Background

Wheat and rice are two staple cereal crops that feed humans. However, their production is frequently affected by many fungal pathogens [1]. Among them, Fusarium graminearum and Magnaporthe oryzae are two of the most devastating plant pathogenic fungi, mainly causing Fusarium head blight (FHB) in wheat and blast disease in rice, respectively [2, 3]. FHB epidemics occur in many countries and cause a severe reduction in grain quantity and quality [4, 5]. F. graminearum infects the kernel of wheat, thus leading to a direct yield loss. In addition, the mycotoxins deoxynivalenol (DON) and zearalenone (ZEA) that reside in the wheat grains are dangerous to humans and animals, resulting in further damage to the quality of wheat products [4, 6]. Rice blast is also a global disease that threatens rice production [7]. In China, epidemics often cause yield losses of 40–50% or even up to 100% in severely infected areas [8]. Yield losses range from 50 to 85% in the Philippines [8]. It is estimated that the rice destroyed by this disease is sufficient to feed 60 million people, leading to an economic loss of $66 billion annually [9]. In addition, some lineages of M. oryzae are responsible for the recent wheat blast outbreaks in South America and Southeast Asia [10,11,12].

The discovery, utilization, and deployment of resistant cultivars is an effective approach to control these diseases [13, 14]. Resistant germplasm for FHB is rare in wheat, and only 155 quantitative trait loci (QTL) that contribute to FHB resistance have been identified [5, 15, 16]. Sources of resistant germplasm for rice blast are available, but they are frequently overcome due to the rapid evolution of new pathotypes of the pathogen [8, 17]. Therefore, it is essential to determine the mechanisms of pathogen infection and host defense. To understand host defense responses, many transcriptomic analyses have been conducted in plants infected with F. graminearum or M. oryzae [18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33]. For example, although a majority of wheat cultivars are susceptible to F. graminearum infection, the gene expression profiles in the cold-killed and living wheat heads are obviously distinct and suggest that the host defense responses actively suppress fungal growth during infection [18]. In line with this evidence, many transcriptomic studies showed that genes related to oxidative burst, MAPK signaling, hormone biosynthesis, transcription, secondary metabolism, and other defense associated proteins were altered upon infection [19,20,21,22,23,24,25]. These defense responses are also observed in M. oryzae infection [26,27,28,29,30,31,32] and many other plant-pathogen interactions [33].

To use the host defense mechanisms for resistance breeding, it is important to know which type of defenses were most responsive in the hosts and which were generally fired by the pathogens. However, our knowledge of these differences in host defenses is very limited. Considering wheat and rice have distinct habits, genome size, and gene number, a direct comparison of their differences in host defenses is difficult and may lead biased conclusion. Therefore, developing an efficient model for dissecting these differences is required. A model of cereal species, Brachypodium distachyon, is susceptible to many important cereal pathogens, including F. graminearum and M. oryzae, and serves as a model for studying cereal diseases [34]. Disease development of B. distachyon caused by F. graminearum and M. oryzae is highly similar to that on wheat [35] and rice [36, 37], respectively. Furthermore, transcriptional analyses showed that the expression of the defense-related genes is also similar to those in wheat [38] and rice [34, 36], respectively. In this study, we employed two different pathosystems, B. distachyon-F. graminearum and B. distachyon-M. oryzae, and performed comparative transcriptomics to study the dynamics of host gene expression upon infection. Our analysis showed that, although many similar defense responses have been observed in B. distachyon challenged by F. graminearum or M. oryzae, the expression programs and some specific defense responses are very different in these two pathosystems.

Results and discussion

Distinct global gene expression of Brachypodium infected by two cereal fungi

To investigate the transcriptional dynamics during Brachypodium-fungal interactions, we performed RNA-seq analysis on B. distachyon infected by two cereal pathogens: F. graminearum and M. oryzae. In nature, F. graminearum and M. oryzae mainly infect the spike and leaf of wheat and rice, respectively. We therefore employed the spike and leaf of B. distachyon to mimic their infections. Based on the lesion development (Fig. 1a-b) and infection progress [35,36,37], three stages (24 h, 48 h, and 72 h) of B. distachyon spikes infected by F. graminearum, and three stages (24 h, 48 h, and 96 h) of B. distachyon leaves infected by M. oryzae, were collected. Together with the mock of B. distachyon spike and leaf, eight samples with three biological replicates were sequenced in these two pathosystems (Table S1).

Fig. 1
figure 1

Disease development and differentially expressed genes of B. distachyon infected by F. graminearum or M. oryzae. (a) Disease development of B. distachyon spikes infected by F. graminearum. (b) Disease development of B. distachyon leaves infected by M. oryzae. (c) The relative fungal transcript levels during B. distachyon - fungal interaction. (d) Venn diagram showing the common and differential transcriptional responses of up-regulated (in red) and down-regulated (in green) genes of B. distachyon spikes infected by F. graminearum at different time points. (e) Venn diagram showing common and differential transcriptional responses of up-regulated (in red) and down-regulated (in green) genes of B. distachyon leaves infected by M. oryzae infection at different time points. (f-h) Venn diagram showing the number of up-regulated (red arrows) and down-regulated (green arrows) genes of B. distachyon inoculated with F. graminearum and M. oryzae, respectively, at 24 h (f), 48 h (g), and 72/96 h (h). The statistical significances of the overlapped genes were analyzed by hypergeometric tests. “**”, P < 0.01; “ns”, not significant

Full size image

When mapping to fungal genomes (Fig. 1c, Table S1), only a small number (less than 1%) of fungal mRNA reads were found in the early stages of infection (24–48 h). In the later stage of infection (72 and 96 h for F. graminearum and M. oryzae, respectively), substantial fungal mRNA reads were detected (3–10%). The gradual increase in fungal mRNA reads is highly consistent with the lesion development in B. distachyon. The dramatic increment of fungal mRNA reads in the later stages suggests that the fungi have absolutely broken plant defenses and switched to necrotrophic growth.

For each RNA-seq library, approximately 81.2% of the reads were mapped to the B. distachyon genome (Table S1). Among the 25,532 B. distachyon genes, 20,917 (78.8%) and 18,978 (71.5%) expressed genes were detected in spikes and leaves infected by F. graminearum and M. oryzae, respectively. Multi-dimensional scaling plots (Fig. S1a-b), which illustrate the intrinsic biological variation among samples, showed that samples from the same infection stage were grouped together. Hierarchical clustering (Fig. S1c-d), which was constructed from the distances between the RNA-seq libraries in the experiment, revealed that the gene expression profiles of B. distachyon are distinct at different stages following infection by F. graminearum and M. oryzae.

In comparison to controls, 44.1 and 58.6% of the expressed genes are affected in at least one stage upon infection by F. graminearum and M. oryzae, respectively. Therefore, approximately half of the B. distachyon genes are likely involved in the battle with fungal pathogens, although many genes may be passively up-regulated or down-regulated. Generally, the specific differentially expressed genes (DEGs) for each stage continue to increase with lesion development in both pathosystems (Fig. 1d-e). This is consistent with the colonization of fungi in the host. The majority of DEGs (Fig. 1d-e), especially the DEGs highly specific (≥10 fold) to individual fungi (Table S2), are found in the later stages of infection, suggesting that host physiology is greatly affected and the plant is forced to focus on struggling with the pathogens at those stages.

Examination of the specific DEGs for each pathogen revealed that the mode of gene up-regulation and down-regulation is different in F. graminearum and M. oryzae infections (Fig. 1f-h). While F. graminearum-infected samples have the maximum number of DEGs at 48 h post-inoculation (hpi), M. oryzae-infected samples have the maximum number of DEGs at 96 hpi. These may associate with different biotrophic-necrotrophic progress in these two distinct pathogens. Taken together, these data suggest that many genes are affected upon infection, and the global gene expression of B. distachyon infected by F. graminearum and M. oryzae are very different from each other.

Distinct gene expression program of Brachypodium infected by two cereal fungi

To compare the functional difference carried by gene expression dynamics, we performed gene ontology (GO) enrichment analysis on the DEGs from F. graminearum- and M. oryzae-infected samples at different infection time points. Many of the enriched GO terms are associated with plant-pathogen interaction, suggesting that the host gene expression had been greatly affected by pathogen infections, rather than the developmental changes of leaf and spikelet during the sampling. Throughout all the interactions, genes involved in “response to biotic stimulus” are up-regulated and enriched in F. graminearum- and M. oryzae-infected samples, while genes related to or involved in “photosynthesis” are down-regulated and enriched in F. graminearum- and M. oryzae-infected samples (Fig. 2). This suggests that B. distachyon had switched on plant defense and compromised to basal metabolism during the entire interaction. Similar tradeoffs between photosynthesis and plant defense have also been observed in many other studies [39].

Fig. 2
figure 2

Gene Ontology enrichment analysis of the differentially expressed genes of B. distachyon infected by F. graminearum (a) and M. oryzae (b). The size of the blue/red circle (in the first three columns) represents the number of the enriched genes, while the color represents the corresponding average log2 fold change. Red and blue circles indicate the up-regulation and down-regulation of the enriched gene, respectively. The adjusted P-values are illustrated with the size of the green circles. In (a-b), only adjusted P-values less than 0.05 are shown

Full size image

While some of the B. distachyon genes were enriched in both fungal infections at the equivalent stages, some were enriched in only one of the interactions (Fig. 2). For example, genes involved in “drug transmembrane transport” were enriched in the early stage of both fungal infections, whereas genes involved in “histone acetylation” and “glutamine family amino acid metabolic process” are enriched solely in F. graminearum- or M. oryzae-infection, respectively.

Some defense-related genes were up-regulated and enriched in both F. graminearum- or M. oryzae-infection, but some appeared at a different stage in the other pathosystem (Fig. 2). For example, genes related to “polyamine metabolic process”, “jasmonic acid metabolic process”, and “shikimate metabolic process”, which are known to be involved in plant defenses [40,41,42], are enriched in both F. graminearum- or M. oryzae-infection. During the B. distachyon-F. graminearum interaction, genes involved in the “polyamine metabolic process” are up-regulated at 48 hpi, but they are up-regulated at 96 hpi in B. distachyon-M. oryzae interaction. While genes involved in the “jasmonic acid metabolic process” are enriched at 48 hpi by F. graminearum, they are enriched throughout the M. oryzae interaction. Unlike the early stage (24 hpi), which involves genes related to the “shikimate metabolic process” in B. distachyon-M. oryzae interaction, these genes began to be enriched at 48 hpi in B. distachyon-F. graminearum interaction.

Therefore, the expression programs of the defense-related genes are distinct during F. graminearum and M. oryzae infections. It is unclear how the plant orchestrates these different defense programs to deal with various pathogens. However, most likely the different pathogen-associated molecular patterns (PAMPs) are detected by distinct receptors on the cell membrane, thus triggering multiple host defenses. Further characterization of the code of PAMPs detection and the pathways leading to subsequent cascades may be crucial for understanding these defense programs.

Distinct gene expression network of Brachypodium infected by two cereal fungi

To further understand the dynamic expression pattern of DEGs at different time points, we clustered the DEGs into different model profiles using STEM software [43]. Ten and twelve significant expression profiles were identified from the F. graminearum- and M. oryzae-infected plants, respectively (Fig. S2). Genes in the same expression profile are potentially co-regulated and more likely to functionally associate with each other [44, 45]. We therefore integrated the clustered gene expression profiles with protein-protein interaction networks. With the help of the STRING database [46], eight from the F. graminearum-infected plants and twelve from the M. oryzae-infected plants were extracted, and most of them were comprised by several modules (Fig. 3, Fig. S3–5).

Fig. 3
figure 3

Protein-protein interaction networks of the up-regulated genes in B. distachyon infected by M. oryzae. The up-regulated genes in each profile were subjected to protein-protein interaction networks analysis. Gene Ontology annotation was used to reveal the function of each submodule of the networks (blue oval). Nodes and edges represent proteins and functional links, respectively

Full size image

Functional annotation of these networks revealed that many of them are associated with plant-pathogen interactions (Fig. 3, Fig. S3–5). For example, during the B. distachyon-F. graminearum interaction, a series of co-expression modules, such as MAPK signaling, plant hormone signaling, gene transcription, and many plant defense-related responses, were up-regulated (Fig. S3). Interestingly, two modules in plant hormone signaling, exhibited distinct gene expression profiles. The genes related to “regulation of jasmonic acid mediated signaling pathway” continue to upregulation all the time, whereas the genes related to “response to ethylene” stop to upregulation in the later stage (Fig. S3).

During the B. distachyon-M. oryzae interaction, except the co-upregulation of MAPK signaling and plant hormone signaling related genes, many distinct networks associated with different plant defense responses were also involved (Fig. 3), such as “exocytosis”, “Golgi vesicle transport”, “chitin catabolism”, “regulation of cell proliferation and anthocyanin biosynthesis”, “response to external stimulus and nitric oxide”, “response to toxic substance”, as well as the “regulation of innate immune response”, “regulation of defense response and cell death”, etc. These data suggest that, although many defense-associated networks are overlapped, substantial networks in response to different fungi are distinct. Consistent with the defense programs, this analysis also showed that different biological processes worked in concert to knit the total plant defense.

MapMan analysis revealed involvement of different pathways in B. distachyon-fungal interactions

To understand the details of B. distachyon defense mechanisms that are activated during the interaction, we mapped individual gene expressions into metabolic pathways by using MapMan [47]. Pathway enrichment analysis based on Wilcoxon tests (Table S3) revealed that many genes encoding receptor-like kinases, WRKY transcription factors, or genes involved in ethylene, jasmonate, aromatic amino acid metabolism, protein synthesis or degradation were remarkably up- or down-regulated during both fungal infections. In the B. distachyon-M. oryzae pathosystem, many additional genes essential for hormone metabolism and secondary metabolism were affected. We therefore examined several major classes of genes responsive to biotic stress (Fig. 4). The results showed that a large number of them were altered during the interactions.

Fig. 4
figure 4

B. distachyon defense mechanisms activated during the interaction. In each step, the dynamics of the gene expression are illustrated with a heatmap. For each heatmap, the first four columns are B. distachyon spikes inoculated with F. graminearum at 0 h, 24 h, 48 h, and 72 h; the second four columns are B. distachyon leaves inoculated with M. oryzae at 0 h, 24 h, 48 h, and 96 h. The expression levels are mapped to color from green to red, to represent a lower to a higher abundance of the transcripts

Full size image

Pattern detection

When pathogens initiate contact with the plant, some conserved PAMPs are perceived via many receptor-like kinases (RLKs) and receptor-like proteins in the plasma membrane [48]. According to the MapMan analysis, 454 genes belong to the RLK family. Interestingly, many of the 192 cysteine-rich RLKs (with the DUF 26 domain) genes and LRK10-like RLKs were significantly up-regulated in both F. graminearum and M. oryzae infections (Table S3–4). Although the roles of cysteine-rich RLKs are unknown in B. distachyon-fungal interactions, they were shown to have a positive role in regulating anti-bacterial resistance [49]. The expression levels of some receptor-like kinases (RLK) genes were increased (Table S4), which may allow the generation of a strong danger signal. At the same time, some were suppressed and blocked by F. graminearum or M. oryzae when infecting B. distachyon (Table S4). This result suggests that, during the first encounter, neither the plant nor the pathogen has an advantage. They are both compromised. The behavior of RLK genes may be shaped by their co-evolution, since unsuccessful pathogens and plants with inferior fitness may be eliminated by strong constraints from each other.

MAPK signaling

Mitogen-activated protein kinase (MAPK) cascades play crucial roles in plant defense against pathogens. Activation of MAPK cascades is one of the earliest responses after the plant senses pathogen-associated molecular patterns (PAMPs) [50, 51]. In turn, a successful pathogen may suppress host MAPK cascades via their armed effectors [52, 53]. During the interaction of B. distachyon with these two fungi, different MAPK genes were up-regulated at later stages (Fig. 4). This result suggests that different MAPK may be employed by B. distachyon to translate pathogen attack signals from the two cereal pathogens and also suggests that these two pathogens have succeeded in suppressing the earliest defense cascades by MAPK, which may further facilitate the infection.

Phytohormone signaling

Phytohormones, such as salicylic acid (SA), jasmonate (JA), and ethylene (ET), are major plant hormones that orchestrate plant immunity [54,55,56]. Consistent with the brief biotrophic phase of F. graminearum and M. oryzae in infection [57, 58], expression of genes involved in SA signaling seems to have changed very little upon infection. Notably, many genes related to JA and ET signaling were significantly up-regulated (Fig. 4), as observed in the wheat- F. graminearum interaction [59]. These data suggest that the JA and ET pathways are critical for responding pathogens like F. graminearum and M. oryzae.

While the number of specifically induced ET-related genes are comparable in F. graminearum and M. oryzae infection, the number of specifically induced JA-related genes upon F. graminearum infection are obviously lower (Fig. 4). Close examination showed that behaviors of JA biosynthesis genes in F. graminearum-and M. oryzae-infected B. distachyon are quite different (Fig. 5). In the first two steps, the number of induced JA biosynthesis genes (LIPOXYGENASE, and ALLENE OXIDASE SYNTHASE) were consistently lower in F. graminearum infection than those in M. oryzae infection. In particular, Bradi1g15840, which encodes the only allene oxidase cyclase (AOC), was down-regulated during the B. distachyon-F. graminearum interaction. Such fine-tuning may lead to less 12-OPDA production and an insufficient final level of JA metabolites. Therefore, JA biosynthesis is likely to be suppressed in the B. distachyon-F. graminearum interaction.

Fig. 5
figure 5

The response of JA biosynthesis genes in B. distachyon upon F. graminearum and M. oryzae infection. The responses of the genes involved in the JA biosynthesis is illustrated with a heatmap. For each gene, the log2 fold change (logFC) level is mapped to color from gray to red, to represent the down-regulation and up-regulation of gene expression. The asterisks indicate no changes in expression

Full size image

Auxin (AUX), abscisic acid (ABA), and brassinosteroid (BR) have also been shown to be involved in phytohormone signaling [60, 61]. Remarkable differences in AUX- or ABA-responsive gene expression occurred between B. distachyon infected with F. graminearum and M. oryzae, respectively (Fig. 4). Each of the two fungi caused a similar number of specifically up-regulated genes, suggesting different downstream signaling pathways may be activated upon either F. graminearum or M. oryzae infection, respectively. This pattern is not observed in BR signaling. Interestingly, we found that some BR-responsive genes that were up-regulated in M. oryzae-infection were down-regulated in F. graminearum-infection, suggesting that BR signaling is likely to be inhibited during the B. distachyon-F. graminearum interaction.

Transcription factors

Early plant defense signals often result in the activation of downstream transcription factor genes to enhance defense-related gene expression [62]. Many transcription factor genes, such as ERF, bZIP, WRKY, MYB genes were up-regulated upon the infection by these two cereal fungal pathogens, respectively [24, 29]. Interestingly, most of the WRKY transcription factor genes that were up-regulated in B. distachyon-F. graminearum interaction are distinct from the ones up-regulated in B. distachyon-M. oryzae interaction (Fig. 4). This may be a result of different MAPK signaling pathways employed by the host plant. The distinct MAPK signaling, and distinct up-regulation of WRKY transcription factor genes suggest that B. distachyon has evolved different mechanisms to deal with these two different fungi.

Secondary metabolism

The final defenses are mediated by many defense-related genes, such as PR genes and secondary metabolism genes [63,64,65]. Genes involved in the metabolism of phenylpropanoid (82), isoprenoid (71), and flavonoid (56) comprise three-quarters of the 275 genes related to secondary metabolism (Table S5). MapMan enrichment showed that phenylpropanoid metabolism genes are induced, while genes involved in isoprenoid and simple phenol metabolism are repressed in the B. distachyon-M. oryzae interaction (Table S3). Phenylpropanoid metabolism leads to the biosynthesis of lignin and plays an important role in plant defense [66, 67]. In the infected Brachypodium, transcriptional changes in this pathway mainly occurred in the genes encoding phenylalanine ammonia lyase (PAL), 4-coumarate CoA ligase (4CL), hydroxycinnamoyl-CoA shikimate/quinatehydroxy cinnamoyl transferase (HCT), caffeoyl-CoA O-methyltransferase (CCoAOMT), and cinnamyl alcohol dehydrogenase (CAD) (Fig. 6), which are known to play different roles in regulation of the resistance to diverse pathogens [67,68,69,70,71,72,73,74,75]. Many components of the phenylpropanoid pathway are targeted by different effectors [72, 76, 77], suggesting that the pathogens recognize the importance of this pathway and have evolved corresponding strategies to deal with the defense conferred by this pathway. Upon F. graminearum and M. oryzae infection, many upstream genes of phenylpropanoid metabolism, such as PALs, were generally up-regulated, while some of the downstream genes, such as CADs, showed diverse expression patterns (Fig. 6). These data suggest that, although plant defense from the phenylpropanoid pathway was initiated, the products that execute the defense were likely to be attenuated by these two cereal fungi.

Fig. 6
figure 6

The response of phenylpropanoid metabolism genes in B. distachyon upon F. graminearum and M. oryzae infection. The responses of the genes involved in phenylpropanoid metabolism are illustrated with heatmaps. For each gene, the log2 fold change (logFC) level is mapped to color from gray to red, to represent the down-regulation and up-regulation of gene expression. The asterisks indicate no changes in expression

Full size image

Protein degradation by the ubiquitin-proteasome system

The ubiquitin-proteasome system (UPS) selectively degrades functional proteins carrying the specific ubiquitination signal, thus providing an efficient and rapid strategy to control many different cellular processes, including responses to plant-pathogen interactions [78,79,80,81]. The specific ubiquitination signals of the substrate proteins are attached by E1, E2, and E3 conjugation cascades, and the modified substrates are finally degraded by the 26S proteasome [82]. Consistent with GO and MapMan enrichment analyses, a close examination revealed that many genes involved in the UPS pathway were affected (Fig. 7, Table S3). Interestingly, we found that nearly all the proteasome genes were up-regulated in the M. oryzae infection, while they were essentially unchanged in the F. graminearum infection (Fig. 7). Almost every step of the host defense mechanism is regulated by UPS, such as plant hormone signaling and programmed cell death [78, 79, 83]. In turn, this pathway is frequently targeted by different viral, bacterial, fungal, and oomycete pathogens [83,84,85]. Therefore, UPS is a central hub as well as a battleground for plant-pathogen interaction. Our findings indicated that gene expression in the UPS is likely manipulated during infection and the up-regulation of proteasome genes is likely switched-off by F. graminearum.

Fig. 7
figure 7

The expression of genes involved the UPS degradation pathway in Brachypodium during infection. The expression dynamics of E1, E2, E3, DEB, and proteasome genes in the UPS degradation pathway are illustrated with a heatmap. A row in the heatmap indicates a gene, and a column represents a time point of Brachypodium infection samples. The gene expression is mapped to color from green to red, representing a lower to a higher abundance of the transcripts

Full size image

Conclusions

F. graminearum and M. oryzae are two of the most devastating plant pathogenic fungi. In this study, we studied the transcriptome dynamics of B. distachyon infected by these two fungal pathogens. We found that numerous defense-related genes were induced and showed distinct temporal dynamics of expression across the infection time course, including genes encoding RLK, MAPK cascade, phytohormone signaling pathways, and so on. Although the defense systems against the two pathogens were conserved, distinct expression programs and specific defense responses were frequently observed during the infections, suggesting their distinct interaction patterns. In particular, some genes related to JA signaling and the 26S proteasome pathway are likely specifically inhibited or manipulated upon infection by F. graminearum. Our study provides valuable insight into the interactions between B. distachyon and two important cereal pathogens.

Methods

Plant growth

The seeds of B. distachyon were sourced from our laboratory (NWAFU-Purdue Joint Research Center, Northwest A&F University). B. distachyon ecotype Bd21 was used for F. graminearum and M. oryzae infection. Bd21 seeds were germinated on damp filter paper at 4 °C in the dark. After one week, they were transferred into pots in a growth chamber at 23 °C ± 2 °C with a 16-h light/8-h darkness photoperiod.

F. graminearum inoculation

The wild-type F. graminearum strain PH-1 was routinely cultured on potato dextrose agar (PDA) plates at 25 °C. To prepare the conidial suspensions, cultures were grown for 3–7 days in carboxymethyl cellulose (CMC) medium with shaking at 175 rpm. The resulting conidia were suspended in sterile distilled water at a final concentration of 105 conidia/mL. Flowering heads of Bd21 plants were inoculated with conidial suspensions at the second spikelet from the base. Inoculated spikes were capped with a plastic bag for 24 h to maintain moisture. The samples were collected at 0, 24, 48, and 72 hpi and stored at − 80 °C.

M. oryzae inoculation

The wild-type M. oryzae strain Guy11 was cultured on oat meal agar at 22 °C for 7 days. Conidia were harvested with 0.25% (w/v) gelatin and suspended at a final concentration of 104 conidia/mL. The resulting conidial suspension was sprayed evenly on the leaf surface of five-leaf-stage seedlings. Inoculated plants were kept in a moisturizing chamber to maintain 100% air humidity for 24 h and then transferred to a growth chamber at 23 °C ± 2 °C with a 16-h light/8-h darkness photoperiod. The leaf samples were collected at 0, 24, 48, and 96 hpi and stored at − 80 °C.

RNA-sequencing

The spikelet and leaves inoculated with F. graminearum and M. oryzae, respectively, were used for RNA-seq analysis. TruSeq Stranded mRNA Sample Prep Kit was used for RNA extraction and library construction. High-throughput sequencing was performed on an Illumina Hiseq 2000 machine at Novogene (Beijing, China).

Gene expression analysis

The genome of B. distachyon [86] was obtained from the MIPS genome database. F. graminearum and M. oryzae genomes [87, 88] were downloaded from the Fungal Genome Initiative (FGI) site of Broad Institute. After removing low-quality sequences, the clean RNA-seq reads were aligned to genomes with Tophat2 (v2.0.9, “--mate-inner-dist 80 --mate-std-dev 70 --min-intron-length 20 -I 3000 --microexon-search --min-segment-intron 20 --max-segment-intron 3000 --b2-sensitive”) [89]. The program featureCounts (v.1.4.4, “-p -t exon -g gene_id”) [90] was subsequently used to summarize the counts of mapped reads for genomic features. To filter out weakly expressed genes, only those genes with a minimum expression level of 1 count per million in at least 3 of the 24 libraries were included in the analysis.

To assess the biological variability between samples, hierarchical clustering (with the top 1000 genes) was conducted with Pearson correlation distance measure and the pairwise average-linkage method. Multi-dimensional scaling plots were generated by using R package limma [91] with default setting. Differential gene expression analyses were performed for B. distachyon with the default setting in R package edgeR [92]. Benjamini-Hochberg method was used to correct for multiple comparisons. Genes were considered differentially expressed between conditions with a false discovery rate (FDR) of below 0.05.

Genes with similar dynamic expression patterns were profiled using the STEM software [43] with a significance of 0.05. The FPKM values of genes were calculated with Cuffnorm (v2.2.1, with default setting). Heatmaps were carried out with heatmap.2 function in the gplots package and hierarchical clustering was conducted with the same distance measure mentioned above.

Functional enrichment analysis and protein interaction network analysis

Gene Ontology enrichment analysis was performed using the BINGO (v2.44, with default setting) [93] plugin for Cytoscape (v3.2.0, with default setting) [94] and the Ontologizer software (v2.1) [95]. The term-for-term approach was used combined with Benjamini-Hochberg correction for multiple comparisons with a threshold of 0.05. The annotation of B. distachyon was obtained from the PLAZA website (http://bioinformatics.psb.ugent.be/plaza). Protein associated networks of B. distachyon were constructed in the STRING (v10, with default setting) database.

Pathway analyses

Pathway analyses were performed with MapMan (v3.5.1) [47]. To identify those BINs significantly affected by the infection, we calculated the induction factor of all genes in a BIN and compared the average induction factor of a BIN to that of all other BINs by the Wilcoxon rank-sum test with FDR < 0.05.