Cross-kingdom synthetic microbiota supports tomato suppression of *Fusarium* wilt disease

Tomato Fusarium wilt disease (FWD), caused by Fusarium oxysporum f. sp. lycopersici (FOL), is a major constraint on greenhouse tomato production and a growing global concern due to monoculture and agrochemical overuse. Chemical control has drawbacks including health and environmental risks and disruption of beneficial microbiota. Rhizosphere microbial diversity and composition are linked to plant disease outcomes, with healthy plants often harboring richer, more interconnected communities that can suppress pathogens. The study asks whether tomato-associated cross-kingdom (bacterial and fungal) synthetic communities can suppress FWD more effectively than single-kingdom communities, and investigates how microbiome diversity, keystone taxa, and community interactions relate to disease suppression and plant immunity. The authors hypothesize that fungi, together with bacteria, play critical roles in maintaining plant health and resistance, and that SynComs enriched from natural field (NF) microbiota can confer robust FWD suppression.

Prior work shows plant microbiomes contribute to nutrient uptake, stress tolerance, and pathogen suppression via antimicrobial production, resource competition (e.g., iron), biofilm formation, and activation of plant immune defenses. Healthy tomato rhizospheres exhibit higher diversity and more modular and connected co-occurrence networks than diseased plants, with enrichment of beneficial taxa such as Bacillus, Pseudomonas, Streptomyces, Trichoderma, and Penicillium. Synthetic communities (SynComs) enable causal tests of microbiome functions, yet most studies focus on bacteria, underrepresenting fungi despite their critical symbioses with plants and interactions with bacteria. Culturomics using diverse media and high-throughput identification can bridge sequencing surveys with functional validation, challenging the notion that only ~1% of microbes are culturable. Keystone taxa that co-occur broadly can structure networks and disease outcomes. Machine learning biomarkers and meta-analyses have linked specific taxa to FWD health status, but functional verification with cultured consortia, especially cross-kingdom, has been limited.

• Sampling: Tomato plants from greenhouse-grown (GH) and natural field-grown (NF) environments were collected at four sites in two distant Chinese provinces (Shandong: Luojiazhuang, Changyi; Heilongjiang: Zhaoyuan, Lindian). 10–16 plants per site were sampled. Soil physicochemical properties (pH, TC, TN, TP, AP, K, Fe) were measured.
• Rhizosphere microbiome profiling: Rhizosphere components were recovered from roots (PBS + Tween wash, filtration, centrifugation). Total DNA was extracted. Amplicon sequencing targeted bacterial 16S rRNA V3–V4 (338F/806R) and fungal ITS1 (ITS1F/ITS2R) regions using Illumina MiSeq (2×250 bp). Reads were denoised to zOTUs (UNOISE3). Community analyses included alpha/beta diversity, ANOSIM/PERMANOVA, beta dispersion, RDA/Mantel tests for environmental associations, and co-occurrence networks (Spearman |r|>0.7, P<0.001). NetShift identified potential driver/keystone taxa; random-forest classified NF vs GH biomarkers; EdgeR identified NF-enriched zOTUs.
• Culturomics: From NF rhizospheres, 4992 bacterial isolates on five media (e.g., 1/10 TSA, TWYE, TYG, LB, beef extract peptone) and 1011 fungal isolates on four media (1/10 PDA, 1/4 RBM, CMA, MEA) were obtained. Bacteria were high-throughput barcoded by 16S V4 (505F/806R) on NovaSeq; fungi identified by ITS Sanger sequencing (ITS1/ITS4). After de-replication, 209 unique bacterial and 197 unique fungal strains were retained, covering 53.7% and 56.3% of rhizosphere zOTUs at genus level (and 72.9% and 60% at family level) with relative abundance >0.1%.
• Antagonism assays: 209 bacteria and 197 fungi were screened against FOL in vitro (triplicate). Inhibition assessed by reduced FOL colony radius and inhibition zone on TSA/PDA plates vs controls; 53 bacteria and 47 fungi showed strong inhibition and were taxonomically assigned.
• SynCom design and gnotobiotic assays: Based on NF-enriched taxa, keystones, biomarkers, and antagonism data, 105 bacteria (93 species) and 100 fungi (74 species) were selected to assemble SynComs. Eight treatments: CK (axenic), Bac (bacteria only), Fun (fungi only), CrossK (bacteria+fungi, biomass ratio 4:1), each with/without FOL (CKFOL, BacFOL, FunFOL, CrossKFOL). Germ-free tomato seedlings (cv. Zhongza9) were grown in sterilized artificial soil in sterile chambers. SynCom inocula were standardized (bacteria ~10^7 cells g−1 soil; fungi ~10^6 conidia mL−1). FOL soils contained 10^6 conidia g−1.
• Longitudinal tracking: Rhizosphere samples were collected at multiple time points (1–42 days post inoculation). Amplicons were mapped to custom reference databases (full-length 16S/ITS of SynCom members) using closed-reference OTU assignment (usearch_global, 97%). Dynamics, alpha/beta diversity, and pairwise correlations (Pearson) across time were evaluated.
• Disease phenotyping and pathogen quantification: Disease index scored over 6 weeks; FOL quantified by qPCR targeting SIX genes. Plant growth (height, fresh weight) recorded.
• Host response assays: RT-qPCR of SA- and JA-pathway marker genes (PR1a and LOX/PDF1) across time points; normalization to β-actin. RNA-seq profiled differentially expressed genes (DEGs) under Bac, Fun, CrossK vs CK; GO enrichment analyzed.
• Metagenomics: Shotgun metagenomics (NovaSeq 150 bp PE) assessed functional genes at days 1 and 14; KEGG Orthology (KO), CAZy (carbohydrate-active enzymes), and Resfams (antibiotic resistance) differential abundance analyses performed.
• Statistics: ANOSIM/PERMANOVA for beta diversity; Kruskal–Wallis with Dunn’s post hoc for disease/FOL levels; one-way ANOVA with Tukey HSD for growth; two-way ANOVA for RT-qPCR; Mantel/RDA for environmental correlations.

• Disease incidence and diversity: Greenhouse-grown (GH) tomatoes showed significantly higher FWD incidence than natural field-grown (NF) plants across provinces. Microbial alpha diversity (bacteria and fungi) was significantly higher in NF than GH (Kruskal–Wallis P<0.01). FOL levels were significantly higher in GH rhizospheres (Wilcoxon P<0.001), indicating a negative correlation between microbial diversity and FOL abundance.
• Community composition and networks: Rhizosphere communities clustered by environment and geography (ANOSIM: bacteria R=0.6317; fungi R=0.8876; P=0.001). NF networks (bacterial and fungal) had significantly higher degree and closeness centrality (Mann–Whitney P<0.001), more nodes/edges, higher modularity, and longer average path length than GH networks, indicating greater complexity and stability.
• Keystone taxa and biomarkers: NetShift identified NF driver taxa including Acremonium, Bacillus, Pseudomonas, Paenarthrobacter, Penicillium, Gemmatimonas, Rhizobium, and Sporocytophaga. Random-forest classification defined 16 bacterial and 18 fungal biomarker genera distinguishing NF vs GH with lowest cross-validation error. EdgeR identified 151 bacterial and 133 fungal zOTUs enriched in NF rhizospheres (FDR<0.05).
• Culturomics and antagonism: Culture collection covered 53.7% (bacteria) and 56.3% (fungi) of rhizosphere taxa at genus level (>0.1% relative abundance). 53 bacterial and 47 fungal isolates strongly inhibited FOL in vitro (~25% of unique isolates for each kingdom), spanning Bacillus, Enterobacter, Pseudomonas, Serratia, Acremonium, Aspergillus, Cladosporium, Penicillium, Trichoderma, Mortierella, etc.
• SynCom dynamics and efficacy: Bacterial communities in SynComs were initially variable and stabilized by ~21 days; fungal communities were more stable throughout. Cross-kingdom SynComs (CrossKFOL) yielded the lowest disease index over 42 days and the lowest FOL levels compared to FunFOL, BacFOL, and CKFOL (Kruskal–Wallis P<0.05). Taxonomic end-states (day 42) showed Bacillaceae, Pseudomonadaceae, Enterobacteriaceae, Rhizobiaceae, etc., dominant in bacterial SynComs, and Trichocomaceae, Hypocreaceae, Mortierellaceae, etc., in fungal SynComs.
• Host immunity: All SynComs upregulated SA- (PR1a) and JA- (LOX) pathway marker genes vs CK (two-way ANOVA P<0.05), with CrossK showing the strongest induction overall. PR1a induction peaked early (1–4 dpt) and declined; LOX induction increased later (peaking around 28 dpt). RNA-seq revealed robust transcriptional responses: CrossK significantly altered 781 transcription factor genes (log2FC>1, P<0.05) and uniquely upregulated 305 genes; Bac and Fun uniquely upregulated 175 and 253 genes, respectively. GO enrichments: CrossK—ABA responses and regulation of phosphatases; Bac—JA signaling and multiple abiotic stress responses; Fun—phosphorus/carbohydrate metabolism and phosphorylation pathways.
• Functional metagenomics: At day 14, KO functions differed from day 1 with 159 pathways enriched (FDR<0.05). CrossK SynComs showed higher abundance of CAZy functions (e.g., chitinase, endoglucanase, pectin methylesterase, lipases), potentially inhibiting fungal pathogens. Bacterial-associated chitin/alginate lyases were higher in CrossK and Bac vs Fun. Distinct antibiotic resistance efflux-related functions were enriched in each SynCom type, reflecting differing ecological strategies.
• Environmental drivers: Soil TC, TP, AP, Fe, and pH significantly associated with bacterial composition; TC, TN, TP, and available K with fungal composition, explaining ~13.9% (bacteria) and ~15.5% (fungi) of variance.

The study demonstrates that richer, more interconnected rhizosphere microbiomes in field conditions correlate with lower FWD incidence and lower FOL loads, supporting the concept that disease suppression is an emergent property of diverse microbial consortia. Network analyses and identified keystone taxa suggest NF communities harbor structural properties conducive to resilience and pathogen suppression. By integrating culturomics with multi-omics, the authors functionally verified that SynComs derived from NF-enriched taxa suppress FWD, with the cross-kingdom (bacteria+fungi) SynComs outperforming single-kingdom consortia. Mechanistically, cross-kingdom consortia activated both SA- and JA-mediated defenses with distinct temporal dynamics, enriched ABA-related signaling, and provided a broader arsenal of cell-wall–degrading enzymes (CAZy), which can directly impair fungal pathogens. Bacterial SynComs preferentially activated JA/stress pathways and encoded chitin/alginate-degrading functions, while fungal SynComs enriched phosphorus and carbohydrate metabolism pathways that may bolster host nutrition and defense readiness. The observed early instability of bacterial communities versus the relative stability of fungal communities suggests differential colonization dynamics and trophic interactions, with fungi potentially acting as early stabilizers and resource modulators. Overall, the findings substantiate that cross-kingdom interactions are central to microbiome-mediated disease suppression and plant immune modulation, advancing beyond correlative omics to causal, culture-based validation.

Natural field (NF) rhizosphere microbiota, characterized by higher diversity and complex interconnections, are associated with reduced Fusarium wilt disease (FWD) in tomato. Reconstituted synthetic communities verified this functional capacity: cross-kingdom SynComs combining bacteria and fungi provided the strongest suppression of FOL and disease, while also eliciting robust, temporally coordinated SA/JA immune responses and enriching functional pathways (e.g., CAZy enzymes) relevant to pathogen control. The work expands culturable microbiome coverage and establishes an inter-kingdom SynCom framework for sustainable disease management in tomato, offering a blueprint for designing microbiome-based biocontrols. Future research should incorporate rare, low-abundance taxa, refine SynCom composition and dosing, and validate performance in greenhouse and field agricultural settings.

• Low-abundance microorganisms, which may play key functional roles in recruitment and pathogen suppression, were underrepresented or absent in the SynComs. Advanced cultivation (e.g., microfluidics, diffusion chambers, cell sorting) could improve recovery of rare/uncultured taxa.
• SynCom suppression efficacy was demonstrated under controlled laboratory/gnotobiotic conditions; validation in greenhouse and real-world agricultural production systems is needed to confirm robustness and scalability.