Typical brown lesions occurred on the FLSs after 5 days of inoculation in the pot experiment, and the mock inoculation treatment had no symptoms of infection. Representative FLSs for DNA extraction of endophytes were then collected. The surface sterilization of FLSs and the extraction of endophytes were carried out as suggested in a previous study (Araujo et al., 2002; Ruiz-Pérez and Zambrano, 2017). Briefly, (1) each FLS was washed with 70% alcohol for 1 min; (2) the FLS was washed with 2% NaClO for 3 min; (3) the FLS was washed with 70% alcohol for another 30 s; (4) the FLS was washed with sterile water for 30 s, 3–4 times in total; (5) to ensure the success of the sterilization process, 100 μl of the water used in the fourth step was spread on different agar plates; (6) the agar medium was incubated for 2 weeks in the dark in an incubator at 25°C; and (7) the water (described in step 4) was used for the first PCR labeling of the 16S rRNA gene or ITS region to control DNA-removal success. Surface-sterilized plant material was ground in liquid nitrogen using a sterile mortar and pestle, and the powdered plant material was stored at −20°C for DNA extraction. DNA was extracted using a soil DNA kit, which was extracted according to the manufacturer’s instructions, and the concentration and purity of DNA were detected by a NanoDrop One.

Using genomic DNA as a template, according to the selection of sequencing regions, PCR amplification was performed using primers with barcodes and Premix Taq (TaKaRa). Sixteen SV4 primers (515F and 806R) were used to identify bacterial and archaeal diversity; ITS1 primers (ITS5-1737F and ITS2-2043R) were used to identify fungal diversity. The PCR system is shown in Table 1.

GeneTools Analysis software (Version 4.03.05.0, SynGene) was used to compare the concentrations of PCR products, calculate the required volume of each sample according to the principle of equal mass, and mix the PCR products. An EZNA^® Gel Extraction Kit was used to recover PCR mixed products, and TE buffer elution was used to recover target DNA fragments. Library construction was performed according to the standard procedure of the NEBNext^® Ultra^™ DNA Library Prep Kit for Illumina^®. The constructed amplicon library was subjected to PE250 sequencing using the Illumina Hiseq2500 platform. Data analysis was performed according to the method suggested by Leff et al. (2015).

Typical symptoms of sheath rot appeared on FLSs 8 days after S. oryzae inoculation in the hydroponic experiment. At this time, fresh samples were collected from the shoots and roots of the treatment inoculated with S. oryzae and the mock treatment. The fresh samples used for comparing the sensitivity of WT and aos mutants to potassium deficiency were collected at 45 days after commencing K-starvation treatment. The collected samples were snap-frozen in liquid nitrogen and stored in a − 80°C ultralow temperature freezer. The internal standard 10-dihydro-JA (DHJA; OlChemIm) for JA measurement was purchased from Sigma, and the content of JA was determined according to the method described by Huang et al. (2021). The ultrahigh liquid chromatography-electrospray ionization tandem mass spectrometry method was used for the measurement of total JA content. The dry samples of shoots and roots were collected and dried in an oven at 65°C and then ground and passed through a 0.1 mm sieve. These dry samples were used for the determination of nutrients. Additionally, the relative contents of oleic acid and linoleic acid were measured through metabolome profiling as described in our previous study (Zhang et al., 2021b).

Shoot and root samples used for measurements of elemental contents were collected in the hydroponic experiment 8 days after S. oryzae infection. The inductively coupled plasma-mass spectrometry (ICP-MS) method suggested by McLoughlin et al. (2018) was used for the measurements of elemental contents. Briefly, 0.15 g of dry shoot and root samples were digested in a mixture of HNO₃ and HClO₄ (4 HNO₃:1 HClO₄), after which the digested solution was dissolved in 100 ml of ultra-pure water. Double filter paper was used to filter the solution, which was then used for ICP-MS profiling. Calibration curves were built based on a multielement standard (Ultra Scientific, Providence, RI, United States).

The measured parameters were analyzed by descriptive statistical analyses. Two-way analysis of variance was performed to reveal statistically significant differences among different treatments. Data analyses were performed in SPSS 19.0 (SPSS, Inc., Chicago, IL, United States). Graphics were created using Origin 9.0 software (OriginLab Corporation, Northampton, MA, United States). Community composition differences were assessed using permutation multivariate analysis of variance (PERMANOVA) with the vegan package in R 3.5.1.

Potassium (K) deficiency promoted the infection of S. oryzae, which reflected that the largest length of the lesions occurred in the −K + I treatment (Figure 1A). Additionally, S. oryzae infection significantly altered the endophyte community structure in leaf sheaths. Principal component analysis (PCA) of the endophyte community showed that infection with S. oryzae significantly increased the differences in the endophyte community between K-sufficient and K-deficient rice (Figure 1B). Compared with the heathy plants, S. oryzae infection increased the relative abundance of Ascomycota by an average of 12.5%. Under the condition of S. oryzae inoculation, compared with the +K + I treatment, K deficiency increased the relative abundance of Ascomycota in leaf sheaths by 52.3% and reduced the relative abundance of Basidiomycota by 26.5% (Figure 2A). Similarly, S. oryzae infection also altered the bacterial communities in the flag leaf sheath (FLS). Compared with the −K + I treatment, the K supply significantly increased the relative abundance of Proteobacteria by 63.4%, and the relative abundance of Cyanobacteria was obviously decreased (Figure 2B). K deficiency profoundly affected the alpha diversity of the fungal community in the FLS during S. oryzae infection. Compared with the heathy plant, S. oryzae infection decreased the Shannon and Simpson indexes of fungi community by an average of 16.1 and 13.9%, respectively. Compared with the +K + I treatment, K-deficient rice infected with S. oryzae significantly decreased the Shannon and Simpson indexes of fungi community by 27.7 and 25.0%, respectively (Table 2).

S. oryzae infection increased the differences in fungal and bacterial species at the genus level between K-deficient and K-sufficient rice. When compared with the +K + I treatment, K deficiency significantly increased the relative abundance of Coniosporium, Chaetomium, Didymella, and Cephalotheca, and all the other fungal species were decreased (Figure 3A). Compared with +K treatment, K-sufficient rice infected with S. oryzae significantly increased the abundance of Burkholderia, Stenotrophomonas, and Allorhizobium (Figure 3B). However, K deficiency profoundly decreased the alpha diversity of the bacterial community in the FLS during S. oryzae infection. The Chao1 index was used to reflect the diversity of the endophyte microbial community after rice leaf sheaths were inoculated with S. oryzae. Compared with the +K + I treatment, K-deficient rice infected with S. oryzae significantly decreased the Chao 1 index by 16.7% (Table 2).

S. oryzae infection significantly affected oleic acid and linoleic acid metabolic processes in FLS. Compared with the heathy plant, S. oryzae infection increased the relative content of oleic acid and linoleic acid by an average of 127.4 and 326.3%, respectively. Compared with +K + I treatment, K-deficient rice infected with S. oryzae increased the oleic acid and linoleic acid content by 82.0 and 261.5%, respectively (Figure 4). Oleic acid and linoleic acid are the synthetic precursors of jasmonic acid (JA), which is initially synthesized from linoleic acid and is localized in chloroplasts. S. oryzae infection decreased the JA content of leaves, FLSs, and roots by averages of 14.3, 24.1, and 12.7%, respectively. Notably, K deficiency amplifies the effect of S. oryzae infection on JA content. Compared with −K-treated rice, K-deficient rice infected with S. oryzae decreased the JA contents of leaves, FLSs, and roots by 22.5, 24.9, and 17.1%, respectively, while the corresponding values of +K + I treatment were decreased by 5.5, 23.1, and 6.6%, respectively (Figure 5).

Allene oxide synthase (aos) is the key enzyme that controls the synthesis of JA, and aos mutants show significantly lower JA content than the wild type (WT), especially under the condition of K deficiency. Additionally, aos mutants are more sensitive to potassium deficiency. The root length and the root dry weight of aos1 and aos3 were obviously lower than those of WT under the condition of K deficiency (Figure 6). aos gene mutation also inhibited K⁺ uptake. Compared with WT, the K⁺ content of aos1 and aos3 decreased by averages of 10.7 and 17.7%, respectively. ICP-MS was used to analyze the changes in the ionomes of roots and shoots after S. oryzae infection (Figure 7). The results demonstrated that S. oryzae infection resulted in a significant decrease in nutrient uptake. S. oryzae infection significantly reduced the contents of K, Si, S, Mo, and Fe in the shoots. In contrast, the infection increased the content of most nutrients in the root parts, such as Ca, Fe, Mg, and Cu. Compared with the K-sufficient treatment (+K), K deficiency resulted in lower S, K, and Mn contents in roots under both diseased and healthy conditions. Likewise, K deficiency decreased the contents of Mn, K, Si, and S in the shoots of diseased plants, while the contents of these nutrients exhibited no significant differences in healthy plants.

The causal agent of sheath rot disease was first identified as Acrocylindrium oryzae in Taiwan in 1922 (Mew and Gonzales, 2002). Acrocylindrium oryzae infection induces the typical symptoms of leaf sheath rot and causes server yield loss. In 1975, researchers formally established the genus Sarocladium, which covers potential pathogens, such as phytopathogens, mycoplasmas, and saprophytes (Gams and Hawksworth, 1975; Giraldo et al., 2015). According to the classification, S. oryzae belongs to Hypomycetes and Ascomycota. Previous studies have demonstrated that phytotoxins such as cerulenin produced by S. oryzae can inhibit the development of other fungi (Gnanamanickam and Mew, 1991). Consistent with our result, S. oryzae infection significantly decreased the abundance of other endophytic fungi in FLSs (Figure 2; Tables 2 and 3). Additionally, helvolic acid secreted during S. oryzae infection shows strong antibacterial activity against Gram-positive bacteria (Tschen et al., 1997). A similar phenomenon was observed in the present study: the abundance of endophytic bacterial communities in leaf sheaths obviously decreased after S. oryzae infection, especially in K-deficient rice (Figure 3). During the infection process, S. oryzae will interact with other pathogenic microorganisms, such as the synergistic effect of pathogenic bacteria and other mites, which results in sterile grains (Karmakar, 2008; Hummel et al., 2009). Moreover, bacterial sheath brown rot caused by Gluconococcus rhizogenes may also complicate S. oryzae infection, which ultimately results in leaf sheath rot (Venkataraman et al., 1987). All these results demonstrate that the infection process of pathogens that cause leaf sheath rot is complex. Our results further confirm that S. oryzae is the major pathogenic fungi of sheath rot disease, and it also interacts with other endophytes. After S. oryzae successfully invaded the leaf sheath, the pathogens inhibited the growth of other endophytic fungi and their ability to obtain ecological niches, which was conducive to its rapid colonization and finally causes sever yield loss.

S. oryzae infection alters the endophyte community of the leaf sheath predominantly by secreting phytotoxin and altering the metabolism process of host plant, while the K level of the leaf sheath mediated those process. A previous study showed that K deficiency increased the competition capacity of pathogens for nutrients (Holzmueller et al., 2007). In the present study, the relative abundance of Ascomycota was sharply increased during the infection; therefore, S. oryzae shows a strong competitive capacity for niches in the K-starved leaf sheaths. Notably, Pseudomonas spp. shows the potential to inhibit the infection of S. oryzae (Surya et al., 2019). In the present study, sufficient K supply significantly increased the abundance of Pseudomonas spp. during S. oryzae infection. Thus, it was speculated that Pseudomonas spp. might play an important role in K-sufficient rice against S. oryzae infection. However, the mechanism of antagonism between pathogenic bacteria and endophytes needs to be further studied. Because the growth rate of S. oryzae in K-deficient leaf sheaths was faster than that in K-sufficient rice, the development of S. oryzae preempted the niche for other fungi, which explains why the endophyte community abundance was lowest in the −K + I treatment (Tables 2 and 3). The decreased diversity of the community also indicated the instability of the community, this result supports our hypothesis mentioned in the introduction that K deficiency reduces endophyte community stability in diseased leaf sheaths, which was beneficial for the colonization and development of S. oryzae.

During the interaction between host plants and microorganisms, the host plants provide a stable environment, nutrients, and energy for the survival of microorganisms, while the microorganism directly or indirectly impacts the host plant through its own metabolites (Dreyfuss and Chapela, 1994). The colonization of endophytes will inevitably cause numerous responses of the host plant, including stimulation of plant defense responses, gene expression, and metabolism, among others (Pang et al., 2018). Our previous study also demonstrated that S. oryzae infection profoundly alters the metabolic process of host plants, especially the lipid metabolism process, the infection of which significantly increases the accumulation of lipid metabolism products (Zhang et al., 2021b). Similarly, based on our data, S. oryzae infection sharply increased the accumulation of oleic acid and linoleic acid, and the linoleic acid metabolism pathway was predominantly affected by S. oryzae infection (Figure 4; Supplementary Figure S1), which is consistent with a previous study showing that S. oryzae infection inhibited fatty acid synthesis by secreting cerulenin (Bigirimana et al., 2015). Because linoleic acid is the precursor for JA synthesis, its accumulation could suppress the JA synthesis process. However, JA is crucial for nutrient uptake, especially for K⁺ and boron, among others (Armengaud et al., 2004; Huang et al., 2021). Thus, the low JA level caused by S. oryzae infection (especially for K-starved rice) significantly suppressed the uptake of K⁺ (Figure 7), which might be due to the expression of K⁺ uptake-related genes was inhibited by low JA level (Supplementary Figure S2). Additionally, ionomics-detection technology has been used in the study of plant diseases and the interaction of plants and fungi. Ramos et al. (2009) used ion-selective electrode technology to explore the interaction between fungi and Arabidopsis thaliana and found that Ca²⁺ has a signaling role in fungi promoting ion uptake in plant roots. Li et al. (2012) used ion scanning electrode technology to confirm that ectomycorrhizal fungi regulated the Na⁺/K⁺ balance in poplar, which improved the salt tolerance of salt-sensitive poplars. K⁺ is the most abundant cation in higher plants and plays an important role in maintaining the electrochemical balance of the cytoplasm, catalyzing enzymatic reactions, regulating the osmotic balance of cells, and maintaining the turgor pressure of cells (de Bang et al., 2021). A previous study indicated that the fungal pathogen Magnaporthe oryzae suppresses the K⁺ absorption process (Shi et al., 2018). Similarly, in the present study, we found that S. oryzae infection not only significantly reduced the absorption capacity of K⁺ but also suppressed the uptake of other nutrients by the root system (Figure 7). This process might correlate with the inhibition of JA in the host plant. S. oryzae infection significantly decreased the JA content of leaves, FLSs, and roots in this study (Figure 5). Finally, low JA levels in the host plant significantly decreased the uptake of Mn, Si, S, Fe, and Ca (Figure 7). aos mutation significantly decreased the JA content of host plants, which makes rice more sensitive to K-deficiency stress (Li et al., 2017). In this study, we found that the aos mutation significantly suppressed the expression of K⁺ uptake channel genes such as OsAKT1 and OsHAK5 (Supplementary Figure S2). Under the condition of K deficiency, aos mutants exhibited the weakest K⁺ absorption capacity, which exacerbated the negative impacts of S. oryzae infection on K⁺ uptake by the host plant (Figure 7).