Transcription factor ART1 mediates starch hydrolysis and mycotoxin production in Fusarium graminearum and F. verticillioides

Summary

Molecular mechanisms underlying the responses to environmental factors, such as nitrogen, carbon and pH, involve components that regulate the production of secondary metabolites, including mycotoxins. In this study, we identified and characterized a gene in the FGSG_02083 locus, designated as FgArt1, which was predicted to encode a Zn(II)₂Cys₆ zinc finger transcription factor. An FgArt1 deletion mutant of F usarium graminearum exhibited impaired starch hydrolysis as a result of significantly reduced α‐amylase gene expression. The deletion strain was unable to produce trichothecenes and exhibited low Tri5 and Tri6 expression levels, whereas the complemented strain showed a similar ability to produce trichothecenes as the wild‐type strain. In addition, FgArt1 deletion resulted in impairment of germination in starch liquid medium and reduced pathogenicity on flowering wheat heads. To investigate the roles of the FgArt1 homologue in F . verticillioides, we deleted the FVEG_02083 gene, and the resulting strain showed defects in starch hydrolysis, similar to the FgArt1 deletion strain, and produced no detectable level of fumonisin B ₁. Fum1 and Fum12 expression levels were undetectable in the deletion strain. However, when the FvArt1‐deleted F . verticillioides strain was complemented with FgArt1 , the resulting strain was unable to recover the production of fumonisin B ₁, although FgArt1 expression and starch hydrolysis were induced. Thus, our results suggest that there are different regulatory pathways governed by each ART1 transcription factor in trichothecene and fumonisin biosynthesis. Taken together, we suggest that ART1 plays an important role in both trichothecene and fumonisin biosynthesis by the regulation of genes involved in starch hydrolysis.

Introduction

Fusarium graminearum is an economically important plant pathogen that causes diseases in major cereal crops, such as maize, wheat and barley (Leslie and Summerell, 2006). During infection of plants, this pathogen produces various mycotoxins, including deoxynivalenol (DON), nivalenol (NIV) and their acetylated derivatives, which are harmful to humans and animals (Desjardins, 2006). The presence of these mycotoxins in edible grains has become a critical issue in food safety because the toxins are heat stable and are not degraded during food processing (Bullerman and Bianchini, 2007). To date, one of the best strategies for the control of contamination of food by trichothecenes is to limit their biosynthesis before harvesting crops (Magan and Aldred, 2007). Although considerable efforts have been made in the detection and regulation of mycotoxins in infected crops, there is still limited information about the regulatory mechanisms of trichothecene biosynthesis.

Over the past decade, the trichothecene biosynthetic pathway has been studied extensively, and genes (Tri genes) involved in the formation of trichothecenes have been well characterized (Alexander et al., 2009; Desjardins, 2006). In F. graminearum, most of the Tri genes are located within a cluster on chromosome 2, whereas Tri1, Tri16 and Tri101 are located in different chromosomal regions (Alexander et al., 2009; Brown et al., 2004; Gale et al., 2005). In addition to functional studies of genes encoding enzymes that catalyse biochemical reactions in trichothecene formation, TRI10 and TRI6 have been determined to be required for trichothecene biosynthesis and to coordinate the transcriptional expression of pathway‐specific Tri genes (Seong et al., 2009).

Trichothecene biosynthesis in F. graminearum has been investigated in the presence of various environmental factors, such as nitrogen and carbon sources, pH and reactive oxygen species (ROS) (Audenaert et al., 2010; Gardiner et al., 2009; Jiao et al., 2008; Miller and Greenhalgh, 1985). Related to nitrogen sources, it has been shown to be suppressed by ammonium, which is preferentially utilized over other nitrogen sources, whereas nitrate and arginine induce DON biosynthesis (Gardiner et al., 2009). Furthermore, the expression of AREA (FGSG_08634), which is a global regulator of nitrogen metabolism, is responsible for the repression of trichothecene production by ammonium, and AREA has also been shown to interact with TRI10 (Hou et al., 2015; Min et al., 2012). With regard to pH as an environmental signal, trichothecene production by F. graminearum is induced only under acidic pH conditions (Merhej et al., 2011). Deletion of the Pac1 gene, which plays a key role in pH regulatory mechanisms, has been shown to result in poor growth under neutral and alkaline pH, but to have no effect on growth or trichothecene production under acidic conditions. However, constitutive expression of Pac1 leads to strong repression of Tri genes and toxin production regardless of pH, indicating that Pac1 negatively regulates Tri gene expression and trichothecene production in F. graminearum (Merhej et al., 2011). Similarly, these regulatory mechanisms mediated by AreA and Pac1 have been observed in fumonisin biosynthesis in F. verticillioides (Flaherty et al., 2003; Kim and Woloshuk, 2008).

Several studies have been performed to discern the effects of carbon sources on trichothecene production by F. graminearum (Jiao et al., 2008; Kawakami et al., 2014; Miller and Greenhalgh, 1985). Jiao et al. (2008) have found that, although the F. graminearum H3 strain grows equally well on medium when glucose or sucrose are supplied as the carbon source, DON production is poor in the presence of glucose alone. However, sucrose supports DON production without a discernible concentration effect. In addition, increasing amounts of both glucose and sucrose do not affect DON production, suggesting that carbon catabolite repression is not a key regulatory mechanism of DON production (Jiao et al., 2008). In F. verticillioides, considerable efforts have been made to elucidate the fumonisin regulatory mechanisms associated with carbon metabolism (Bluhm and Woloshuk, 2005; Bluhm et al., 2008; Kim and Woloshuk, 2011; Kim et al., 2011; Shim et al., 2003). In maize kernels, F. verticillioides has been shown to produce 10‐ to 20‐fold greater fumonisin B₁ (FB₁) levels in endosperm tissues than in germ tissues, and an α‐amylase gene (FVEG_07545)‐disrupted mutant has been demonstrated to exhibit low levels of FB₁ production in endosperms (Bluhm and Woloshuk, 2005; Shim et al., 2003). Hxk1 (a putative hexose kinase; FVEG_00957) and Fst1 (a putative hexose transporter; FVEG_08441) have been demonstrated to be involved in carbon utilization and to be required for fumonisin biosynthesis (Kim and Woloshuk, 2011; Kim et al., 2011). Recently, Niu et al. (2015) have shown that several genes involved in fumonisin biosynthesis and starch degradation are down‐regulated following Fst1 deletion. Thus, the utilization of starch or other carbohydrates by fungal pathogens is important for their secondary metabolism and colonization on plants; however, the signal transduction networks in F. graminearum that regulate trichothecene biosynthesis in response to carbohydrate availability remain relatively unclear.

The purpose of this study was to identify a transcriptional regulator in F. graminearum, which is involved in starch hydrolysis and has effects on trichothecene production. Here, we identified and characterized FGSG_02083, designated as FgArt1, which was predicted to encode a Zn(II)₂Cys₆ zinc finger transcription factor. An FgArt1 deletion mutant exhibited defects in starch hydrolysis, trichothecene production and pathogenicity. In addition, to investigate the role of the FgArt1 homologue in F. verticillioides, we deleted the FVEG_07545 gene and found that the resulting strain also showed defects in starch hydrolysis and produced no detectable FB₁. Taken together, our results show that the ART1 transcription factor plays important roles in both trichothecene and fumonisin biosynthesis by regulating the genes involved in starch hydrolysis.

Results

Starch induces trichothecene production in F. graminearum

To investigate the effects of carbon sources on trichothecene production, DON and 15‐acetyldeoxynivalenol (15‐ADON) were analysed in cultures in which F. graminearum wild‐type strain Z‐3639 was grown in liquid agmatine minimal medium (AMM) containing glucose, sucrose, maltose, amylopectin, cellulose or starch as the sole carbon source. We found that sucrose and starch supported high levels of trichothecene production, but that no detectable trichothecenes were present in the cultures supplemented with glucose, maltose, amylopectin or cellulose (Fig. 1). However, dry weight measurements indicated that the growth of F. graminearum Z‐3639 strain was largely similar in liquid AMM containing glucose, sucrose, maltose, amylopectin or starch, but that it differed in the presence of cellulose (Fig. 1). In addition to the finding of the importance of sucrose to trichothecene biosynthesis, which is consistent with previous observations (Jiao et al., 2008; Kim et al., 2014), the finding that starch promotes the production of a considerable amount of these mycotoxins gave rise to the question of whether transcriptional regulatory genes involved in starch hydrolysis affect their biosynthesis. Previously, we generated a mutant library, in which 657 F. graminearum genes encoding putative transcription factors were individually deleted, and 26 mutants were found to exhibit defective DON production (Son et al., 2011b). Thus, we investigated the growth and starch hydrolysis of 26 mutants on starch medium, and consequently found that an FGSG_02083 deletion strain exhibited impairments in these processes (Table S1, see Supporting Information).

Figure 1.

Fungal growth and trichothecene production of the F usarium graminearum Z‐3639 strain. A spore suspension from the strain was inoculated into liquid agmatine minimal medium (AMM) containing various carbon sources. After 7 days of incubation, the fungal dry mass and trichothecene level were determined using data obtained from three independent biological replicates. 15‐ADON, 15‐acetyldeoxynivalenol; AP, amylopectin from corn; CMC, carboxymethylcellulose; DON, deoxynivalenol; Glu, glucose; Mal, maltose; MDW, mycelial dry weight; n.d., not detected; Suc, sucrose.

Identification of Zn(II)₂Cys₆ transcription factor FgArt1

Computational analysis was performed to understand the predicted functions of FGSG_02083, which has an open reading frame (ORF) of 2380 bp in size interrupted by a 56‐bp intron. BLASTp and InterPro analyses predicted that it encodes 656 amino acids, and it was annotated as the putative transcriptional activator AMYR, which contains all of the predicted features of a transcription factor, including a conserved Zn(II)₂Cys₆ binuclear motif (IPR001138; residues 4–48) and a fungal transcription factor domain (IPR007219; residues 102–270) (Fig. S1A, B, see Supporting Information). In this study, we designated the FGSG_02083 gene as FgArt1, or F. graminearum amylase regulatory transcription factor 1. FgART1 also contains a nuclear localization signal at the N‐terminal region (KKGPKGSR; residues 44–51). With the exception of these domains, no other known motifs exist in FgART1.

BLASTMatrix was used to perform comparative analysis of FgART1 homologues in other fungi, showing that it is relatively conserved among species of the subphylum Sordariomycetes of the Ascomycota compared with the phyla Oomycota and Basidiomycota and the subphyla Saccharomycotina and Pezizomycotina (Fig. S1C). Although FgART1 was annotated as a transcriptional regulator of amylase, it showed relatively low sequence identity with AMYR in Aspergillus nidulans (29%) and BGLR in Trichoderma reesei (41%), which have been functionally characterized as regulators of α‐amylase and β‐glucosidase, respectively (Nitta et al., 2012; Tani et al., 2001b). Phylogenetic analysis showed that AMYR of A. nidulans was more similar to the proteins FGSG_00069 and FGSG_03702 than to FgART1; FGSG_00069 and FGSG_03702 were also annotated as Zn(II)₂Cys₆‐type transcription factors (Fig. 2A). Notably, transcriptional regulators, such as AMYR of Aspergillus spp., are known to be located within a small cluster containing amylolytic genes (Gomi et al., 2000), but FgART1 was not found to be clustered with amylolytic genes (data not shown). In contrast, several AMYR homologues of F. graminearum, such as FGSG_00069, FGSG_03702 and FGSG_03892, were present in these types of clusters (Fig. 2B). To investigate the functional conservation of AMYR homologues in F. graminearum, we explored the ability of starch hydrolysis from FGSG_00069, FGSG_03702 and FGSG_03892 deletion strains. The results showed that the mutants had similar capacities for starch hydrolysis as the wild‐type strain, suggesting that the FGSG_00069, FGSG_03702 and FGSG_03892 genes do not regulate amylolytic gene expression (Fig. 2C).

Figure 2.

Comparison of putative amylase regulators in fungi. (A) Phylogenetic analysis of FgART1. The tree was generated using mega5 with 1000 bootstraps based on alignments of the full amino acid sequences of the homologues. Abbreviations for the fungal species are as follows: A n, A spergillus nidulans; A o, A spergillus oryzae; F g, F usarium graminearum; F v, F usarium verticillioides; M o, M agnaporthe oryzae; S c, S accharomyces cerevisiae; T r, T richoderma reesei. (B) Schematic gene organization of AMYR homologues in F . graminearum compared with the AMYR cluster of A spergillus spp. FGSG_00069, FGSG_03702 and FGSG_03892 were determined to be representative loci by a blastp search of A . nidulans  AMYR. (C) Starch hydrolysis of F . graminearum strains. Three‐day‐old culture medium containing starch as the carbon source was stained with iodine solution. FgWT, F . graminearum wild‐type strain; Δ00069, FGSG_00069 deletion mutant; Δ03702, FGSG_03702 deletion strain; Δ03892, FGSG_03892 deletion strain; Mock, non‐inoculated medium.

Growth of strain ΔFgArt1 on various carbon sources

Visual assessments of culture plates showed that the ΔFgArt1 strain grew normally on complete medium (CM) and minimal medium (MM) containing sucrose or maltose (Fig. 3A). However, the mutant exhibited reduced aerial mycelia on MM compared with the wild type when provided with glucose or starch as the sole carbon source, although there was no significant difference in colony diameter. When the ΔFgArt1 strain was grown in liquid MM containing starch, a growth defect was apparent, with reduced germination compared with the wild type; the number of germinated spores of the ΔFgArt1 strain was 10% of that of the wild‐type (Fig. 3B, C). These observations suggest that the ΔFgArt1 strain has a defect in the perception or utilization of starch, particularly when grown in liquid medium with shaking. To confirm the effects of FgArt1 on fungal development, we generated a complementary strain by co‐transforming the FgArt1 gene and the hygromycin resistance gene into a protoplast of the ΔFgArt1 strain, and the resulting strain exhibited an indistinguishable phenotype compared with the wild type on CM and MM in the presence of various carbon sources (Figs 3 and S2, see Supporting Information).

Figure 3.

Growth and development of ΔFgArt1 strain in response to starch. (A) Vegetative growth of each strain on complete medium (CM) and minimal medium (MM) plates containing glucose, sucrose, maltose or starch. Photographs were taken at 4 days after inoculation. FgWT, F usarium graminearum wild‐type strain; ΔFgArt1 ,  FgArt1 deletion strain; HK025, complemented strain of ΔFgArt1. (B) F usarium graminearum strains were grown in starch liquid medium with shaking at 150 rpm. Photographs were taken at 4 days after inoculation. (C) Quantification of germinated spores. A spore suspension was inoculated into liquid MM containing different carbon sources, and the numbers of germinated spores were counted at 24 h after inoculation. Glu, glucose; ST, starch; Suc, sucrose.

Regulation of amylolytic genes by FgART1

Given that FgART1 has a nuclear localization signal and a zinc finger DNA‐binding domain, it was hypothesized that FgART1 is localized to nuclei. To clarify this hypothesis, we observed green fluorescent protein (GFP) signals in a complementary strain, generated by transformation of an FgArt1::gfp fusion construct under the control of the native promoter into the ΔFgArt1 strain, designated as HK025 (Fig. S2). For further confirmation of the nuclear localization of FgART1, the HK025 strain was outcrossed with a mat1r strain that contained red fluorescent protein (RFP) fused to histone H1 protein in a MAT1‐1 deletion background (Son et al., 2011a), with the expectation that FgART1::GFP and hH1::RFP would co‐localize to nuclei. We selected 20 progeny from the resulting perithecia from the outcrossing that showed resistance to both hygromycin and geneticin, and we observed that all of the progeny expressed GFP and RFP in mycelial nuclei (Fig. 4A). In addition, there were no differences in GFP signal intensity when the strain was grown on various carbon sources (data not shown).

Figure 4.

FgART1 acts as a transcriptional regulator. (A) Cellular localization of FgART1. For the microscopic observations, HK026 strain carrying FgART1::GFP and hH1::RFP was grown in liquid minimal medium (MM) containing sucrose. The representative example shows the localization of FgART1 fused with green fluorescent protein (GFP) and histone H1 fused with red fluorescent protein (RFP). DIC, differential interference contrast; scale bar, 20 μm. (B) Expression of amylolytic genes in F usarium graminearum strains. To measure gene expression, all strains were grown in liquid complete medium (CM) for 4 days and were then resuspended in liquid MM containing starch. Total RNA was extracted at 0, 2 and 4 days after resuspension. FGSG_03703, FGSG_03890, FGSG_06278 and FGSG_13805 were annotated as putative α‐glucosidase genes, and FGSG_03842 was annotated as a putative α‐amylase gene. FgWT, F . graminearum wild‐type strain; ΔFgArt1, FgArt1 deletion strain; HK025, complemented strain of ΔFgArt1. (C) Starch hydrolysis of F . graminearum strains on MM starch plates. After 4 days of incubation, culture plates were examined by staining with iodine solution. The white arrows indicate the regions of starch hydrolysis.

To investigate whether FgArt1 is involved in starch hydrolysis by regulating amylolytic genes, we measured the expression of five genes predicted to encode enzymes involved in starch saccharification. In both the wild‐type and HK025 strains, the expression of two amylolytic genes, FGSG_03842 (a putative α‐amylase) and FGSG_06278 (a putative 1,4‐α‐glucosidase), was highly induced on MM containing starch as the sole carbon source, but this induction was not observed in the ΔFgArt1 strain (Fig. 4B). To determine whether the decreased amylolytic gene expression in the ΔFgArt1 strain corresponded to a lower level of starch hydrolysis, we grew the wild‐type, ΔFgArt1 and HK025 strains on plates containing starch and then stained the plates with an iodine solution; the enzymatic degradation of starch caused areas of the plate to remain unstained (Fig. 4C). Consistent with the increase in amylolytic gene expression, the sizes of the unstained areas beneath mycelia of the wild‐type and HK025 strains were much greater than those beneath the ΔFgArt1 colonies (Fig. 4C). These results indicate that FgArt1 is involved in starch hydrolysis by regulating amylolytic gene expression.

FgArt1 deletion has effects on virulence and trichothecene biosynthesis

Our observation that the ΔFgArt1 mutant exhibits impaired germination and starch hydrolysis in the presence of starch (Figs 3 and 4) gave rise to the question of whether the mutant would show reduced virulence during wheat infection. To assess the role of FgART1 in virulence on wheat heads, wheat spikelets were point inoculated with a conidial suspension of the wild‐type, ΔFgArt1 or HK025 strain. After 21 days of incubation, we observed that the wild‐type and HK025 strains showed normal head blight symptoms, such as bleaching (Fig. 5A). In contrast, the ΔFgArt1 strain was restricted to infection sites and was unable to spread from the rachis to adjacent spikelets in the head (Fig. 5). By 5 days after infection, symptoms appeared with slight discoloration of the infected spikelets (data not shown). By 7 days after infection, symptoms became clearly visible on spikelets next to infection sites in the wild‐type and HK025 strains, whereas the ΔFgArt1 strain did not spread to neighbouring spikelets (Fig. 5A). From 7 to 19 days after infection, the numbers of diseased spikelets in heads inoculated with the wild‐type and HK025 strains increased, whereas the spread of disease symptoms by the ΔFgArt1 strain was significantly slower (Fig. 5B). Furthermore, we sectioned the first rachis above the inoculated spikelet, which was attached to the uninoculated spikelet, at 5 days after infection, and found that the wild‐type and HK025 strains had heavily colonized all cell types in the rachis vasculature. However, hyphae of the ΔFgArt1 strain were barely detectable within the rachis (Fig. 5C). Thus, the reduced ability of the ΔFgArt1 strain to spread within wheat heads indicates that FgART1 plays an important role in the progression of head blight after initial colonization.

Figure 5.

Virulence assay of F usarium graminearum strains. (A) Flowering wheat heads were inoculated with conidia of the wild‐type Z‐3639 (FgWT), ΔFgArt1 mutant and complemented strain HK025. Mock indicates a negative control inoculated with water. The white arrows show the inoculated points, and images were captured at 3 weeks after inoculation. (B) The number of diseased spikelets was counted every 2 days after 7 days of inoculation (DAI). (C) Cross‐sections of rachis internodes of wheat spikelets. The images show the vascular bundles of rachises attached to uninoculated spikelets, which are the first spikelets above the inoculated spikelets. The rachis internodes were sectioned at 5 days after inoculation and stained with toluidine blue for anatomical examination. The black arrows indicate intracellular fungal hyphae. mx, metaxylem; ph, phloem; scale bars, 100 μm.

To determine whether FgArt1 deletion affects trichothecene biosynthesis, we measured trichothecene production in the wild type, ΔFgArt1 and HK025 strains grown in liquid AMM, with sucrose provided as the sole carbon source. As a result, DON and 15‐ADON were not detectable in the cultures of the ΔFgArt1 strain, but trichothecenes accumulated in the culture media inoculated with the wild‐type and HK025 strains (Fig. 6A). In addition, we investigated whether FgART1 is required for the expression of the trichothecene synthesis genes Tri5 (FGSG_03537) and Tri6 (FGSG_03536), which encode a trichodiene synthase and a pathway‐specific transcriptional regulator of trichothecene production, respectively (Seong et al., 2009). In the ΔFgArt1 strain, Tri5 and Tri6 expression was not detectable or was reduced by more than 50‐fold (Fig. 6B) compared with the wild‐type strain. The HK025 strain exhibited similar Tri5 and Tri6 expression to the wild‐type strain. Together, these results show that the decrease in trichothecene biosynthesis is attributable to FgArt1 deletion.

Figure 6.

Trichothecene production in F usarium graminearum strains. (A) For analysis, trichothecenes were extracted from 7‐day‐old cultures in liquid agmatine minimal medium (AMM). 15‐ADON, 15‐acetyldeoxynivalenol; DON, deoxynivalenol; n.d., not detected. (B) Expression of the trichothecene biosynthetic genes Tri5 and Tri6 was measured in F . graminearum strains grown in AMM for 4 days. FgWT, F . graminearum wild‐type strain; ΔFgArt1, FgArt1 deletion strain; HK025, complemented strain of ΔFgArt1.

FvArt1, an FgArt1 homologue in F. verticillioides, is also required for starch hydrolysis and FB ₁ biosynthesis

To functionally characterize the role of the FgArt1 homologue in F. verticillioides, the locus (FVEG_07545; FvArt1) was deleted via the double crossover method, in which the predicted coding region of the gene was replaced with a geneticin resistance cassette (Fig. S2). The FvArt1 gene of F. verticillioides has an ORF of 3596 bp in size interrupted by a 50‐bp intron, and is predicted to encode a 670‐amino‐acid protein. Like FgART1, the protein encoded by FvArt1 contains a conserved Zn(II)₂Cys₆ binuclear motif (IPR001138; residues 9–55) and a fungal transcription factor domain (IPR007219; residues 111–293) (Fig. S1A, B). Similar to the ΔFgArt1 strain, an FvArt1‐deleted strain, ΔFvArt1, was unable to hydrolyse starch when grown on starch plates, although the mutant showed similar growth in colony diameter compared with the wild‐type strain (Fig. 7A). On whole kernels of maize, the ΔFvArt1 strain produced approximately 20% of the ergosterol produced by the wild type (Fig. 7B); ergosterol is commonly used as a quantitative measurement of fungal growth. However, when grown on germ tissues separated from whole kernels, the growth of ΔFvArt1 was similar to that of the wild‐type and complemented strains (data not shown). Thus, our results indicate that FvArt1 plays an important role in kernel colonization, particularly on endosperm tissues of maize kernels.

Figure 7.

Defects in starch hydrolysis and fumonisin B₁ (FB ₁) production caused by deletion of the FgArt1 homologue in F usarium verticillioides. (A) Starch hydrolysis assay of F . verticillioides strains. The strains were grown on minimal medium (MM) starch plates for 4 days, and starch hydrolysis was investigated by staining with iodine solution. FvWT, F . verticillioides wild‐type strain; ΔFvArt1, FvArt1 deletion strain; HK061 and HK114, integrated strain of FgArt1 into ΔFvArt1. The white arrows indicate the regions of starch hydrolysis. (B) Growth and FB ₁ production by F . verticillioides strains. Ergosterol and FB ₁ were extracted from each strain grown on maize kernels for 7 days, with three independent biological replicates. Growth on maize kernels is represented by the ergosterol concentration. (C) Expression of fumonisin biosynthetic genes (Fum1 and Fum12) and the Art1 gene. To analyse the transcript levels of the genes in each strain, total RNA was isolated from cultures grown in BSAP liquid medium (containing 3 g KH₂PO₄, 0.3 g MgSO₄, 5 g NaCl, 1 g BSA and 20 g amylopectin per litre) for 2 days after resuspension of 5‐day‐old yeast extract peptone dextrose (YEPD) cultures. n.d., not detected.

To investigate the role of FvArt1 in FB₁ production, we evaluated the ability of the mutant strain to produce FB₁ on maize kernels. The ΔFvArt1 strain produced approximately 34‐fold less FB₁ compared with the wild‐type strain (Fig. 7B). When the FB₁ level was normalized to fungal growth, the ΔFvArt1 strain was found to produce over six‐fold less FB₁ than the wild‐type strain. For further confirmation, we investigated the expression of the fumonisin synthesis genes Fum1 (FVEG_00316) and Fum12 (FVEG_00323), which encode a polyketide synthase and a cytochrome P450 monooxygenase for fumonisin biosynthesis, respectively (Proctor et al., 2003). In the ΔFvArt1 strain, the Fum1 and Fum12 expression levels were reduced by more than 50‐fold compared with the wild‐type strain (Fig. 7C). Taken together, these results show that the decrease in fumonisin biosynthesis is attributable to FvArt1 deletion.

FgArt1 expression in the ΔFvArt1 strain

To determine whether the FgArt1 gene can complement the fumonisin‐deficient phenotype in the ΔFvArt1 strain, we amplified FgArt1 cDNA from the F. graminearum wild‐type strain Z‐3639 and co‐transformed the FgArt1 cDNA and pBCATPH carrying the Hyg gene into the ΔFvArt1 strain (Fig. S2B, D). We then identified two strains based on antibiotic resistance and polymerase chain reaction (PCR) screening: one with resistance to both hygromycin and geneticin (strain HK061) and another with resistance only to hygromycin (strain HK114), indicating that the HK061 strain was an ectopic strain, whereas the HK114 strain was generated by replacement of geneticin with FgArt1 (Table 1). These two strains showed similar ability for starch hydrolysis compared with the wild‐type strain, indicating that FgArt1 functionally restored this activity in ΔFvArt1 (Fig. 7A). However, assessment of FB₁ production revealed that these two strains did not produce comparable amounts of FB₁ to the wild‐type strain, which corresponded to Fum1 and Fum12 expression (Fig. 7B). For further confirmation that FB₁ production was not restored in the HK061 and HK114 strains, we examined FgArt1 expression and found that it was induced in the HK114 and HK061 strains, but not in the wild‐type or ΔFvArt1 strain (Figs 7C and S2D). Taken together, our results suggest that amylase regulators have an effect on the production of mycotoxins, such as the trichothecenes and fumonisins of Fusarium spp.

Table 1.

F usarium graminearum and  F . verticillioides strains used in this study

Strain	Genotype	Source or reference
Z‐3639	Fusarium graminearum wild type	Bowden and Leslie (1999)
7600	Fusarium verticillioides wild type	Fungal Genetics Stock Center, Kansas City, KS, USA
ΔFgArt1	ΔFgArt1::gen	Son et al. (2011b)
mat1r	Δmat1‐1::gen, hH1::hH1‐rfp‐hyg	Son et al. (2011a)
HK025	ΔFgArt1::FgArt1‐gfp‐hyg	This study
HK026	ΔFgArt1::FgArt1‐gfp‐hyg, hH1::hH1‐rfp‐gen	This study
ΔFvArt1	ΔFgArt1::gen	This study
HK061	ΔFvArt1::gen, FgArt1, hyg	This study
HK114	ΔFvArt1::FgArt1, hyg	This study

Discussion

Because trichothecenes are known virulence factors, most studies of pathogenesis have focused on the initial stages of wheat and barley infection at anthesis (Gardiner et al., 2010; Hallen‐Adams et al., 2011; Ilgen et al., 2009). For example, in F. graminearum strains expressing GFP under the control of the Tri5 promoter, GFP expression has been observed in colonized developing seeds at 4 days after inoculation (Ilgen et al., 2009). In addition, several studies have measured the expression of genes involved in trichothecene production, showing that their expression is induced at early stages after inoculation in plants, corresponding to the accumulation of DON, which can be measured at early stages (Gardiner et al., 2010; Guldener et al., 2006; Hallen‐Adams et al., 2011). Based on the findings of these studies, the environment within seeds during the early stages of development seems to be conducive to DON production. However, relatively little information is available on how the later stages of seed development impact DON production and trichothecene pathway gene expression.

Considering that F. graminearum mainly causes disease on starch grains, we cannot exclude the possibility that starch or degrading products available to F. graminearum play critical roles in the regulation of trichothecene biosynthesis during the colonization of plants. This hypothesis was supported by our observation that starch induced trichothecene production. However, this fungus did not produce trichothecenes, in this study, when amylopectin was provided; amylopectin is one of the two components of starch, and is known to induce FB₁ in F. verticillioides (Bluhm and Woloshuk, 2005). Similarly, it has been reported that F. graminearum poorly produced trichothecenes when amylopectin or amylose was supplemented as a carbon source (Jiao et al., 2008; Kawakami et al., 2014). Given that amylopectin and amylose are unable to induce trichothecenes, the question of how starch induces trichothecene production remains unclear, although a similar observation has been reported by Miller and Greenhalgh (1985). In F. graminearum, the presence of sucrose, 1‐kestose or nystose as a carbon source is known to result in a high level of trichothecene biosynthesis, which is a common characteristic among DON producers (Jiao et al., 2008; Kawakami et al., 2014). 1‐Kestose and nystose are fructooligosaccharides which can be degraded to sucrose and fructose by exoinulinases secreted by fungi (Jiao et al., 2008). The observation that F. graminearum produces trichothecenes during infection of wheat spike tissues after anthesis is probably related to the composition of the wheat spike tissues; the amounts of sucrose and fructooligosaccharides in wheat spike tissues drastically increases until the 10th day after anthesis (Jansen et al., 2005; Proctor et al., 1995; Takahashi et al., 2001). Moreover, Jiao et al. (2008) suggested that, when considering the small amounts of trichothecene produced by glucose and fructose, α‐glycoside structures included in sucrose and fructooligosaccharides might be key factors for the induction of trichothecenes.

The degradation of starch by amylases has been shown to affect aflatoxin and fumonisin production in Aspergillus flavus and F. verticillioides, respectively (Bluhm and Woloshuk, 2005; Fakhoury and Woloshuk, 1999). Recently, with regard to the genes involved in carbohydrate catabolism, two putative genes encoding hexokinase have been identified and characterized in F. graminearum (Zhang et al., 2015). Of these two genes, deletion of FgHxk1 has been shown to result in a dramatically decreased level of DON production compared with the wild type, which is similar to a previous observation that the Δhxk1 mutant of F. verticillioides produces 80% less FB₁ than the wild type (Kim and Woloshuk, 2011; Kim et al., 2011). Moreover, an FgHxk1 over‐expression strain has been shown to exhibit a remarkably high level of DON production (Zhang et al., 2015). These results suggest that hexokinases encoded by Hxk1 play an important role in the transduction of external environmental cues for the regulation of secondary metabolites, such as trichothecenes and fumonisins. Thus, the elucidation of additional mechanisms of carbohydrate detection and utilization in mycotoxigenic fungi is essential for a better understanding of how carbon sources affect trichothecene biosynthesis in F. graminearum.

In this study, we functionally characterized FgArt1 of F. graminearum, which plays a role as a regulator of amylolytic genes, and showed that the FgArt1 homologue of F. verticillioides is also involved in starch hydrolysis. Similarly, the mutant strains ΔFgArt1 and ΔFvArt1 produce low levels of trichothecenes and FB₁, respectively, compared with the wild‐type strain. A few putative FgArt1 orthologues of several fungal species have been identified and characterized, including BGLR in T. reesei, MAL13 in Saccharomyces cerevisiae and AMYR in A. nidulans and A. oryzae, which function as regulators of β‐glucosidase, maltase and α‐amylase, respectively (Chang et al., 1988; Gomi et al., 2000; Nitta et al., 2012; Tani et al., 2001b). Phylogenetic analysis has revealed that FgART1 and its putative orthologues form a separate cluster from the AMYR group, consistent with previous reports (Chung et al., 2013; Nitta et al., 2012). The putative AMYR orthologues, FGSG_00069 and FGSG_03702, are clustered in the AMYR group, whereas FgART1 is clustered with BGLR and MoCOD1, which have been characterized previously. In particular, AMYR orthologues characteristically cluster with amylolytic genes. Although FgART1 is clustered in the same group as BGLR and MoCOD1, deletion of the BglR gene results in a significant yield of cellulase during growth on cellobiose, but we observed no difference in cellulose activity between the F. graminearum wild‐type and ΔFgArt1 strains when grown on cellobiose plates (data not shown). In addition, Chung et al. (2013) have shown that MoCOD1 plays an important role in conidiation and pathogenicity rather than as a regulator of extracellular enzymes. Thus, it is likely that FgART1 constitutes a new functionally diverse group that is distinct from the AMYR group.

In filamentous fungi, secondary metabolite production is tightly regulated by environmental signals and intracellular signalling pathways (Brakhage, 2013). Of these regulatory mechanisms, G proteins of F. graminearum have been shown to function in toxin production, as well as asexual and sexual development and virulence (Yu et al., 2008). Two Gα protein mutants have been demonstrated to produce higher levels of DON compared with the wild type. Furthermore, the deletion of FgFlbA and FgRgsA, which regulate the activity of G proteins, has been shown to enhance DON production in F. graminearum (Park et al., 2012). Recently, Ubl1, which encodes a UBR‐Box/RING domain E3 ubiquitin ligase of F. verticillioides, has been identified, and its deletion has been shown to result in a substantial reduction in starch hydrolysis despite robust growth (Ridenour et al., 2014). In addition to the involvement of Ubl1 in amylolysis, the UBL1 protein physically interacts with the Gα1 and Gα2 proteins, providing a possibility that the G proteins of F. graminearum are involved in responsiveness to starch in the external environment. In Penicillium decumbens, Hu et al. (2013) have demonstrated an important regulatory role of PGA3, or Gα protein subunit 3, in the expression of polysaccharide‐degrading enzymes, especially amylase. Deletion of Pga3 results in reduced amylase activity, corresponding to amylase gene expression, and a low level of intracellular cyclic adenosine monophosphate (cAMP) concentration during growth on starch. Furthermore, the authors showed that the expression of AmyR, which regulates amylase gene expression, is significantly reduced in the ΔPga3 strain. Amylase production was restored in the Δpga3 strain by supplementation of exogenous cAMP, suggesting an essential role of PGA3 in amylase synthesis via the maintenance of a sufficient cAMP level. From these observations, we hypothesize that the Gα protein pathway, including cAMP production, may mediate FgART1 and amylase gene expression, thereby regulating trichothecene biosynthesis. Thus, the possibility that FgArt1 impacts trichothecene biosynthesis through the modulation of G protein signalling warrants future investigation.

During the initiation and development of disease, F. graminearum secretes various extracellular enzymes that have been hypothesized to be involved in nutrition acquisition during host infection (Wanjiru et al., 2002). The identification of extracellular proteins has been attempted in this fungus grown under various conditions (e.g. in hop cell wall medium and in planta) (Paper et al., 2007; Phalip et al., 2005). The disruption of Fgl1, which encodes a secreted lipase of F. graminearum, results in a reduction in extracellular lipolytic activity in culture, in addition to virulence to both wheat and maize (Salomon et al., 2012). In addition, the severity of disease caused by the wild‐type strain is strongly reduced following supplementation with ebelactone B, a known lipase inhibitor. In contrast, a xylanase gene of F. graminearum has been shown to be unnecessary for virulence, although it is expressed during wheat infection (Sella et al., 2013). In this study, we observed that the ΔFgArt1 strain exhibited reduced extracellular amylolytic activity on the starch plate and produced a low level of trichothecenes, which have been identified as important virulence factors. The loss of these abilities in the ΔFgArt1 strain seems to be attributable to its limited spread within the wheat head. However, given that nutrient starvation is known to induce genes related to pathogenesis (Trail et al., 2003), we cannot exclude the possibility that these genes might be expressed in the ΔFgArt1 strain as a result of the loss of the ability to use starch as a carbon source. This hypothesis could be supported by the following observations: the ergosterol level in the ΔFvArt1 strain grown on maize kernels was 20% of the wild‐type level, whereas the growth of the ΔAmy1 strain, which is a deletion strain of the α‐amylase gene (an orthologue of FGSG_03842 regulated by FgArt1) in F. verticillioides, was comparable with that of the wild type on maize kernels (Bluhm and Woloshuk, 2005). Thus, we suggest that ART1 transcription factors not only regulate amylase genes, but also other genes related to pathogenesis or development.

To further explore the function of FgArt1, we generated a deletion strain of FvArt1, an FgArt1 homologue in F. verticillioides, and found that this gene is involved in amylolysis and FB₁ production. When FgArt1 was expressed in the ΔFvArt1 strain, amylolysis activity was restored compared with the wild type. However, the level of FB₁ production by the FgArt1‐expressed strain was similar to that by the ΔFvArt1 strain. To clarify this finding, we generated two different FgArt1‐expressing strains, one with insertion of FgArt1 into the original FvArt1‐deleted site, and the other with arbitrary insertion of FgArt1 into the genome of the ΔFvArt1 strain. The two strains showed the same phenotype, including low levels of FB₁ production, although FgArt1 expression was highly induced. Considering the different properties of fumonisins and trichothecenes, as well as the enzymatic and cellular compartmentalization for the biosynthesis of secondary metabolites, a possible explanation for the non‐complementation of FB₁ production is that F. graminearum and F. verticillioides have unique regulatory mechanisms for the biosynthesis of trichothecenes and fumonisins, respectively. In addition, we found that upstream sequences of the start codon (within 500 bp) of trichothecene and fumonisin pathway‐specific genes have putative ART1 DNA‐binding sites, based on binding sequences of 5′‐CGGN₈(C/A)GG‐3′, as described previously (Tani et al., 2001a): one for Tri4 (cytochrome p450 monooxygenase), two for Fum1 (polyketide synthase) and one for Fum7 (alcohol dehydrogenase). Considering the function of these genes which are possibly contacted by ART1, this finding supports our hypothesis that FgArt1 and FvArt1 have unique regulatory mechanisms for the biosynthesis of trichothecenes and fumonisins, respectively. Therefore, the elucidation of mycotoxin‐specific regulatory mechanisms is needed to fill the gaps in the current knowledge.

Experimental Procedures

Fungal strains, media and starch hydrolysis assay

As wild‐type strains, F. graminearum strain Z‐3639 (Bowden and Leslie, 1999) and F. verticillioides strain 7600 (M3125; Fungal Genetics Stock Center, Kansas City, KS, USA) were used in this study, which produce relatively high concentrations of trichothecenes and fumonisins, respectively. All transgenic strains (Table 1) derived from the wild‐type strain were maintained in CM and stored as conidial suspensions in 20% glycerol at −80 °C according to the instructions in the Fusarium Laboratory Manual (Leslie and Summerell, 2006). For the analysis of trichothecene production, conidia induced in carboxymethylcellulose (CMC) liquid medium (Cappellini and Peterson, 1965) were inoculated onto AMM (Gardiner et al., 2009). For the analysis of ergosterol and FB₁ production, F. verticillioides strains were inoculated into autoclaved maize kernels (kindly provided by Dr Kyoung‐Su Kim, Kangwon National University, Chuncheon, South Korea). To determine the ability of the fungal strains to hydrolyse starch, cultures of strains grown on starch plates for 4 days were stained with an iodine solution (0.5% iodine and 5% potassium iodide). The starch plate was modified from MM by the addition of 1% corn starch instead of sucrose. For the analysis of Fum gene expression, F. verticillioides strains were grown in yeast extract peptone dextrose (YEPD) medium for 5 days, and the resulting mycelia were resuspended in BSAP liquid medium containing 3 g KH₂PO₄, 0.3 g MgSO₄, 5 g NaCl, 1 g BSA and 20 g amylopectin per litre (Kim and Woloshuk, 2011).

Nucleic acid manipulations

For the isolation of genomic DNA (gDNA), each strain was grown in 25 mL of liquid CM at 25 °C for 4 days on a rotary shaker (150 rpm), and gDNA was extracted using the cetyltrimethylammonium bromide (CTAB) procedure, as described by Leslie and Summerell (2006). To measure transcript levels, total RNA was extracted using an Easy‐Spin Total RNA Extraction Kit (iNtRON Biotechnology, Seongnam, South Korea). First‐strand cDNA was synthesized from total RNA using SuperScript III First‐Strand Synthesis SuperMix (Invitrogen, Carlsbad, CA, USA). All PCR primers used in this study were obtained from Neo Probe (Daejeon, South Korea) and are listed in Table S2 (see Supporting Information). The primers were diluted to 100 μm in sterilized water and stored at −20 °C. The nucleotide and protein sequences were obtained from the Fusarium Comparative Database of the Broad Institute (http://www.broadinstitute.org/annotation/genome/fusarium_group), the Fusarium graminearum Genome Database (FGDB; Wong et al., 2011) and the Aspergillus Genome Database (AspGD; Cerqueira et al., 2013). Sequences were aligned using the Clustal Omega algorithm (Sievers et al., 2011), and phylogenetic trees were generated using the neighbour‐joining method with the mega5 program (Tamura et al., 2011). Nuclear localization signals were predicted using NLStradamus (Ba et al., 2009).

Genetic manipulation and fungal transformation

For complementation of the ΔFgArt1 strain, intact copies of FgArt1, including the 5′‐flanking region, were amplified from F. graminearum wild‐type Z‐3639 strain with the primer pair FgArt1‐5F and FgArt1‐5R. A PCR product that includes GFP and a hygromycin resistance cassette (Hyg) was amplified from a pIGPAPA plasmid using the primer pair GFP‐F1 and HYG‐F1 (Horwitz et al., 1999). The 3′‐flanking region of the FgArt1 ORF was amplified with the primer pair FgArt1‐3F and FgArt1‐3R. Three resulting amplicons were fused using double‐joint (DJ) PCR (Yu et al., 2004), and the final PCR construct was obtained with the nested PCR primer pair FgArt1‐5N and FgArt1‐3N. Subsequently, the final PCR products were transformed into a protoplast of the ΔFgArt1 strain, and the resulting strains were selected using hygromycin antibiotic resistance and PCR analyses.

FvArt1 deletion constructs for F. verticillioides were also created by DJ PCR. The 5′‐ and 3′‐flanking regions of the FvArt1 ORF were amplified using the primer pairs FvArt1‐5F/FvArt1‐5R and FvArt1‐3F/FvArt1‐3R, respectively, and were then fused to a geneticin resistance cassette (Gen) amplified from pII99 using the primers Gen‐F and Gen‐R (Hong et al., 2010). The resulting PCR product was used as a template for the final PCR with the primer pair FvArt1‐5N and FvArt1‐3N. To generate F. verticillioides strains expressing FgArt1, the FgArt1 coding region, which was amplified from F. graminearum Z‐3639 cDNA using the primers FgArt1‐cF and FgArt1‐cR, was fused to the 5′‐ and 3′‐flanking regions of the FvArt1 ORF and amplified by the primer pairs FvArt1‐5F/FvArt1‐c5 and FvArt1‐c3/FvArt1‐3R, respectively. The resulting amplicons were co‐transformed with pBCATPH carrying the hygromycin selection marker into a protoplast of the FvArt1 deletion strain (Kim et al., 2005).

Outcrossing and microscopic observations

To observe the co‐localization of FgART1 with a nuclear protein, the HK025 strain carrying GFP was outcrossed with a mat1r strain that contained RFP fused to histone H1, as generated previously (Son et al., 2011a). To perform outcrossing, aerial mycelia of the mat1r strain grown on carrot agar medium for 5 days were fertilized with 1 mL of conidial suspension from the HK025 strain. After sexual induction, the plates were incubated for 7 days in constant near‐UV light (wavelength, 365 nm; HKiv Import & Export Co., Ltd., Xiamen, China) at 25 °C. Ascospores carrying both FgART1‐GFP‐HYG and hH1‐RFP‐GEN were selected using antibiotic resistance and confirmed by PCR. Localization of FgART1 in fungal nuclei was observed in cultures grown in MM supplemented with various carbon sources. Microscopic observations were performed with a DE/Axio Imager A1 microscope (Carl Zeiss, Oberkochen, Germany) using the filter set 38HE (excitation 470/40; emission 525/50) for GFP and the filter set 15 (excitation 546/12; emission 590) for RFP.

Pathogenicity assay and histological observations

Pathogenicity assay was performed as described previously (Brown et al., 2010) using the susceptible wheat (Triticum aestivum) cultivar, Eunpamil, as the host plant. Briefly, at the first appearance of anther extrusion, 5 μL of conidial suspension (1 × 10⁶ conidia/mL), harvested from a 5‐day‐old culture in CMC medium, were placed in the floral cavity between the palea and lemma of the first two florets within the two middle spikelets. Inoculated plants were placed in a humidity chamber for 3 days and were then kept in a glasshouse at ambient humidity. Spikelets with disease symptoms were counted until 3 weeks after inoculation. The data presented in this study were obtained from six biological replicates with two repetitions, and images were captured that are representative of these replications.

To observe fungal colonization of the wheat head, the first rachis above the inoculated spikelet, which was attached to the uninoculated spikelet, was collected and fixed in 2% glutaraldehyde (Sigma‐Aldrich, St. Louis, MO, USA) in 1 × PBS solution [10 mm phosphate buffer (pH 7.2), 138 mm NaCl and 3 mm KCl] at 4 °C overnight. After washing three times in PBS solution, the tissues were dehydrated through a gradient series (10%–100%) of ethanol and finally embedded in LR white resin (Sigma‐Aldrich). The embedded tissues in 100% LR white resin were polymerized in a PELCO UVC2 Cryo Chamber (Ted Pella, Inc., Redding, CA, USA) at 4 °C. Thin sections (thickness, 0.5 μm) were prepared using a microtome (Leica Biosystems, Nussloch, Germany) and stained with 0.1% toluidine blue. The sections were observed under a compound microscope (Leica Biosystems) using bright‐field illumination.

Quantitative reverse transcription‐polymerase chain reaction (qRT‐PCR) analysis

To analyse gene expression, qRT‐PCR was performed using SsoFast EvaGreen Supermix (Bio‐Rad, Hercules, CA, USA) and a CFX96 Real‐Time PCR system (Bio‐Rad). Each reaction contained 10 μL of EvaGreen Supermix, 500 nm each of forward and reverse primers, cDNA template generated from 2 μg of total RNA and nuclease‐free water added to a final volume of 20 μL. The PCR cycling conditions were as follows: 30 s at 95 °C (one cycle), and 5 s at 95 °C followed by 5 s at 60 °C (40 cycles). The experiments were repeated twice with three replicates each. The expression of each gene was normalized to that of CYP1 (FGSG_07439) and TUB (FVEG_05512) for F. graminearum and F. verticillioides, respectively, and calculated as a fold change based on the 2^−ΔΔCt method. The conditions for semi‐qRT‐PCR were as follows: 2 min at 94 °C, followed by 25 cycles of 15 s at 94 °C, 30 s at 60 °C and 30 s at 72 °C.

Trichothecene analysis

To analyse DON and 15‐ADON production, conidial suspensions (1 × 10⁶/mL) of each strain were inoculated into AMM, and the cultures were incubated for 7 days at 25 °C under stationary conditions. Trichothecenes were extracted from 250 μL of culture filtrates by mixing with 250 μL of an ethyl acetate–methanol solution (4 : 1, v/v) (He et al., 2007). Extracts were dried and derivatized with Sylon BTZ (BSA (N,O‐bis[trimethylsilyl] acetamide) + TMCS (Trimethylchlorosilane) + TMSI (N‐trimethylsilyimidazole), 3:2:3; Supelco, Bellefonte, PA, USA). Sequentially, 200 μL of n‐hexane and 200 μL of distilled water were added to the derivatized samples. The reaction was left standing until the two layers separated, and 2 μL of the upper layer was analysed for trichothecene production using a Shimadzu QP5050 gas chromatograph‐mass spectrometer (GC‐MS) (Shimadzu, Kyoto, Japan) with an SP2330 column (30 m × 0.32 mm inside diameter; 0.25 μm film thickness; Supelco), with helium as the carrier gas. The column temperature was increased from 120 to 280 °C at a rate of 8 °C/min, and the holding time was 10 min at 280 °C. Quantification was performed by comparing the peak areas of the samples with those of external standards of DON and 15‐ADON (Sigma‐Aldrich).

Fumonisin and ergosterol analyses

FB₁ and ergosterol were extracted from cultures grown on maize kernels as described previously (Kim and Woloshuk, 2011; Kim et al., 2011). Briefly, FB₁ was extracted from ground maize kernels (200–500 mg) with 2 mL of acetonitrile–water (1 : 1, v/v). The extracts were passed through equilibrated C‐18 solid phase extraction columns (Agilent Technologies, Santa Clara, CA, USA), and FB₁ was eluted with 2 mL of acetonitrile–water (7 : 3, v/v). Analysis was performed with a liquid chromatograph‐mass spectrometer (LC‐MS) system (Hewlett Packard HP‐1100 Series, Palo Alto, CA, USA) equipped with a Luna C‐18 column (250 mm × 4.6 mm; Phenomenex, Madrid, Spain) and MS detector with an electrospray ionization (ESI) probe. LC separation was performed using gradient elution with water as mobile phase A and methanol as phase B, with both containing 0.5% formic acid. After an isocratic step of 65% B for 4 min, it was gradually increased to 95% B over 4 min and held for 16 min. Mass spectra were obtained by scanning from m/z 300 to 800 using selected ion monitoring. FB₁ was quantified by comparisons of peak areas with those of an FB₁ standard (Sigma‐Aldrich).

For the analysis of ergosterol, it was extracted from ground kernel tissue (200–500 mg) overnight at room temperature in 2 mL of chloroform–methanol (2 : 1, v/v). Analysis was performed using a high‐performance liquid chromatography (HPLC) system (Waters, St. Maple, Milford, MA, USA) with a UV detector (Waters) set to monitor at 282 nm and a 4.6 U ODS column (250 mm × 4.6 mm; Phenomenex). Compounds were eluted with 100% methanol at a flow rate of 1.0 mL/min, and the ergosterol levels were determined by comparing the peak areas of the samples with a standard curve generated from HPLC‐grade ergosterol (Sigma‐Aldrich).

Supporting information

Fig. S1 Sequence alignment and distribution of FgART1 homologues in fungi.

Fig. S2 Strategies for the deletion and complementation of Art1 in Fusarium graminearum and F. verticillioides.

Table S1 Growth and starch hydrolysis of the Fusarium graminearum strains.

Table S2 Primers used in this study.