Microbiome homeostasis on rice leaves is regulated by a precursor molecule of lignin biosynthesis

Su, Pin; Kang, Houxiang; Peng, Qianze; Wicaksono, Wisnu Adi; Berg, Gabriele; Liu, Zhuoxin; Ma, Jiejia; Zhang, Deyong; Cernava, Tomislav; Liu, Yong

doi:10.1038/s41467-023-44335-3

Download PDF

Article
Open access
Published: 02 January 2024

Microbiome homeostasis on rice leaves is regulated by a precursor molecule of lignin biosynthesis

Nature Communications volume 15, Article number: 23 (2024) Cite this article

15k Accesses
6 Citations
275 Altmetric
Metrics details

Subjects

Abstract

In terrestrial ecosystems, plant leaves provide the largest biological habitat for highly diverse microbial communities, known as the phyllosphere microbiota. However, the underlying mechanisms of host-driven assembly of these ubiquitous communities remain largely elusive. Here, we conduct a large-scale and in-depth assessment of the rice phyllosphere microbiome aimed at identifying specific host-microbe links. A genome-wide association study reveals a strong association between the plant genotype and members of four bacterial orders, Pseudomonadales, Burkholderiales, Enterobacterales and Xanthomonadales. Some of the associations are specific to a distinct host genomic locus, pathway or even gene. The compound 4-hydroxycinnamic acid (4-HCA) is identified as the main driver for enrichment of bacteria belonging to Pseudomonadales. 4-HCA can be synthesized by the host plant’s OsPAL02 from the phenylpropanoid biosynthesis pathway. A knockout mutant of OsPAL02 results in reduced Pseudomonadales abundance, dysbiosis of the phyllosphere microbiota and consequently higher susceptibility of rice plants to disease. Our study provides a direct link between a specific plant metabolite and rice phyllosphere homeostasis opening possibilities for new breeding strategies.

Differential selection of yield and quality traits has shaped genomic signatures of cowpea domestication and improvement

Article 22 April 2024

Towards establishing a fungal economics spectrum in soil saprobic fungi

Article Open access 18 April 2024

Spatial co-transcriptomics reveals discrete stages of the arbuscular mycorrhizal symbiosis

Article Open access 08 April 2024

Introduction

In nature, plants and their associated microbes, collectively known as the microbiota, form functional entities that rely on each other¹. The microbiota contributes to aspects such as disease resistance^2,3,4,5, stress tolerance^6,7, and nutrient acquisition^8,9. Successful recruitment and maintenance of a sufficient abundance of specific microbial members determines the outcome of plant-microbiota interactions^10,11. Thus, understanding the principles driving microbiome assembly in crop plants has become one of the main pursuits of present studies in order to integrate microbiome functioning into sustainable crop production.

The plant microbiota can be compartmentalized into the rhizosphere (root-soil interface), phyllosphere (leaf surface) and endosphere (internal tissues)¹². Studies revealed that host genetics¹³ and metabolic processes¹⁴ participate in the recruitment of specific microbial taxa in the rhizosphere microbiome, and consequentially shape microbiome assembly. In barley, a specific genomic locus was shown to be associated with the recruitment of specific bacterial taxa¹⁵. Deepening analysis revealed three genes in this locus, including a Nucleotide-Binding-Leucine-Rich-Repeat (NLR) receptor as primary candidates involved in shaping the microbiome¹⁵. In tomato plants, the identification of specific quantitative trait loci (QTLs) provided further evidence for the importance of host genetics in microbiome assembly¹⁶. In addition, specific plant root exudates were demonstrated to steer rhizosphere microbiome assembly toward the host’s needs for defense and nutrition¹⁷. The plant phyllosphere represents the largest biological surface on earth and widespread interface for interactions between plants and their microbiota¹⁸. Identification of specific genetic components shaping the phyllosphere microbiome is still sparsely understood, although the plant genotype has been repeatedly demonstrated to impact the phyllosphere microbiome in consistent ways across geographically separated sites^19,20. Model plants like Arabidopsis²¹ and non-model plants like perennial wild mustard²² as well as cereal crops such as wheat and barley²³ were implemented to increase our understanding of the phyllosphere microbiome.

It is known that external factors such as microbial inocula and climate conditions can alter phyllosphere microbiomes²⁴, however, the plant genotype has become the central focus in terms of seeking approaches to engineer plant microbiome for agricultural purposes²⁵. Recent studies have revealed that defense responses and cell wall integrity are likely involved in phyllosphere microbiome assembly¹⁹. Arabidopsis cuticle mutants and ethylene signaling mutants were shown to have a different phyllosphere microbiome compared to wild-type plants²⁰. The phosphate starvation response pathway²⁶ and pattern-triggered immunity²⁷ were also shown to take part in shaping the Arabidopsis phyllosphere microbiome. A common strategy to investigate host genetic effects on microbiome assembly is based on amplicon sequencing of microbial marker genes coupled with genome-wide association studies (GWAS)²⁸. Such studies have resulted in the identification of many putatively causal genes with effects on the relative abundance of plant-associated microbes^19,21,22. In the case of the Arabidopsis phyllosphere microbiome, specific microbial hubs were successfully linked to host genomic loci that are related to specialized metabolite biosynthesis²¹. Genes involved in the synthesis of sinapoyl glucose, sinapoyl malate, glucosinolates are assumed to participate in shaping the Arabidopsis phyllosphere microbiome. Nevertheless, there is still lack of direct evidence bridging the host genetic background with the recruitment of specific microbial members. So far, phenotyping phyllosphere microbiome compositions via GWAS was exclusively conducted on the basis of amplicon sequencing of microbial marker genes¹⁷. However, metagenomic data provides better means to more accurately conduct such analyses, mainly because it is less prone to over- or underestimate the abundance of certain microbial taxa.

In this work, with the employment of phyllosphere metagenomes from 110 rice accessions of the Rice Diversity Panel II core collection (C-RDP-II)²⁹, we perform GWAS experiments to link bacterial abundances with single nucleotide polymorphisms (SNPs) in the rice genome. Rice genetic variations are shown to be significantly associated with members of four predominant phyllosphere bacterial orders, Pseudomonadales, Burkholderiales, Enterobacterales, and Xanthomonadales. To unravel a prevailing mechanism of how host genetics affect phyllosphere microbiome assembly, we implement mutants and over-expression constructs of a candidate gene associated with Pseudomonadales and assess the resulting microbiome shifts. Furthermore, we analyze rice metabolites in rice leaves that are regulated by the candidate gene. We discover that the compound 4-hydroxycinnamic acid, also known as p-coumaric acid and a precursor in lignin biosynthesis, is required for the assembly and homeostasis of the rice phyllosphere microbiome. Overall, our study provides clear evidence for host metabolite-driven phyllosphere microbiome assembly.

Results

Rice phyllosphere microbiomes are specific to genotypes

A large-scale dataset was implemented to assess microbiome structures in the phyllosphere of rice plants. The dataset was obtained via shotgun-based metagenomics²⁹. It is based on 110 accessions of the Rice Diversity Panel II core collection (C-RDP-II) that were all grown in the northwest of Hunan Province, China (28°38′09″ N, 112°0′57″ E). For comparative analyses, the rice accessions were separated into three main groups: 56 indica, 36 japonica and 18 unassigned (Supplementary Data 1). Rarefaction analysis confirmed that the implemented approach sufficiently captured microbial diversity in the indica and japonica groups (Supplementary Fig. 1). Processing of the metagenomes using Kraken2 and Bracken³⁰ generated 6,862 species-level taxonomic units. Bacteria (comprising over 94% of the total taxonomic units) were the prevalent microbiome constituents in the rice phyllosphere (Supplementary Data 2).

We found that bacterial communities were specific to rice subspecies. Bacterial communities obtained from indica and japonica varieties formed two distinct clusters, as indicated by the unconstrained principal coordinate analysis (PCoA) of Bray-Curtis distances (Fig. 1a, P < 0.001, R² = 0.03, PERMANOVA with Adonis test). This suggests that the genotypes of indica and japonica varieties are connected to distinct phyllosphere microbiome compositions and structures. Furthermore, we assessed the within-sample diversity (α-diversity) and found a significant difference between indica and japonica varieties (Fig. 1b and Supplementary Fig. 2). The phyllosphere microbiome of indica was more diverse than that of japonica (Fig. 1b and Supplementary Fig. 2), indicating that the former was colonized by more bacterial species. We also observed significant differences in the phyllosphere microbiome structures between indica and japonica varieties at the phylum, order and genus levels (Supplementary Fig. 3, 4, Fig. 1c, Supplementary Table 1), providing further evidence that bacteria are influenced by the rice genotypes.

**Fig. 1: Phyllosphere microbiome comparisons of *indica* and *japonica* rice varieties.**

The predominant bacterial orders of both indica and japonica rice plants were comprised of Pseudomonadales (57% and 72% of relative abundance for indica and japonica, respectively), followed by Enterobacterales (19% and 7%), Sphingomonadales (5% and 5%), Burkholderiales (3.7% and 2.1%), Xanthomonadales (1.8% and 3.5%), Flavobacteriales (3.4% and 1.1%), Rhizobiales (1.5% and 1.3%), Caulobacterales (1.3% and 1.7%) and Bacillales (1.7% and 0.8%) (Fig. 1c). Compared to japonica, indica varieties had a higher relative abundance of Enterobacterales, Flavobacteriales and Burkholderiales, while japonica varieties had a higher relative abundance of Pseudomonadales and Xanthomonadales (Tukey’s HSD test, false discovery rate (FDR) adjusted P < 0.05) (Fig. 1c, Supplementary Table 2). Interestingly, 46 genera from Enterobacterales (accounting for 92% of the identified genera within this order), 17 genera from Burkholderiales (27.4% of the identified genera) and 5 genera from Pseudomonadales (83% of the identified genera) occurred at significantly different relative abundances in indica and japonica varieties (Supplementary Fig. 4, Supplementary Data 2). In contrast, only the genus Aerosticca from Xanthomonadales and only the genus Chryseobacterium from Flavobacteriales showed such differences in relative abundance between the varieties. Therefore, we hypothesized that a specific genetic background of the rice varieties may play a role in shaping their phyllosphere microbiomes.

Identification of metabolic pathways linked to phyllosphere microbiome

With recent advances in sequencing and bioinformatics, genome-wide association studies (GWAS) have become an increasingly promising approach for identifying genetic factors associated with important agronomic traits in plants²⁸. To identify potential rice genetic factors that are linked to the phyllosphere microbiome, we conducted GWAS. The analysis targeted associations between rice genomes and the relative abundance of bacterial species within the respective metagenomes. Our analysis led to the identification of 2,667 non-redundant single nucleotide polymorphisms (SNPs) located in 235 loci that were significantly associated with 496 bacterial species (hereafter referred to as GWAS-associated species) at a genome-wide significance threshold of P ≤ 0.00001 (Fig. 2c, Supplementary Data 3).

**Fig. 2: GWAS with rice phyllosphere microbiome.**

We found that the majority of GWAS identified species belonged to Pseudomonadales (associated with 129 loci, accounting for 30.24% of total GWAS-associated species) Burkholderiales (associated with 100 loci, accounting for 21.37%), Xanthomonadales (associated with 47 loci, accounting for 8.87%) and Enterobacterales (associated with 76 loci, accounting for 8.67%) at the bacterial order level (Fig. 2a, Supplementary Data 4). To separate loci according to bacterial orders associated with them, we searched for locus clusters using Bray-Curtis distances based on GWAS-associated species distribution of each locus. Our results showed that 91.06% of the total SNP loci could be divided into four clusters: Cluster1 (containing 38.30% of total SNP loci and associated with Pseudomonadales), Cluster2 (containing 31.91% of total SNP loci and associated with Burkholderiales), Cluster3 (containing 12.77% of total SNP loci and associated with Enterobacterales) and Cluster4 (containing 8.09% of total SNP loci and associated with Xanthomonadales) (Fig. 2a, Supplementary Fig. 5, Supplementary Data 4). These associations suggested that members of four bacterial orders are particularly responsive to certain genetic backgrounds of the host plant.

To further elucidate how host genetic variations affect the phyllosphere microbiome, especially members of the bacterial orders that were highly responsive, genes with SNPs located in the associated loci were annotated and enriched into rice pathways (Supplementary Data 4). We found that metabolic pathways (dosa01100) and secondary metabolic process (GO:0019748) were prevalent in gene-enriched KEGG (Kyoto Encyclopedia of Genes and Genomes) (Supplementary Fig. 6) and GO (Gene Ontology) (Supplementary Fig. 6) annotations, respectively. Furthermore, we observed that SNP-associated genes in each locus cluster were enriched in specific metabolic pathways (Fig. 2b). For instance, Enterobacterales-associated rice genes were significantly enriched in Proteasome (dosa03050), Lysine degradation (dosa00310), Biosynthesis of amino acids (dosa01230), Ubiquitin mediated proteolysis (dosa04120), SNARE interactions in vesicular transport (dosa04130), Cysteine and methionine metabolism (dosa00270), Phosphatidylinositol signaling system (dosa04070), and Glutathione metabolism (dosa00480) pathways (Fisher’s exact test, P < 0.05, Fig. 2b, Supplementary Data 4). Similarly, Burkholderiales-associated rice genes were significantly enriched in Pentose and glucuronate interconversions (dosa00040), Protein export (dosa03060), Phenylalanine metabolism (dosa00360) and Arginine and proline metabolism (dosa00330) pathways (Fisher’s exact test, P < 0.05, Fig. 2b, Supplementary Data 4). Pseudomonadales-associated rice genes were significantly enriched in Plant-pathogen interaction (dosa04626), Stilbenoid, diarylheptanoid and gingerol biosynthesis (dosa00945), Plant hormone signal transduction (dosa04075), Ether lipid metabolism (dosa00565), Glycerophospholipid metabolism (dosa00564), MAPK signaling pathway-plant (dosa04016) and Ubiquinone and other terpenoid-quinone biosynthesis (dosa00130) pathways (Fisher’s exact test, P < 0.05, Fig. 2b, Supplementary Data 4). Xanthomonadales-associated rice genes were significantly enriched in Zeatin biosynthesis (dosa00908) and Amino sugar and nucleotide sugar metabolism (dosa00520) (Fisher’s exact test, P < 0.05, Fig. 2b, Supplementary Data 4). Notably, there were also overlapping pathways among different bacterial orders. For example, Enterobacterales- and Pseudomonadales-associated rice genes were enriched in the Sulfur metabolism pathway (dosa00920), while Burkholderiales-, Pseudomonadales- and Xanthomonadales-associated rice genes were enriched in the Phenylpropanoid biosynthesis pathway (dosa00940) (Fig. 2b, Supplementary Data 4). These results provided key insights into host metabolic pathways potentially involved in the assembly of the phyllosphere microbiota. Since the Phenylpropanoid biosynthesis pathway was associated with three predominant bacterial orders, we expect a key role of this pathway with respect to host-microbiota interactions.

We therefore further investigated Burkholderiales-, Pseudomonadales- and Xanthomonadales-associated rice genes that were enriched in the Phenylpropanoid biosynthesis pathway (Fig. 2c, d). We found that a majority of prx (peroxidase) genes, which are involved in lignin biosynthesis, were associated with Burkholderiales, Pseudomonadales and Xanthomonadales (Fig. 2d, Supplementary Data 4). In addition, the 4CL1 (4-coumarate-CoA ligase, Os08g0245200) and C4H2 (cinnamate-4-hydroxylase, Os05g0320700) genes, which are involved in the synthesis of cinnamoyl-CoA and 4-hydroxycinnamoyl-CoA, respectively, were found to be specifically associated with Burkholderiales. On the other hand, OsPAL02 (phenylalanine/tyrosine ammonia-lyase, Os04g0518100), OsCAD4 (cinnamyl alcohol dehydrogenase, Os11g0622800) and OsCCR21 (cinnamoyl-CoA reductase, Os02g0180700) genes, which are involved in the synthesis of 4-hydroxycinnamic acid, 4-hydroxycinnamyl alcohol and 4-hydroxycinnamyl aldehyde, respectively, were specifically associated with Pseudomonadales (Fig. 2d, Supplementary Data 4). This indicated potential differences in host genetic control of the phyllosphere microbiota. Collectively, our analysis suggested that compounds that are part of lignin biosynthesis or their precursors may be participating in regulating microbiota assembly, and more importantly, that certain compounds may exert effects on specific bacterial taxa.

Identification of leaf metabolite-associated genes that drive microbiome assembly

Given the common occurrence and already known significance of members from the Pseudomonadales order in the plant phyllosphere, we took particular interest in deciphering how plant genetics may be involved in the enrichment of it. Pseudomonadales were not only highly abundant in the present study, but also accounted for the largest number of associated loci in the GWAS approach, indicating a robust connection to plant genetics. To verify rice genetic factors that affect the assembly of Pseudomonadales, we performed gene-editing to mutate a host gene shown to be specifically associated with it. OsPAL02 was selected as the target gene due to the strong association Pseudomonadales (P-value: 1.33E-09) and because there are no paralogous genes in locus 158 and other loci identified by GWAS. Using 16S rRNA gene fragment amplicon sequencing, we showed that the rice OsPAL02 gene mutant (hereafter referred to as OsPAL02-KO), which lacks phenylalanine/tyrosine ammonia-lyase, showed a significant change in the relative abundance of Pseudomonadales (Fig. 3a). Intriguingly, natural variation of SNPs in OsPAL02 was different between indica and japonica rice (Supplementary Fig. 7), suggesting a correlation between Pseudomonadales abundance and this gene.

**Fig. 3: *OsPAL02* is linked to phyllosphere microbiome assembly and has implications for plant health.**

To further investigate the detailed role of OsPAL02 in shaping the rice leaf microbiota, we carried out 16S rRNA gene fragment amplicon sequencing to profile the leaf microbiota of wild-type (WT) rice variety ZH11 (japonica), the loss-of function mutant, OsPAL02-KO and the over-expression construct, OsPAL02-OE. Under greenhouse conditions, we observed that the community structure in the phyllosphere of WT, OsPAL02-OE and OsPAL02-KO lines were dominated by Pseudomonadales, Enterobacterales, Sphingomonadales, Xanthomonadales, Flavobacteriales, Paenibacillales, Rhizobiales, Sphingobacteriales and Burkholderiales (Fig. 3a). This result was consistent with initial analysis of the leaf microbiome, indicating that field observations were reproducible in the greenhouse. Unconstrained PCoA (Fig. 3b, P < 0.001, R² = 0.32, PERMANOVA) and constrained principal coordinate analysis (CPCoA) (Supplementary Fig. 8, 23.7% of total variance was explained by the plant genotype, P < 0.001, PERMANOVA) of Bray-Curtis distances revealed that the WT, OsPAL02-OE and OsPAL02-KO phyllosphere microbiomes formed separate clusters, indicating that OsPAL02 affected their bacterial communities. Further analysis of differences in phyllosphere microbiomes between WT, OsPAL02-OE and OsPAL02-KO revealed significant variations at the order level (Fig. 3a). The relative abundance of Pseudomonadales showed a significant decrease in OsPAL02-KO and increase in OsPAL02-OE, compared to WT plants respectively (Fig. 3c). The relative abundance of Xanthomonadales, Flavobacteriales and Burkholderiales showed a significant increase in OsPAL02-KO compared to WT or OsPAL02-OE plants (Fig. 3d, Supplementary Table 3, Supplementary Data 5). Notably, the Shannon index (alpha-diversity) of OsPAL02-KO was significantly higher than that of OsPAL02-OE and WT (Supplementary Fig. 8). Similar observations were made when two additional OsPAL02-KO and OsPAL02-OE lines were analyzed (Supplementary Fig. 9, Supplementary Tables 4 and 5, Supplementary Data 5). The observation that OsPAL02-KO harbored higher alpha-diversity and lower relative abundance of Pseudomonadales than wild-type ZH11 (japonica) followed the same trend observed with indica and japonica varieties (Fig. 1b, c). This finding indicated that differences in OsPAL02 were sufficient to mimic microbiome variations between indica and japonica plants. Conclusively, the gene OsPAL02 was shown to contribute to the recruitment of Pseudomonadales and shape the bacterial community structure in the rice phyllosphere.

In the phenylpropanoid biosynthesis pathway, the gene OsPAL02 encodes a phenylalanine/tyrosine ammonia-lyase which catalyzes the transformation of L-tyrosine into 4-hydroxycinnamic acid (4-HCA) as well as the transformation of L-phenylalanine into trans-cinnamic acid, which are both precursors for the biosynthesis of lignin. When OsPAL02 variation between japonica and indica was analyzed, the protein structures of haplotype 1 (mainly present in indica) and haplotype 5 (mainly present in japonica) showed an amino acid exchange at site 134. This result indicated that OsPAL02 catalytic activity may differ between japonica and indica plants (Supplementary Fig. 7). We, therefore, hypothesized that the products of OsPAL02 impose a selective force on the recruitment of phyllosphere microbiota members, especially those within the bacterial order Pseudomonadales.

To investigate this hypothesis, we performed metabolite analyses with rice leaves of WT, OsPAL02-OE, and OsPAL02-KO, in order to identify differences attributable to OsPAL02. The results showed that WT, OsPAL02-OE, and OsPAL02-KO metabolite profiles were separated into three clusters based on Bray-Curtis distances and confirmed that OsPAL02 was involved in the biosynthesis of cinnamic acid and its derivatives (Fig. 4a, Supplementary Data 6). The data was also subjected to unconstrained PCoA and the results showed that the first principal coordinates PCo1, which accounts for 53.89% of the total variance, separated the WT, OsPAL02-OE, and OsPAL02-KO plants, indicating substantial differences in metabolites present in rice leaves (Fig. 4b, P < 0.001, R² = 0.60, PERMANOVA). This was also observed with CPCoA (Supplementary Fig. 10) where 46.1% of total variance was explained by the plant genotype (P < 0.001, PERMANOVA). Further analysis revealed that 12 metabolites showed significant differences between WT, OsPAL02-OE, and OsPAL02-KO (Supplementary Data 6). Therein, 4-HCA showed significantly higher intensity in OsPAL02-OE and lower in OsPAL02-KO compared to WT plants (Fig. 4c), suggesting that 4-HCA is the key metabolite regulated by OsPAL02 and may thus contribute to phyllosphere microbiome assembly. LC-MS analysis was conducted to detect 4-HCA physiological concentrations in rice leaves. As expected, the concentration of 4-HCA was significantly higher in the three OsPAL02-OE lines and lower in the three OsPAL02-KO lines when compared to WT plants (Supplementary Fig. 11, Supplementary Table 6, Supplementary Data 5). The concentration of 4-HCA was additionally determined in five Pseudomonadales-enriched japonica and five Xanthomonadales-enriched indica varieties. We found that the concentration of 4-HCA in the analyzed japonica varieties was significantly higher than in the indica varieties (Supplementary Fig. 12, Supplementary Table 7). A correlation analysis indicated that the concentration of 4-HCA shows a positive relationship with the relative abundance of Pseudomonadales and a negative relationship with the abundance of Xanthomonadales in the leaves of the analyzed rice varieties (Supplementary Fig. 12). These observations provided further evidence that 4-HCA is the key metabolite regulated by OsPAL02 and contributes to phyllosphere microbiome variations between indica and japonica varieties.

**Fig. 4: Metabolite profiles of *OsPAL02* mutants.**

Apart from the altered phyllosphere microbiomes, OsPAL02-KO developed severe disease symptoms compared with WT and OsPAL02-OE after leaf cutting (Fig. 3e). This phenomenon was not observed before the leaf cutting (Supplementary Fig. 13), indicating that the absence of OsPAL02 caused the dysbiosis in the rice phyllosphere microbiome and consequently led to the vulnerability of plants to pathogens on leaves. Unsurprisingly, the fresh weight of WT and OsPAL02-OE leaves was significantly higher than that of OsPAL02-KO; fresh weight of OsPAL02-OE leaves showed no significant difference to WT after leaf cutting (Fig. 3f). To determine whether disease symptoms were caused by the microbiota shift in OsPAL02-KO, we grew WT and OsPAL02-KO plants under gnotobiotic conditions and conducted the leaf cutting under controlled conditions preventing contamination. The OsPAL02-KO plants were indistinguishable from WT plants, and the fresh weight showed no significant differences among the three rice lines (Supplementary Fig. 14). This observation indicated that the disease symptoms in OsPAL02-KO might result from a microbiota dysbiosis. Combining the results of the metabolite analysis and the above assays, we proposed that OsPAL02 might be responsible for producing 4-HCA and maintaining a functional microbiota.

To validate the function of 4-HCA, rice leaves of OsPAL02-KO plants were sprayed with 4-HCA. As we expected, the application of 4-HCA on OsPAL02-KO plants resulted in a phyllosphere microbiome showing no significant difference to WT plants, and their fresh weights also did not differ (Supplementary Fig. 15, Supplementary Table 8). These results demonstrated that 4-HCA prevents dysbiosis in the phyllosphere microbiota of OsPAL02-KO plants. Overall, our data suggested that OsPAL02-synthesiszed 4-HCA plays a key role in maintaining plant health.

Implications of 4-HCA and members of Pseudomonadales for phyllosphere microbiome homeostasis

Although we observed that 4-HCA is involved in microbiome assembly and prevents dysbiosis in the rice phyllosphere, the underlying mechanism remained unclear. To investigate potential interactions of 4-HCA with rice leaf-associated bacteria, we isolated bacteria using the limiting dilution method³¹. Bacterial strains were isolated from pooled leaf samples of WT, OsPAL02-OE, and OsPAL02-KO plants grown under greenhouse conditions. Taxonomic assignments of bacterial strains were conducted by Sanger sequencing of their 16S rRNA genes, and non-redundant strains were identified based on gene similarity <97%. In total, we obtained a collection of 77 non-redundant strains that contained 14 bacterial orders, mainly Pseudomonadales (22.08% of the total non-redundant strains), Bacillales (19.48%), Enterobacterales (15.58%), Sphingomonadales (9.1%) and Flavobacteriales (7.79%). They covered 62.5% of the orders (relative abundance > 0.1%) detected with amplicon sequencing analyses (Supplementary Fig. 16, Supplementary Data 7). The isolate collection enabled us to further study interactions between 4-HCA and phyllosphere microbiota members through targeted assays and synthetic community (SynCom) approaches.

Since the disease symptoms after leaf cutting were observed not only in OsPAL02-KO plants but also in some WT and OsPAL02-OE plants (Fig. 3e), we hypothesized that the pathogenic agent might be present within the isolate collection. To identify the causal agent(s) of the disease symptoms, we tested the pathogenicity of the individual isolated strains. We inoculated single strains onto germ-free WT seedlings and monitored the plant disease occurrence. We then identified Xanthomonas oryzae strain TJ1 (hereafter referred to as TJ1), as a causative pathogen of bacterial leaf-blight disease on rice (Supplementary Fig. 17). We then tested the pathogenicity of TJ1 on OsPAL02-KO and OsPAL02-OE plants. As expected, the inoculation of TJ1 caused severe disease symptoms on OsPAL02-KO, meanwhile, OsPAL02-OE were less affected in terms of disease occurrence and fresh weight, compared to WT and OsPAL02-KO. (Supplementary Fig. 17). We also classified disease symptoms into four grades (1, healthy, to 4, dead), and showed that increasing disease severity correlated with decreasing plant fresh weight (Supplementary Fig. 17). Furthermore, we re-isolated bacteria from TJ1-inoculated WT, OsPAL02-KO and OsPAL02-OE plants. Sequencing of their 16S rRNA genes confirmed 100% sequence similarity with TJ1 (Supplementary Fig. 17). Additional isolation experiments showed that TJ1 abundance in OsPAL02-OE plants was significantly lower than in the OsPAL02-KO mutant and WT plants (Supplementary Fig. 17). This provided further evidence that TJ1-induced disease resulted in the observed reduced average plant weight. Overall, the isolate TJ1 was confirmed to be a rice pathogen, and it was demonstrated to cause lighter disease symptoms in OsPAL02-OE than that observed with OsPAL02-KO and WT plants.

We then hypothesized that 4-HCA at certain concentrations may contribute to inhibition of TJ1 and reduce disease symptoms. We grew TJ1 in liquid culture with different concentrations of 4-HCA. We found that 4-HCA at certain concentrations can effectively inhibit TJ1 growth, and the growth of TJ1 was completely inhibited when the concentration of 4-HCA reached 2 mM (Fig. 5c, d). This result indicated that 4-HCA can directly inhibit the growth of the pathogen; the 50% inhibitory concentration (IC50) of TJ1 was 1 mM.

**Fig. 5: Effects of 4-HCA on phyllosphere bacterial growth and rice pathogen TJ1.**

Next, we conducted in vitro assays to test the influence of 4-HCA on strains from Pseudomonadales, Xanthomonadales and Burkholderiales. These bacterial strains were grown in liquid R2A medium supplemented with 2 mM 4-HCA; the same concentration inhibited the growth of pathogen TJ1. We found that most of the Pseudomonadales strains grew faster in the presence of 2 mM 4-HCA, whereas all Xanthomonadales and Burkholderiales strains were significantly inhibited (Fig. 5a). Detailed analysis of growth changes (OD₆₀₀) indicated that all Pseudomonadales strains, except Acinetobacter sp. 21, showed a significantly higher growth rate than Xanthomonadales and Burkholderiales strains in the presence of the 2 mM 4-HCA (Fig. 5a, Supplementary Fig. 18, Supplementary Data 7). The in vitro assays clarified the effect of 4-HCA on the growth of strains from the orders Pseudomonadales, Xanthomonadales and Burkholderiales. They also shed further light on the phyllosphere microbiome assembly discrepancies between OsPAL02-OE, OsPAL02-KO, and WT plants (Fig. 3a) and indicated a pivotal role of 4-HCA therein.

We observed severe disease symptoms in TJ1-inoculated WT plants under gnotobiotic conditions (Supplementary Fig. 17), interestingly, this phenomenon did not occur when plants were grown under greenhouse conditions, where they were exposed to various microbes (Fig. 3e, f). Therefore, we speculated that the effect of 4-HCA at least partially relies on microbiota homeostasis. We performed in vitro inhibition assays with TJ1 and 22 isolates representing leaf-associated Pseudomonadales, Xanthomonadales and Burkholderiales members, on R2A medium with and without 1 mM 4-HCA (Fig. 5a). The results revealed that under the 1 mM 4-HCA condition, all Pseudomonadales strains, including Acinetobacter sp. 21, strongly inhibited TJ1 (Fig. 5a). Conversely, 75% of the Pseudomonadales strains lost the ability to inhibit TJ1 when 4-HCA was absent (Fig. 5a). Within Burkholderiales, only Delftia sp. 12, inhibited TJ1 in the presence of 4-HCA and lost this inhibition activity when 4-HCA was absent (Fig. 5a).

Overall, 4-HCA-dependent inhibition mostly occurred among strains from Pseudomonadales. This observation also connected 4-HCA to microbial functioning in the phyllosphere. Furthermore, we designed a synthetic community (SynCom, Fig. 5a) consisting of Pseudomonadales strains, and then subjected it to specific assays to evaluate its efficacy against the pathogen TJ1. Under gnotobiotic conditions, WT, OsPAL02-KO and OsPAL02-OE plants showed no significant differences in fresh weight when treated with either 10 mM MgCl₂ (control, Fig. 6a, b) or the SynCom suspension (Supplementary Fig. 19). This result confirmed that the SynCom did not have a negative impact on rice plants. Then, we treated the rice plants with SynCom + TJ1 suspension and TJ1 suspension only. In WT and OsPAL02-OE plants, the SynCom showed protective effects against TJ1 (Fig. 6c, d). In contrast, no protection was observed with OsPAL02-KO plants, as indicated by strong disease symptoms and significantly lower leaf fresh weight (Fig. 6c, d). When only TJ1 suspension was applied, WT and OsPAL02-KO plants both showed more pronounced disease symptoms and significantly lower leaf fresh weight compared to OsPAL02-OE plants (Fig. 6e, f), suggesting that without the SynCom, the protection effect against TJ1 is lost. Evidently, an interplay between OsPAL02 and Pseudomonadales members was required for the SynCom-associated plant health maintenance. Combined with the aforementioned results, a conclusion can be drawn that OsPAL02-synthesized 4-HCA is important for the enrichment of bacterial members of the Pseudomonadales and drives phyllosphere microbiome assembly. Microbiome assembly regulated by 4-HCA-regulated is of great importance to maintaining rice phyllosphere homeostasis, which is directly connected to plant health.

**Fig. 6: SynCom inhibits TJ1 under 4-HCA presence in vivo.**

Discussion

Plants are colonized by highly diverse, tissue-specific microbial communities, the plant microbiota. The plant microbiota is known to complement host functioning and to be an important factor for enhancing resilience under abiotic as well as biotic stress. Plants can assemble and maintain certain microbiota structures, which is vital for plant health. A dynamically balanced state of the microbiota is known as ‘homeostasis’³². Disrupting homeostasis often results in a prevalence of detrimental microbiota members, leading to shifted microbiota structures and plant disease occurrence, also known as dysbiosis³³. It is currently sparsely understood how plants assemble and maintain a functional (homeostatic) microbiota, especially in regards to their phyllosphere³³. Previous research has mostly focused on microbe-microbe interactions that result in homeostasis and can explain certain colonization patterns^34,35,36. Here, we show that specific host-microbe interplay is involved in shaping phyllosphere bacteria prevalence in rice plants. Research conducted in the past has provided indications for active host regulation of the phyllosphere microbiota. The plant immune system was demonstrated to be required for maintaining microbiota homeostasis^37,38,39,40. Mutations of immunity-related genes in Arabidopsis have been shown to result in a dysbiotic phyllosphere microbiota^27,41. In the present study, we show that rice plants deploy a potentially conserved mechanism to shape the phyllosphere microbiome in a wide range of genotypes. Except for immunity-related mechanisms, plants can also employ metabolites as a selective force to assemble candidate microbiota members to serve their interests¹⁷. Various plant-specific metabolites can fulfill the purpose of maintaining microbiota homeostasis on leaves⁴², or suppress bacterial pathogens⁴³. Nevertheless, fundamental questions remain unaddressed related to which plant genetic components are devoted to control microbiota assembly, and how this takes place on metabolite level.

Genome-wide association studies (GWAS) have proven to be a highly useful tool to shed further light on plant-microbe interactions^28,44. In a recent study, researchers showed that the abundance of phyllosphere microbial hubs in Arabidopsis correlated with multiple specialized metabolic pathways affecting both bacterial and fungal communities^19,21. It should be highlighted the previous studies were solely based on amplicon sequencing of microbial marker genes in the phyllosphere. While such approaches are valuable, this method is also known to be prone to specific biases, because marker genes of certain microbial groups are more likely to be amplified than those of others¹⁷. Therefore, microbiome structures are not always accurately captured. In the present study, a large-scale phyllosphere metagenome dataset was combined with GWAS and facilitated more precise insights into associations between rice plants and their phyllosphere microbes. For one, it provided further evidence for the role of the plant immune system in shaping its microbiota, such as the MAPK signaling pathway. More importantly, it allowed us to pinpoint specific metabolic pathways and genes which we identified as suitable targets to dissect phyllosphere microbiome assembly at molecular level. Furthermore, the high resolution of taxonomic data obtained here was advantageous to connect host genetic traits to specific bacterial taxa, which was critical to validate the effects of a distinct gene and metabolite on microbiome assembly.

The present study revealed that primarily bacteria belonging to the order Pseudomonadales are enriched by 4-HCA in the rice phyllosphere. A similar observation was obtained with poplar tree endophytes and down-regulation of cinnamoyl-CoA reductase; it caused the enrichment of 4-HCA in plant tissues and consequently increased the abundance of Pseudomonas⁴⁵. This genus is a ubiquitous member of the plant microbiota^46,47. Previous studies have shown that it is an important component of a functional microbiota that can shield off pathogens and prevent diseases in plants. Pseudomonas were often identified as core microorganisms associated with both the plant rhizosphere and phyllosphere^35,36,48,49, and as key players in many disease-suppressive soils^50,51. Plants benefit from Pseudomonas as they can produce a wide range of antimicrobial compounds⁵², induce plant systemic disease resistance⁵³, and establish an array of chemical dialogues with plants^37,54,55. Our identification of plant genetic components controlling Pseudomonas could provide an ideal target for engineering the phyllosphere microbiota of various crop plants. We also showed that Xanthomonas, a highly prevalent phytopathogen, is negatively affected by 4-HCA. It is therefore very likely that plants protect themselves against pathogens by relying on their immune system as well as specific metabolites that not only allow them to directly antagonize pathogens but also to maintain homeostasis in the phyllosphere microbiota.

In recent years, plant microbiome studies have facilitated the identification of various disease-reducing or even disease-preventing microorganisms⁵⁶. Leaves are common entry points of various, highly devastating pathogens. The present study shows that metabolite-driven regulation of the phyllosphere microbiota is a hidden mechanism to support plant hosts against pathogen attacks. Previous studies that assessed the rhizosphere microbiome have resulted in similar findings. For example, iron-mobilizing coumarins in Arabidopsis roots were linked to shaping the rhizosphere microbiota by inhibiting proliferation of certain Pseudomonas species via a redox-mediated mechanism⁵⁴. Re-shaping of the rhizosphere microbiome by the coumarin scopoletin was shown to have consequences for plant health⁵⁷. Similarly, the benzoxazinoid breakdown product 6-methoxy-benzoxazolin-2-one (MBOA) was found to regulate rhizosphere microbiota assembly of cereal crops⁵⁸.

The present findings related to plant metabolite-driven phyllosphere homeostasis and their implications for host health will serve as a basis for targeted breeding strategies. Previous studies have provided certain indications, mostly via correlation analyses, that plant metabolites play an important role in maintaining a functional microbiota⁴². However, it remained unresolved if the identified compounds are the main drivers of the observed structural changes of the microbial communities. In the present study, knock-out as well as over-expression constructs allowed us to specifically attribute microbiome changes to 4-HCA in the rice phyllosphere. A lack of 4-HCA was linked to dysbiosis of the phyllosphere microbiome which results in higher susceptibility of rice plants to disease. Current resistance breeding strategies known as ‘R gene breeding strategies’ are increasingly challenged by rapidly evolving phytopathogens. This problem could be addressed by identifying and engineering microbiome-shaping genes, which may become known as ‘M gene breeding’.

Methods

Chemicals, reagents and analytical instruments

The chemicals, reagents and analytical instruments used in this study are listed as follows: ethanol (EtOH), sodium hypochlorite (NaOCl), Triton X-100, methanol (MeOH), 4-hydroxycinnamic acid (4-HCA), MgCl₂, Graphitized carbon black (GCB), Acetonitrile (ACN) and other organic solvents was purchased from Merck (Shanghai, China); agar powder, agarose gel, PBS buffer, Kanamycin, Hygromycin, Murashige and Skoog (MS), Linsmaier–Skoog (LS), Reasoner’s 2 A (R2A) and tryptic soy broth (TSB) medium were purchased from Solarbio (Beijing, China); BsaI, T4 DNA Ligase, Takara Ex Taq were purchased from Takara (Beijing, China); TransScript® II All-in-One First-Strand cDNA Synthesis SuperMix for PCR, TransTaq® DNA Polymerase High Fidelity (HiFi), EasyPure® Quick Gel Extraction Kit a were purchased from TransGen Biotech (Beijing, China); FastPure Plant DNA Isolation Mini Kit was purchased from Vazyme Biotech (Nanjing, China); MagPure Soil DNA LQ Kit was purchased from Magen (Shanghai, China); TIANprep Plasmid Midi kit was purchased from TIANGEN (Beijing, China); Agencourt AMPure XP beads was purchased from Beckman Coulter (Pasadena, USA); Qubit dsDNA assay kit was purchased from Yeasen (Shanghai, China); TRIzol was purchased from Invitrogen (Carlsbad, USA).

The analytical equipment used included a UHPLC (1290 Infinity LC, Agilent Technologies, USA) coupled to a quadrupole time-of-flight mass (TripleTOF 6600, AB Sciex, USA), and a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, USA).

Rice phyllosphere metagenomes

In order to investigate potential influence of rice genotypes on the phyllosphere microbiome, we obtained shotgun-sequenced metagenomes of 110 rice varieties (three replicates for each variety). Details about this dataset were included in a dedicated data publication (Supplementary Data 1)²⁹. The 110 rice varieties were selected from the Rice Diversity Panel II core collection (C-RDP-II), which can be grouped into ‘japonica’ (including 27 tropical japonica, 7 temperate japonica, and 2 admixed japonica) and ‘indica’ (including 32 indica, 22 aus, 2 admixed indica) varieties as defined by McCouch et al.⁵⁹. All varieties were planted in Taojiang County, Hunan Province, China (28°38′09″ N, 112°0′57″ E). Rice leaves were harvested at the booting stage (September 5th, 2020). The samples were enriched for their bacterial fraction, metagenomic DNA was extracted and subjected to high-throughput sequencing²⁹. Quality filtering of raw metagenomic data was described in the data publication²⁹.

Bioinformatics analyses of the metagenomes

Quality-filtered data were subjected to the taxonomic classification using Kraken2 v2.0.9³⁰. A species abundance table was generated with Bracken v2.6.0³⁰. The cumulative-sum scaling normalization method was applied with R package metagenomeSeq⁶⁰. A principal coordinates analysis (PCoA) based on Bray Curtis distances was performed using the R package vegan⁶¹. Alpha diversity analysis was carried out using aforementioned R package vegan. Differences in beta diversity were accessed using a pairwise Adonis test and permutational analysis of variance (PERMANOVA, 999 permutations) using R package amplicon⁹. Analysis of differential species abundance was performed using Wilcoxon rank sum and Tukey’s HSD tests based on species with an average relative abundance >0.001% in the assessed rice varieties. Species were defined as significantly different abundant if the P-value was <0.05. Diversity and relative species abundance data were visualized by using R package amplicon. Bacterial taxonomy was visualized with GraPhlAn v.0.9.7⁶². Boxplots and scatter diagrams were visualized by using the R package ggplot2 v.2.2.1⁶³.

GWAS analysis

The genome-wide association study was based on 110 C-RDP-II rice accessions represented by a 168,699-SNP dataset (filter removal MAF < 0.05) (Supplementary Data 8). This dataset was extracted from the C-RDP-II collection that encompasses a 700,000-SNP dataset⁵⁹. We defined ‘traits’ as the enrichment levels of the 6,862 identified species in the phyllosphere of the 110 rice varieties (Supplementary Data 3). Tassel5.0 (https://www.maizegenetics.net/tassel) was used for GWAS and an analysis pipeline was set up on a Linux system to analyze the 6,862 ‘traits’ individually (scripts are available at Code Availability). The GWAS method was based on a rice genome-wide association analysis approach using a high-density SNP array⁶⁴. We used a mixed linear model (MLM) that integrated the kinship matrix (K) with population structure (Q). We screened for significantly associated loci using the following criteria: a genomic region with ≤300 kb containing at least two significant SNPs (with a P-value threshold of ≤0.00001). We filtered for non-redundant loci associated with at least one enriched microorganism. The GWAS-identified loci were further analyzed for candidate genes.

Locus clustering and pathway enrichment

GWAS-identified loci can be associated with multiple bacterial species from different orders. For each locus, we calculated the proportion of associated species from the respective order. The proportion was later subjected to clustering locus visualization. Clustering analysis based on Bray Curtis distance between all loci was conducted using R package vegan. The clustering of loci was visualized by using R package circlize⁶⁵. For each of the locus clusters, gene annotation and Gene ID conversion were conducted using The Rice Annotation Project Database (RAPDB, https://rapdb.dna.affrc.go.jp/), and GO or KEGG enrichment analyses were conducted using R package clusterProfiler⁶⁶. GO enrichment analysis was conducted using org.Osativa.eg.db (https://github.com/xuzhougeng/org.Osativa.eg.db) database and significant enrichment of pathways was noted if the P-value was <0.05 (Fisher’s exact test). KEGG enrichment was analyzed using Oryza sativa japonica (RAPDB) KEGG Gene Database (Release 105.0 + /03-07) and significant enrichment of pathways was noted if the P-value was <0.05 (Fisher’s exact test). GO and KEGG enrichment pathways were visualized by using ggplot2.

Generation of CRISPR-edited and gene overexpression rice mutants

The CRISPR–Cas9 system was used to generate OsPAL02 (Os04g0518100) mutants according to a high-efficiency monocot genome editing method⁶⁷. Briefly, specific primers containing the OsPAL02 target site sequences (CCATGCGCTTCACCTCGTCCAGG) were ligated into pEGCas9Pubi-H cassettes. To construct guide RNA genes, rice small nuclear RNA U6 promoter was amplified from pYLsgRNA-OsU6a vector⁶⁷ using primer pairs U6-1/ U6-2. The DNA sequence encoding the gRNA scaffold was amplified from pYLsgRNA-OsU6a vector⁶⁷ using primer pair gRNA-1/gRNA-2. The PCR product of the U6 promoter was fused with the DNA fragment encoding the gRNA scaffold by overlapping PCR using primer pair sgRNA-1/sgRNA-2. The U6 promoter–gRNA fragment was cloned into the pEGCas9Pubi-H vector to form the pEGCas9Pubi-H-Os04g0518100 construct via BsaI (Takara, Beijing, China) restriction digestion followed by ligation using T4 DNA Ligase (Takara, Beijing, China) (Supplementary Fig. 20). After transformation into Escherichia coli DH5-alpha, the resulting constructs were purified with TIANprep Plasmid Midi kit (TIANGEN, Beijing, China) for PCR amplification using primer pairs KO-1/KO-2 and sequenced via Sanger sequencing (Sangon Biotech, Shanghai, China) (Supplementary Fig. 20). The validated constructs were stored at −20 °C for subsequent use in rice protoplast transformation.

The OsPAL02-overexpression constructs were generated according to a transgenic rice construction method⁶⁸. Briefly, for the OsPAL02 overexpression vector construct, total RNA of rice variety ZH11 leaves were extracted using TRIzol, and complementary DNA (cDNA) was synthesized from the total RNA using cDNA Synthesis SuperMix kit (TransGen Biotech, Beijing, China). The OsPAL02 coding sequence was amplified by PCR using primer pairs PAL02-1/PAL02-2. The amplified fragment was cloned into the pEGOEPubi-H vector, which contained the maize ubiquitin promoter and a hygromycin resistant gene, to form the pEGOEPubi-H-OsPAL02 overexpression construct via BsaI (Takara, Beijing, China) restriction digestion followed by ligation using T4 DNA Ligase (Takara, Beijing, China) (Supplementary Fig. 21). After transformation into Escherichia coli DH5-alpha, the resulting constructs were purified with TIANprep Plasmid Midi kit (TIANGEN, Beijing, China) for PCR amplification using primer pairs OE-1/OE-2 and sequenced via Sanger sequencing (Sangon Biotech, Shanghai, China) (Supplementary Fig. 21). The validated constructs were stored at −20 °C for subsequent use in rice protoplast transformation.

The resulting vectors were introduced into Agrobacterium tumefaciens (strain EHA105) and then into rice by Agrobacterium-mediated transformation⁶⁹. The transformation was performed using a rice transformation method⁶⁹. Briefly, positive transgenic bacteria were selected by using kanamycin for selection. Hygromycin was used to select putative transgenic plants. All the selected putatively transgenic seedlings were cultivated in the greenhouse or field for further identification.

To identify transgenic plants, genomic DNA was extracted from approximately 30 mg of leaf tissue using FastPure Plant DNA Isolation Mini Kit (Vazyme Biotech, Nanjing, China) according to the manufacturer’s instructions. The isolated and purified genomic DNA was used for PCR amplification of CRISPR/Cas9 target sites or hygromycin gene using primer pairs KO2-1/KO2-2 or H-1/H-2, respectively. Amplified fragments were detected using agarose gel electrophoresis, purified using EasyPure® Quick Gel Extraction Kit (TransGen Biotech, Beijing, China) and sequenced with sanger sequencing (Sangon Biotech, Shanghai, China) to identify OsPAL02-KO and OsPAL02-OE transgenic plants. All related primers are listed in Supplementary Table 9.

Plant material, growth conditions and treatments

Hunan Hybrid Rice Research Center (HHRRC) provided seeds of the ZH11 rice variety for the experiments and genetic transformation. ZH11 was implemented as the wild-type (WT) and to generate OsPAL02-KO (knockout) and OsPAL02-OE (overexpressing) plants. WT, OsPAL02-OE and OsPAL02-KO plants were cultured under relative humidity set at 80%, temperature at 28 °C and 13 h light cycle in a greenhouse or were grown at a field site of HHRRC in Changsha, China. Mature seeds of the WT, OsPAL02-OE and OsPAL02-KO plants were harvested, dried, and stored in a refrigerator at 4 °C.

Rice plants were grown in Taojiang soil filtrate mixed with greenhouse potting soil (PINDSTRUP substrate, Denmark; autoclaved twice) in an air-circulating greenhouse for controlled experiments. For Taojiang soil filtrate preparation, the soil was collected from the same rice field as for the metagenomic experiments, Taojiang County (28°38′09″ N, 112°0′57″ E), Hunan Province, China. The top 10–20 cm of field soil was collected and sieved (3-mm sieve) to remove rocks and other debris, dried at room temperature. Then, the soil was diluted 10-times using sterile water (1:10 w/v), followed by shaking for 20 min at 180 rpm/min and sonicated for 5 min at a frequency of 30 kHz at 4 °C. The suspensions were subjected to centrifugation for 1 min at 42 g, 4 °C. The supernatant was collected and used as the Taojiang soil filtrate. Greenhouse soil preparations were made by using 100 mL Taojiang soil filtrate together with 100 mL half-strength Murashige and Skoog (MS) medium solution (pH 5.8) and mixed with 200 g of greenhouse potting soil (PINDSTRUP substrate, Denmark; autoclaved twice). Rice seeds were surface-sterilized using 75% ethanol (EtOH) for 2 min and 7% sodium hypochlorite (NaOCl) supplemented with 0.2% Triton X-100 three times for 8 min, following washed with sterile water six times. Sterile seeds were further subjected to microbiota inoculation from filed soil by immersing them in Taojiang soil filtrate for 1 d at 4 °C and 4 d at 28 °C in the dark. Subsequently, all rice seeds at the early germination stage were transplanted into the pots with the aforementioned soil preparation. Each plant was watered with 5 mL of sterile water twice a week and sterile half-strength MS solution once a week after germination. Top leaves of 3-week-old plants were cut (1 cm) using sterile scissors (surface-sterilized using 75% EtOH and washed with sterile water six times for each plant) for identifying OsPAL02-KO or OsPAL02-OE constructs 5-week-old plants were harvested for collection of phyllosphere bacteria, extraction of leaf metabolites, determination of fresh weight and other analyses.

For the complementary 4-HCA assay, WT and OsPAL02-KO plants were surface-sterilized, cold-stratified, germinated and transferred to the same greenhouse as mentioned before. Each 2-week-old OsPAL02-KO plant was sprayed with 2 mL of sterile 1 mM 4-HCA (dissolved in water). Each 2-week-old plant of WT plant was sprayed with 2 mL sterile water. Watering of the seedlings and identification of the genotype followed the same procedure as described before. 5-week-old plants were harvested for microbiome, phenotype and fresh weight analyses.

For gnotobiotic plant growth, WT, OsPAL02-KO and OsPAL02-OE seeds were surface-sterilized with the same procedure as described before, and then cold-stratified and germinated in sterile water. Subsequently, all rice seeds at early germination stage were transferred into sterile bottles with autoclaved (three times for 45 min each, with 24 h storage at room temperature in between) soil. Sterility of rice seeding was routinely monitored by plating samples of plants and culture substrate on R2A plates. Watering of seedlings and identification of the genotype were as described above, but under sterile conditions. 5-week-old plants were harvested for phenotype and fresh weight analyses.

Phyllosphere microbiome analyses of WT and mutant plants

5-week-old plants cultivated under greenhouse conditions were harvested for profiling of the phyllosphere microbiome (six replicates per genotype, each replicate contained three plants leaves). Sample collection was conducted according to a rice phyllosphere bacteria enrichment method²⁹. Total DNA was extracted using MagPure Soil DNA LQ Kit (Magen, Shanghai, China) according to the manufacturer’s instructions. Quality and quantity of DNA was verified using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, USA) and agarose gel electrophoresis, respectively. Extracted DNA was diluted to a concentration of 1 ng/μL and stored at −20 °C until further processing. PCR amplification of bacterial 16S rRNA gene fragments (V3–V4 region) was performed using Takara Ex Taq (Takara, Beijing, China) and the barcoded primers 343 F and 798 R listed in Supplementary Table 9. Amplicons were visualized using agarose gel electrophoresis and purified using Agencourt AMPure XP beads (Beckman Coulter, Pasadena, USA) twice. After purification, the DNA was quantified using Qubit dsDNA assay kit (Yeasen, Shanghai, China). Equal amounts of purified DNA were pooled for sequencing on the NovaSeq 6000 platform (Illumina Inc, USA) at Shanghai OEbiotech (Shanghai, China).

The 16S rRNA gene fragment sequences were processed using QIIME2 v.2021.4⁷⁰. Paired-end reads were detected and the adapters were removed using Cutadapt v.2.1⁷¹. After trimming, paired-end reads were filtered for low-quality sequences, denoised, merged and clustered using DADA2 v.2020.2.0⁷². Amplicon sequence variants (ASVs) and a feature table were generated using QIIME2. All ASVs were annotated using the Silva v138.1 reference databases⁷³. Diversity and differential abundance analyses were performed using the same methods as described above.

Rice leaf metabolite extraction

5-week-old WT, OsPAL02-KO and OsPAL02-OE plants cultivated under greenhouse conditions were harvested for extraction of leaf metabolites (six replicates per genotype, each replicate included three plants leaves). Rice leaves were immediately frozen in liquid nitrogen and ground into fine powder with a mortar and pestle. Leaf metabolites were extracted from 80 mg of each sample using 1000 μL methanol (MeOH)-acetonitrile (ACN)-ultrapure water (2:2:1 v/v/v). The homogenates were shaken at 200 rpm/min for 20 min, followed by sonication at a frequency of 30 kHz for 20 min. After centrifugation at 19000 g, 4 °C for 20 min, the supernatant was dried to remove the organic solvent using a vacuum centrifuge. The samples were re-dissolved using 100 μL ACN-ultrapure water (1:1 v/v) and centrifuged at 19000 g, 4 °C for 15 min, followed by filtering through a 0.22-μm filter before analysis with a liquid chromatography–tandem mass spectrometry (LC–MS/MS) system.

LC-MS/MS conditions

The samples were separated by ultra-performance liquid chromatography (UHPLC) on an ACQUITY UPLC BEH C-18 column (1.7 µm, 2.1 mm× 100 mm, Waters, USA); the column temperature was 40 °C. The flow rate was set at 0.4 mL/min, and the injection volume was 2 μL. The gradient elution procedure was as follows: 0-0.5 min, 5% MeOH/H₂O (v/v); 0.5-10 min, linearly changed to 100% MeOH; 10-12 min, MeOH was maintained at 100%; 12.0-12.1 min, MeOH/H₂O linearly changed to 5%; 12.1-16 min, MeOH/H₂O was maintained at 5%. In MS only acquisition, the instrument was set to acquire over the m/z range 60-1000 Da, and the accumulation time for the TOF MS scan was set at 0.20 s/spectrum. In auto MS/MS acquisition, the instrument was set to acquire over the m/z range 25-1000 Da, and the accumulation time for product ion scan was set at 0.05 s/spectrum. During the whole analysis, the sample were kept in an automatic sampler at 4 °C. In order to avoid influences caused by the fluctuations of the instrument, a random sequence was used for the analysis of samples.

Cinnamic acid and derivatives profiling

The raw MS data (wiff.scan files) were converted to MzXML files using ProteoWizard MSConvert⁷⁴. Peak picking and peak grouping were conducted by using R package XCMS v3.20.0⁷⁵. Annotation of adducts and calculating hypothetical masses for the group were performed with R package CAMERA (Collection of Algorithms of MEtabolite pRofile Annotation) v3.1-5⁷⁶. In the extracted ion features, only the variables with > 50% of nonzero measurement values in at least one group were kept. Identification of cinnamic acid and derivatives was performed by comparing high-accuracy m/z value using MS/MS spectra with an in-house cinnamic acid and derivatives database established with available standards.

After normalization for total peak area intensity, clustering analysis of metabolites were performed with R package ComplexHeatmap⁷⁷. A principal coordinates analysis (PCoA) of Bray-Curtis distances was performed using R package vegan, and a pairwise Adonis test and PERMANOVA (999 permutations) were performed using R package amplicon. Analysis of differential intensity of total peak areas was performed using Tukey’s HSD test. Features were visualized by using ggplot2.

Quantification of 4-HCA in rice leaves

Rice plants that were cultivated for five weeks under greenhouse conditions were harvested for the extraction of leaf metabolites (15 replicates per genotype, each replicate included leaves of three plants). Rice leaves were immediately frozen in liquid nitrogen and ground into fine powder with a mortar and pestle. Leaf metabolites were extracted from 0.6 g of each sample using 20 mL ACN. The homogenates were shaken at 200 rpm/min for 20 min, followed by sonication at a frequency of 30 kHz for 20 min. After centrifugation at 19000 g, 4 °C for 20 min, the supernatant was obtained and mixed with 30 mg GCB. The suspension was shaken at 200 rpm/min for 20 min and centrifuged at 19000 g, 4 °C for 15 min, followed by filtering through a 0.22-μm filter before analysis with a liquid chromatography (LC; 1290 Infinity LC, Agilent Technologies, USA) –tandem mass spectrometry (MS/MS; AB SCIEX 4500Q, AB Sciex, USA) system. The standard concentration of 4-HCA was prepared at 0.01, 0.05, 0.1, 0.5, 1, 5, 10 mg/L. The standard preparations were also mixed with 30 mg GCB, then shaken, centrifuged and filtered before LC–MS/MS experiments. The standard curve was calculated with the linear regression method by using ggplot2 (Supplementary Fig. 22). Significance of differences for 4-HCA concentrations was assessed using Tukey’s HSD test. Features were visualized with ggplot2.

Isolation of rice leaf bacteria

Rice plants cultivated under the aforementioned greenhouse conditions were used for bacterial isolations via the limiting dilution method³¹. To obtain a collection of representative rice leaf colonizing bacteria, we isolated bacteria from pooled WT, OsPAL02-KO and OsPAL02-OE leaves. Plants were harvested after five weeks. The leaves of three plants of each sample type were pooled to increase diversity of present bacteria. Rice leaves were washed twice in sterile PBS buffer (0.02 M, pH 7.0) on a shaker for 5 min at 180 rpm/min to remove small particles and loosely attached microbes. The leaves were ground into a homogeneous suspension using 10 mM sterile MgCl₂, followed by sonification for 5 min at a frequency of 30 kHz. Homogenized leaves were left to sediment for 15 min and the supernatants were diluted, distributed and cultivated in 96-well microtiter plates in 1:10 tryptic soy broth (TSB)-water (v/v) and 1:2 Reasoner’s 2 A (R2A)-water (v/v) for 20 d at room temperature. Then bacteria were purified by continuous streaking. Cultivated bacteria were identified by Sanger sequencing with 27 F and 1492 R primers (Supplementary Table 9) as well as used for the preparation of glycerol stocks for the culture collection. Taxonomic identity was assessed using NCBI BLASTN (https://blast.ncbi.nlm.nih.gov/Blast.cgi). Non-redundant bacterial isolates were selected by 16S rDNA sequence homology <97%. The taxonomic profiling of non-redundant bacterial isolates was visualized with GraPhlAn.

Screening for opportunistic rice pathogens

To remove seed-epiphytic microbes, ZH11 seeds were subjected to surface-sterilization as described above. The resulting seeds were placed on TSB agar plates (ten seeds per plate) to confirm that their surface is epiphyte-free by checking for the absence of visible colony growth surrounding them (24 h, 28 °C). Individual bacterial strains from the culture collection were picked from R2A agar plates and suspended in 10 mM MgCl₂. The final OD₆₀₀ values of bacterial suspensions were adjusted to 0.01. Subsequently, ten sterilized rice seeds were subjected to bacterial inoculation with the individual isolates by soaking them in 20 mL of bacterial cell suspension in a 9-cm (diameter) Petri dish at 28 °C until their germination. Ten seeds at the early germination stage were transplanted into a 20-cm (height) sterile glass bottle with a filter and permeable membrane containing 200 mL of Linsmaier–Skoog (LS) medium solidified with 0.8% agar. For each treatment, 20 μL bacterial cell suspension was applied directly to each leaf of the one-week-old seedlings. They were incubated in a plant growth incubator (28 °C, 80% relative humidity, 13-h photoperiod). All seedling leaves were cut 1 cm from the top after 2 weeks. The phonotype, disease index and fresh weight of the seedlings were monitored and measured for a total of 3 weeks. Only Xanthomonas oryzae sp. TJ1 (TJ1) showed disease symptoms from all tested bacteria. The phenotype assay with OsPAL02-KO and OsPAL02-OE plants inoculated with TJ1 was conducted in the same way as described above. Bacteria were also re-isolated from the inoculated WT, OsPAL02-KO and OsPAL02-OE plants. TJ1 was identified by Sanger sequencing with 27 F and 1492 R primers (Supplementary Table 9). Bacterial colonization was quantified by plate counting. All isolates were tested in triplicates with three technical repeats.

Effect of 4-HCA on leaf bacterial growth

The main product of OsPAL02 is 4-HCA (4-hydroxycinnamic acid), therefore the effects on the growth of leaf bacteria were tested using in vitro bioassays with pure 4-HCA¹⁴. Single colonies were picked from R2A agar plates and grown overnight in R2A medium on a shaker at 180 rpm/min, 28 °C. Overnight cultures were used to collect bacteria with centrifugation at 600 g, 4 °C for 10 min. The collected bacteria were resuspended to OD₆₀₀ = 0.01 using R2A medium, followed by addition of sterile 1 M 4-HCA solution (dissolved in ethanol, and filtered through a 0.22-μm filter) to reach different concentrations. The same amount of the solvent EtOH (filtering through a 0.22-μm filter) was added as control. The effect of 4-HCA on the opportunistic pathogen’s (TJ1) growth was assessed by using R2A medium supplemented with 0, 1, 2 or 4 mM 4-HCA. We found that R2A medium supplemented with 2 mM 4-HCA completely inhibited TJ1. Therefore, this concentration of 4-HCA was used to assess growth effects on isolated members of Pseudomonadales, Xanthomonadales and Burkholderiales in the culture collection. After supplementation with 4-HCA or solvent, 200 µl of each microbial suspension was aliquoted into clear 96-well flat-bottom, polystyrene tissue culture plates. The plates were incubated at 28 °C on an orbital shaking platform at 80 rpm/min. The optical density (OD₆₀₀) was measured every 2 hours at 600 nm and monitored over 24 hours (Supplementary Data 7). All isolates were tested in triplicates with three technical repeats.

Assessment of TJ1 inhibition by leaf bacteria

All isolates belonging to Pseudomonadales, Xanthomonadales and Burkholderiales were screened for inhibition of TJ1 using a binary interaction assessment method²⁷. Bacterial isolates were individually cultured on R2A plates at 28 °C. A sterile 1-µL plastic loop with collected bacteria was suspended in 3 mL R2A medium. To incorporate bacteria into the medium, 2.6 mL of bacterial suspension were added to 40 mL of molten R2A-agar or R2A-agar supplemented with 1 mM 4HCA pre-cooled to 42 °C, gently mixed and then poured into Petri dishes (9 cm). A mock treatment was prepared with 2.6 mL filtrate of the same bacterial suspension, through a 0.22-μm filter. Single colonies of TJ1 were picked from R2A agar plates and grown overnight in R2A medium on a shaker at 180 rpm/min, 28 °C, followed by the collection of bacteria via centrifugation at 600 g, 4 °C for 10 min. For the inhibition target, TJ1 suspension was adjusted to OD₆₀₀ = 0.1 using 10 mM sterile MgCl₂. A total of 10 μL TJ1 suspension was spotted onto paper discs at the agar plates as well as, 10 μL filtrate of TJ1 suspension, which was filtered through a 0.22-μm filter, as control. All isolates were tested in triplicates with three technical repeats, and their inhibition ability was recorded after 2d at 28 °C.

Synthetic community assembly and plant leaf inoculation

A synthetic community (SynCom) was designed to contain Pseudomonadales isolates on the basis of the observed improved growth when exposed to 4-HCA (shown in Fig. 5a) and prepared using a SynCom preparation method⁴¹. Sixteen bacterial isolates and TJ1 were individually grown on R2A agar plates, and then single colonies were re-streaked on fresh R2A plates and grown at 28 °C. A sterile 1-µL plastic loop with material of each isolate was used to resuspend them in 1 mL of sterile 10 mM MgCl₂. Tubes containing the resuspended bacteria were vortexed for 5 min, followed by sonification for 2 min at a frequency of 30 kHz to disperse cell aggregates. The SynCom inoculum was obtained by combining 16 bacterial isolates in equal suspension volumes. The optical density (OD₆₀₀) of the SynCom inoculum and TJ1 was adjusted to 0.02.

For plant inoculation, rice seeds of WT, OsPAL02-KO and OsPAL02-OE were surface-sterilized as described above. Each sterilized seed was germinated for 5 d at 28 °C with 800 μL of the SynCom or sterile 10 mM MgCl₂ (as control) in a Petri dish, followed by direct application of 50 μL TJ1 or sterile 10 mM MgCl₂ (as control) to each seed. Then rice seedlings were transplanted into sterilized bottles containing LS agar. They were incubated in a plant growth incubator (28 °C, 80% relative humidity, 13-h photoperiod). Leaves were cut from 3-week-old plants using sterile scissors (surface-sterilized using 75% EtOH and washed with sterile water six times for each plant) to identify their genotype. After 2 weeks, the phenotype of each plant was recorded, and their fresh weight was measured. Different plant genotypes were grown in the same bottle to minimize bottle variations, and treatments were blinded after inoculum preparation to avoid unconscious biases.

Reporting summary

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

Data availability

Raw sequence data (16S rRNA gene fragment sequencing) generated in this study have been deposited in the Genome Sequence Archive of the BIG Data Center⁷⁸, Chinese Academy of Sciences under accession PRJCA016320. The phyllosphere metagenomes of 110 rice genotypes were deposited in the European Nucleotide Archive (ENA) database under the project PRJEB45634. The 4-HCA-associated pathway is available under the pathway map “dosa00940” in the KEGG Pathway database [https://www.kegg.jp/pathway/dosa00940]. Taxonomic classifications of shotgun-sequenced metagenomes used in this study are available in the MiniKraken2_v1_8GB database [https://ccb.jhu.edu/software/bracken/]. Taxonomic classifications of 16S rRNA gene fragments used in this study are available in the Silva v138.1 reference database [https://www.arb-silva.de/documentation/release-1381/]. Rice gene annotations used in this study are available in the Rice Annotation Project Database (RAPDB) [https://rapdb.dna.affrc.go.jp/]. GO function annotations used in this study are available at Github [https://github.com/xuzhougeng/org.Osativa.eg.db]. All isolated bacterial strains and transformed rice lines were deposited in the State Key Laboratory of Hybrid Rice and the Institute of Plant Protection of Hunan Academy of Agricultural Sciences (Changsha, China). Other data generated in this study are provided in the Supplementary Data files. Source data are provided with this paper.

Code availability

Scripts employed in the computational analyses are available at Github [https://github.com/kanghouxiang105/micro-GWAS] and Zenodo [https://zenodo.org/records/10115039]⁷⁹.

References

Hassani, M. A., Durán, P. & Hacquard, S. Microbial interactions within the plant holobiont. Microbiome 6, 1–17 (2018).
Article Google Scholar
Durán, P. et al. Microbial interkingdom interactions in roots promote Arabidopsis survival. Cell 175, 973–983.e914 (2018).
Article PubMed PubMed Central Google Scholar
Kwak, M.-J. et al. Rhizosphere microbiome structure alters to enable wilt resistance in tomato. Nat. Biotechnol. 36, 1100–1109 (2018).
Article CAS Google Scholar
Carrión, V. J. et al. Pathogen-induced activation of disease-suppressive functions in the endophytic root microbiome. Science 366, 606–612 (2019).
Article ADS PubMed Google Scholar
Ritpitakphong, U. et al. The microbiome of the leaf surface of Arabidopsis protects against a fungal pathogen. N. Phytol. 210, 1033–1043 (2016).
Article CAS Google Scholar
Lau, J. A. & Lennon, J. T. Rapid responses of soil microorganisms improve plant fitness in novel environments. Proc. Natl Acad. Sci. USA 109, 14058–14062 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Castrillo, G. et al. Root microbiota drive direct integration of phosphate stress and immunity. Nature 543, 513–518 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Bakker, P. A., Pieterse, C. M., de Jonge, R. & Berendsen, R. L. The soil-borne legacy. Cell 172, 1178–1180 (2018).
Article CAS PubMed Google Scholar
Zhang, J. et al. NRT1. 1B is associated with root microbiota composition and nitrogen use in field-grown rice. Nat. Biotechnol. 37, 676–684 (2019).
Article CAS PubMed Google Scholar
Berendsen, R. L., Pieterse, C. M. & Bakker, P. A. The rhizosphere microbiome and plant health. Trends Plant Sci. 17, 478–486 (2012).
Article CAS PubMed Google Scholar
Edwards, J. et al. Structure, variation, and assembly of the root-associated microbiomes of rice. Proc. Natl Acad. Sci. USA 112, E911–E920 (2015).
Article CAS PubMed PubMed Central 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
Deng, S. et al. Genome wide association study reveals plant loci controlling heritability of the rhizosphere microbiome. ISME J. 15, 3181–3194 (2021).
Article 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
Escudero-Martinez, C. et al. Identifying plant genes shaping microbiota composition in the barley rhizosphere. Nat. Commun. 13, 3443 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Oyserman, B. O. et al. Disentangling the genetic basis of rhizosphere microbiome assembly in tomato. Nat. Commun. 13, 3228 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Liu, H., Brettell, L. E., Qiu, Z. & Singh, B. K. Microbiome-mediated stress resistance in plants. Trends Plant Sci. 25, 733–743 (2020).
Article CAS PubMed Google Scholar
Vorholt, J. A. Microbial life in the phyllosphere. Nat. Rev. Microbiol. 10, 828–840 (2012).
Article CAS PubMed Google Scholar
Horton, M. W. et al. Genome-wide association study of Arabidopsis thaliana leaf microbial community. Nat. Commun. 5, 5320 (2014).
Article ADS PubMed Google Scholar
Bodenhausen, N., Bortfeld-Miller, M., Ackermann, M. & Vorholt, J. A. A synthetic community approach reveals plant genotypes affecting the phyllosphere microbiota. PLoS Genet. 10, e1004283 (2014).
Article PubMed PubMed Central Google Scholar
Brachi, B. et al. Plant genetic effects on microbial hubs impact host fitness in repeated field trials. Proc. Natl Acad. Sci. USA 119, e2201285119 (2022).
Article CAS PubMed PubMed Central Google Scholar
Wagner, M. R. et al. Host genotype and age shape the leaf and root microbiomes of a wild perennial plant. Nat. Commun. 7, 12151 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Sapkota, R., Knorr, K., Jørgensen, L. N., O’Hanlon, K. A. & Nicolaisen, M. Host genotype is an important determinant of the cereal phyllosphere mycobiome. N. Phytol. 207, 1134–1144 (2015).
Article CAS Google Scholar
Gong, T. & Xin, X. F. Phyllosphere microbiota: Community dynamics and its interaction with plant hosts. J. Integr. Plant. Biol. 63, 297–304 (2021).
Article PubMed Google Scholar
Zhan, C., Matsumoto, H., Liu, Y. & Wang, M. Pathways to engineering the phyllosphere microbiome for sustainable crop production. Nat. Food 3, 997–1004 (2022).
Article PubMed Google Scholar
Finkel, O. M. et al. The effects of soil phosphorus content on plant microbiota are driven by the plant phosphate starvation response. PLoS Biol. 17, e3000534 (2019).
Article CAS PubMed PubMed Central Google Scholar
Chen, T. et al. A plant genetic network for preventing dysbiosis in the phyllosphere. Nature 580, 653–657 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Beilsmith, K. et al. Genome‐wide association studies on the phyllosphere microbiome: embracing complexity in host–microbe interactions. Plant J. 97, 164–181 (2019).
Article CAS PubMed Google Scholar
Su, P. et al. Recovery of metagenome-assembled genomes from the phyllosphere of 110 rice genotypes. Sci. Data 9, 254 (2022).
Article MathSciNet CAS PubMed PubMed Central Google Scholar
Lu, J. et al. Metagenome analysis using the Kraken software suite. Nat. Protoc. 17, 2815–2839 (2022).
Article CAS PubMed PubMed Central Google Scholar
Zhang, J. et al. High-throughput cultivation and identification of bacteria from the plant root microbiota. Nat. Protoc. 16, 988–1012 (2021).
Article CAS PubMed Google Scholar
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
Sohrabi, R., Paasch, B. C., Liber, J. A. & He, S. Y. Phyllosphere microbiome. Annu. Rev. Plant. Biol. 74, 539–568 (2023).
Article CAS PubMed Google Scholar
Schäfer, M., Vogel, C. M., Bortfeld-Miller, M., Mittelviefhaus, M. & Vorholt, J. A. Mapping phyllosphere microbiota interactions in planta to establish genotype–phenotype relationships. Nat. Microbiol. 7, 856–867 (2022).
Article PubMed PubMed Central Google Scholar
Helfrich, E. J. et al. Bipartite interactions, antibiotic production and biosynthetic potential of the Arabidopsis leaf microbiome. Nat. Microbiol. 3, 909–919 (2018).
Article CAS PubMed PubMed Central Google Scholar
Carlström, C. I. et al. Synthetic microbiota reveal priority effects and keystone strains in the Arabidopsis phyllosphere. Nat. Ecol. Evol. 3, 1445–1454 (2019).
Article PubMed PubMed Central Google Scholar
Song, Y. et al. FERONIA restricts Pseudomonas in the rhizosphere microbiome via regulation of reactive oxygen species. Nat. Plants 7, 644–654 (2021).
Article CAS PubMed Google Scholar
Ma, K.-W. et al. Coordination of microbe–host homeostasis by crosstalk with plant innate immunity. Nat. Plants 7, 814–825 (2021).
Article CAS PubMed PubMed Central Google Scholar
Eichmann, R., Richards, L. & Schäfer, P. Hormones as go‐betweens in plant microbiome assembly. Plant J. 105, 518–541 (2021).
Article CAS PubMed PubMed Central Google Scholar
Hacquard, S., Spaepen, S., Garrido-Oter, R. & Schulze-Lefert, P. Interplay between innate immunity and the plant microbiota. Annu. Rev. Phytopathol. 55, 565–589 (2017).
Article CAS PubMed Google Scholar
Pfeilmeier, S. et al. The plant NADPH oxidase RBOHD is required for microbiota homeostasis in leaves. Nat. Microbiol. 6, 852–864 (2021).
Article CAS PubMed PubMed Central Google Scholar
Xu, P. et al. Temporal metabolite responsiveness of microbiota in the tea plant phyllosphere promotes continuous suppression of fungal pathogens. J. Adv. Res. 39, 49–60 (2022).
Article PubMed Google Scholar
Maier, B. A. et al. A general non-self response as part of plant immunity. Nat. Plants 7, 696–705 (2021).
Article CAS PubMed PubMed Central Google Scholar
Wang, Y. et al. GWAS, MWAS and mGWAS provide insights into precision agriculture based on genotype-dependent microbial effects in foxtail millet. Nat. Commun. 13, 5913 (2022).
Article ADS CAS PubMed Google Scholar
Beckers, B. et al. Lignin engineering in field-grown poplar trees affects the endosphere bacterial microbiome. Proc. Natl Acad. Sci. USA 113, 2312–2317 (2016).
Article ADS CAS PubMed Google Scholar
Shalev, O., Ashkenazy, H., Neumann, M. & Weigel, D. Commensal Pseudomonas protect Arabidopsis thaliana from a coexisting pathogen via multiple lineage-dependent mechanisms. ISME J. 16, 1235–1244 (2022).
Article CAS PubMed Google Scholar
Levy, A. et al. Genomic features of bacterial adaptation to plants. Nat. Genet. 50, 138–150 (2018).
Article CAS Google Scholar
Vogel, C. M., Potthoff, D. B., Schäfer, M., Barandun, N. & Vorholt, J. A. Protective role of the Arabidopsis leaf microbiota against a bacterial pathogen. Nat. Microbiol. 6, 1537–1548 (2021).
Article CAS PubMed PubMed Central Google Scholar
Wang, N. R. et al. Commensal Pseudomonas fluorescens strains protect Arabidopsis from closely related Pseudomonas pathogens in a colonization-dependent manner. mBio 13, e02892–02821 (2022).
Article CAS PubMed Central Google Scholar
Tao, C. et al. Bio-organic fertilizers stimulate indigenous soil Pseudomonas populations to enhance plant disease suppression. Microbiome 8, 1–14 (2020).
Article MathSciNet Google Scholar
Shalev, O. et al. Commensal Pseudomonas strains facilitate protective response against pathogens in the host plant. Nat. Ecol. Evol. 6, 383–396 (2022).
Article PubMed PubMed Central Google Scholar
Sarniguet, A., Kraus, J., Henkels, M. D., Muehlchen, A. M. & Loper, J. E. The sigma factor sigma s affects antibiotic production and biological control activity of Pseudomonas fluorescens Pf-5. Proc. Natl Acad. Sci. USA. 92, 12255–12259 (1995).
Article ADS CAS PubMed PubMed Central Google Scholar
Beskrovnaya, P. et al. Comparative genomics identified a genetic locus in plant-associated Pseudomonas spp. that is necessary for induced systemic susceptibility. mBio 11, e00575–00520 (2020).
PubMed PubMed Central Google Scholar
Voges, M. J., Bai, Y., Schulze-Lefert, P. & Sattely, E. S. Plant-derived coumarins shape the composition of an Arabidopsis synthetic root microbiome. Proc. Natl Acad. Sci. USA. 116, 12558–12565 (2019).
Article ADS PubMed PubMed Central Google Scholar
Koprivova, A. et al. Root-specific camalexin biosynthesis controls the plant growth-promoting effects of multiple bacterial strains. Proc. Natl Acad. Sci. USA. 116, 15735–15744 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Cernava, T. & Berg, G. The emergence of disease‐preventing bacteria within the plant microbiota. Environ. Microbiol. 24, 3259–3263 (2022).
Article PubMed PubMed Central 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).
Article CAS PubMed PubMed Central Google Scholar
Hu, L. et al. Root exudate metabolites drive plant-soil feedbacks on growth and defense by shaping the rhizosphere microbiota. Nat. Commun. 9, 2738 (2018).
Article ADS PubMed PubMed Central Google Scholar
McCouch, S. R. et al. Open access resources for genome-wide association mapping in rice. Nat. Commun. 7, 10532 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Paulson, J. N., Stine, O. C., Bravo, H. C. & Pop, M. Differential abundance analysis for microbial marker-gene surveys. Nat. Methods 10, 1200–1202 (2013).
Article CAS PubMed PubMed Central Google Scholar
Oksanen, J. et al. Vegan: Community Ecology Package. R package version 1.17-4. http://CRAN.R-project.org/package=vegan (2010).
Asnicar, F., Weingart, G., Tickle, T. L., Huttenhower, C. & Segata, N. Compact graphical representation of phylogenetic data and metadata with GraPhlAn. PeerJ 3, e1029 (2015).
Article PubMed PubMed Central Google Scholar
Wickham, H. ggplot2: Elegant Graphics For Data Analysis (Springer, 2016).
Kang, H. et al. Dissection of the genetic architecture of rice resistance to the blast fungus Magnaporthe oryzae. Mol. Plant Pathol. 17, 959–972 (2016).
Article CAS PubMed PubMed Central Google Scholar
Gu, Z., Gu, L., Eils, R., Schlesner, M. & Brors, B. Circlize implements and enhances circular visualization in R. Bioinformatics 30, 2811–2812 (2014).
Article CAS PubMed Google Scholar
Wu, T. et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innov. (Camb.) 2, 100141 (2021).
CAS Google Scholar
Ma, X. et al. A robust CRISPR/Cas9 system for convenient, high-efficiency multiplex genome editing in monocot and dicot plants. Mol. Plant. 8, 1274–1284 (2015).
Article CAS PubMed Google Scholar
Dai, S. et al. Transgenic rice plants that overexpress transcription factors RF2a and RF2b are tolerant to rice tungro virus replication and disease. Proc. Natl Acad. Sci. USA 105, 21012–21016 (2008).
Article ADS CAS PubMed PubMed Central Google Scholar
Lin, Q. et al. Prime genome editing in rice and wheat. Nat. Biotechnol. 38, 582–585 (2020).
Article CAS PubMed 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
Martin, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet. J. 17, 10–12 (2011).
Article 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
Quast, C. et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 41, D590–D596 (2012).
Article PubMed PubMed Central Google Scholar
French, W. R. et al. Wavelet-based peak detection and a new charge inference procedure for MS/MS implemented in ProteoWizard’s msConvert. J. Proteome Res. 14, 1299–1307 (2015).
Article CAS PubMed Google Scholar
Benton, H. P., Want, E. J. & Ebbels, T. M. Correction of mass calibration gaps in liquid chromatography–mass spectrometry metabolomics data. Bioinformatics 26, 2488–2489 (2010).
Article CAS PubMed Google Scholar
Kuhl, C., Tautenhahn, R., Bottcher, C., Larson, T. R. & Neumann, S. CAMERA: an integrated strategy for compound spectra extraction and annotation of liquid chromatography/mass spectrometry data sets. Anal. Chem. 84, 283–289 (2012).
Article CAS PubMed Google Scholar
Gu, Z. Complex heatmap visualization. iMeta 1, e43 (2022).
Article Google Scholar
BIG Data Center Members. Database resources of the BIG data center in 2018. Nucleic Acids Res. 46, D14–D20 (2018).
Article Google Scholar
Su, P. et al. Microbiome homeostasis on rice leaves is regulated by a precursor molecule of lignin biosynthesis. https://zenodo.org/records/10115039 (2023).

Download references

Acknowledgements

P.S. and Y.L. were supported by a grant from the National Key R&D Program of China (2022YFD1400700); H.K. was supported by grants from the National Key R&D Program of China (2021YFC2600400) and the National Natural Science Foundation (32261143468); P.S. was additionally supported by a grant from the Key R&D Program of Hunan Province (2022NK2014); D.Z. was supported by a grant from the LONGPING HIGH-TECH INDUSTRY ZONE (LongPing committee issue 2022-18). We appreciate the support of Bin Liu and Junliang Zhao from the Rice Research Institute, Guangdong Academy of Agricultural Sciences, China for providing the RDP-II seeds used in this study. We also want to thank OE Biotech Co., Ltd (Shanghai, China) for their technical support during high-throughput sequencing experiments.

Author information

These authors contributed equally: Pin Su, Houxiang Kang, Qianze Peng.

Authors and Affiliations

State Key Laboratory of Hybrid Rice and Institute of Plant Protection, Hunan Academy of Agricultural Sciences, Changsha, 410125, China
Pin Su, Deyong Zhang & Yong Liu
State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, 100193, China
Houxiang Kang
National Center of Technology Innovation for Saline-Alkali Tolerant Rice in Sanya City, Sanya, 572024, China
Qianze Peng & Deyong Zhang
College of Tropical Crops, Hainan University, Haikou, 570228, China
Qianze Peng & Deyong Zhang
Institute of Environmental Biotechnology, Graz University of Technology, Graz, 8010, Austria
Wisnu Adi Wicaksono, Gabriele Berg & Tomislav Cernava
Leibniz Institute for Agricultural Engineering and Bioeconomy (ATB), Potsdam, 14469, Germany
Gabriele Berg
Institute for Biochemistry and Biology, University of Potsdam, Potsdam, 14476, Germany
Gabriele Berg
Longping Branch, College of Biology, Hunan University, Changsha, 410082, China
Zhuoxin Liu & Jiejia Ma
School of Biological Sciences, Faculty of Environmental and Life Sciences, University of Southampton, Southampton, SO17 1BJ, UK
Tomislav Cernava

Authors

Pin Su
View author publications
You can also search for this author in PubMed Google Scholar
Houxiang Kang
View author publications
You can also search for this author in PubMed Google Scholar
Qianze Peng
View author publications
You can also search for this author in PubMed Google Scholar
Wisnu Adi Wicaksono
View author publications
You can also search for this author in PubMed Google Scholar
Gabriele Berg
View author publications
You can also search for this author in PubMed Google Scholar
Zhuoxin Liu
View author publications
You can also search for this author in PubMed Google Scholar
Jiejia Ma
View author publications
You can also search for this author in PubMed Google Scholar
Deyong Zhang
View author publications
You can also search for this author in PubMed Google Scholar
Tomislav Cernava
View author publications
You can also search for this author in PubMed Google Scholar
Yong Liu
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

T.C., D.Z., Y.L., and P.S. conceived the idea for the study. P.S. and Q.P. performed the bioinformatic analyses and visualized the data. H.K. conducted the GWAS analyses. T.C., P.S., Q.P., and H.K. analyzed the data. W.A.W. conducted formal data analysis and review. G.B. conducted review. P.S., Q.P., Z.L., and J.M. conducted the bacterial strain isolation, rice mutant construction and the microbiological experiments. The manuscript was written by T.C., P.S., Q.P., and H.K. All authors read and approved the final manuscript.

Corresponding authors

Correspondence to Deyong Zhang, Tomislav Cernava or Yong Liu.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Davide Bulgarelli and the other, anonymous, reviewer(s) 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

Peer Review File

Description of Additional Supplementary Files

Supplementary Data 1

Supplementary Data 2

Supplementary Data 3

Supplementary Data 4

Supplementary Data 5

Supplementary Data 6

Supplementary Data 7

Supplementary Data 8

Reporting Summary

Source data

Source Data

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 license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license 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 license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Su, P., Kang, H., Peng, Q. et al. Microbiome homeostasis on rice leaves is regulated by a precursor molecule of lignin biosynthesis. Nat Commun 15, 23 (2024). https://doi.org/10.1038/s41467-023-44335-3

Download citation

Received: 21 June 2023
Accepted: 08 December 2023
Published: 02 January 2024
DOI: https://doi.org/10.1038/s41467-023-44335-3

This article is cited by

Disease resistance through M genes
- Chengfang Zhan
- Mengcen Wang
Nature Plants (2024)
Microbiome homeostasis on rice leaves is regulated by a precursor molecule of lignin biosynthesis
- Pin Su
- Houxiang Kang
- Yong Liu
Nature Communications (2024)
Genetic dissection of ten photosynthesis-related traits based on InDel- and SNP-GWAS in soybean
- Dezhou Hu
- Yajun Zhao
- Deyue Yu
Theoretical and Applied Genetics (2024)

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.