Characterization of the microbial communities in wheat tissues and rhizosphere soil caused by dwarf bunt of wheat

Dwarf bunt of wheat, which is caused by Tilletia controversa J.G. Kühn, is a soil-borne disease which may lead up to an 80% loss of yield together with degradation of the quality of the wheat flour by production of a fishy smell. In this study, high-throughput sequencing technology was employed to characterize the microbial composition of wheat tissues (roots, spikes, first stem under the ear, and stem base) and rhizosphere soil of wheat varieties that are resistant and susceptible to T. controversa. We observed that the soil fungal community abundance and diversity were higher in resistant varieties than in susceptible varieties in both inoculated and uninoculated wheat, and the abundances of Sordariomycetes and Mortierellomycetes increased in the resistant varieties infected with T. controversa, while the abundances of Dothideomycetes and Bacteroidia increased in the susceptible varieties. Regarding the bacteria present in wheat tissues, the abundances of Chloroflexi, Bacteroidetes, Gemmatimonadetes, Verrucomicrobia and Acidobacteria in the ear and the first stem under the ear were higher than those in other tissues. Our results indicated that the abundances of Sordariomycetes, Mortierellomycetes, Leotiomycetes, Chryseobacterium and Massilia were higher in T. controversa-infected resistant varieties than in their controls, that Dothideomycetes, Bacteroidia, Nocardioides and Pseudomonas showed higher abundances in T. controversa-infected susceptible varieties, and that Curtobacterium, Exiguobacterium, Planococcus, and Pantoea may have higher abundances in both T. controversa-infected susceptible and resistant varieties than in their own controls.


Results
Detection of T. controversa in wheat leaves. Based  Sequencing data statistics and evaluation. Through Illumina MiSeq sequencing, 1,709,173 optimized fungal sequences were obtained from 24 rhizosphere soil samples, and 1,532,304 effective sequences were obtained; a total of 1,944,678 optimized bacterial sequences were obtained from 84 samples, including 24 rhizosphere soil samples, 24 root samples, 12 spike samples, 12 stem base samples and 12 first stem under the ear samples, and 475,524 effective sequences were obtained. The dilution curve according to the Shannon index reached a plateau ( Supplementary Fig. S2), indicating that the quantity of sequencing data was sufficient for subsequent experiments. According to the similarity level, all sequences were divided into OTUs, and sequences with 97% or greater similarity to OTU representative sequences were selected for statistical analysis of biological information. A total of 1721 fungal OTUs and 3824 bacterial OTUs were obtained, and species classification analysis was performed (Supplementary Table S1).
Microbial diversity of rhizosphere soil. A comparison of richness and diversity indexes, which are the average of three samples, demonstrated the differences in microorganisms among soil samples. For the fungal community (Supplementary Table S2), the Shannon index indicated that the diversity of the resistant varieties Mianyang 26/Yumai 47 (CK26, 3.77; R26, 4.47) and Yumai 49 * 4/Lankao dwarf 8 (CK49, 3.53; R49, 4.33) was higher in T. controversa-inoculated wheat than in uninoculated wheat, and the diversity of A-45 (CK45, 3.43; S45, 3.56) was also higher in inoculated wheat; however, the diversity of A-44 (CK44, 2.64; S44, 2.62) decreased slightly. Regardless of whether inoculated wheat or uninoculated wheat was used, the diversity of resistant varieties was higher than that of susceptible varieties. In terms of species richness, the Ace and Chao indexes showed that the richness of the infected wheat Yumai 49 * 4/Lankao dwarf 8 (Ace: CK49, 820.46, R49, 899.84; Chao, CK49, 793.82, R49, 899.41) and A-45 (Ace: CK45, 344.68, S45, 769.54; Chao: CK45, 342.11, S45, 762.84) increased compared with the control, while the abundance of A-44 (Ace: CK44, 391.57, S44, 335.47; Chao: CK44, 364.00, S44, 323.75) was observed to decrease. The trends of the Ace and Chao index in Mianyang 26/Yumai 47 (Ace: CK26, 933.55, R26, 912.39; Chao: CK26, 854.66, R26, 901.53) were inconsistent. The fungal diversity of the resistant varieties inoculated with T. controversa was higher than that of uninoculated wheat, while the fungal communities of the susceptible varieties were notably diverse. We also found that among the uninoculated wheat varieties, the richness of fungi in the samples of Mianyang 26/Yumai 47 (Ace: 933.55; Chao: 854.66) and Yumai 49 * 4/Lankao dwarf 8 (Ace: 820.46; Chao: 793.82) were relatively high. In wheat inoculated with T. controversa, the richness of fungi in the samples of Mianyang 26/Yumai 47 and Yumai 49 * 4/Lankao dwarf 8 was also higher than that in the two susceptible varieties. These results indicate that regardless of whether plants were inoculated with T. controversa, the diversity and richness of the fungal community of the resistant varieties was higher than that of the susceptible varieties.
In Supplementary Table S3 Supplementary Fig. S3, we performed statistical analysis on the diversity index of the endophytic bacteria of Mianyang 26/Yumai 47 and A-45, and we observed that in the resistant variety Mianyang 26/Yumai 47, there were significant differences between the three diversity index (Shannon, Ace, and Chao) values of the ears and stem bases, and the Shannon index also showed significant difference between the ear and the first stems under the ear. However, there was no significant difference among the three diversity indexes of endophytic bacteria in the susceptible variety A-45. Therefore, the impact of T. controversa on wheat endophyte communities differs among different tissue components and plant varieties.
Microbial community of wheat rhizosphere soil. The fungal community in the soil ( Fig. 1  www.nature.com/scientificreports/ The bacterial community in the soil ( Fig. 2) was more diverse than the fungal community. As shown in Fig. 2A, Proteobacteria, Actinobacteria, Bacteroidetes, Firmicutes, Chloroflexi, Verrucomicrobia, Gammaproteobacteria and Acidobacteria were the main components of the microbial community. The abundance of Proteobacteria, Firmicutes and Bacteroidetes in inoculated A-44 was higher than that in plants with the other seven treatments, while Actinobacteria exhibited the lowest abundance, and no Chloroflexi was found. As shown in Fig. 2B, the dominant classes in all soil community samples were Actinobacteria, Gammaproteobacteria, Alphaproteobacteria, Bacteroidia, Deltaproteobacteria, Bacilli, Verrucomicrobiae, Gemmatimonadetes, TK10 and Chloroflexia. Compared with the control, the abundance of bacteria in the communities of inoculated resistant varieties Mianyang 26/Yumai 47 and Yumai 49 * 4/Lankao dwarf 8 did not exhibit notable changes. For A-44, the bacterial community in the soil of the inoculated plants changed compared to the control. The abundance of Actinobacteria and Alphaproteobacteria in the inoculated plants was lower than that in the uninoculated plants, and Gammaproteobacteria, Bacteroidia and Bacilli were much more abundant in the inoculated plants than in the uninoculated plants. In A-45, the abundance of Actinobacteria, Bacteroidia, Deltaproteobacteria, Gemmatimonadetes and TK10 was higher in inoculated plants, and the abundance of Gammaproteobacteria and Alphaproteobacteria was lower in inoculated plants, compared with uninoculated plants. The abundance of Actinobacteria decreased in infected resistant cultivars, while the abundance of Bacteroidia increased in infected susceptible cultivars. In Fig. 2C, a study of the differences between PCoA groups showed that uninoculated A-44 and A-45 formed a cluster and separated from other groups. At the same time, there were clear differences in the diversity of inoculated A-44. This finding shows that the diversity of the bacterial community in the rhizosphere soil of the normally susceptible wheat variety is different from that of the inoculated susceptible wheat variety, and it is also different from that of the normal resistant variety.
Composition and difference in entophytic bacteria in tissues of wheat. We observed that Proteobacteria, Firmicutes and Actinobacteria were the dominant bacteria in wheat tissue samples (Fig. 3). The abundance of Proteobacteria in the roots and ears of Mianyang 26/Yumai 47 was higher than that in controls, while it was decreased in the first stem under the ear and stem base compared with the controls (Fig. 3A). The  www.nature.com/scientificreports/ stem base had a higher abundance of Firmicutes than the control. Except for the first stem under the ear, the abundance of Actinobacteria in other tissues was decreased in the inoculated plants (Fig. 3A). The diversity of wheat roots, ears and the first stem under the ears was higher in the susceptible variety A-45 than in the resistant variety (Fig. 3B). The roots of the inoculated wheat exhibited the lowest abundance of Actinobacteria and the highest abundance of Firmicutes in all treatments. The abundances of Chloroflexi, Gemmatimonadetes, Verrucomicrobia and Acidobacteria in the ear and the first stem under the ear of A-45 were considerably higher than those in other organs (Fig. 3B). We applied principal coordinate analysis (PCoA) and Bray-Curtis metrics to compare the differences and similarities in the bacterial community diversity between the resistant variety Mianyang 26/Yumai 47 and the susceptible variety A-45 (Fig. 4). For the inoculated Mianyang 26/Yumai 47 plants, the first and second principal component analysis variables (PC1 was 59.91% and PC2 was 33.02%) accounted for 92.93% of the cumulative variance of all samples (Fig. 4A). There were significant differences in the microbial composition of the roots, spikes, first stem under the ear and base of the stem. Compared with the other treatments, the FS group had significant differences. For the uninoculated Mianyang 26/Yumai 47 plants, PC1 was 78.23% and PC2 was 15.16%, accounting for 93.39% of the cumulative variance of all samples (Fig. 4B). There were large differences between the groups, and the SB group yielded better results. There were also significant differences between groups in inoculated A-45 (Fig. 4C) and uninoculated A-45 (Fig. 4D). The results showed that regardless of whether resistant varieties or susceptible varieties were present, the microbial compositions of wheat roots, ears, first stem under the ear and base of the stems were different.

Network analysis. We established a coexpression network relationship between healthy wheat and diseased
wheat to visualize the relationship between samples and microorganisms. According to taxonomic annotations, the bacterial microorganisms in all groups were mainly composed of Proteobacteria, Bacteroidetes, Firmicutes and Actinobacteria (Fig. 6) Table S8). We also performed taxonomic annotations on fungi. The fungal microorganisms in all groups were mainly composed of Ascomycota, Basidiomycota, Glomeromycota and Mortierellomycota (Fig. 7). We found that there were more edges in the susceptible varieties than in the resistant varieties, indicating that the fungal interaction network in the susceptible varieties was more abundant (Supplementary Table S9). The NetShift method (https ://web.rniap ps.net/netsh ift) was further utilized to identify the key taxa in the network according to the network interaction differences between healthy wheat and inoculated wheat microbial communities 27,28 . Figure 8A shows that in resistant wheat, the potential bacterial flora related to the interaction of wheat and T. controversa primarily consisted of Massilia, Acidovorax, Pseudoxanthomonas, MND1, Paeniglutamicibacter, Altererythrobacter, Sphingomona, Acidibacter, Micrococcaceae, Erwinia, Curtobacterium, and Frigoribacterium, and our data did not capture the potential fungal flora. Figure 8B,C showed that in susceptible wheat, the important bacterial "drivers" in the interaction between wheat and T. controversa were MND1, Xanthomonadaceae, Altererythrobacter, Sphingomonas, Actinobacteria, Pantoea, Nocardioides, Devosia, Phyllobacterium, Erwinia, Micrococcaceae, Paeniglutamicibacter, Lechevalieria, Flavobacterium, and Massilia. The potential fungal "drivers" primarily consisted of Hypocreales, Monodictys, Ilyonectria, Fusarium, and Nectriaceae.

Discussion
In this study, the microbial communities of wheat varieties resistant and susceptible to dwarf bunt, including wheat tissues of roots, ears, first stem under the ear, and stem base, as well as the microbial community in rhizosphere soil, were analysed. The results may facilitate the control of dwarf bunt in wheat and may establish a foundation for future research exploring the interactions among plant, pathogen and microbial communities. communities, we observed that the abundance of Ascomycota and Basidiomycota detected in this study was higher than that of other fungi, and this phenomenon was also observed in strawberry 29 . Second, we determined that in both inoculated plants and control plants, the richness of fungal communities in resistant varieties was higher than that in susceptible varieties. A richer and more diverse community leads to increased competition  www.nature.com/scientificreports/ for resources between species, which has been described as a key factor determining whether pathogens can successfully invade plants 30 .
For the bacterial community in soil samples, Proteobacteria and Actinobacteria were most abundant, and this result was also reported in rice 31 , soybean 32 , and grape 33 . Many members of Proteobacteria can produce important substances that enable bacteria to be adsorbed onto the surfaces of plants and plant roots and promote interactions 34 . Composition and diversity of endophytic bacteria in wheat tissues. We detected bacteria in wheat tissues of roots, ears, first stem under the ear, and stem base. In our research, Proteobacteria, Firmicutes and Actinobacteria were the predominant bacteria in wheat, which was also reported by Robinson et al. 35 and Wemheuer et al. 36 . Proteobacteria and Firmicutes were the two most abundant phyla in healthy wheat and grasses, respectively. We also observed that there was no significant difference between the diversity and richness of susceptible varieties and resistant varieties. However, Xu et al. 37 reported that the diversity of the bacterial communities of resistant varieties was higher than that of susceptible varieties in mulberry trees. Based on the heatmap, the abundance of Exiguobacterium, Planococcus and Pantoea associated with susceptible varieties was lower than that in the other treatment groups in the root and stem base.
The results of this study showed that the microbial diversity of bacteria in the rhizosphere was higher than that in the roots which, as demonstrated by the root surface of plants, has a gating effect and is selective for the entry of bacteria 38 . In general, endophytic microbial communities are more specific than rhizosphere soil microbial communities because plants allow less adaptive bacteria to enter and survive in plants 39 . These endophytic microorganisms can interact with plants and have antagonistic effects on pathogens 40 .
Potential active microbial community in wheat tissues and rhizosphere soil. In this study, the abundances of some microbial populations in the susceptible and resistant varieties increased after T. controversa inoculation. For example, in the rhizosphere soil, after inoculation with T. controversa, the abundances of Sordariomycetes and Mortierellomycetes in resistant varieties increased compared with those in the control. In infected susceptible plants, the abundances of Dothideomycetes and Bacteroidia increased after inoculation. In the resistant and susceptible varieties, after inoculation, the abundance of Leotiomycetes increased.
The root is an important organ for the entry of soil microorganisms into the plant, and the ear is targeted by T. controversa after infection. Therefore, we focused on the changes in microorganisms in these two plant tissues to further characterize the role played by microbial communities in dwarf bunt-resistant wheat varieties. In wheat tissues, we found large differences in the microbial composition of different parts of wheat, indicating that the potential indicator flora was also different. In the wheat roots, we found that the abundances of bacteria in the two groups of resistant varieties were different after inoculation. However, the changes in bacterial populations in the two susceptible varieties were more consistent. In susceptible wheat infected by T. controversa, the microbial communities in the ear and the first section under the ear undergo similar changes. The abundances of Chloroflexi, Bacteroidetes, Gemmatimonadetes, Verrucomicrobia and Acidobacteria in the ear and the first section under the ear of susceptible plants were higher than those in other tissues and were higher than those in resistant plants. This result indicated that the bacteria of these phyla could be associated with the formation of wheat disease spikes. Further analysis of the changes in bacteria in wheat ears at the genus level demonstrated that in both resistant and susceptible varieties, the abundance of Pantoea and Curtobacterium was higher in the inoculated plants; therefore, these two genera could be useful indicators for T. controversa. Notably, the abundances of Chryseobacterium and Massilia in the spikes of the resistant varieties were higher than that of the uninoculated wheat, while the abundances of Nocardioides and Pseudomonas were higher in the susceptible varieties.
The co-occurrence network analysis of the microbial communities of plants and soil may be employed to strengthen disease management and identify candidate microorganisms that affect plant health. Based on NetShift analysis, we found some potential microorganisms related to wheat-T. controversa interactions. Most of these microorganisms or their classes have been discussed above to further characterize their effects, such as Massilia, Sphingomonas, Paeniglutamicibacter, Curtobacterium, Frigoribacterium, Pantoea, Nocardioides, Lechevalieria, Alphaproteobacteria, Actinobacteria, Bacteroidia, Sordariomycetes and Dothideomycetes.
In summary, Sordariomycetes, Mortierellomycetes, Leotiomycetes, Chryseobacterium and Massilia were determined to be more abundant in inoculated resistant varieties than in control plants (Figs. 2, 6), and Dothideomycetes, Bacteroidia, Nocardioides and Pseudomonas were observed to be more abundant in susceptible varieties after infection with T. controversa compared with the control (Figs. 2, 3, 6), and Curtobacterium, Exiguobacterium, Planococcus, and Pantoea exhibited higher abundances in both the susceptible and resistant varieties after inoculation than in the controls (Fig. 6).

Materials and methods
Source and inoculation of fungal materials. T. controversa was provided by Blair Goates, United States Department of Agriculture (USDA), Agricultural Research Service (ARS), Aberdeen, Idaho, USA. The soil mix (Klasmann-deilmann, Germany) was placed in an autoclave (ZEALWAY, GR110DA, China) and autoclaved at 121 °C for 30 min to kill microorganisms and insects. Seeds were seeded in a 20-cm-diameter pot with sterilized soil at a depth of approximately 3 cm, with 15 seeds in each pot. The seeds were cultured in an ultralow temperature biochemical incubator (Percival, ARC-36VL-lt, USA) for 24 h under full light with a light intensity of 300 mol/m 2 /s. After sowing, the soil surface was covered with teliospores at a density of 3 × 10 6 /m 2 . At the initial stage of seedlings, the seeds were allowed to germinate and grow to the one-leaf stage (Z11) 41 at 4 °C. When the seedlings grew to the jointing stage (Z31), the temperature in the biochemical incubator was adjusted to 18 °C, DNA extraction and PCR amplification. Total soil and plant tissue DNA were extracted using a FastDNA SPIN Soil kit (MP Biomedicals, Solon, USA). A NanoDrop 2000 UV-vis spectrophotometer (Thermo Scientific, Wilmington, USA) was utilized to determine the DNA concentration and purity, and 1% agarose gel electrophoresis was used to determine the DNA extraction quality. The primers used to amplify the hypervariable regions V5-V7 of the bacterial 16S rRNA gene were 799F (5′-AACMGGA TTA GAT ACC CKG-3′) and 1193R (5′-ACG TCA TCC CCA CCT TCC -3′). The primers used to amplify the internal transcribed spacer (ITS) region of ribosomal DNA were ITS1F (5′-CTT GGT CAT TTA GAG GAA GTAA-3′) and ITS2R (5′-GCT GCG TTC TTC ATC GAT GC-3′). The programme for PCR amplification was as follows: initial denaturation at 95 °C for 3 min; 27 cycles of denaturing at 95 °C for 30 s, annealing at 55 °C for 30 s and extension at 72 °C for 45 s; a single extension at 72 °C for 10 min; and a final extension at 4 °C. The reaction system included 4 μL 5 × FastPfu buffer, 2 μL 2.5 mM dNTPs, 0.8 μL forward primer (5 μM), 0.8 μL reverse primer (5 μM), 0.4 μL FastPfu Polymerase, 0.2 μL BSA, 10 ng template DNA, and ddH 2 O up to 20 μL. The reactions were performed in triplicate. The PCR product was extracted from a 2% agarose gel and purified with an AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA) based on the manufacturer's instructions and quantified using a Quantus Fluorometer (Promega, USA).
Illumina MiSeq sequencing and data analysis. Based on the standard protocols by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China), we used an Illumina MiSeq platform (Illumina, San Diego, USA) to pool purified amplicons in equimolar amounts and performed paired-end sequencing (2 × 300).
The original sequencing data were processed by Trimmomatic software and spliced by FLASH software: (i) the 300-bp reads were truncated at any site with an average quality score of < 20 over a 50-bp sliding window, and the truncated reads shorter than 50 bp were discarded. Reads containing ambiguous characters were also discarded; (ii) only overlapping sequences longer than 10 bp were assembled according to their overlapping sequence. The maximum mismatch ratio of the overlap region was 0.2. Reads that could not be assembled were discarded. (iii) Samples were distinguished according to the barcode and primers, and the sequence direction was adjusted for exact barcode matching and 2 nucleotide mismatches in primer matching.
Using UPARSE software (version 7.1 http://drive 5.com/upars e/), OTUs were clustered based on 97% similarity, and single sequences and chimaeras were removed during the clustering process. The RDP classifier (http:// rdp.cme.msu.edu/) was utilized to classify and annotate each sequence. The Silva database (SSU123) was utilized to compare 16S rRNA sequences, and Unite (Release 6.0 http://unite .ut.ee/index .php) was utilized to compare Table 1. Sample collection information. The samples for each variety of wheat were collected from both inoculated and uninoculated plants.