Ste2 receptor-mediated chemotropism of Fusarium graminearum contributes to its pathogenicity against wheat

Fusarium Head Blight of wheat, caused by the filamentous fungus Fusarium graminearum, leads to devastating global food shortages and economic losses. While many studies have addressed the responses of both wheat and F. graminearum during their interaction, the possibility of fungal chemotropic sensing enabling pathogenicity remains unexplored. Based on recent findings linking the pheromone-sensing G-protein-coupled receptor Ste2 to host-directed chemotropism in Fusarium oxysporum, we investigated the role of the Ste2 receptor and its downstream signaling pathways in mediating chemotropism of F. graminearum. Interestingly, a chemotropic response of growing hyphae towards catalytically active Triticum aestivum ‘Roblin’ cultivar secreted peroxidases was detected, with deletion of STE2 in F. graminearum leading to loss of the observed response. At the same time, deletion of STE2 significantly decreased infection on germinating wheat coleoptiles, highlighting an association between Ste2, chemotropism and infection by F. graminearum. Further characterization revealed that the peroxidase-directed chemotropism is associated with stimulation of the fungal cell wall integrity mitogen-activated protein kinase signaling cascade. Altogether, this study demonstrates conservation of Ste2-mediated chemotropism by Fusarium species, and its important role in mediating pathogenicity.


Results
Fusarium graminearum exhibits chemotropism towards chemical stimuli. A quantitative chemotropism assay 5 was used to assess the abilities of different compounds to induce directional hyphal growth in F. graminearum (Supplemental Figure S1a). Cell concentrations of 0.25 million macroconidia per mL of aqueous agar media were found to be the most suitable for quantifying hyphae. Higher cell concentrations resulted in intertwining hyphae that could not be counted discreetly, while concentrations lower than 0.25 million per mL yielded insufficient numbers of cells on the scoring line. Growth of conidia exposed to gradients of various compounds was monitored microscopically and an optimum period of 14 h was chosen for counting hyphae. For consistency, only conidia with single hypha were included; those with more than one germinating hypha were excluded from the count.
Nutrients with nitrogen and/or carbon sources were screened first for their ability to induce chemotropism in wild type F. graminearum (Fig. 1a). The chemotropic responses to nutrients were compared with a doublenegative control plate where water was added to both the test and control wells. Among nutrients, a significant chemotropic response was induced by methionine where ~ 10% more macroconidia grew hyphae towards methionine compared to the water control (Fig. 1a, Supplemental Figure S1b). Responses towards other nutrients were highly variable, and none of them elicited a response as robust or significant as methionine. Interestingly, it was observed that exposure of wild type macroconidia to any of the nitrogen-containing compounds, including glutamate, aspartate, ammonium sulfate and betaine, caused rapid hyphal growth yielding long, intertwined hyphae, compared to exposure to other nutrients. Nonetheless, with the exception of methionine, this rapid growth was not found to be significantly directional and thus was not deemed chemotropic in its nature.
Scientific RepoRtS | (2020) 10:10770 | https://doi.org/10.1038/s41598-020-67597-z www.nature.com/scientificreports/ With FgSte2 being an α-pheromone receptor, it was expected that exposure of F. graminearum to the α-pheromone would stimulate signaling through the receptor and ultimately a chemotropic response towards it. Hence, chemically synthesized α-pheromone peptides of S. cerevisiae and F. graminearum were next screened against wild type F. graminearum macroconidia (Fig. 1b). The two α-pheromone peptides are of similar size, but differ in sequence and thus, specificity. Exposure to Fg α-pheromone (Supplemental Figure S1c) stimulated a robust chemotropic response compared to the double-negative water control, while Sc α-pheromone induced a weaker and more variable response. To validate that it is the α-pheromone peptide that elicits directional growth, the F. graminearum α-pheromone was proteolyzed with proteinase K, resulting in a complete loss of chemotropism (Fig. 1b).
Fusarium graminearum exhibits positive chemotropism toward the wheat head and secreted wheat peroxidases. Macroconidia of F. graminearum are dispersed onto the wheat head prior to initiation of infection, implying that any wheat-derived chemoattractant inducing chemotropism would likely originate from the wheat head. This provided a rationale for investigating the wheat head alone in inducing directional hyphal growth in wild type F. graminearum. Preliminary chemotropism assays using wheat heads of cultivars having different susceptibilities towards F. graminearum, including the highly susceptible 'Roblin' , moderately resistant 'Wuhan' , and highly resistant 'Sumai3' cultivars, were conducted (Supplemental Figure S2a). While each of the wheat heads elicited chemotropic responses in wild type F. graminearum, these preliminary studies do not suggest any correlation between susceptibility and the intensity of the chemotropic response (Supplemental Figure S2b). Experiments were subsequently limited to the most susceptible cultivar 'Roblin' . To facilitate further investigation into the nature of the host molecules that induce this response, exudate from the spikelets of 'Roblin' was extracted and tested for chemotropic effect. The response to 'Roblin' exudate (Supplemental Figure S1d) was similar to that seen for the intact 'Roblin' wheat head (Fig. 2a), confirming that the 'chemoattractant' is likely a water-soluble molecule derived from the wheat head. A concentration-dependent chemotropic response was seen towards 'Roblin' exudate, with a more concentrated exudate inducing a stronger chemotropic response in F. graminearum (Supplemental Figure S2c).
The role of one or more of these peroxidases in eliciting chemoattraction was further examined by treatment of the 'Roblin' exudate with the peroxidase-specific inhibitor salicylhydroxamic acid (SHAM), prior to assessing chemotropism (Fig. 2c). The observed elimination of any chemotropic response toward the SHAM-inhibited (a) Directed hyphal growth of wild type F. graminearum towards a gradient of specified nutrient sources after 14 h exposure; MeOH methanol, Gluc glucose, Glyc glycerol, Gal galactose, Met methionine, Asp aspartate, Glu glutamate, Bet betaine, (NH 4 ) 2 SO 4 ammonium sulfate (compared to solvent control, ****P < 0.0001). (b) Directed hyphal growth of F. graminearum towards a gradient of α-pheromone of S. cerevisiae (Sc) and F. graminearum (Fg), either untreated (C) or treated with Proteinase K (Prot K) (compared to untreated control, ****P < 0.0001). Data represent the average from at least three experiments. n = 500 hyphae. Error bars represent standard deviation. Graphs were plotted using Graphpad Prism version 6.01 (https ://www.graph pad.com). The figure was compiled using Adobe Illustrator CC 2015 (https ://www.adobe .com).
Scientific RepoRtS | (2020) 10:10770 | https://doi.org/10.1038/s41598-020-67597-z www.nature.com/scientificreports/ 'Roblin' exudate by F. graminearum validates the importance of active peroxidases secreted by the wheat head in this system. While wheat peroxidases are valid candidates to contribute to the stimulation of chemotropism, HRP was tested in the chemotropism assay as a more reliable and simplified assay system. Conservation of essential catalytic residues between HRP and the identified wheat peroxidases provides a rationale for similar chemotropic responses of F. graminearum towards the two stimuli (Supplemental Figure S3a). As expected, exposure to HRP induced a robust chemotropic response in wild type F. graminearum (Fig. 2c, Supplemental Figure S1e). This HRP-induced chemotropism was completely abolished either by proteinase K treatment or by boiling at 100 °C. Inhibition of HRP with SHAM also eliminated the chemotropic response. These findings emphasize that the chemoattractant is not the peroxidase itself, but the product of a peroxidase-catalyzed reaction.
Finally, to confirm that this observed response was indeed chemotropism and not a growth speed or subjective bias, the lengths and angles of the hyphae growing towards test (HRP) and control (water) compounds in the chemotropism plate assay were measured (n = 300, Fig. 3a-c). This analysis showed no significant differences in the lengths, and thus no growth speed bias, of hyphae growing towards HRP compared to those growing towards the water control (Fig. 3b). To avoid subjective bias, only hyphae with angles of 0 to approximately 45° were counted and used for calculation of the chemotropic index. The angles of hyphae that were counted as growing towards the HRP and water control were measured (Fig. 3c) and a similar distribution of angles was observed for both HRP and the water control. No significant difference was observed between the cosine of angles of these hyphae with the average cosine of 0.87, corresponding to an angle of approximately 28° (Fig. 3c).

Deletion of F. graminearum STE2 results in loss of chemotropic response.
To investigate the role of FgSte2 in chemotropism, the previously annotated STE2 gene 27,34,35 was deleted through homologous recombination using ATMT. Three positive hygromycin-resistant transformants (Fgste2Δ-1, Fgste2Δ-3, Fgste2Δ-5) were further confirmed by PCR amplification across the upstream junction of integration of the knockout cassette with primers P13 (located in the upstream genomic DNA region outside of the knockout cassette) + P16 (within the hygromycin coding region) (Supplemental Table 1, Supplemental Figure S4a). Furthermore, PCR analysis with primers internal to the STE2 coding region (Primers P11 and P12-Supplemental Table 1) showed a complete absence of bands. To validate that any phenotypic and chemotropic changes observed are solely due to the deletion of STE2 and not any off-target genetic defects, a complement strain was constructed by re-introducing the STE2 gene into the Fgste2Δ-5 strain (Supplemental Figure S4b). The complemented geneticin-resistant strain was verified by PCR using geneticin-specific and genomic DNA-specific primers to amplify across the junction of the cassette. Sequencing of the genomic DNA showed that the complemented gene was integrated into the native STE2 locus.
The three Fgste2Δ strains and one Fgste2Δ + STE2 strain were assayed against compounds that elicited significant chemotropic responses in wild type F. graminearum, specifically Fg α-pheromone, 'Roblin' exudate, HRP and methionine. A chemotropic index of essentially zero was determined for Fg α-pheromone for the Fgste2Δ-5 mutant, indicating random hyphal growth and an inability to sense the pheromone peptide, confirming that deletion of FgSTE2 eliminates all chemotropic response to α-pheromone (Fig. 4a). Next, it was observed that the Fgste2Δ-5 mutant exhibited no chemotropism towards 'Roblin' exudate, signifying the FgSte2 receptor is responsible for mediating chemotropism towards the 'Roblin' exudate (Fig. 4a). The same result was observed upon www.nature.com/scientificreports/ exposure of the Fgste2Δ-5 mutant to a gradient of HRP. Interestingly, however, the Fgste2Δ-5 mutant exhibited a robust response towards methionine, comparable to that observed in wild type strain, indicating that the response to nutrients is mediated independently of the FgSte2 receptor. All three Fgste2Δ transformants exhibited similar chemotropic responses towards the compounds tested (Supplemental Figure S5). Re-introduction of STE2 into the Fgste2Δ-5 mutant restored chemotropic responses towards Fg α-pheromone, 'Roblin' exudate and HRP in F. graminearum, confirming the role of the FgSte2 receptor in sensing these stimuli (Fig. 4a).
Deletion of STE2 has no effect on vegetative growth on solid and liquid media. Vegetative growth of Fgste2Δ-5 and Fgste2Δ + STE2 strains were assessed on PDA and SNA plates (Fig. 4b). On both media, the growth, colony color, and morphology of Fgste2Δ-5 was comparable to wild type F. graminearum. Additionally, growth of all three Fgste2Δ mutants and Fgste2Δ + STE2 in liquid CMC and PDB media was found to be comparable to wild type. Interestingly, conidia of Fgste2Δ-5 mutants are significantly shorter than those of wild type (Fig. 4c,d).

Deletion of STE2 leads to decreased virulence on wheat coleoptiles. To assess whether FgSte2
plays a role in pathogenicity, the three Fgste2Δ mutants and the Fgste2Δ + STE2 complement strain were assessed in the coleoptile infection assay and compared to wild type F. graminearum. The coleoptile assay was selected as a pathosystem to study infection of wheat by F. graminearum as it represents a very fast and reliable method, that yields simple and easily quantifiable results. Past studies have shown this coleoptile assay to yield results comparable to wheat head infection assays [36][37][38] , validating the effectiveness of this assay as a means to assess infection of wheat by F. graminearum. Pathogenicity was quantified by measuring the extent of the lesion formed on the 'Roblin' coleoptile stalk after 10 days of incubation with F. graminearum conidia (Fig. 5a). Inoculation of coleoptiles with wild type F. graminearum resulted in dense mycelial growth originating from the wound site, as well as formation of a 5.3 ± 1.9 mm lesion on the stalk. Conversely, all three Fgste2Δ strains showed a significant decrease in the extent of infection of the coleoptile (Fig. 5b). As expected, reintroduction of a wild type copy of STE2 restored pathogenicity of F. graminearum (Fig. 5c).
Exposure of F. graminearum to peroxidase leads to CWI MAPK signaling. As Ste2-mediated stimuli has been shown to be transduced through MAPK signaling pathways, immunoblotting was used to monitor the phosphorylation of FgGpmk1 of the invasive growth MAPK pathway and FgMgv1 of the CWI MAPK pathway upon exposure to HRP (Fig. 6a). Untreated wild type F. graminearum exhibited a basal level of both phosphorylated Mgv1 and Gpmk1 (Fig. 6b). Exposure to HRP for 1 h resulted in a 2.5-fold increase in phosphorylation of FgMgv1 compared to the control (Fig. 6c), whereas no significant difference in FgGpmk1 phosphorylation was observed (Fig. 6d).
To further understand the signal transduction pathway initiated by FgSte2 activation, chemotropism was assessed for F. graminearum lacking selected genes involved in, and/or associated with, the CWI MAPK pathway. A deletion mutant of the MAPK in the CWI pathway, Fgmgv1Δ 39 (kindly provided by Dr. Rajagopal  www.nature.com/scientificreports/ Subramaniam), lacked all chemotropic response towards Fg α-pheromone (Fig. 7a), signifying that this pathway is involved in relaying the signal from pheromone-activated FgSte2. The Fgmgv1Δ mutant was also unable to sense and grow towards 'Roblin' exudate, or HRP, further supporting the fact that F. graminearum requires this pathway for sensing and mediating chemotropism towards the host. However, the Fgmgv1Δ strain retained the ability to sense and grow towards methionine, reiterating that the pathway responsible for chemotropism towards the host is separate from that involved in nutrient response. A second MGV1 mutant strain of F. graminearum with constitutively overexpressed MGV1 (FgMGV1-OX) was generated and characterized (Supplemental Figure S6a, c). Despite the expression of FgMGV1 in this mutant being four times higher than wild type (Supplemental Figure S6b), the chemotropic responses observed in FgMGV1-OX towards Fg α-pheromone, 'Roblin' exudate, HRP, and methionine did not differ significantly from those observed in wild type F. graminearum (Fig. 7a).
Bmh1 and Bmh2 are 14-3-3 adaptor proteins known to associate with the CWI MAPK signaling pathways in S. cerevisiae 40 . Orthologues of these proteins in filamentous fungi have been implicated in a variety of processes, including cell cycle progression and cell growth 41,42 , germ tube development and growth 43 , and even repression of secondary metabolite production 44,45 . Single deletion mutants of these orthologous F. graminearum 46 proteins, Fgbmh1Δ and Fgbmh2Δ, were tested in the chemotropism plate assay (Fig. 7b). When exposed to Fg www.nature.com/scientificreports/ α-pheromone, 'Roblin' exudate or HRP, no chemotropic responses were observed in either Fgbmh1Δ or Fgbmh2Δ (Fig. 7b). Lastly, both Fgbmh1Δ and Fgbmh2Δ retained robust, wild type-like responses to methionine.

Discussion
Since the first report of gene disruption in F. graminearum in 1995 47 , several studies have been implemented to investigate the role of candidate virulence factors and other fungal genes involved in the interaction between F. graminearum and its cereal hosts. However, knowledge about the fungal cell surface receptors that sense and mediate chemotropic growth of F. graminearum has been lacking. The primary objective of this research was to understand the mechanisms underlying host sensing and chemotropism by F. graminearum. Many of the nutrients tested in our study have previously been assessed in the context of growth stimulation and mycotoxin induction in F. graminearum, but not chemotropism. Unexpectedly, F. graminearum exhibited only weak and variable chemotropic responses towards the carbon sources tested, in contrast to the strong response that was observed in F. oxysporum toward glucose 5 . While glucose and galactose have been shown to stimulate growth in F. graminearum, they are not important for pathogenic traits such as mycotoxin synthesis 48 , consistent with the findings reported here that they do not serve as chemotropic stimuli. Among nitrogencontaining compounds, methionine induced the most robust chemotropic response, as well as extensive hyphal growth and branching in wild type F. graminearum. Methionine uptake has been demonstrated to be involved in the induction of trichothecene mycotoxins 49 . Although F. oxysporum did not respond to methionine 5 , other filamentous fungi such as Achlya bisexualis 50,51 and Achlya ambisexualis 52 exhibit methionine-driven chemotropism with similar hyphal branching patterns as observed for F. graminearum. In agreement with our observations, aspartate and glutamate have previously been shown to induce growth in F. graminearum 49 . Interestingly, previous work also identified wheat anthers 53 , and specifically choline and betaine, as selectively stimulating F. graminearum conidial growth in a plate assay [54][55][56][57] . The lack of chemotropic response towards betaine herein, suggest that while it is likely a growth stimulant of F. graminearum, it does not induce chemotropism. Phospho-and total MAPK for both were probed for in an untreated control (C) and HRP-induced (HRP) condition in wild type F. graminearum. For normalization of quantification, α-tubulin was used. Molecular weights of detected proteins are indicated on the blot. Images were cropped using ImageJ (https ://image j.nih.gov/ij/). (c,d) Quantification analysis was performed using ImageJ software. The intensity of pMgv1 and pGpmk1 bands were normalized to tubulin, and the ratio of intensities of induced compared to uninduced samples were determined (compared to uninduced sample, *P < 0.05). Data represents the average of three independent experiments. Error bars represent standard deviation. Graphs were plotted using Graphpad Prism version 6.01 (https ://www.graph pad.com). The figure was compiled using Adobe Illustrator CC 2015 (https ://www.adobe .com).
Scientific RepoRtS | (2020) 10:10770 | https://doi.org/10.1038/s41598-020-67597-z www.nature.com/scientificreports/ For a long time, research on fungal G-protein-coupled receptors was largely limited to the Ste2p receptor of S. cerevisiae, which was used as a model system to study GPCR signaling and mating in yeast. Over the past two decades, pheromone receptors in other fungi, including filamentous fungi like F. graminearum 27 , were identified and their roles in mating or sexual reproduction characterized. Pheromone-induced Ste2-and Ste3-mediated chemotropism has been observed in S. cerevisiae and Neurospora crassa 58 , respectively, and both of these organisms rely on this process for mating and sexual reproduction. More recently, Ste2-and Ste3-mediated chemotropism towards α-and a-pheromone was demonstrated in F. oxysporum 5,25 . Furthermore, this pair of receptors has been shown to be involved in the regulation of autocrine pheromone signaling and conidial germination in F. oxysporum 25 . While the most obvious role for Ste2 in mating has been extensively studied in heterothallic organisms, its relevance in homothallic fungi such as F. graminearum remains enigmatic. Despite not needing a partner for sexual reproduction, F. graminearum can only undergo sexual reproduction when both Ste2 and Ste3 are co-expressed; deletion of either MAT gene results in an obligate heterothallic strain that can be outcrossed 59 . Our results highlight that even though F. graminearum does not need to mate to reproduce, the Ste2 receptordriven chemotropism mechanism observed in other fungi is maintained in this species.
In addition to an α-pheromone response, FgSte2 also contributed to the detection of and mediation of chemotropism towards the activity of peroxidases from both wheat and horseradish. This demonstrates the ability of FgSte2 to recognize multiple ligands and reveals the conserved role of Ste2 in a second Fusarium species in hostsensing. Specifically, FgSte2 responds to the catalytic product of a class III peroxidase secreted from the wheat head. Production of reactive oxygen species (ROS) and secreted peroxidases is a universally utilized strategy by plants to defend themselves against pathogens [60][61][62] . Indeed, wheat infected with F. graminearum 63,64 and other pathogens such as Puccinia triticina 65 show increased expression of peroxidases up to 48 h after infection. Additionally, a previous study from our group 33 showed a significant upregulation of three of the four wheat peroxidases identified in the 'Roblin' exudate in wheat infected with F. graminearum. Despite the higher amounts of peroxidase secreted by F. graminearum-infected wheat, the significantly lower extent of infection by the Fgste2Δ strains placed directly on the wound site emphasizes the importance of the FgSte2 receptor in pathogenicity.
F. graminearum encodes three MAPK proteins orthologous to those found in S. cerevisiae. These are FgMgv1, FgGpmk1 and FgHog1 with ascribed function in cell wall integrity and remodelling 28 , pathogenicity and invasion 29,30 , and osmotic stress response 31 , respectively. All three MAPKs have been implicated in pathogenicity in filamentous fungal pathogens (reviewed by di Pietro et al. 32 ). The Ste2-mediated response to HRP was found to be governed by the cell wall integrity pathway, similar to that observed in F. oxysporum 5  and Fgbmh2Δ strains of F. graminearum towards a gradient of the indicated chemical stimuli (versus water control, ****P < 0.0001). n = 500 hyphae. Data represents the average of at least three replicates. Error bars represent standard deviation. Graphs were plotted using Graphpad Prism version 6.01 (https ://www.graph pad. com). Figure was compiled  Unlike root-colonizing fungi that exhibit chemotropism towards compounds diffusing through the soil, F. graminearum conidia are dispersed from overwintering crop debris onto the wheat head through various physical means and are thus already on the wheat head prior to initiation of chemotropism. Based on the findings presented herein, two mechanisms of peroxidase-stimulated chemotropism of F. graminearum are proposed. First, F. graminearum secretes cell wall degrading enzymes (CWDE) that would stimulate an increase in peroxidase secretion by the wheat to reinforce the cell wall at that site 61,70 . Higher concentrations of peroxidase-derived chemoattractant could increase the affinity or probability of F. graminearum invasion. Second, these aforementioned sites of increased peroxidase secretion would result in the release or diffusion of peroxidase-derived product and attract more distal conidia to invade at these same sites. This is supported by our finding that conidia of Fgste2Δ strain placed directly on a wound site displayed a significantly lower extent of infection when compared to the wild type strain. Elements involved in relaying the peroxidase-induced stimulus from FgSte2 have also been implicated in pathogenicity of wheat; deletion of FgMgv1 28 and FgBmh2 46 result in a significant reduction of pathogenicity. While FgBmh1 has been reported to be dispensible for F. graminearum infection of wheat 46 , it appears to be involved in mediating chemotropic response towards the wheat peroxidase-derived product. FgBmh1 and FgBmh2 have been shown to be involved in sensing nitrogen-containing compounds ammonium nitrate and sodium nitrate 46 , however, deletion of either protein does not affect the response of F. graminearum towards methionine, reiterating that methionine uptake occurs through a different mechanism.

. Consistent
Conventional methods of controlling F. graminearum include application of fungicides and use of more resistant wheat cultivars 71 . More recently, biological control agents 72,73 ,small RNA interference 74,75 and generation of cultivars overexpressing genes conferring F. graminearum resistance [76][77][78] have emerged as potential methods of managing F. graminearum. In any case, existing disease management methods have their limitations 71,79 and increasing fungicide resistance, limited FHB-resistant wheat cultivars and changing climate conditions are confounding factors in keeping FHB under control. Understanding the mechanism of infection by F. graminearum is essential and will potentially result in more specific targets for fungal inhibition to reduce the devastating consequences of this fungal disease.

Materials and methods
Fungal strains, culture conditions and maintenance. Fungal strains used in this study are listed in Table 1. Macroconidia from all strains were obtained through cultures in liquid carboxymethylcellulose (CMC) medium 80 at 28 °C with shaking at 170 rpm in the dark. Routine maintenance of strains was done on Potato Dextrose Agar (PDA) plates. Plugs of F. graminearum strains grown on SNA 81 or conidial suspensions were stored long-term at − 80 °C in 15% glycerol. For macroconidia harvest, the liquid CMC cultures were filtered through four layers of sterilized cheesecloth and the filtrate was centrifuged at 3,400 g for 10 min at 4 °C. The macroconidia were washed in sterile water twice, resuspended in 1-3 mL of sterile water, and quantified with a hemocytometer. www.nature.com/scientificreports/ Wheat growth conditions. Wheat cultivars 'Roblin' , 'Sumai3' and 'Wuhan' were grown in an AC-60 growth chamber (Enconair) at the greenhouse facility in the Dept. of Biology at Queen's University. Wheat seeds were kindly provided by Dr. Thérèse Ouellet (AAFC, Ottawa, Canada). Growth light and temperature conditions were 20 °C day, 16 °C night, with a 16 h photoperiod (750 µmol photons/m 2 × s). Soil was made of 1:1:1 topsoil, sand, Pro-Mix. Fertilizer solution "20-20-20" at a concentration of 2 g/L was administered weekly.
Quantitative chemotropism plate assay. Chemotropism assays were performed as described previously 5 , with minor modifications. Briefly, fresh F. graminearum macroconidia were mixed with 0.5% (w/v) water agar to a final concentration of 2.5 × 10 5 spores per mL and plated in a Petri dish. A scoring line was drawn down the middle of the plate and two wells were made 5 mm away and parallel to the scoring line. Equal volumes (50 µL) of sterile water and test compound were pipetted into the control well and test well, respectively. Tested compounds were: 50% (v/v) methanol (MeOH), ammonium sulfate ((NH 4 ) 2 SO 4 ), glucose (Gluc), glycerol (Glyc), galactose (Gal), all at 50 mM; methionine (Met), sodium aspartate (Asp), sodium glutamate (Glu), all at 295 mM; and 0.1 M betaine. Chemotropic response of F. graminearum towards wheat was tested for each of three cultivars. To measure chemotropism towards wheat, the flowering wheat head still attached to the live plant was placed directly into the test well containing sterile water (as shown in Supplemental Figure S2a). Plates were incubated for approximately 14 h at 22 °C in the dark. The number of germinating hyphae growing towards the test (N test ) or control compound (N cont ) were counted under the Nikon SMZ1000 microscope and a chemotropic index was calculated as C.I. = Ntest−Ncontrol Ntest+Ncontrol × 100% . While only hyphae with angles of approximately 45° or less with respect to the direction of the gradient of test or control compounds were included in the count, no strict criteria for inclusion of hyphae based on length was used. For each compound, a minimum of 500 macroconidia per plate were counted. All experiments were repeated at least three times. Statistical analyses were conducted using Student's t-test and one-way ANOVA on GraphPad Prism version 6.
Commercially available horseradish peroxidase (HRP) was assayed at a concentration of 4 µM. To study the chemoattractive nature of HRP, the enzyme was inhibited by salicylhydroxamic acid (60 mM) (SHAM) for 5 min, heat-denatured at 95 °C for 10 min or proteolyzed by proteinase K (1 mg/mL) for 30 min at room temperature and then assayed. Synthetic F. graminearum (Fg) (WCTWKGQPCW) and S. cerevisiae (Sc) (WHWLQLK-PGQPMY) α-pheromone peptides were synthesized. Pheromones were reconstituted in 50% (v/v) methanol in water and used in the chemotropism assay at a final concentration of 378 µM (final methanol concentration 2.5%). Fg α-pheromone was treated with proteinase K solution (1 mg/mL) for 30 min. The reaction was stopped with 1 mM phenylmethylsulfonyl fluoride (PMSF) and tested in the chemotropism assay.
For hyphal length and angle measurements, light microscopy images of chemotropism plates containing HRP as the test compound were taken on an Olympus SZX10 microscope fitted with a DP27 camera. Hyphal lengths and angles, with respect to the HRP gradient, of at least 300 germinating conidia were measured using ImageJ 82 . The experiment was performed twice. Statistical analysis was conducted using Student's t-test.
'Roblin' exudate production and identification of associated wheat peroxidases. Two flowering 'Roblin' wheat heads still attached to the plant were submerged in 25 mL of sterile water and incubated for 48 h at room temperature. Exudate was concentrated either 200-or 300-fold using an Amicon Ultra centrifugal filter (Millipore, cutoff 3 kDa) and stored at 4 °C until further usage in chemotropism or peroxidase activity assays and for protein sequencing. The exudates were assayed directly for peroxidase activity in 96-well plates at 22 °C as described previously 83 . The formation of the pyrogallol oxidation product (extinction coefficient, ε 420 , 4,400 M -1 cm -1 ) was measured spectrophotometrically at 420 nm.
Concentrated 'Roblin' exudate was run on a 10% SDS-polyacrylamide gel and stained with Coomassie blue. The resolved bands corresponding to proteins of molecular weight 34 and 37 kDa were cut out of the gel and sent to Mass Spectrometry Research, SPARC Biocentre, Hospital for Sick Children, where they were subjected to tryptic digestion. The peptide fragments were analyzed by LC-MS/MS and used to identify the proteins present in the corresponding bands. Data was analyzed using the Scaffold 4 software.
Fungal genomic DNA isolation. Mycelia were collected from two-day-old F. graminearum liquid cultures through filtration, and ground into a fine powder in liquid nitrogen. Genomic DNA (gDNA) was then isolated from the ground tissue using the E.Z.N.A Fungal DNA Mini Kit and eluted in sterile water.

Construction of vectors.
Vectors used for generating the various mutant strains were constructed using Uracil-Specific Excision Reagent (USER) technology 84,85 . All cloning, sequencing and screening primers used in this study are listed in Supplementary Table 1.
To construct the vector for generating the STE2 deletion mutant (Fgste2Δ), flanking regions of the STE2 gene, which constitute the homologous recombination sequences (HRS) were amplified by polymerase chain reaction (PCR) with primers P1-P4 from wild type F. graminearum gDNA using Pfu Cx Turbo Hotstart polymerase (Agilent Technologies). The amplified fragments were cloned into the pRF-HU2 vector 85 flanking a hygromycin B phosphotransferase (HPH) gene using USER enzyme mix (NEB). Correct orientation of the inserts in the plasmid was confirmed through PCR and DNA sequencing (P5-P10).
The vector used to generate the STE2 complement (Fgste2Δ + STE2) was constructed by amplifying the STE2 gene, with 1 kb upstream and downstream flanking regions, using primers P17-P18 and cloned into the linearized pRF-GU vector upstream of the geneticin-resistance gene, aminoglycoside 3′phosphotransferase (GEN) 85 . The generated vector was verified by PCR (P11-P12 and P26-P27) and sequenced with primers P19-P25. www.nature.com/scientificreports/ A similar method was employed in generating the MGV1 over-expression vector using the pRF-HU2E vector which is designed for in locus overexpression driven by the Aspergillus nidulans GAPDH promoter 85 , was used. In this case, primers P28-P31 were used to generate the expression cassette in pRF-HU2E.

Agrobacterium tumefaciens-mediated transformation (ATMT). Vectors were transformed into
Agrobacterium tumefaciens strain LBA4404 by electroporation, and transformants were confirmed by PCR. The reagents and protocol required for ATMT were described by Frandsen 86 .
Correct integration of the knockout cassette and replacement of STE2 with HPH was confirmed by PCR amplification, across the upstream junction formed between the genomic DNA and knockout cassette following homologous recombination with primers P13-P16. Insertion of the complementation cassette was confirmed by PCR with primers P11-P12 and P26-P27. The site of integration of the complementation cassette was confirmed in a similar manner to that of the knockout strain using primers P26-P16 and sequencing of the region of cassette integration into the genomic DNA.
Correct insertion of the MGV1 overexpression cassette into F. graminearum was corroborated using the primers P32-P37 and the MGV1 sequence was confirmed through sequencing at the Genome Quebec service lab (Montréal, Québec, Canada). This was followed by confirmation of single copy in locus insertion of the MGV1 transgene by quantitative PCR (qPCR). gDNA was isolated from five-day-old F. graminearum mycelium culture on PDA plate using the DNeasy Plant Mini Kit (QIAgen). qPCR was employed on the gDNA of the MGV1 overexpression strains (FgMGV1-OX) and the wild type strain to determine the copy number of MGV1 (amplified with primers P38-P39) and the housekeeping gene ß -TUBULIN (primers P40-P41) which served as a control. qPCR was carried out using PerfeCTa SYBR Green SuperMix Low ROX (Quantabio) on a Quant Studio 6 Flex Real-Time PCR System (Applied Biosystems) as described previously 87 . Copy number estimation was calculated against the standard curve, which was generated using the MGV1 and ß-TUBULIN genes (previously amplified from the wild type gDNA).
The overexpression of MGV1 was verified through reverse-transcription qPCR (RT-qPCR). RNA extraction, cDNA synthesis and qPCR were carried out as previously described 87 . The qPCR reaction was performed in 10 µL with PerfeCTa SYBR Green SuperMix Low ROX (Quantabio) and primers P38-P41. Three biological replicates with three technical replicates were included in each reaction along with the negative controls. Standard curve calculations were used to normalize the data to housekeeping genes and to estimate the relative expression of MGV1 in FgMGV1-OX compared to wild type using the QIAgen's Relative Expression Software Tool (REST) (p < 0.05).

Coleoptile infection assay.
Infection of germinating coleoptiles with the various STE2 mutants was carried out as previously described 88 . Briefly, 16 'Roblin' seeds per strain to be tested were placed on ½ MS media in 0.7% (w/v) agar in water in autoclaved Magenta boxes and stratified overnight at 4 °C in the dark. The Magenta boxes were then placed in the growth chamber and coleoptiles were grown until at least 1 cm in height, where 12-16 seeds germinated per strain. Sterile scissors were used to cut 1 mm off the top of the coleoptile and a cotton soaked with macroconidial suspension (2 × 10 5 spores per mL) was wrapped around the wound site. The Magenta boxes were placed in the growth chambers to allow symptom development. After ten days, the length of coleoptile stalks infected were measured for each strain. The experiment was performed twice. Statistical analysis was conducted by one-way ANOVA.
Conidial length quantification. Conidia of wild type and STE2 mutants of F. graminearum were imaged using an Olympus IX83 inverted microscope fitted with a 20 × objective and an Andor Zyla 4.2 Plus camera controlled by the cellSens software. Two hundred conidia were measured per strain and conidial lengths were quantified using Image J. Statistical analysis was conducted by one-way ANOVA.
Immunoblotting of MAPKs. Approximately 10 5 spores of wild type F. graminearum were inoculated in 20 mL of liquid Potato Dextrose Broth (PDB) and grown for 24 h at 28 °C in the dark. The growing culture was exposed to 0.05 µM HRP or water control for 1 h, and the cells were then lysed as previously described 89 with some modifications. Briefly, the mycelia were collected by filtering the culture through Whatman filter paper, finely ground in liquid nitrogen and resuspended in 1 mL of protein extraction buffer (1 M NaCl, 50 mM sodium phosphate, pH 8.0, 50 mM NaF, 1 mM PMSF, 0.2% β-mercaptoethanol, protease inhibitor cocktail, and phosphatase inhibitor cocktail). The cell lysate was homogenized by vortexing and then centrifuged at 13,000 × g. The supernatant was transferred to a fresh microfuge tube and total protein concentration was quantified with Bradford assay 90 . Twenty micrograms of total protein of each sample were loaded and resolved on a 12% SDS polyacrylamide gel and transferred to a polyvinylidene difluoride (PVDF) membrane by wet electroblotting at 400 mA for 2 h. The membranes were blocked for 1 h in 5% (w/v) non-fat dried skim milk in TBST (50 mM Tris, pH 7.5, 150 mM NaCl, 0.05% (v/v) Tween 20) at 4 °C. The membranes were subsequently incubated with either anti-p44/42 MAP kinase (1:1,000 dilution, M5670, Millipore Sigma) or anti-phospho p44/42 MAP kinase (1:1,000 dilution, CST #9101, Cell Signaling Technology) primary antibodies. The membranes were then incubated with anti-rabbit IgG secondary antibodies (1:5,000 dilution, 7074S, Cell Signal Technology). Pierce Enhanced Chemiluminescent substrate was added to the membranes and the emitted light was captured on an x-ray film. The same membranes were then re-probed for α-tubulin (1:1000, sc53030, Santa Cruz Biotechnology) as a loading control. Quantification was performed using ImageJ. The experiment was repeated three times with independent sets of samples and analyzed by Student's t-test.