Tapping the rhizosphere metabolites for the prebiotic control of soil-borne bacterial wilt disease

Soil-borne plant diseases cause major yield losses and conventional chemical controls are often ineffective and environmentally costly. Biocontrol strategies using microbial inoculants can lack stability due to poor adaptation and competition with native microbiomes. Manipulating resident rhizosphere microbiomes in situ via targeted metabolites (prebiotics) offers a promising alternative. The rhizosphere is a key interface where metabolites can recruit beneficial microbes to antagonize pathogens or induce plant defenses. Rhizosphere metabolite profiles change upon pathogen attack and can be linked to disease suppressiveness. However, rational selection of plant prebiotic metabolites to shape the rhizosphere microbiome for disease control is lacking. This study hypothesizes that metabolites enriched in healthy plant rhizospheres under pathogen pressure can serve as prebiotics to stimulate commensal microbes (not positively correlated with R. solanacearum), thereby suppressing bacterial wilt. The work aims to: (i) identify differential rhizosphere metabolites between healthy and diseased tomato plants, (ii) test their effects on disease outcomes and microbiomes, (iii) assess utilization by pathogen versus commensals in vitro and in situ (SynCom), and (iv) evaluate cross-crop efficacy and metagenomic functional shifts.

Prior studies have established prebiotics in gut systems and suggested analogous strategies in plant systems where plant growth-promoting rhizobacteria (PGPR) act like probiotics and organic substrates act as prebiotics. Organic amendments (e.g., composts) and specific root exudate components can modulate soil microbiomes. Specific metabolites such as malic acid recruit Bacillus subtilis in Arabidopsis, and fatty acids/amino acids can enrich beneficial Pseudomonas in soils. Pathogen challenge can alter rhizosphere metabolites and recruit beneficial microbes, contributing to disease-suppressive soils. Single-compound amendments can decrease microbial diversity or disturb microbiomes, while increasing the variety of applied compounds can alleviate diversity declines. Small carbohydrates (e.g., xylose, inositol) have been linked to stress tolerance and biocontrol signaling in multiple plant systems, and inositol has been implicated in R. solanacearum inhibition.

• Field sampling and metabolomics: Tomato plants (cv. HeZuo 903) were grown in a bacterial wilt-conducive field with historical high disease incidence. At fruiting stage, rhizosphere samples from 12 healthy and 12 diseased plants (pooled into four biological replicates per condition) were collected. Non-targeted GC-TOF-MS profiling identified 216 metabolites. Generalized linear models (GLM, mvabund) and random forest classification (tenfold cross-validation) identified metabolites differing between health states.
• Metabolite selection: Nineteen metabolites were selected based on differential abundance and availability: 11 enriched in healthy rhizospheres (sucrose, fructose, mannose, xylose, ribose, inositol, lactic acid, gluconolactone, ribitol, melibiose, maltose) and eight enriched in diseased rhizospheres (benzoic acid, malonic acid, 4-hydroxybenzoic acid, phytanic acid, 4-hydroxyphenyl ethanol, glycerol, glutamine, monostearin).
• Greenhouse prebiotic cocktail assay: A cocktail of the 11 healthy metabolites (equimolar, applied weekly for 4 weeks at 1 µmol g−1 soil per dose; 5 mL of 10 mM per metabolite mix per application) was tested in tomato under six treatments: PRS (prebiotics+pathogen), PW (prebiotics+water), WRS (water+pathogen), WW (water only), NPRS (non-prebiotics+pathogen), NPW (non-prebiotics+water). Pathogen inoculum: R. solanacearum (5×10^8 CFU per 50 g soil per plant). Disease incidence was monitored up to 16 days post inoculation.
• Microbiome profiling: 16S rRNA gene amplicon sequencing (V4, 515F/806R) and qPCR quantified total bacteria (16S rRNA gene) and pathogen (fliC gene). Community diversity (Shannon), structure (Bray-Curtis NMDS, PERMANOVA), phylum-level composition, and co-occurrence networks (ggClusterNet) were analyzed. Genera were categorized by correlation with R. solanacearum.
• In vitro carbon utilization: R. solanacearum growth was assessed on each metabolite (10 mM) in inorganic salts medium. A culture collection of 158 bacterial isolates from tomato rhizosphere was established and tested for growth on the same metabolite panel to compare utilization by commensals vs. pathogen.
• In situ SynCom co-culture: Sterilized soil microcosms with sterile tomato seedlings were inoculated with a synthetic community (SynCom) of the 158 isolates and R. solanacearum. Prebiotic (seven-metabolite subset excluding those efficiently utilized by pathogen: ribose, lactic acid, xylose, mannose, maltose, gluconolactone, ribitol) mixtures were applied once at three concentrations (1 µM, 100 µM, 10 mM); a non-prebiotic mixture (10 mM) and water were controls. Pathogen and SynCom populations were enumerated on SMSA and TSA, respectively, on days 1, 3, 5.
• Sterile defense assay: Under sterile soil conditions without resident microbiota, plants were irrigated with prebiotics and challenged with pathogen to test direct induction of plant defenses.
• Cross-crop greenhouse assay: Tomato (Micro-Tom), pepper (Yinchuan Cavel), and eggplant (Chunqiumoqie) were challenged with crop-specific R. solanacearum formae speciales. Treatments: WRS, PRS (10 mM prebiotics), NPRS (10 mM non-prebiotics), applied weekly for 4 weeks. Disease incidence recorded over 30 days.
• Shotgun metagenomics: Rhizosphere soils from the three crops (prebiotic vs. non-prebiotic treatments) were sequenced (Illumina NovaSeq 6000; >30 GB per sample). Assembly (MEGAHIT), ORF prediction (MetaGeneMark), gene catalog construction, and taxonomic/functional annotation (NCBI nr via DIAMOND; KEGG). Diversity, PCoA/MRPP, and GSVA pathway enrichment analyses were performed. Random forest linked functional pathways to commensal vs. pathogen-associated taxa.
• Field trial: In a continuously cropped tomato field (>50% historical incidence), prebiotic (10 mM seven-metabolite mix) vs. non-prebiotic (10 mM) were applied biweekly (200 mL per plant) for 2 months. Disease incidence and qPCR quantification of pathogen and total bacteria were measured at fruiting.
• Statistics: Normality (Shapiro-Wilk), homogeneity (Levene’s), PERMANOVA (Adonis, Bray-Curtis, permutation=999), diversity metrics (vegan), differential analyses (Wilcoxon/Dunn’s), regression analyses (R stats), LEfSe, and network metrics (ggClusterNet).

• Rhizosphere metabolome differences: PCA/Adonis showed distinct metabolite profiles between healthy and diseased tomato rhizospheres (R=0.785, p=0.026); PCA axes explained 44.68% and 16.34% variance. Sugars were more abundant in healthy rhizospheres (34.8%) vs. diseased (4.75%), while long-chain organic acids were higher in diseased (51.9%) vs. healthy (21.9%).
• Differential metabolites: 79 metabolites differed (GLM, p<0.05): 25 enriched in healthy (e.g., sucrose, fructose, mannose, xylose, ribose, inositol, lactic acid, gluconolactone, ribitol, melibiose, maltose) and 54 enriched in diseased (selected eight: benzoic acid, malonic acid, 4-hydroxybenzoic acid, phytanic acid, 4-hydroxyphenyl ethanol, glycerol, glutamine, monostearin). Random forest prioritized top markers; 19 were selected for experiments.
• Disease suppression in tomato greenhouse: Prebiotic cocktail reduced bacterial wilt incidence to 32.0% (PRS) vs. 65.6% in WRS. Non-prebiotics increased incidence by 22.4% over WRS (≈88.0%). Pathogen fliC copies were 8.5% lower in PRS vs. WRS; NPRS increased pathogen by 8.3%. Total bacterial abundance increased in PRS and PW. Pathogen load was negatively correlated with total bacterial abundance in prebiotic-treated samples (R=-0.91; p<0.05; R^2=0.8).
• Microbiome shifts: Prebiotics increased species richness and Shannon diversity (Dunn’s p<0.05) with and without pathogen. NMDS and PERMANOVA indicated significant compositional differences (stress=0.13; p=0.001; R=0.687). Prebiotics explained ~28% and pathogen ~7% of variance (~41% combined). Actinobacteria increased and Proteobacteria decreased under prebiotics. Network stability, reduced by pathogen and non-prebiotics, was restored by prebiotics. Of 491 genera, 474 were not positively correlated with R. solanacearum; their absolute abundances increased with prebiotics.
• Utilization assays: R. solanacearum efficiently utilized sucrose, fructose, melibiose, and inositol but not seven other healthy-enriched metabolites (ribose, lactic acid, xylose, mannose, maltose, gluconolactone, ribitol). Rhizosphere isolates (n=158) collectively utilized prebiotics more efficiently than non-prebiotics.
• SynCom microcosms: A seven-metabolite prebiotic mix at 10 mM significantly reduced pathogen CFUs and increased total bacterial CFUs; negative correlations between SynCom and pathogen were observed under prebiotic treatments across concentrations (R=-0.58; p<0.05; R^2=0.33), but not in water or non-prebiotic controls.
• Cross-crop efficacy: In greenhouse tests with tomato, eggplant, and pepper, 1 µmol g−1 soil prebiotic mix reduced disease incidence across all crops over 30 days. For pepper, PRS had 2.7% incidence vs. 13.3% (WRS) and 22.8% (NPRS). Prebiotics decreased pathogen abundance and increased total bacterial abundance across crops.
• Field validation: In a high-incidence tomato field, prebiotics reduced disease incidence to 25.0% vs. 77.2% with non-prebiotics and 51% in control; pathogen abundance decreased with prebiotics (qPCR).
• Metagenomic functions: Prebiotics increased gene richness (significant in tomato and eggplant; Dunn’s p<0.05) and shifted functional profiles (MRPP p=0.001). GSVA showed enrichment in carbon metabolism pathways (galactose metabolism, starch/sucrose metabolism, glycosaminoglycan degradation) and degradation of autotoxins (toluene, xylene, fluorobenzoate). Random forest indicated these functions were primarily driven by commensal microbes in prebiotic treatments; only few functions were driven by microbes positively correlated with R. solanacearum in non-prebiotic treatments.

The study demonstrates a strategy to mine rhizosphere metabolites enriched in healthy plants under pathogen pressure and deploy them as prebiotics to engineer the rhizosphere microbiome for disease suppression. Prebiotic application did not directly induce plant defenses under sterile conditions, supporting a mechanism mediated by stimulation of commensal microbes that expand overall bacterial abundance and diversity, occupy niches, and limit pathogen proliferation (targeted starvation/competition). Microbiome analyses showed that prebiotics increased diversity and altered composition (e.g., increased Actinobacteria), restored network stability, and selectively increased taxa not positively associated with the pathogen. Metagenomics linked prebiotics to enhanced carbon metabolism and autotoxin degradation pathways driven by commensals, suggesting improved ecosystem functioning that disfavors pathogen persistence. Efficacy across Solanaceae crops and validation in a field with high disease pressure highlight translational potential. Compared to inoculating beneficial strains, prebiotics leverage native commensals better adapted to local environments, potentially offering greater stability and breadth of protection.

This work establishes a pipeline for identifying and applying rhizosphere-derived prebiotic metabolites to control soil-borne bacterial wilt. Seven small molecules (ribose, lactic acid, xylose, mannose, maltose, gluconolactone, ribitol) were identified that are poorly utilized by R. solanacearum but support commensal growth. Applied as soil drenches, they increased bacterial abundance/diversity, enriched beneficial functional pathways, and suppressed pathogen populations, reducing disease in tomato, pepper, and eggplant, including in a high-incidence field. The approach offers an environmentally friendly, microbiome-informed alternative to chemical control or unstable inoculants. Potential future directions include: expanding metabolite discovery across diverse crops/soils and pathogens; optimizing doses, timing, and combinations for efficacy and minimal disturbance; mechanistic dissection of key commensal taxa and metabolites; integrating prebiotics with other sustainable practices; and large-scale field trials to assess robustness, economics, and environmental impacts.

• Mechanistic dependence on resident microbiota: Under sterile conditions, prebiotics did not reduce disease, indicating efficacy relies on the presence of commensal communities; performance may vary with soil microbiome composition.
• SynCom representativeness: The 158-isolate SynCom, while phylogenetically broad, may not fully capture natural rhizosphere complexity and interactions.
• Context specificity: Experiments were conducted in specific soils and environmental conditions (greenhouse and selected field sites); generalizability across soil types, climates, and management systems requires further validation.
• Metabolite selection constraints: From many differential metabolites, 19 were advanced based on availability; other potentially effective compounds may have been excluded. Some healthy-enriched metabolites (e.g., sucrose, fructose, inositol, melibiose) were efficiently utilized by the pathogen and were excluded from the final prebiotic mix, which may limit spectrum.
• Dosage considerations: High concentrations (e.g., 10 mM in SynCom assays) were most effective; field-feasible dosages and application regimes need optimization to balance efficacy and microbiome stability.