2.1 Composition and diversity of root exudates
Using untargeted metabolomics, 1325 distinct compounds were identified in root exudates. Then, we classified rhizosphere metabolites into 10 groups based on the HMDB database annotation at the Super Class level(Wishart et al. 2022) (Supplementary Data Fig. S1). Depending on the number and relative abundance of compounds, lipids and organic acids were predominant in all treatments (Fig. 1A, Supplementary Data Fig. S1). For both pepper and tomato, the relative abundance of lipids and lipid-like compounds in diseased rhizosphere soil exceeded that in healthy soil (Fig. 1A).
Principal coordinate analysis showed that the structure of root exudate was significant differences between healthy and diseased plants, as well as between different plant species. PERMANOVA analysis confirmed that species (F species = 19.15, P < 0.001) had more effect on the composition of root exudates than the pathogen (F pathogen = 13.27, P < 0.001, Fig. 1B, Supplementary Data Table S1). Additionally, the diversity of diseased plant root exudates was significantly lower compared to that of healthy plants, both in pepper and tomato (Fig. 1C and D). Differential compound analysis showed that PHR and PDR were significantly enriched with 103 and 219 compounds, respectively (P < 0.05, Fig. 1E), and THR and TDR were significantly enriched with 115 and 244 compounds, respectively (P < 0.05, Fig. 1F). In the healthy rhizosphere soil of various host plants, only 10 identical compounds were enriched, but 92 identical compounds were enriched in the diseased rhizosphere soil of various host plants, of which 28 were lipids and lipid-like molecules (Fig. 1G). Although species and pathogens can cause variations in overall exudation patterns, the results revealed that two Solanaceae plants infected with the same disease were enriched with a higher number of the same compounds and reduced the diversity of exudates.
2.2 Soil microbial community composition in rhizosphere
In total, 7,593,358 16S rRNA reads were provided from 64 samples, which were classified as 33,830 bacterial ASVs. Proteobacteria, Actinobacteria, Chlorofexi, Bacteroidetes, Acidobacteria, Firmicutes, Myxococcota, Planctomycetes, and Verrucomicrobia were largely dominant among the bacteria (Fig. 2A, Supplementary Data Table. S2). Our results also implied that the variance of pepper and tomato at the phylum level corresponded between treatments and that there was a much higher relative abundance of Proteobacteria in the soil of diseased plants than in healthy plants.
Then we conducted a two-tailed test for genera with relative abundance greater than 1% to reveal the significantly enriched bacterial genera in the healthy rhizosphere and diseased rhizosphere of pepper and tomato, respectively. The results showed that Ralstonia, a typical pathogen of bacterial wilt, was significantly enriched in TDR and PDR, with a relative abundance of 22.73% and 31.98%, respectively (Fig. 2B and C). However, THR and PHR were enriched with different bacterial genera. Bacillus, Rhodanobacter, Pseudolabrys, Gemmatimonas, Sphingomonas, and JG30−KF−AS9 were significantly enriched in THR, and Chitinophaga, Streptomyces, Devosia, and Bacillus were significantly enriched in PHR (Fig. 2B and C). Meanwhile, our correlation analysis uncovered a significant negative correlation between the relative abundance of Ralstonia and bacterial genera, which were notably enriched in the PHR and THR, respectively (Supplementary Data Fig. S2). These results suggested that different Solanaceae plants will accumulate different microbial genera in the rhizosphere to prevent plant diseases when facing the invasion of the same diseases.
2.3 Soil microbial community structure and diversity in rhizosphere
Principal coordinate analysis showed significant differences in bacterial community structure between the healthy and diseased plants in rhizosphere, but not in bulk soil. PERMANOVA analysis confirmed that compartments had the greatest impact on bacterial communities (F compartment = 28.83, P < 0.001), followed by plant species (F plant species = 18.51, P < 0.001), and the pathogen (F pathogen = 3.90, P = 0.002, Fig. 3A, Supplementary Data Table S3). Additionally, as shown by beta dispersion, the bacterial communities in the diseased plants were more variable in comparison to the healthy plants (Fig. 3B). Further correlation analysis displayed a significant positive association between the variance of the bacterial community on the first axis and the relative abundance of Ralstonia (Fig. 3C). That suggested that higher pathogen abundance may drive the variation of community structure.
Bacterial alpha diversity in peppers and tomatoes, including Shannon and Simpson, displayed similar patterns of change. Healthy plants showed higher bacterial diversity in rhizosphere soil than diseased plants, while all plant rhizospheres had significantly higher bacterial diversity than bulk soil (Fig. 3D, Supplementary Data Fig. S3A). In addition, the correlation analysis found that bacterial alpha diversity was negatively correlated with the relative abundance of the Ralstonia (Fig. 3E, Supplementary Data Fig. S3B). These results implied that the community structure and diversity of different Solanaceae plants had the same pattern of change after infection with the same disease. The enrichment of pathogens in the rhizosphere changed the community structure and reduced the alpha diversity.
2.4 Soil microbial functioning potential in rhizosphere
Further, PICRUSt2 was implemented to predict the functioning potential of soil bacterial communities in peppers and tomatoes. As shown by principal coordinate analysis, the functional potential of healthy and diseased rhizosphere soils was significantly different, but not between species, especially between THR and PHR (Fig. 3F, Supplementary Data Table. S4). This suggested that when different solanaceous plants encountered the same pathogen invasion, although they had enriched different microbial genera to resist diseases, their functional potential was similar.
Additional investigation on the KEGG pathway enrichment of KO functional categories showed the existence of 15 and 12 significantly enriched pathways in HR and DR correspondingly (P < 0.001, Fig. 4). Pathways associated with genetic information replication, repair, and Protein export, such as “DNA replication”, “Homologous recombination”, “Mismatch repair”, and “Protein export” all showed significantly higher relative abundance in the HR. “Pentose phosphate pathway”, “Peptidoglycan biosynthesis”, “Photosynthesis”, “Phenazine biosynthesis”, and “Lysine biosynthesis”, those biosynthetic pathways were significantly enriched in the HR. The metabolism of cysteine, methionine, purine, and pyrimidine was also enriched in the HR. Conversely, the degradation pathways of aminobenzoate, benzoate, caprolactam, dioxin, polycyclic aromatic hydrocarbon, xylene, and other toxic substances were significantly enriched in DR. The relative abundance of metabolic pathways, including “Drug metabolism - cytochrome P450”, “Phenylalanine metabolism”, “Propanoate metabolism”, “Retinol metabolism”, “Synthesis and degradation of ketone bodies”, were notably higher in the DR (Fig. 4).