Generation of a dual F. graminearum-wheat co-expression network

F. graminearum floral infections can be divided into symptomatic or symptomless stages of infection. Disease spread through the rachis internodes (RI) can be further broken down to four different key stages of infection. Namely early symptomless (RI7-8), late symptomless (RI5-6), early symptomatic (RI3-4), and late symptomatic (RI1-2) infection (Fig 1A). A spatio-temporal transcriptomics dataset of F. graminearum floral infection of the susceptible wheat cultivar Bobwhite, which distinguishes between these key distinct stages, was previously generated [31]. This dataset also included spikelet tissue (SP) sampled at 3 (early symptomatic) and 7 days post-infection (dpi) (late symptomatic). The WGCNA framework [40,30] was used to construct a dual co-expression network to model fungal pathogen/crop interaction in wheat using this dataset (Fig 1B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Dual RNA-seq dataset and bioinformatics pipeline used for constructing the dual co-expression network.

A. Schematic illustration depicting the symptomatic (yellow) and symptomless (green) stages of Fusarium graminearum infection of wheat spikes denoted as stages A through D, corresponding to tissue samples collected for generating the RNA-seq data published in Dilks et al. 2019 [31]. F. graminearum hyphae growing in either the apoplast or inside the wheat cells are depicted in blue. B. Summary outlining the bioinformatics pipeline used for processing raw reads and constructing the dual RNA-seq network. The dual RNA-seq reads were initially processed together (processes depicted as white squares) before being separated to generate two distinct weighted gene co-expression networks. The bioinformatic pipelines are annotated accordingly, with yellow indicating the wheat reads-only pipeline and blue indicating the fungal reads-only pipeline. Annotation includes Gene Ontology terms (GO), Trait Ontology terms (TO), unique Gene Set Enrichment Analysis (GSEA), and PHI-base phenotypes. The modules from the two separate networks are then correlated to each other by their eigengene values to form the dual co-expression network.

https://doi.org/10.1371/journal.ppat.1012769.g001

Normalised counts were used to generate two distinct networks: one for F. graminearum and another for T. aestivum. The F. graminearum network consisted of 10,189 genes organised into 18 modules (with 2629–60 genes per module), while the T. aestivum network consisted of 47,458 genes distributed among 25 modules (with 23063–83 genes per module) (Figs 2A, S1 and S1 File). Both networks met scale free model criteria at their selected soft thresholding power (S2A and S2B Fig). The examination of module quality statistics found that each module within both networks were of a high quality (Z-Summary > 10), with the exception of F16 (Z-Summary = 9.67), which still markedly surpasses the minimum Z-Summary score of > 2 [41] (S2C Fig). This indicates a substantial preservation of modules compared to a random selection of all network genes. Additionally, preservation statistic calculations confirmed that all modules maintain preservation (Z-summary > 2) across both networks with all modules of the wheat network and the majority of the fungal modules (11/18) having strong preservation (Z-summary > 10) (S2D Fig). These findings suggest a consistent preservation of within-network topology across modules [41]. For each module, a single summarised expression pattern, the eigengene value, was calculated. The fungal and wheat modules were correlated by their eigengene expression values, and modules displaying significant correlation (p ≤ 0.001) formed the dual co-expression network (Fig 2A and 2B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Dual Fungal-Wheat co-expression network.

A. Network summarising significant co-expression patterns (p ≤ 0.001) between fungal modules (blue nodes) and wheat modules (yellow nodes). Positive correlations are depicted as black edges, while negative correlations are shown as grey edges. R-squared values are indicated next to edges, with edge width corresponding to the value. B. Wheat modules sizes. C. Fungal module sizes (S1 File). D. Wheat module enrichment. Significant Plant Trait Ontology (TO) enrichment results (p ≤ 0.05) for the five most significant enrichment terms for all wheat modules in the network. Higher significance is indicated by brighter yellows. Sizes of the circles correspond to the enrichment ratio. E. Fungal module enrichment. Significant (p ≤ 0.05) Biological Processes (BP) and Gene Ontology (GO) enrichment results for the five most significant terms of all fungal modules in the network. Higher significance is indicated by darker blues. Size of the circles correspond to the enrichment ratio.

https://doi.org/10.1371/journal.ppat.1012769.g002

To gain insight into the function of individual modules, a Gene Ontology (GO) enrichment analysis was performed for both network sets (Figs 2D, 2E and S1). To confirm these enrichment patterns were not due to chance, a random network was generated for both the fungal and wheat datasets. No significant enrichment was found for the random wheat network and fungal network.

Among the eight wheat modules within the dual co-expression network, five of them were significantly enriched for disease resistance genes (TO:0000112, p ≤ 0.05) and one was specifically enriched for wheat stripe rust resistance genes (TO:0020055) (Fig 2D), suggesting the wheat modules in the network are needed for plant defence. One of these wheat modules, W12, was significantly enriched in the GO terms detoxification (GO:0098754; p = 7.13 x 10⁻⁷) and response to toxic substances (GO:0009636; p = 2.11 x 10⁻⁶). This module was correlated to the fungal module F12, which was enriched in genes belonging to the trichothecene biosynthesis (TRI) gene cluster (TRI3, TRI4, TRI11, TRI12, and TRI14) (p = 1.92 × 10⁻⁴) and for the GO term terpenoid biosynthesis (GO:0016114; p = 0.00085) (Fig 2E and Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Function of correlated expression between wheat and fungal modules.

This table illustrates the relationship between wheat and fungal gene expression at different stages of infection, detailing the associated functions and key fungal genes involved.

https://doi.org/10.1371/journal.ppat.1012769.t001

The stage specific expression patterns of each module was examined (Figs 3, S3 and S4). Notably, the expression patterns of the TRI gene enriched module F12 aligned with its predicted function in DON biosynthesis. The module was most highly expressed in the late symptomless stage of infection. Expression of this module then rapidly decreased during the symptomatic stages of infection. Module F12 therefore appears to be positioned specifically at the transition between the late symptomless stage and the early symptomatic stage. The production of the DON mycotoxin is essential for the transition to the extensive symptomatic stage [52,53]. DON inhibits protein translation, which then eventually leads to cell death and the bleached phenotype distinctive of symptomatic F. graminearum infection [54,55]. High expression of module F12 in the symptomless stage is also supported by previous data which found that genes involved in mycotoxin biosynthesis are highly expressed in symptomless wheat tissue [56]. The correlation with the wheat module W12 therefore implies that detoxification genes in the module are being expressed in response to production of fungal mycotoxins.

Interestingly, the fungal module F10 contains genes that are highly expressed in the earliest and latest stages of F. graminearum infection, but not intermediate stages (Fig 3). The fungal module F10 includes the Killer toxin 4 genes (KP4L-1, KP4L-2, and KP4L-3). The KP4L genes are necessary for virulence and expressed during both self and non-self interactions (Table 1). It is suggested that KP4L proteins provide F. graminearum with a competitive advantage when occupying new niches [47], which would explain their expression during the earliest stage of infection. High expression during late infection may be necessary for intraspecific interactions, when the fungus is coordinating growth at a high fungal density.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Stage-specific expression of modules in the dual co-expression network.

A. Expression of modules across stages of F. graminearum infection. Illustration depicting symptomatic (yellow) and symptomless (green) stages of infection (A through D) annotated with specific modules (W or F) from the dual co-expression network that were highly expressed at specific stages. B. Eigengene summarised expression of fungal modules and C. wheat modules. Eigengene summarised expression plots illustrating the expression patterns of genes in wheat modules across different stages of infection as illustrated in panel A, along with spikelet tissue (SP) at 3 and 7 dpi.

https://doi.org/10.1371/journal.ppat.1012769.g003

The stress-responsive mitogen-activated protein kinase FgOS-2 is a key regulator in F. graminearum and acts upstream of the ATF/CREB-activating transcription factor FgAtf-1 (Table 1). Both FgOS-2 and FgAtf-1 cluster in module F10. These proteins are involved in broad functions, including secondary metabolite production, sexual reproduction, and stress tolerance [48]. Module F10 also contains two hydrophobin genes, FgHyd3 and FgHyd5. FgHyd3 is necessary for attachment to hydrophobic surfaces, while both genes are necessary for the production of aerial mycelia (Table 1). These genes are likely to play a crucial role during early infection for surface attachment and are possibly expressed again during the late stage of infection to facilitate the production of aerial mycelia.

The fungal module F10 is correlated with the wheat module W06 (R = 0.85, p = 6 x 10⁻⁶), which is enriched in protein catabolism (GO:0010498; p = 1.60 x 10⁻¹⁹) and autophagy (GO:0006914; p = 2.31 x 10⁻⁴) genes (Table 1). Autophagy plays a dual role in plant immunity where it is involved in immune signalling and programmed cell death to restrict pathogen spread, but also in response to pathogen induced necrotic cell death [57]. Therefore, it is likely these genes are expressed during early infection as an immediate immune response and then expressed again in highly colonised tissue for late-stage necrotrophic damage control.

Wheat genes in module W12 are expressed in response to DON production

To validate the correlation between modules F12 and W12, expression of wheat genes in the detoxification module W12 in response to F. graminearum infection with and without DON was examined (Fig 4A). This was achieved by inoculating wheat plants with either the wild-type F. graminearum reference strain PH-1 or the DON-deficient ΔFgtri5 mutant strain, which was generated in the PH-1 background. Since TRI5 is essential for DON biosynthesis, the primary function of the TRI gene cluster, the ΔFgtri5 mutant effectively acts as a representation of the function of the entire cluster. Expression of three wheat genes was studied, including two phenylalanine ammonia-lyases (PAL1 and PAL2; TraesCS4A02G401300 and TraesCS2D02G377200) which were annotated with the term disease resistance (TO:0000112), and a predicted transmembrane exporter, detoxification gene 16 (DTX16; TraesCS5B02G371100).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Validation of correlation between the trichothecene mycotoxin biosynthesis gene enriched module (F12) and the detoxification gene enriched module (W12).

A. Modules F12 (N = 124) and W12 (N = 451) depicted with significant enrichment annotations and genes with known phenotypes from PHI-base. Three genes listed in red in the PHI-base annotation (grey box) exhibit a loss of pathogenicity phenotype, while the remaining genes display a reduced virulence phenotype when individually deleted in F. graminearum. B. Equal levels of fungal burden were observed in tissue samples (p > 0.05). Absolute quantity of actin cDNA in mock (M), ΔFgtri5, and wild-type (WT)-recovered from the base of the inoculated spikelet and RI1-2 tissue sampled at 3 dpi (N = 3). Significance was determined by a one-way ANOVA followed by Tukey HSD correction. C. Normalised fold change expression of selected W12 wheat genes in mock (M), ΔFgtri5, and WT-recovered base of the inoculated spikelet and RI1-2 tissue sampled at 3 dpi (N = 3). Significance is denoted as * = p < 0.05, ** = p < 0.01, and *** = p < 0.001. Significance was determined by a one-way ANOVA followed by Tukey HSD correction.

https://doi.org/10.1371/journal.ppat.1012769.g004

The base of the inoculated spikelet and first two rachis internodes below the point of inoculation (POI) were sampled at 3 days post inoculation (dpi). Levels of FgActin cDNA were not significantly different between treatments (Fig 4B). Expression of the three wheat genes from module W12 was significantly lower in the ΔFgtri5 infected samples relative to wild-type infection (Fig 4C). This indicates that expression of genes in module W12 is correlated with DON production, thereby supporting the correlated co-expression patterns observed between modules of the two networks.

Dual co-expression networks as a tool to identify key genes necessary for virulence

To identify novel genes necessary for virulence, modules were examined for their potential biological importance based on their patterns of expression. Across the fungal modules, the module F16 is uniquely highly expressed during the earliest stages of infection, with markedly decreased expression at all the other stages of infection. Despite being highly expressed at a critical point for fungal establishment, the module has few characterised genes (6/76) and no significant enrichment annotation. This module is highly correlated to two wheat modules. These are W01 (R = 0.91; p = 5 x 10⁻⁷) and W05 (R = 0.85, p = 2 x 10⁻⁵). W01 is the largest wheat module and is enriched for defence response genes (GO:0006952; p = 3.60 x 10⁻⁰⁸), but also maintenance genes which include photosynthesis (GO:0015979; p = 4.59 x 10⁻²⁹) and RNA modification (GO:0009451; p = 1.42 x 10⁻⁴⁷) GO terms. The wheat module W05 is enriched for disease resistance (TO:0000112, p = 2.55 x 10⁻¹⁷⁸) and for the term oxidative stress (TO: 0002657; p = 3.88 x10^-34), suggesting that genes in the correlated fungal module F16 are possibly required for tolerance to this stress. Four genes in module F16 result in reduced virulence when individually deleted. These are FgNPC1 (sterol trafficking) [42], FgSrp2 (mRNA splicing) [45], and the transcription factors GzCon7 and Gzc2h045 [43] (Table 1 and S2 File). However, no gene deletion mutants exhibiting a loss of pathogenicity have yet been identified within this module, even though the eigengene expression pattern clearly indicates an association with the early establishment of the fungus in this key host tissue.

In WGCNA, genes with high module membership (MM) (i.e. intramodular connectivity) are predicted to be central to the biological functions of the module [30] and this continues to be a metric to identify biologically significant genes within WGCN analyses [58–60]. Following methods by Mateus et al. (2019) [39] key genes were defined as those exhibiting elevated MM within the module, which were also strongly correlated (R > |0.70|) with modules of the partner organism (wheat). In accordance with this, to identify genes in F16 that were likely necessary for virulence, the initial candidate gene list was selected by examining the 15 key genes with the highest MM from the 74 genes in the module (S1 Table and S3 File). Genes were then excluded that were likely to have functional redundancy (i.e. belonged to a gene family or had ancient paralogues within PH-1) to avoid compensatory effects when performing single gene deletion thereby leading to no changes in phenotypes (S1 Table). Ultimately, only two genes met these criteria: FGRAMPH1_01T23707 and FGRAMPH1_01T27545. FGRAMPH1_01T27545 has been previously characterised as the Niemann–Pick type C gene (FgNPC1). FgNPC1 is necessary for sterol trafficking, with its deletion resulting in ergosterol accumulation within the vacuole and a reduced virulence upon wheat infection [42]. Orthologue analysis identified that the FGRAMPH1_01T23707 gene was a 1:1 orthologue of Killer-nine resistant 4 (Knr4) in Saccharomyces cerevisiae [61], therefore the orthologue in F. graminearum is henceforth referred to as FgKnr4.

FgKnr4, a key gene of module F16, is necessary for establishment of fungal infection

T. aestivum cv. Bobwhite was inoculated at anthesis with three independent ΔFgknr4 transformants (Fig 5A and 5B), where FgKnr4 was deleted using a split hygromycin replacement cassette (S5A and S5B Fig). No symptomatic disease progression past the inoculated spikelets was observed with each ΔFgknr4 transformant (Fig 5A and 5B). While the inoculated spikelets developed symptoms, these did not exhibit full bleaching of the spikelet characteristic of FHB infection. Instead, eye-shaped lesions formed akin to those evident following ΔFgtri5 mutant infection [52] (Fig 5C). Plating of surface sterilised wheat dissected into its constituent parts revealed the absence of fungal growth in un-inoculated spikelets (S6 Fig). Nevertheless, browning was noted in the rachis tissue immediately adjacent to the point inoculated spikelet, accompanied by fungal growth. However, this colonisation did not occur past the rachis internode of the 3rd spikelet. These data suggest that, despite entering the rachis, the ΔFgknr4 mutant is unable to grow through the rachis node tissue and re-enter other spikelets. Microscopic examination revealed that while ΔFgknr4 can colonise host tissue, it does so less effectively compared to the wild-type strain (Fig 5D). Despite highly reduced virulence, DON mycotoxin was detected in the inoculated spikelet and adjacent rachis internodes (≥ 0.2 ppm). However, DON was undetectable in the neighbouring uninoculated spikelet (< 0.2 ppm) (S6B Fig). Complementation of the mutant with wild-type FgKnr4 (S5C and S5D Fig) restored virulence to wild-type levels (Fig 5E and 5F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Decreased virulence observed during in planta infection with ΔFgknr4.

A. Wheat spike infection assay done on the susceptible cultivar Bobwhite point inoculated with sterile water only (M), wild-type F. graminearum conidia, or conidia from three independent single gene deletion F. graminearum mutants lacking Knr4 (ΔFgknr4, T1-3). Images were captured at 15 dpi. Scale bar = 1 cm. B. Number of diseased spikelets per wheat spike at 15 dpi (N = 10). Letters indicate significant differences (ANOVA, TukeyHSD p < 0.05). C. Symptom development on the inoculated spikelets and adjacent rachis tissues at 15 dpi. This includes an image of an ΔFgknr4 inoculated spikelet with the outer glumes removed and zoomed in insets of the eye-shaped lesions. D. Ultra-thin 1 μm LR White resin sections stained with 0.1% Toluidine Blue for visualisation of wheat cell walls (light blue) and fungal hyphae (purple) of inoculated spikelet lemma tissue harvested at 7 dpi. Ab = abaxial wheat cell layer, ad = adaxial wheat cell layer, as = air spaces, and h = fungal hyphae. In wheat inoculated with ΔFgknr4 compared to WT, decreased host colonisation is observed, evidenced by (i) reduced fungal hyphae proliferation, particularly in the thicker adaxial cell layer, (ii) collapsed air spaces in wheat cells, and (iii) increased polymer deposition at the wheat cell walls (black arrows), indicating an enhanced defence response. Scale bar = 50 μm. E. Wheat spike infection complementation assay done on the susceptible cultivar Bobwhite treated with conidia either from wild-type F. graminearum, different complemented transformants (ΔFgknr4::KNR4-T1 and T2), the single gene deletion mutant (ΔFgknr4-T1), or sterile water (M). Images were taken at 15 days post inoculation. F. Number of diseased spikelets per wheat spike at 15 dpi (N = 10). Letters indicate significant differences (ANOVA, TukeyHSD p < 0.05).

https://doi.org/10.1371/journal.ppat.1012769.g005

FgKnr4 influences cell wall structure, stress resistance, and growth

In vitro growth of ΔFgknr4 was examined by culturing the fungus on both high or low nutrient agar. In both conditions a decreased growth rate relative to the wild-type was apparent (Figs 6A and S7, S8). In addition to this, conidia of ΔFgknr4 appear smaller than wild-type (S9A and S9B). Despite these morphological differences ΔFgknr4 retains the ability to produce perithecia and ascospores, albeit 8 days later than the wild-type (S9C, S9D, S9E and S9F Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Cell wall stress sensitivity and abnormal cell wall morphology of ΔFgknr4.

A. Dilution series of wild-type (WT), ΔFgknr4, and ΔFgknr4::KNR4 strains on Synthetic Nutrient Agar (SNA) and half-strength Potato Dextrose Agar (0.5 PDA) with and without the addition of calcofluor white (CFW), hydrogen peroxide (H₂O₂), and sodium chloride (NaCl). The dilution series begins at 1 x 10⁶ and continues with 10-fold dilutions. Images taken after 3 days of growth at room temperature. This experiment was replicated twice with similar results. B. Abnormal chitin deposition patterns in ΔFgknr4 conidia. Chitin-stained in conidia visualised using Wheat Germ Agglutinin Alexa Fluor 488 Conjugate (WGA). Scale bar = 50 μm. C. TEM imaging of wild-type and ΔFgknr4 conidia, showing differences in cell wall structure D. Western blot of proteins extracted from, ΔFgknr4 and ΔFgknr4::KNR4 mycelium incubated without (C) or with (T) the addition of 200 μg/ml calcofluor white (CFW) for 24 h. Phospho-p44/42 MAPK (Erk1/2) and p44/42 MAPK (Erk1/2) antibodies were used to detect phosphorylated and total MGV1, respectively. Amido black total protein staining was performed to compare protein loading. E. Ratio of phosphorylated MAPK/total MAPK based on quantification of band intensity. Significance is denoted as * = p ≤ 0.05. Significance was determined by a one-way ANOVA followed by Tukey HSD correction.

https://doi.org/10.1371/journal.ppat.1012769.g006

The FgKnr4 (F16) module was correlated with the wheat module W05, which exhibits a significant enrichment in the term oxidative stress (TO: 0002657; p = 3.88 x10^-34) that encompasses a total of 1143 genes. Among these genes are two respiratory burst oxidase homologues (RBOH), specifically a predicted homolog of RBOF (TraesCS1A02G347700) and RBOHE (TraesCS5D02G222100), along with predicted catalase homologues, CAT3 (TraesCS7B02G473400) [62,63], and two CAT4 genes (TraesCS5B02G023300, TraesCS5D03G079400) [64]. This would suggest a necessity of a functional FgKnr4 gene in oxidative stress tolerance that is critical during this early infection stage, where the fungus confronts various reactive oxygen species produced by the plant in response to fungal infection.

Stresses encountered by the fungus during in planta infection were mimicked in vitro using chemical stressors. As predicted by the correlation to wheat module W05, ΔFgknr4 had increased susceptibility to oxidative stress (H₂O₂) (Figs 6A, S7B, S7C and S8). The ΔFgknr4 mutant was also susceptible to other stresses faced during in planta infection including osmotic stress (1.5M NaCl), and calcofluor white induced cell wall damage compared to the wild-type and complemented strains (Figs 6A, S7B, S7C and S8). These susceptibilities may be due to changes in the cell wall structure of the ΔFgknr4 strain. Corroborating this hypothesis, staining for chitin found an irregular deposition of chitin on the ΔFgknr4 conidial cell wall, specifically at the tips and septa of the conidia (Figs 6B and S10). Furthermore, an irregular cell wall structure was observed upon transmission electron microscopy (TEM) analysis of the ΔFgknr4 conidia, indicative of an abnormal cell wall composition (Figs 6C and S11).

The involvement of FgKnr4 in cell wall metabolism was further studied by examining its effect on the cell wall integrity pathway (CWI). The fungal CWI pathway is triggered in response to various stresses (e.g. oxidative stress, osmotic pressure, cell wall damage) [65] and in F. graminearum is activated through the phosphorylation of the MAP-kinase (MAPK) FgMGV1 [66,67]. A Western blot was run on mycelium samples grown with and without a cell wall stress (calcofluor white). Constitutive activation of MGV1 in the absence of stress and increased phosphorylation under stress was observed in ΔFgknr4 when compared to the wild-type (Fig 6D and 6E). This finding is consistent with previous observations in S. cerevisiae [68]. This reinforces the biological function of FgKnr4, suggesting an involvement in fungal stress responses and cell wall morphology in F. graminearum.

Although the module F16 lacked significant enrichment for ontology terms and unique gene sets, 8/15 of the candidate genes within the module were related to cell division and ATP binding genes through connections based on protein similarity and orthology (S1 Table and S12 Fig). Four of the candidate genes were also part of a predicted protein-protein interaction subnetwork of genes which relate to cell cycle and ATP binding within module F16 (Fig 7A). To determine if FgKnr4 was involved in the cell cycle, as suggested by the shared genes in the module, the mutant was tested for susceptibility to benomyl, a microtubule inhibitor commonly used to identify mutations affecting cell division [69]. The ΔFgknr4 mutant was more susceptible to benomyl than the wild-type and complemented strain, with a greater inhibition when grown both in liquid culture (Fig 7B) and in solid media (Fig 7C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. FgKnr4 is associated with cell cycle genes and sensitive to the microtubule inhibitor benomyl.

A. Protein-protein interaction subnetwork of S. cerevisiae orthologs of genes in module F16 relating to cell cycle and ATP binding. Each node is the BioGRID ID of an S. cerevisiae protein with each grey vertex representing a direct protein-protein interaction recorded on the database. The corresponding F. graminearum ID is provided next to each protein node. Red nodes indicate that this protein is an orthologue of a candidate gene from the F16 module. Black borders indicate proteins involved in ATP binding, while green borders denote proteins associated with cell cycle processes, as identified through Knetminer and/or BioGRID. B. Relative growth inhibition in Potato Dextrose Broth (PDB) containing 0.5 μM benomyl compared to growth in PDB without the stress. Cultures were inoculated with either wild-type (WT), ΔFgknr4, or ΔFgknr4::KNR4. Optical density readings were taken 48 hours post-inoculation. Each biological replicate (N = 3) consists of three technical replicates. Letters indicate significant differences (ANOVA, TukeyHSD p < 0.01) C. Dilution series of wild-type (WT), ΔFgknr4, and ΔFgknr4::KNR4 strains on half-strength Potato Dextrose Agar (0.5 PDA) with and without the addition of benomyl. Images of plates treated with 1 μg/ml benomyl were captured on both a lightbox and a dark background. The lightbox images more clearly reveal the increased susceptibility of ΔFgknr4, which is obscured by the growth of aerial hyphae in the dark background images. The dilution series begins at 1 x 10⁶ and continues with 10-fold dilutions. Images taken after 3 days of growth at room temperature.

https://doi.org/10.1371/journal.ppat.1012769.g007

The orthologous gene in the wheat pathogen Zymoseptoria tritici is also important for cell wall integrity and virulence on wheat

Analysis of the Knr4 protein conservation found that orthologues were highly distributed across the Dikarya, occurring in both Ascomycota and Basidiomycota (Fig 8). Notably, no orthologues of the gene were found in other Eukaryotes or basal fungal lineages, highlighting its specificity to the lifestyle of the Dikarya. This high level of conservation across fungi suggests that phenotypes observed in F. graminearum may also be conserved in other economically significant pathogenic fungi.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Distribution of Knr4 orthologues across eukaryotes reveals exclusive presence in fungi.

A phylogenetic tree depicting the distribution of Knr4 orthologues across Eukaryota, with the positions of F. graminearum and Z. tritici highlighted in red. Different taxonomic levels are indicated in various colours as specified in the legend, alongside the percent distribution of orthologues between Ascomycota and Basidiomycota. Evolutionary distances between species or taxa are denoted by an internal scale (range 0–3.5). Bootstrapping confidence values are depicted as pale blue circles, with increasing size corresponding to higher confidence.

https://doi.org/10.1371/journal.ppat.1012769.g008

The orthologous Knr4 gene in another wheat fungal pathogen Z. tritici (ZtKnr4, Mycgr3G105330) was disrupted to test for conserved gene function. Despite the phylogenetic distance between the two fungi, the FgKnr4 and ZtKnr4 proteins share 43.5% pairwise identity. Alphafold3 was successfully used to accurately model the core region of the Knr4 protein, as evidenced by the high similarity between the predicted core structure of ScKnr4 and its X-ray crystal structure (S13A and S13B Fig). Structural predictions for both FgKnr4 and ZtKnr4 also revealed a strong similarity in their core regions, with a root-mean-square deviation (RMSD) of 1.060, indicating a high degree of structural conservation (S13C, S13D and S13E Fig).

Mirroring the phenotype observed in F. graminearum, reduced virulence (chlorosis but limited to no necrosis) was observed when wheat leaves were inoculated with ΔZtknr4 (Figs 9A and S14). In addition to this the ΔZtknr4 mutant was susceptible to calcofluor white induced cell wall stress and exhibited reduced hyphal branching (Fig 9B and 9C). These results highlight the potential of employing the Fusarium-wheat dual co-expression approach to gain insights into fungal-plant interactions, both within Fusarium species and across the fungal kingdom.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Functional characterisation of the Zymoseptoria tritici ΔZtknr4 gene deletion mutant.

A. Detached wheat leaves inoculated with wild-type (WT), Z. tritici (IPO323), ΔZtknr4 mutant strain, and two complemented strains (ΔZtknr4::KNR4-T1 and T2). Images taken at 20 dpi. B. WT Z. tritici, the ΔZtknr4 mutant and two complemented strains (ΔZtknr4::KNR4-T1 and T2) spot inoculated onto YPD agar with (bottom) and without (top) calcofluor white (CFW). Dilution series begins at 1 x 10⁵ and continues with 10-fold dilutions. Images taken after 3 days of growth at room temperature (RT). C. WT Z. tritici, the ΔZtknr4 mutant and two complemented strains (ΔZtknr4::KNR4-T1 and T2) spot inoculated onto 1% Tap Water Agar (TWA). Images taken after 10 days of growth at room temperature (RT).

https://doi.org/10.1371/journal.ppat.1012769.g009

Gene co-expression network analysis

RNA-seq reads from Dilks et al. (2019) [31] were provided by Dr Neil Brown (European Nucleotide Archive: PRJEB75530). Read quality was assessed with FastQC v. 0.11.9 (Andrews, 2010, [102] https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Reads were mapped to a combined Fusarium–wheat genome, consisting of v. 5 of the Fusarium graminearum PH-1 genome [103] and the high confidence (HC) transcripts of the v. 2.1 of the International Wheat Genome Sequencing Consortium (IWGSC) Triticum aestivum genome cultivar Chinese Spring [104]. Genome indexing and read alignment were performed using STAR aligner 2.7.8a [105]. Soft clipping was turned off to prevent reads incorrectly mapping to similar regions of the highly duplicated hexaploid wheat genome. Reads were filtered using the filterByExpr function part of the R package Edge R v.3.32.1 [106]. Fragments Per Kilobase per Million mapped fragments (FPKM) was calculated using the GenomicFeatures v.1.50.4 R package [107]. For the WGCN analysis raw counts were normalised separately for fungal and wheat reads by performing a variance stabilising transformation (VST) using the DESeq2 v 1.30.1 R package [108] in R (v4.0. 2, https://www.r-project.org/).

The VST normalised counts were filtered to remove any excessive missing values using the function goodSamplesGenesMS in the WGCNA R package [30]. Standard methods were implemented to generate the network using the WGCNA R package, with the following parameters. A signed-hybrid network was constructed using the filtered counts. The soft thresholding power (β) was uniquely selected per network according to scale free model criteria [40], where β = 9 for the fungal network and β = 18 for the wheat network (S2 Fig). A deepSplit of 3 was paired with a standard cutheight of 0.25. A minimum module size of 50 was selected to minimise potential transcriptional noise when assigning modules using smaller datasets [109,110]. The function multiSetMEs from the WGCNA package was used to calculate module eigengene expression. Module eigengenes with similar expression profiles were then merged.

Module quality and preservation was calculated using the function modulePreservation present in the WGCNA R package [30,41]. When calculating module preservation, the original wheat or fungal network was considered the reference network. Then 50 different test networks were created, each built upon randomly resampling (with replacement) a proportion of samples from the original dataset. The average preservation metrics (i.e. Z-score) between the original network and the 50 test networks was calculated for both the fungal and wheat networks.

Module enrichment an annotation

Gene ontology (GO) annotations of the v. 5 PH-1 genome (GCA_900044135.1) were generated using Blast2GO v .5 [111]. Enrichment was calculated using a background set of all genes present in the fungal network. GO annotations for the IWGSC v.2.1 genome were generated by performing a BLASTx search on the NCBI nb database using DIAMOND v 2.0.13-GCC-11.2.0 [112], then Blast2GO v.5 was used to annotate the BLAST hits with GO terms. GO term enrichment was calculated for each high level GO ontology (Biological Process, Molecular Function and Cellular Component) using the R package topGO v 2.46.0 [113].

Plant Trait Ontology (TO) [114] enrichment analysis was performed using gene-trait annotations derived from the KnetMiner knowledge graph (release 51) for wheat [115] and KnetMiner notebooks available at: https://github.com/Rothamsted/knetgraphs-gene-traits/. Predicted effectors were determined using EffectorP v.3.0 [116]. Alongside this, predictions to identify extracellularly localised proteins were done using SignalP v6.0 [117]. Custom F. graminearum gene set enrichment of the network modules was calculated by performing a Fisher’s exact test using all the genes in the fungal network as the background gene set. A Benjamini-Hochberg (BH) correction was calculated for both GO and custom enrichments [118]. Modules were deemed significantly enriched if P-corr < 0.05.

Gene lists included in the Gene Set Enrichment Analysis (GSEA) consisted of predicted secreted effector proteins, alongside known gene families associated with virulence, such as biological metabolite clusters (BMCs) [119], polyketide synthases [120], protein kinases [121] and transcription factors [43]. Due to their well-established importance in F. graminearum pathology, a separate enrichment for genes of the TRI gene cluster was also performed.

Annotation from PHI-base was obtained by mapping genes to version PHI-base (v4.16) annotation using UniProt gene IDs and any through Decypher Tera-Blast P (TimeLogic, Inc. Carlsbad, California, USA) (E-value = 0) against the PHI-base (v4.16) BLAST database [95,96].

The Knetminer network of F16 key genes was generated by providing all 15 key genes as input in the web interface of Fusarium graminearum Knetminer version 51 (https://knetminer.com/Fusarium_graminearum/). The protein-protein interaction subnetwork was created by using Orthologous Matrix (OMA) [122] to identify orthologues of F. graminearum in S. cerevisae. This was then used to map orthologues to S. cerevisae protein-protein interactions on BioGRID version 4.4.236 [123].

Fungal material and growth conditions

F. graminearum strains were cultured and conidia prepared as previously described [12]. Fungal strains were grown in the dark for 3 days on nutrient-rich potato dextrose agar (PDA), nutrient-poor synthetic nutrient agar (SNA; 0.1% KH₂PO₄, 0.1% KNO₃, 0.1% MgSO₄·7H₂O, 0.05% KCL, 0.02% glucose, 0.02% sucrose and 2% agar) and PDA with different cell wall stresses. This included 1–1.8 μg/ml benomyl, 150/100 μg/ml calcofluor white, 1.5mM H₂O₂, and 1.5M NaCl. Plates were point inoculated with 20 μl of 4-fold dilution series starting with 1 x 10⁶ conidia/ml. For growth rate inhibition calculations, plates were point inoculated with 20 μl 1 x 10⁶ conidia/ml water suspensions. Colony diameter was measured at 3d growth and relative inhibition was calculated as (colony diameter in 0.5 PDA with stress)/(colony diameter in 0.5 PDA without stress) for each strain. Sensitivity to benomyl in liquid cultures was tested by mixing 100 μl of 1 x 10⁵ conidia/ml or 100 μl water (control) with 100 μl of Potato Dextrose Broth (PDB) without and with 0.5 μM benomyl in a 96 well plate incubated at 22 °C and 118 rpm for 48 hrs. Optical density (OD) was measured at 600 nm and the readings of fungal inoculated cultures were corrected by subtracting readings from corresponding control wells.

For the growth rate assay, fungi were grown on PDA and measurements were taken after 3, 5, and 7 days. Surface sterilisation of wheat spikes was performed by submerging single wheat spikelets in 1/8 diluted thin bleach for 3 min, followed by three washes with sterile distilled H₂O. Dissection was done using a razor blade to separate the point inoculated spikelets and adjacent spikelets (S6A Fig). Wheat tissue was placed on SNA and images were taken after a 3-day incubation at room temperature in the dark. Perithecia induction was achieved as described in Cavinder et al. (2019) [124]. All plate images were taken using an Olympus OM-D camera with a 60mm ED M.Zuiko macro lens. Conidia and ascospore images were taken using the Axiomager 2 (Zeiss, Oberkochen, Germany) under brightfield illumination. Conidia lengths (N = 50) and perithecia heights (N = 40) were measured using ImageJ [125].

Fusarium graminearum genetic manipulations

The FgKnr4 gene was deleted through split marker-mediated transformation to target fungal gene replacement with the hygromycin resistance cassette by homologous recombination [126]. F. graminearum gene deletion construct assembly and fungal transformation was performed following methods outlined in King et al., 2017 [127]. Primers were designed for the fusion of the 5’ and 3’ constructs using the NEBbuilder Assembly Tool v.1 (https://nebuilderv1.neb.com/). Using the Gibson Master Mix (New England Biolabs, UK) the paired split marker fragments were ligated into the pGEM—T Easy Vector (Promega, UK) then transformed into DH5α competent Escherichia coli (C2987H, New England Biolabs, UK) following the standard manufacturer protocol. Diagnostic PCRs were done using DreamTaq polymerase (ThermoFisher, UK) and standard cycling conditions. For the single gene deletion, in three separate transformants two diagnostic PCRs were done to detect the presence of the replacement cassette flanks (P3-4, P5-6) and the absence of the wild-type gene (P1-2) (S5A and S5B Fig). Complementation was performed following the protocol developed by Darino et al. (2024) [128]. Diagnostic PCRs for the complemented strains involved amplification of the insertion cassette flanks (P7-8; P1-9), testing for the absence of the short 868 bp empty intragenic locus amplicon (P11-P12), and testing for the heterozygosity of the geneticin gene (P13-P14) (S5C and S5D Fig). A full primer list is available in S4 File.

Wheat host inoculation

The susceptible spring wheat (T. aestivum) cultivar, Bobwhite, was grown to anthesis. The 5th and 6th spikelets from the top of the wheat spike were inoculated on both sides using 5 μl of 5 x 10⁵ conidia/ml. Each treatment included 10 separate wheat plants (N = 10). After inoculation, plants were kept in a high humidity chamber for 48 h in the dark. Disease progression was documented every two days by scoring the number of bleached spikelets. At 15 dpi wheat spikelet tissue and the adjacent rachis internode was separated, frozen in liquid nitrogen, and ground to form a fine powder. The presence of DON mycotoxin was assessed using the Deoxynivalenol (DON) Plate Kit (Cat. 20–0016, Beacon Analytical Systems Inc., USA) following standard protocol. This experiment was replicated with three biological replicates per treatment (N = 3). All F. graminearum infected plant images were taken using an Olympus OM-D camera using a 60mm ED M.Zuiko macro lens.

For resin dissection microscopy wheat cv. Bobwhite was inoculated at the 7^th and 8^th true spikelets from the base on both sides using 5 μl 5x10⁵ conidia/ml. After inoculation, plants were kept in a high humidity chamber for 48 h in the dark. Lemma tissues were excised from infected spikelets at 7 dpi, fixed in a 4% paraformaldehyde, 2.5% glutaraldehyde solution with 0.05M Sorensen’s phosphate buffer (NaH₂PO₄:Na₂HPO₄, pH 7.0). Samples then underwent 3 further buffer washes, a subsequent ethanol dehydration protocol (0–100% EtOH) over 48hrs and then LR White resin (TAAB, Laboratories Equipment Ltd, UK) infiltration diluted with dry ethanol at increasing ratios (1:4, 2:3, 3:2, 4:1, 100%). Samples were inserted into capsules (TAAB, Laboratories Equipment Ltd, UK) and resin polymerised at 60°C for 16 hours in a nitrogen oven (TAAB, Laboratories Equipment Ltd, UK). Ultra-thin 1 μm sections of samples were cut on an ultramicrotome (Reichert-Jung, Ultracut) with glass knives, placed onto glass polysine slides (Sigma Aldrich, UK), dried at 70°C, stained with 0.1% (w/v) Toluidine Blue O and mounted in DPX mounting medium (Fisher Scientific). Stained sections were imaged on a Zeiss Axioimager (Axiocam 512 color, Zeiss, Jena, Germany) light microscope with brightfield illumination.

Gene expression of module W12 genes

Bobwhite wheat plants were point inoculated at anthesis with either wild-type PH-1, ΔFgtri5 or water only (Mock) following the protocol outlined in Dilks et al., (2019) [31]. Each experimental condition was replicated in triplicate, with each replicate deriving from three pooled independent wheat spikes. Tissues from the base of the inoculated spikelet and rachis internodes 1 and 2 were sampled and frozen in liquid nitrogen at 3 dpi. Sampling at this timepoint captures the early colonisation of the initially inoculated spikelet, where the macroscopic disease progression of ΔFgtri5 is comparable to the wild type. Frozen samples were ground and RNA was extracted using the Monarch Total RNA Miniprep Kit (NEB, UK). Equal amounts of RNA were used to synthesise cDNA with Revertaid cDNA synthesis kit (ThermoScientific, UK). PowerTrack SYBR Green Master Mix (ThermoScientific, UK) was used for qPCR. To quantify fungal burden, cDNA was used instead of genomic DNA, as the entire tissue sample was required to generate cDNA. Consequently, the housekeeping gene FgActin was selected as a reference for quantifying fungal burden. Each biological replicate included three technical replicates. All primers are provided in S4 File.

Western blot

A 200 μl aliquot of a F. graminearum conidia solution (1 x 10⁶ conidia/ml) was added to 10 ml potato dextrose broth (PDB) at 27°C. Calcofluor white was added to a concentration of 200 μg/ml after 24 h of incubation at 180 rpm. Twenty-four hours after the addition of the stress, mycelium was harvested, flash frozen and freeze dried. To lyse the samples Y-PER Yeast Protein Extraction Reagent (ThermoScientific, UK) was added to the freeze-dried samples at 1.5 ml per 150 mg tissue, alongside Protease Inhibitor Cocktail (100x) (ThermoScientific, UK). Samples were lysed using the FastPrep-24 machine for 20s (MP Biomedical, USA). The supernatant was mixed with 5xSDS loading buffer (National Diagnostics, USA).

Equal amounts of protein (60 μg) were resolved on 8% SDS-PAGE gels (Mini-PROTEAN, Bio-Rad, UK) and transferred onto a nitrocellulose membrane. Immunoblots were performed by standard procedures using the Phospho-p44/42 MAPK (Erk1/2) (cat. #4370) and p44/42 MAPK (Erk1/2) (cat. #9102S) (Cell Signalling Technologies, USA) antibodies at their specified dilutions. The blots were developed using ECL Plus Western Blotting Detection Kit and images were acquired using Odyssey Imaging System (LI-COR Biosciences Ltd, Cambridge, UK).

Microscopic examination of cell walls

To induce conidia formation, 200 μl of frozen conidia (1 x 10⁶) were plated onto PDA plates and the plates were incubated at room temperature under constant illumination. For conventional transmission electron microscopy (TEM), fresh conidia were harvested the same day from the PDA plates and pellets were fixed in a mixture of 2.5% glutaraldehyde and 4% Paraformaldehyde in Sorenson’s buffer (SB) at pH 7.2 overnight at 4°C. The samples were rinsed in SB and post fixed in 1% osmium tetroxide for 60 min at room temperature. The samples were dehydrated for 10 min per step into increasing concentrations of alcohol (30%, 50%, 70%, 90% and final 100%×3). Subsequently, the pure ethanol was replaced with propylene oxide, and the specimens were infiltrated with increasing concentrations (25%, 50%, 75%, and 100%) of Spurr resin mixed with propylene oxide for a minimum of 2 hr per step. The samples were embedded in pure, fresh Spurr resin and polymerised at 60°C for 24 hr. Ultrathin sections (70 nm) were cut using an ultramicrotome (Leica UC7, Germany) and post-stained, first with Uranyless (TAAB Laboratories Equipment Ltd, UK) for 1 min and then with Reynolds lead citrate (TAAB Laboratories Equipment Ltd, UK) for 2 min at room temperature, prior to observation using a Transmission Electron Microscope (Jeol 2100plus, UK) operated at 200 kV.

F. graminearum conidia solution (1 x 10⁶ conidia/ml) was stained with Wheat Germ Agglutinin, Alexa Fluor 488 Conjugate (WGA) (10 μg/ml) for 10 minutes each. Samples were washed three times in sterile distilled water after staining. A ZEISS 780 Confocal Laser Scanning Microscope (ZEISS, Germany) was used to image conidia.

Phylogenetic tree construction

Eggnogmapper-v5 [129] was used to map FgKnr4 to the eggnog Orthologue Group (OG) ENOG502QTAZ and generate the phylogenetic tree. The tree was visualised and annotated using the interactive Tree of Life (iTOL) software [130].

Functional characterisation of the Knr4 orthologue in Z. tritici

Separate analyses using BLASTp (E value = 3.7 x 10⁻¹¹⁰) [131], Orthologous Matrix (OMA) [122], and Eggnogmapper [129] identified a single orthologous sequence in the genome of the Z. tritici isolate IPO323 (https://fungi.ensembl.org/Zymoseptoria_tritici/Info/Index) [14]. The gene has a Rothamsted gene model Id of ZtritIPO323_04g12347 [127,91] and is present on Chromosome 8 at start position 230142 bp. This maps to Mycgr3P105330 in the current genome call on Joint Genome Institute (JGI) Mycocosm [14]. Predicted folded protein structures and confidence plots were generated used the Alphafold3 server [132] (https://alphafoldserver.com/). The core X-Ray structure of ScKnr4 was downloaded from the RCSB Protein Data Bank (accession: 5J1B). Alignment between the ScKnr4 crystal structure and Alphafold3 predicted structure was done using the Pymol v.3.0.3 align function.

Agrobacterium-mediated fungal transformation [133] was performed to generate a series of independent gene disruption mutants of ZtKnr4. Flanking sequences and the hygromycin resistance gene were amplified from either genomic DNA or from plasmid pCHYG and using Phusion polymerase (NEB, UK). Fragments were gel purified using QIAquick Gel Extraction Kit (QIAGEN, UK) and assembled into the backbone (Kpn1 and BamH1 digested) of pCHYG by Gibson Assembly (NEB, UK). The resulting plasmids were transformed into the Agrobacterium strain AgL1 and fungal transformation of isolate IPO323 was performed as per standard protocols [133]. Positive transformants containing a disrupted ZtKnr4 gene were identified by diagnostic PCR (S15 Fig). Complementation of the validated ZtKnr4 mutant was performed through Agrobacterium-mediated transformation with plasmid pCGEN (digested EcoR1 and Kpn1) containing the native gene plus 1 kb upstream (5’) and 300 bp (3’) downstream genomic DNA, amplified by Phusion PCR (NEB, UK).

Attached leaf virulence assays were performed as per standard protocols [134] on wheat cultivar Riband. Leaf blades (N = 3) were inoculated with spore suspensions of 1 x 10⁶ spores / ml in sterile water + 0.05% v:v Tween 20. Final disease assessments were made 20 days after inoculation. In vitro hyphal growth assays were performed following droplet inoculation of spore suspensions onto 1% Tap Water Agar (TWA) plates. Hyphal growth morphologies were determined by light microscopy and / or photography 10 days after inoculation. Calcofluor white sensitivity assays were performed to ascertain changes in cell wall strength. For this, spore suspensions were inoculated onto YPD agar (Formedium, UK) plates (control) and onto YPD agar plates containing 200 μg / ml calcofluor white. Plates were incubated at RT for 8 days and then growth was monitored and recorded by photography. Images of ZtKnr4 in planta and in vitro experiments were taken with a Nikon D3200 camera. Disease levels were quantified using LemnaGrid image analysis system, as described in Chen et al. (2023) [91].