The Supplementary Table 1 presents all the strains used in this study. The culture media used in the study include complete medium (CM) consisting (all in w/v) 0.6% casein acid hydrolysate, 0.6% yeast extract, 1% sucrose, 2% agar (for solid media) or starch yeast medium (SYM) consisting (all in w/v) 1% starch, 0.2% yeast extract, 0.3% sucrose, and 2% agar (Zheng et al., 2015). The incubations were done at 28 °C for 3 days. Previous protocols (Cappellini and Peterson, 1965; Zheng et al., 2015; Fan et al., 2020) were used to assay for conidiation and sexual reproduction.

Previously reported protocols (Hou et al., 2002) were used for the preparation of F. graminearum protoplasts and subsequent transformations. The split-marker approach was used to generate gene replacement constructs for the FgGYP1 (FGSG_17336) deletion mutants. The Supplementary Table 2 presents all the primers used for this purpose. Two gene deletion mutants (ΔFggyp1) were successfully generated and further confirmed by Southern blotting. To reintroduce this gene into the mutant (complementation), a vector harboring FgGyp1-GFP fusion protein was generated by amplifying the coding sequence and native promoter of FgGyp1 using FgGyp1CF and FgGyp1CR primer pair (Supplementary Table 2). A Cloning Kit (Vazyme Biotech Co., Ltd., China) was then used to insert the amplicon into a pKNTG2 vector and the product was sequenced for verification. The vector (pFgGyp1-GFP) was then transformed into the protoplast of ΔFggyp1 mutant and a complemented strain was successfully generated which has similar phenotypes to the PH-1.

The pFgBet3-mCherry vector was constructed by amplification of the 2,573-bp FgBet3 sequence (including its native promoter and coding sequence) using FgBet3CF and FgBet3CR primer pair (Supplementary Table 2). The previously stated cloning kit was used to clone the PCR product into a pKNT-mCherry plasmid and the insertion was confirmed by sequencing. pFgGyp1^ΔN-GFP, pFgGyp1^ΔTBC-GFP, pFgGyp1^ΔC-GFP and pR357K, pR284K constructs were generated by PCR amplification of the sequences using their respective primer pairs (Supplementary Table 2) and pKNTG2 vector was used for the cloning and verified by sequencing.

Flowering wheat heads were used to test for the pathogenicity of the various strains as previously reported (Zheng et al., 2015). Wheat coleoptiles were also used for this assy. Briefly, 2 μl of 100 × 10⁴ cells/ml conidial suspension was inoculated in each coleoptile and the observed symptoms were recorded after 7 days of inoculation. To investigate their DON production abilities, the various strains were cultured in TBI (trichothecene biosynthesis induction) liquid media and incubated for 7 days in the dark at 28°C without shaking, after which the mycelia were separated from the liquid (Gardiner et al., 2009). An ELISA (enzyme linked immunosorbent assay)-based DON assay kit (Beacon Analytical Systems, Saco, ME, United States) was used to check for DON levels in the liquids while the mycelia were quantified after drying. The experiment was repeated three times.

The sequences of FgRab1 cDNA, FgGyp1 cDNA, TBC domain and R357K point mutant of FgGyp1 were cloned into the MBP (Maltose-binding protein) vector pMAL-c2X, using their respective primers listed in Supplementary Table 2, and expressed. The protein products were isolated and purified for GAP activity assay using a GTPase assay kit (Sigma-Aldrich, Catalog Number MAK113) according to the manufacturer’s protocols. GTP hydrolysis was assayed based on previous reports (Xiong et al., 2012; Zheng et al., 2020). Briefly, the recombinant proteins MBP–FgGyp1, MBP–FgGyp1^TBC (TBC), MBP–FgGyp1^R357K (R357), and MBP–FgRab1 were expressed in BL21 Escherichia coli strain and isolated by Amylose resin (Sangon Biotech, NO.C500096) as per the instructions of the manufacturer. A colorimetry-based kit was used (Sigma-Aldrich, Catalog Number MAK113) to assay for the GTPase activity of Gyp1, and FgRab1 was incubated with FgGyp1, FgGyp1^TBC (TBC), FgGyp1^R357K (R357K), and MBP control, respectively, following the instructions of the manufacturer. The experiment was repeated three times.

Aspergillus fumigatus (AfGyp1-XP_749737.1), Candida albicans (CaGyp1-XP_717292.1), Fusarium graminearum (FgGyp1-PCD18761.1), Fusarium oxysporum (FoGyp1-EMT63394.1), Fusarium verticillioides (FvGyp1-EWG40605.1), Magnaporthe oryzae (MoGyp1-ELQ43659.1), Neurospora crassa (NcGyp1-CAC18313.2), Saccharomyces cerevisiae (ScGyp1-NP_014713.1), Schizosaccharomyces pombe (SpGyp1-NP_595314.1), and Ustilago maydis (UmGyp1-XP_011391412.1).

The live cell imaging of F. graminearum mycelia and conidia was performed using an Olympus BX51 (Olympus, Japan) microscope and a laser confocal microscope (Nikon, Japan). The following settings were used. GFP excitation: 488 nm light (Em.525/40 nm); mCherry excitation: 561 nm light (Em. 607/36 nm).

To identify the FgGYP1 gene in F. graminearum genome, we used the budding yeast Gyp1 amino acid sequences as a reference to carry out a BLAST search against the available fungal genomes. We were able to identify a homolog of Gyp1 at the FGSG_17336 locus of F. graminearum genome. FGSG_17336 encodes a protein of 601 amino acid residues and it is 40.53% similar to the yeast Gyp1, and it covers 61.00% of the total length of FgGyp1 and ScGyp1. The phylogenetic relationship of Gyp1 homologs suggest that Gyp1 is conserved in plant pathogenic fungi, especially in Fusarium oxysporum, Fusarium verticillioides, Neurospora crassa, and Magnaporthe oryzae (Supplementary Figure 1). Furthermore, targeted gene replacement strategy was used to delete FgGYP1 gene in PH-1 (wild type) where many positive mutants were identified by PCR screening (Supplementary Figure 2A). Two of these mutants (ΔFggyp1-2 and ΔFggyp1-5) were subjected to Southern blotting for confirmation (Supplementary Figure 2B).

To determine whether FgGyp1 is required for the fungal vegetative growth, we cultured the ΔFggyp1 mutant on CM (complete media) and monitored its growth rate after 3 days. Compared to the PH-1 and ΔFggyp1-C (complemented strain) strains, the ΔFggyp1 growth rate was critically reduced (Figures 1A,B and Table 1). Close microscopic examinations indicated numerous branching in the ΔFggyp1 mutant hyphae which was not observed in PH-1 and ΔFggyp1-C strains (Figure 1A). These results suggest that FgGyp1 is important for F. graminearum vegetative growth and hyphal polarity.

The major F. graminearum inoculums that infect flowering wheat heads are the conidia (Francl et al., 1999). As such, we inoculated the wild type PH-1, ΔFggyp1, and ΔFggyp1-C strains on carboxymethylcellulose (CMC) media at 28°C for 3 days for conidia production. As shown in Table 1, the conidiation of ΔFggyp1 mutant decreased significantly when compared to those produced by PH-1 and ΔFggyp1-C strains, suggesting that FgGyp1 plays an important role in the fungal conidiogenesis. In addition to conidia, the ascospores of F. graminearum are also an important inoculum in its infection cycle. However, sexual reproduction and ascospore formation are not significantly affected by FgGYP1 deletion (Supplementary Figures 3A,B).

To investigate the effect of FgGYP1 deletion on the pathogenicity of the fungus, the mutant and the controls were inoculated on flowering wheat heads under moist condition for 14 days. As shown in Figure 1C, deletion of FgGYP1 gene caused significantly decreased infection capability to wheat heads. The average disease index (diseased spikelets per head) for ΔFggyp1 mutant is less than six while the disease index of PH-1 and ΔFggyp1-C strains are more than 13 (Table 1). Furthermore, infection assays on wheat coleoptiles yielded similar results (Figure 1D). Taken together, our data here demonstrate that FgGyp1 is important for F. graminearum virulence.

In F. graminearum, deoxynivalenol (DON) has been widely studied as an important mycotoxin and virulence factor (Proctor et al., 1995). To investigate whether FgGyp1 is required for DON production, we checked and compared the levels of DON in the PH-1, ΔFggyp1, and ΔFggyp1-C strains cultured in liquid TBI media for 7 days at 28°C in the dark. We found that ΔFggyp1 mutant showed strikingly higher level of DON production than the wild type PH-1 (Table 1, the standard curves for quantification of the DON were showed in Supplementary Figure 4), suggesting that FgGyp1 negatively regulates DON biosynthesis in F. graminearum. However, we found that the trichothecene biosynthesis proteins FgTri1-GFP and FgTri4-GFP retain their normal localizations at the toxisomes of both strains, meaning that FgGyp1 is dispensable for their localizations (Figure 2).

To establish the sub-cellular localization of FgGyp1, we generated and transformed a FgGyp1-GFP-expressing vector into the ΔFggyp1 mutant protoplasts. The positive transformants were then subjected to live cell confocal microscopy. As shown in Figure 3, FgGyp1-GFP fluorescence were clearly visible as puncted structures distributed through the conidia and conidiophores of the fungus at all developmental stages (0.5, 2, and 18 h). In yeast, Gyp1 localizes to the Golgi apparatus (Du and Novick, 2001). To check whether the observed puncta are localized to the Golgi apparatus, several Golgi markers were generated, including the trans-Golgi network (TGN) marker, FgKex2-mCherry, cis-Golgi marker, FgBet3-mCherry, and the medial Golgi marker, mCherry-FgGos1 (Thomas et al., 2018). These constructs were co-transformed with FgGyp1-GFP, respectively, into the ΔFggyp1 mutant and their intracellular localization examined by confocal microscopy. As shown in Figure 4, we found that FgGyp1-GFP partially colocalizes with FgKex2-positive TGN (61.72 ± 4.67% colocalization), FgBet3-positive cis-Golgi (28.79 ± 12.03% colocalization) and FgGos1-positive medial Golgi (14.71 ± 6.48% colocalization) in CM medium. We therefore conclude here that FgGyp1 mainly localizes to the TGN.

Ypt1/Rab1 is a substrate for Gyp1 in yeast (Albert and Gallwitz, 1999). This prompted us to check if FgGyp1 is a GTPase-activating protein (GAP) for Rabs in F. graminearum. To achieve this, we investigated its Rab GAP activity in vitro by cloning and expressing its full-length and FgRab1 in BL21 cells, respectively. We then analyzed FgGyp1 GTP-hydrolyzing potential by quantifying the amount of phosphate moiety released by activated (GTP-bound) FgRab1. As shown in Figure 5, we found that FgGyp1 has higher efficiency in hydrolyzing GTP-bound FgRab1 than the control. Furthermore, we equally found that the TBC domain of FgGyp1 more efficiently hydrolyzes GTP-bound FgRab1 than the control. The arginine 343 residue (Arg343) of the TBC domain of Gyp1 in yeast was shown to be essential for the Gyp1 GAP activity (Du and Novick, 2001). For this reason, we identified the conserved Arginine residue in FgGyp1 (Arg357) and analyzed its role in FgGyp1 GAP activity. We observed that the FgGyp1 GAP activity was significantly decreased when Arg357 residue of FgGyp1 was mutated to lysine (R357K), as shown in Figure 5. Considering these data, we conclude that FgGyp1 catalytically converts an active (GTP-bound) FgRab1 to its inactive (GDP-bound) form through its TBC domain, and Arg357 residue is an important GTP-hydrolyzing residue of FgGyp1.

Since the TBC domain (with its Arg357 residue) is required for FgGyp1 GAP activity, we set to find out to the roles of this domain as well as its Arg357 residue in the biological functions of FgGyp1 in F. graminearum. To unveil this, we generated the following forms of FgGyp1-GFP fusion constructs: Fggyp1^ΔTBC-GFP (lacking the TBC domain), Fggyp1^R357K-GFP (mutation of the arginine 357 residue with lysine), Fggyp1^R284K-GFP (having arginine 284 replaced with lysine, control), Fggyp1^ΔN-GFP (lacking the N-terminal aa 1–280), and Fggyp1^ΔC-GFP (lacking the C-terminal aa 543–601) (Figure 6A). We transformed these constructs into separate protoplasts of ΔFggyp1 mutant and analyzed the phenotypes of each of these mutants and further examined their respective intracellular localizations by confocal microscopy. Figures 6B–F shows that ΔTBC, ΔN, ΔC, and R357K mutants exhibit phenotypes similar to the defects observed in the ΔFggyp1 mutant, including impaired vegetative growth, conidiation, virulence, and DON production. In contrast, the phenotypes of R284K mutant remain similar to those of the controls (Figures 6B–F), suggesting that Arg284 is not required for the function of FgGyp1.

Moreover, the aforementioned mutants were further analyzed for their roles in the subcellular localization of FgGyp1 protein. Figure 7 depicts that Arg284 and Arg357 residues are expendable with respect to FgGyp1 localization since their deletions do not significantly affect the localization of the protein. However, deletions of TBC domain, N-terminal, and C-terminal regions of the FgGyp1 displayed diffused GFP signals in the cytoplasm instead of the normal puncted appearance, suggesting that these domains play an important role to the subcellular localization of FgGyp1. Taken together, we conclude that the TBC domain, N-terminal, and C-terminal regions are all required for FgGyp1 functions and normal localization in F. graminearum. Arg357 residue is equally important for the functions of FgGyp1 but not for its correct localization.

Membranes fusion and scission are known to be regulated by SNARE (soluble N-ethylmaleimide-sensitive factor attachment receptor) proteins during vesicle fission and delivery, respectively (Hong, 2005). In yeast, Snc1 (a SNARE protein) mediates the fusion of vesicles from the Golgi with the plasma membrane (Lewis et al., 2000). We previously demonstrated in F. graminearum that FgSnc1 mediates polarized secretion and fusion of vesicles (Zheng et al., 2018c). To unveil the role of FgGyp1 in vesicles fusion with the plasma membrane, we transformed GFP-FgSnc1 construct into the PH-1 and ΔFggyp1 protoplasts and subsequently observed their cellular localizations. Figure 8 clearly shows that in PH-1, GFP-FgSnc1 is localized on the cell membrane and accumulates at the spitzenkörper (SPK) of the growing hyphal tip (Figures 8A,B and Supplementary Video 1). In the ΔFggyp1 mutant, the fluorescence of GFP-FgSnc1 was similarly detected at the SPK of the growing hyphae but its plasma membrane localization was almost completely abolished (Figures 8A,B and Supplementary Video 2). These results suggest that FgGyp1 is important for FgSnc1-mediated fusion of vesicles with the plasma membrane in F. graminearum.