerated leaf extract and harvest

Macerated leaf extract was diluted 1:4 in 2xYT. Log phase bacteria were inoculated into the diluted leaf extract at a final OD₆₀₀ of 0.01. After 15 min, the cells were pelleted by centrifugation at 5000 × g for 2 min, and the supernatant was removed in the cells were flash frozen in liquid nitrogen.

RNA extraction, RT-qPCR, and analysis

RNA was isolated using RNeasy kit (QIAGENE), and cDNA was prepared using Superscript IV (Invitrogen). The qPCR was preformed using PowerUp SYBR Green (Applied Biosystems) on a VIIA7 machine (Applied Biosystems). Relative gene expression was calculated as −ΔΔCT with recA as the internal control. There was no signal in “no bacteria” control samples. All primers used for RT-qPCR can be found in Supplementary Data 10.

Building a gene tree of all efflux pumps in a 175-member isolate-collection

The inner-membrane subunit of all Gram-negative members of the 185-member isolate collection was identified by the COG identifier COG0841 (Multidrug efflux pump subunit AcrB). Overall, 823 genes were identified. All pairwise protein sequences were aligned and the pairwise distances were measured as a BLOSUM50 dissimilarity score using the seqpdist command (MATLAB R2024a). Next, a neighbor-joining gene tree of all 823 genes were built based on the distance matrix. The tree was plotted using ggtree (R). Shoot clade (ef90 orthologs) and root clade were identified based on the most recent common ancestor of the functionally tested genes from MF376 and WCS417.

Measure the toxicity of selected isothiocyanates

The commercially available isothiocyanates Sulforaphane (4MSO-ITC, Sigma, Cat. S6317) and Iberin (3MSO-ITC, Sigma, Cat. SML3844) were dissolved in dimethyl sulfoxide to a stock concentration of 20 mg/ml. Each of the two compounds were serially diluted 1.5-fold in 2xYT in a microtiter plate. The diluted compounds were inoculated by 5 × 10⁵ bacteria per well, and plates were incubated overnight shaking at 28 °C. After overnight growth, bacterial load was estimated by OD600.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

All sequencing data generated and analyzed in the current study are deposited in publicly available archives. The 16S data is deposited at the sequencing read archive (SRA) as BioProject PRJNA1189262 and the BarSeq barcodes sequencing is deposited at the European nucleotide archive (ENA) as study ERP166428 [https://www.ebi.ac.uk/ena/browser/view/PRJEB82761]. The TnBarSeq results are available at Figshare (https://doi.org/10.6084/m9.figshare.28716422). All other data and scripts are available upon request.

Code availability

Figures were generated with custom R and MATLAB codes. All custom codes used in this study are available upon request.

Change history

25 July 2025
This article was originally published with the incorrect license ‘CC BY-NC-ND 4.0’; it should have been ‘CC BY 4.0’. The original article has been corrected.

References

Trivedi, P., Leach, J. E., Tringe, S. G., Sa, T. & Singh, B. K. Plant–microbiome interactions: from community assembly to plant health. Nat. Rev. Microbiol. 18, 607–621 (2020).
Article CAS PubMed Google Scholar
Fitzpatrick, C. R. et al. The plant microbiome: from ecology to reductionism and beyond. Annu. Rev. Microbiol. 74, 81–100 (2020).
Article CAS PubMed Google Scholar
Vannier, N., Agler, M. & Hacquard, S. Microbiota-mediated disease resistance in plants. PLoS Pathog. 15, e1007740 (2019).
Article CAS PubMed PubMed Central Google Scholar
Finkel, O. M. et al. A single bacterial genus maintains root growth in a complex microbiome. Nature 587, 103–108 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Huang, A. C. et al. A specialized metabolic network selectively modulates Arabidopsis root microbiota. Science 364, eaau6389 (2019).
Article CAS PubMed Google Scholar
Cotton, T. E. A. et al. Metabolic regulation of the maize rhizobiome by benzoxazinoids. ISME J. 13, 1647–1658 (2019).
Article CAS PubMed PubMed Central Google Scholar
Tian, T., Reverdy, A., She, Q., Sun, B. & Chai, Y. The role of rhizodeposits in shaping rhizomicrobiome. Environ. Microbiol. Rep. 12, 160–172 (2020).
Article PubMed Google Scholar
Stringlis, I. A. et al. MYB72-dependent coumarin exudation shapes root microbiome assembly to promote plant health. Proc. Natl. Acad. Sci. USA 115, E5213–E5222 (2018).
Rizaludin, M. S., Stopnisek, N., Raaijmakers, J. M. & Garbeva, P. The chemistry of stress: understanding the ‘cry for help’ of plant roots. Metabolites 11, 357 (2021).
Article CAS PubMed PubMed Central Google Scholar
Salas-González, I. et al. Coordination between microbiota and root endodermis supports plant mineral nutrient homeostasis. Science 371, eabd0695 (2021).
Article PubMed Google Scholar
Wippel, K. et al. Host preference and invasiveness of commensal bacteria in the Lotus and Arabidopsis root microbiota. Nat. Microbiol. 6, 1150–1162 (2021).
Article CAS PubMed PubMed Central Google Scholar
Fitzpatrick, C. R. et al. Assembly and ecological function of the root microbiome across angiosperm plant species. Proc. Natl. Acad. Sci. USA 115, E1157–E1165 (2018).
Lundberg, D. S. et al. Defining the core Arabidopsis thaliana root microbiome. Nature 488, 86–90 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Bulgarelli, D. et al. Revealing structure and assembly cues for Arabidopsis root-inhabiting bacterial microbiota. Nature 488, 91–95 (2012).
Article ADS CAS PubMed Google Scholar
Nakamura, N., Toju, H. & Kitajima, K. Leaf, root, and soil microbiomes of an invasive plant, Ardisia crenata, differ between its native and exotic ranges. Front. Microbiol. 14, 1302167 (2023).
Article PubMed PubMed Central Google Scholar
Song, S., Liu, Y., Wang, N. R. & Haney, C. H. Mechanisms in plant–microbiome interactions: lessons from model systems. Curr. Opin. Plant Biol. 62, 102003 (2021).
Article CAS PubMed Google Scholar
Levy, A. et al. Genomic features of bacterial adaptation toplants. Nat. Genet. 50, 138–150 (2018).
Article CAS Google Scholar
López, J. L. et al. Growth rate is a dominant factor predicting the rhizosphere effect. ISME J. 17, 1396–1405 (2023).
Article PubMed PubMed Central Google Scholar
Vannier, N. et al. Genome-resolved metatranscriptomics reveals conserved root colonization determinants in a synthetic microbiota. Nat. Commun. 14, 8274 (2023).
Article ADS CAS PubMed PubMed Central Google Scholar
Unger, K. et al. Glucosinolate structural diversity shapes recruitment of a metabolic network of leaf-associated bacteria. Nat. Commun. 15, 8496 (2024).
Article CAS PubMed PubMed Central Google Scholar
Wetmore, K. M. et al. Rapid quantification of mutant fitness in diverse bacteria by sequencing randomly bar-coded transposons. mBio 6, e00306-15 (2015).
Article PubMed PubMed Central Google Scholar
Adler, B. A. et al. The genetic basis of phage susceptibility, cross-resistance and host-range in Salmonella. Microbiology 167, 001126 (2021).
Torres, M., Paszti, S. & Eberl, L. Shedding light on bacteria–host interactions with the aid of TnSeq approaches. mBio 15, e00390-24 (2024).
Article PubMed PubMed Central Google Scholar
Liu, Z. et al. A genome-wide screen identifies genes in rhizosphere-associated Pseudomonas required to evade plant defenses. mBio 9, e00433-18 (2018).
Article PubMed PubMed Central Google Scholar
Cole, B. J. et al. Genome-wide identification of bacterial plant colonization genes. PLoS Biol. 15, e2002860 (2017).
Article PubMed PubMed Central Google Scholar
Helmann, T. C., Deutschbauer, A. M. & Lindow, S. E. Distinctiveness of genes contributing to growth of Pseudomonas syringae in diverse host plant species. PLoS ONE 15, e0239998 (2020).
Article CAS PubMed PubMed Central Google Scholar
Wheatley, R. M. et al. Lifestyle adaptations of Rhizobium from rhizosphere to symbiosis. Proc. Natl. Acad. Sci. USA 117, 23823–23834 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Fan, J. et al. Pseudomonas sax genes overcome aliphatic isothiocyanate–mediated non-host resistance in Arabidopsis. Science 331, 1185–1188 (2011).
Article ADS CAS PubMed Google Scholar
Price, M. N. et al. Oxidative pathways of deoxyribose and deoxyribonate catabolism. mSystems 4, e00297-18 (2019).
Article PubMed PubMed Central Google Scholar
Kremer, J. M. et al. Peat-based gnotobiotic plant growth systems for Arabidopsis microbiome research. Nat. Protoc. 16, 2450–2470 (2021).
Article CAS PubMed PubMed Central Google Scholar
Nikaido, H. & Takatsuka, Y. Mechanisms of RND multidrug efflux pumps. Biochim. Biophys. Acta 1794, 769–781 (2009).
Article CAS PubMed Google Scholar
Yamasaki, S., Zwama, M., Yoneda, T., Hayashi-Nishino, M. & Nishino, K. Drug resistance and physiological roles of RND multidrug efflux pumps in Salmonella enterica, Escherichia coli and Pseudomonas aeruginosa. Microbiology 169, 001322 (2023).
Pieterse, C. M. J. et al. Pseudomonas simiae WCS417: star track of a model beneficial rhizobacterium. Plant Soil 461, 245–263 (2021).
Article CAS Google Scholar
Lundberg, D. S. et al. Contrasting patterns of microbial dominance in the Arabidopsis thaliana phyllosphere. Proc. Natl. Acad. Sci. USA 119, e2211881119 (2022).
Article CAS PubMed PubMed Central Google Scholar
Burse, A., Weingart, H. & Ullrich, M. S. The phytoalexin-inducible multidrug efflux pump AcrAB contributes to virulence in the fire blight pathogen, Erwinia amylovora. Mol. Plant Microbe Interact. 17, 43–54 (2004).
Article CAS PubMed Google Scholar
Novelli, M. & Bolla, J.-M. RND efflux pump induction: a crucial network unveiling adaptive antibiotic resistance mechanisms of Gram-negative bacteria. Antibiotics 13, 501 (2024).
Article CAS PubMed PubMed Central Google Scholar
Lomovskaya, O., Lewis, K. & Matin, A. EmrR is a negative regulator of the Escherichia coli multidrug resistance pump EmrAB. J. Bacteriol. 177, 2328–2334 (1995).
Article CAS PubMed PubMed Central Google Scholar
Elkins, C. A. & Nikaido, H. Substrate specificity of the RND-type multidrug efflux pumps AcrB and AcrD of Escherichia coli is determined predominately by two large periplasmic loops. J. Bacteriol. 184, 6490–6498 (2002).
Article CAS PubMed PubMed Central Google Scholar
Bodilis, J. et al. HME, NFE, and HAE-1 efflux pumps in Gram-negative bacteria: a comprehensive phylogenetic and ecological approach. ISME Commun. 4, ycad018 (2024).
Article PubMed PubMed Central Google Scholar
Fahey, J. W., Zalcmann, A. T. & Talalay, P. The chemical diversity and distribution of glucosinolates and isothiocyanates among plants. Phytochemistry 56, 5–51 (2001).
Article CAS PubMed Google Scholar
Lister, C. & Dean, C. Recombinant inbred lines for mapping RFLP and phenotypic markers in Arabidopsis thaliana. Plant J. 4, 745–750 (1993).
Article CAS Google Scholar
Singer, T. et al. A high-resolution map of Arabidopsis recombinant inbred lines by whole-genome exon array hybridization. PLoS Genet. 2, e144 (2006).
Article PubMed PubMed Central Google Scholar
Alonso-Blanco, C. et al. 1,135 Genomes reveal the global pattern of polymorphism in Arabidopsis thaliana. Cell 166, 481–491 (2016).
Article Google Scholar
Lambrix, V., Reichelt, M., Mitchell-Olds, T., Kliebenstein, D. J. & Gershenzon, J. The Arabidopsis epithiospecifier protein promotes the hydrolysis of glucosinolates to nitriles and influences Trichoplusia ni herbivory. Plant Cell 13, 2793–2807 (2001).
Article CAS PubMed PubMed Central Google Scholar
Sønderby, I. E., Geu-Flores, F. & Halkier, B. A. Biosynthesis of glucosinolates – gene discovery and beyond. Trends Plant Sci. 15, 283–290 (2010).
Article PubMed Google Scholar
Katz, E. et al. Genetic variation, environment and demography intersect to shape Arabidopsis defense metabolite variation across Europe. eLife 10, e67784 (2021).
Article CAS PubMed PubMed Central Google Scholar
Fernández-Bolaños, J. G. & López, Ó. in Heterocycles from Carbohydrate Precursors, Vol. 7 (ed. El Ashry, E. S. H.) 67–100 (Springer Berlin Heidelberg, 2006).
Russ, D., Fitzpatrick, C. R., Teixeira, P. J. P. L. & Dangl, J. L. Deep discovery informs difficult deployment in plant microbiome science. Cell 186, 4496–4513 (2023).
Article CAS PubMed Google Scholar
Wentzell, A. M. & Kliebenstein, D. J. Genotype, age, tissue, and environment regulate the structural outcome of glucosinolate activation. Plant Physiol. 147, 415–428 (2008).
Article CAS PubMed PubMed Central Google Scholar
Chhajed, S. et al. Glucosinolate biosynthesis and the glucosinolate–myrosinase system in plant defense. Agronomy 10, 1786 (2020).
Article CAS Google Scholar
Kerwin, R. E. et al. Epistasis × environment interactions among Arabidopsis thaliana glucosinolate genes impact complex traits and fitness in the field. N. Phytol. 215, 1249–1263 (2017).
Article CAS Google Scholar
Kerwin, R. et al. Natural genetic variation in Arabidopsis thaliana defense metabolism genes modulates field fitness. eLife 4, e05604 (2015).
Article PubMed PubMed Central Google Scholar
Wang, W. et al. An arabidopsis secondary metabolite directly targets expression of the bacterial type III secretion system to inhibit bacterial virulence. Cell Host Microbe 27, 601.e7–613.e7 (2020).
Article Google Scholar
Moore, B. D., Andrew, R. L., Külheim, C. & Foley, W. J. Explaining intraspecific diversity in plant secondary metabolites in an ecological context. N. Phytol. 201, 733–750 (2014).
Article Google Scholar
Kessler, A. & Kalske, A. Plant secondary metabolite diversity and species interactions. Annu. Rev. Ecol. Evol. Syst. 49, 115–138 (2018).
Article Google Scholar
Basak, A. K. et al. ER body‐resident myrosinases and tryptophan specialized metabolism modulate root microbiota assembly. N. Phytol. 241, 329–342 (2024).
Article CAS Google Scholar
Hur, J., Yoo, M., Shin, D. & Lee, S. Inhibition of nitric oxide production corresponds to the sulforaphane content in Korean sheperd’s purse (Capsella bursa-pastoris) and related species in BV-2 cell. Food Sci. Biotechnol. 22, 1085–1089 (2013).
Article CAS Google Scholar
Farag, M. A. & Motaal, A. A. A. Sulforaphane composition, cytotoxic and antioxidant activity of crucifer vegetables. J. Adv. Res. 1, 65–70 (2010).
Article Google Scholar
Wittstock, U. & Burow, M. Glucosinolate breakdown in Arabidopsis: mechanism, regulation and biological significance. Arab. Book 8, e0134 (2010).
Article Google Scholar
Barth, C. & Jander, G. Arabidopsis myrosinases TGG1 and TGG2 have redundant function in glucosinolate breakdown and insect defense. Plant J. 46, 549–562 (2006).
Article CAS PubMed Google Scholar
Shroff, R. et al. Quantification of plant surface metabolites by matrix‐assisted laser desorption–ionization mass spectrometry imaging: glucosinolates on A rabidopsis thaliana leaves. Plant J. 81, 961–972 (2015).
Article CAS PubMed Google Scholar
Jeschke, V., Weber, K., Moore, S. S. & Burow, M. Coordination of glucosinolate biosynthesis and turnover under different nutrient conditions. Front. Plant Sci. 10, 1560 (2019).
Article PubMed PubMed Central Google Scholar
Humphrey, P. T. & Whiteman, N. K. Insect herbivory reshapes a native leaf microbiome. Nat. Ecol. Evol. 4, 221–229 (2020).
Article PubMed PubMed Central Google Scholar
Lieberman, T. D. et al. Parallel bacterial evolution within multiple patients identifies candidate pathogenicity genes. Nat. Genet. 43, 1275–1280 (2011).
Article CAS PubMed PubMed Central Google Scholar
Rouveix, B. Clinical implications of multiple drug resistance efflux pumps of pathogenic bacteria. J. Antimicrob. Chemother. 59, 1208–1209 (2007).
Article CAS PubMed Google Scholar
Helmann, T. C., Ongsarte, C. L., Lam, J., Deutschbauer, A. M. & Lindow, S. E. Genome-wide transposon screen of a Pseudomonas syringae mexB mutant reveals the substrates of efflux transporters. mBio 10, e02614–e02619 (2019).
Article CAS PubMed PubMed Central Google Scholar
Gohl, D. M. et al. Systematic improvement of amplicon marker gene methods for increased accuracy in microbiome studies. Nat. Biotechnol. 34, 942–949 (2016).
Article CAS PubMed Google Scholar
Callahan, B. J. et al. DADA2: high-resolution sample inference from Illumina amplicon data. Nat. Methods 13, 581–583 (2016).
Article CAS PubMed PubMed Central Google Scholar
McMurdie, P. J. & Holmes, S. phyloseq: An R package for reproducible interactive analysis and graphics of microbiome census data. PLoS ONE 8, e61217 (2013).
Article ADS CAS PubMed PubMed Central Google Scholar
Janssen, S. et al. Phylogenetic placement of exact amplicon sequences improves associations with clinical information. mSystems 3, e00021-18 (2018).
Article PubMed PubMed Central Google Scholar
Bolyen, E. et al. Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2. Nat. Biotechnol. 37, 852–857 (2019).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

This work was supported by NSF grant IOS-1917270 to J.L.D. and C.D.J. J.L.D. is an Investigator of the Howard Hughes Medical Institute, supported by the HHMI. D.R. gratefully acknowledges funding by EMBO Long Term Fellowship (ALTF 743–2019). C.R.F. gratefully acknowledges funding by a Natural Sciences and Engineering Research Council of Canada postdoctoral fellowship (532852-2019). We thank Rex Malstrom and Adam Deutschbauer, DOE-JGI and DOE-LBNL, for providing bacterial strains used in this work. We thank Dr. Detlef Weigel, Dr. Jonathan M. Conway, Dr. Trent R. Northen, and Dr. Katherine Louie for technical discussions. We thank Dr. Omri Finkel, Dr. Karnelia Paul, Dr. Adam Rosenthal, and Dr. Sarah R. Grant for critical comments and reading of the manuscript and the Dangl lab microbiome team for comments throughout the course of this project. This article is subject to HHMI’s Open Access to Publications policy. HHMI lab heads have previously granted a nonexclusive CC BY 4.0 license to the public and a sublicensable license to HHMI in their research articles. Pursuant to those licenses, the author-accepted manuscript of this article can be made freely available under a CC BY4.0 license immediately upon publication.

Author information

Authors and Affiliations

Department of Biology, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA
Dor Russ, Connor R. Fitzpatrick, Chinmay Saha, Theresa F. Law, Corbin D. Jones & Jeffery L. Dangl
Howard Hughes Medical Institute, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA
Dor Russ, Connor R. Fitzpatrick, Chinmay Saha, Theresa F. Law & Jeffery L. Dangl
Integrative Program for Biological and Genome Sciences, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA
Corbin D. Jones
Department of Genetics, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA
Corbin D. Jones
Department of Plant Sciences, University of California Davis, Davis, CA, USA
Daniel J. Kliebenstein

Authors

Dor Russ
View author publications
Search author on:PubMed Google Scholar
Connor R. Fitzpatrick
View author publications
Search author on:PubMed Google Scholar
Chinmay Saha
View author publications
Search author on:PubMed Google Scholar
Theresa F. Law
View author publications
Search author on:PubMed Google Scholar
Corbin D. Jones
View author publications
Search author on:PubMed Google Scholar
Daniel J. Kliebenstein
View author publications
Search author on:PubMed Google Scholar
Jeffery L. Dangl
View author publications
Search author on:PubMed Google Scholar

Contributions

D.R., C.R.F., and J.L.D. conceived and designed the experiments. D.R., C.R.F., C.S., and T.L. performed the experiments. D.R., C.R.F., D.J.K., and C.D.J. analyzed the data. D.R. and J.L.D. wrote the manuscript with feedback from all authors.

Corresponding author

Correspondence to Jeffery L. Dangl.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Thibault Meyer and the other anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Dataset 1

Dataset 2

Dataset 3

Dataset 4

Dataset 5

Dataset 6

Dataset 7

Dataset 8

Dataset 9

Dataset 10

Reporting summary

Transparent Peer Review file

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Russ, D., Fitzpatrick, C.R., Saha, C. et al. An effluent pump family distributed across plant commensal bacteria conditions host- and organ-specific glucosinolate detoxification. Nat Commun 16, 5699 (2025). https://doi.org/10.1038/s41467-025-61266-3

Download citation

Received: 09 April 2025
Accepted: 18 June 2025
Published: 01 July 2025
DOI: https://doi.org/10.1038/s41467-025-61266-3

Subjects

Abstract

Similar content being viewed by others

Introduction

Results

A. thaliana and B. distachyon assemble distinct microbial communities

Dozens of bacterial genes affect association with plants

Most MF376 colonization genes are host- and organ-specific

Two efflux pump systems promote association with either A. thaliana Col-0 roots or shoots

Homologs of the A. thaliana Col-0-associated, shoot-specific ef90 efflux pump are prevalent among Arabidopsis-associated bacteria

The plant-derived antimicrobial detoxified by saxF/ef90 is rare among angiosperms and even across A. thaliana accessions

The saxF/ef90 efflux pump protects commensal bacteria from shoot-derived aliphatic isothiocyanates

Discussion

Methods

Plant material

Microbiome analysis in wild soil

Soil sample and plant growth

Harvest and DNA extraction

Sequencing library prep

Community assembly analysis

Gnotobiotic system, bacterial inoculum, and plants’ growth

Approach

Pot preparation

Seed preparation

Bacteria preparation

Seed sowing and bacteria inoculation

Plant growth

Harvest, preparation of sequencing libraries, and sequencing

Samples harvest

Sequencing library preparation

Sequencing analysis and identification of differentially abundant mutants

Basic analysis

Identification of differentially abundant mutants

Engineer antibiotic-resistant derivatives of P. bryophila MF376

Approach

Cloning plasmids pMo130_par_ TetR and pMo130_par_ ChlR

Selecting antibiotic resistance derivatives of P. bryophila MF376

Gene deletion in P. bryophila MF376

Approach

Cloning plasmid pMo130_d90

Deleting operon ef90 from MF376

Competition assay between WT and d90 mutant derivative of P. bryophila MF376

Macerated leaf extract preparation

Bacterial preparation

Competition assay, plating, and imaging

Comparing the growth of WT P. simiae WCS17 and an ef90 homolog mutant in macerated leaf extract

Quantifying ef90 expression in A.t. leaves and in macerated leaf extract

Infiltration and harvest

Inoculation in macerated leaf extract and harvest

RNA extraction, RT-qPCR, and analysis

Building a gene tree of all efflux pumps in a 175-member isolate-collection

Measure the toxicity of selected isothiocyanates

Reporting summary

Data availability

Code availability

Change history

25 July 2025

References

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding author

Ethics declarations

Competing interests

Peer review

Peer review information

Additional information

Supplementary information

Rights and permissions

About this article

Cite this article

Share this article

Search

Quick links