Redox signalling from NADPH oxidase targets metabolic enzymes and developmental proteins in Fusarium graminearum

Summary

NADPH oxidase (NOX) is one of the sources of reactive oxygen species (ROS) that modulates the activity of proteins through modifications of their cysteine residues. In a previous study, we demonstrated the importance of NOX in both the development and pathogenicity of the phytopathogen Fusarium graminearum. In this article, comparative proteomics between the wild‐type and a Nox mutant of F. graminearum was used to identify active cysteine residues on candidate redox‐sensing proteins. A two‐dimensional gel approach based on labelling with monobromobimane (mBBR) identified 19 candidate proteins, and was complemented with a gel‐free shotgun approach based on a biotin switch method, which yielded 99 candidates. The results indicated that, in addition to temporal regulation, a large number of primary metabolic enzymes are potentially targeted by NoxAB‐generated ROS. Targeted disruption of these metabolic genes showed that, although some are dispensable, others are essential. In addition to metabolic enzymes, developmental proteins, such as the Woronin body major protein (FGSG_08737) and a glycosylphosphatidylinositol (GPI)‐anchored protein (FGSG_10089), were also identified. Deletion of either of these genes reduced the virulence of F. graminearum. Furthermore, changing the redox‐modified cysteine (Cys³²⁵) residue in FGSG_10089 to either serine or phenylalanine resulted in a similar phenotype to the FGSG_10089 knockout strain, which displayed reduced virulence and altered cell wall morphology; this underscores the importance of Cys³²⁵ to the function of the protein. Our results indicate that NOX‐generated ROS act as intracellular signals in F. graminearum and modulate the activity of proteins affecting development and virulence in planta.

Introduction

Redox proteomics is a growing field of enquiry with emphasis on proteins that undergo reversible and irreversible modifications by reactive oxygen species (ROS). Consequences of such modifications have diverse impacts in human, plant and fungal health, disease and development. The origin of ROS is equally diverse and includes mitochondria, chloroplasts and peroxisomes, where they form as a normal byproduct of metabolism, as well as from the action of the NADPH enzyme complex or NOX (NADPH oxidase). NOX catalyses the production of superoxide radicals by transferring an electron from NADPH to oxygen. It has been studied intensively since its discovery in mammalian leucocytes, where it was initially thought to produce bactericidal ROS rather than secondary messengers (Aguirre and Lambeth, 2010; Takemoto et al., 2010).

During plant–pathogen interactions, the production of ROS in plants is associated with the hypersensitive response, which limits the spread of pathogens (Torres et al., 2006). ROS produced in plants do not kill directly the invading pathogens, but act as secondary messengers to activate the proteins involved in immunity. Studies in plants and, more recently, in fungi, have suggested that spatiotemporal relationships between ROS generation and interaction with other signalling molecules may be important to understand the various functions performed by ROS (Marschall and Tudzynski, 2016; Ryder et al., 2013; Segal and Wilson, 2018; Tudzynski et al., 2012). NOX enzymes have been found in all filamentous fungi as NoxA, NoxB (homologues of the mammalian gp91^phox) and often also NoxC, where they are involved in many biological processes, including virulence and development. Fusarium graminearum, Claviceps purpurea, Botrytis cinerea and Alternaria alternata lacking one or more NOX enzymes all exhibit reduced virulence (Giesbert et al., 2008; Segmüller et al., 2008; Takemoto et al., 2010; Tudzynski et al., 2012; Wang et al., 2014; Yang and Chung, 2012). A study in Aspergillus nidulans was one of the earliest to document the role of NOX in fungal development (Lara‐Ortíz et al., 2003). Subsequent studies in non‐pathogenic fungi, such as Neurospora crassa and Podospora anserina, have demonstrated that developmental processes, such as hyphal elongation and fruiting body formation, are linked to NOX activity (Cano‐Dominguez et al., 2008; Malagnac et al., 2004).

Modifications of proteins by ROS directly affect their structure and, consequently, their function. Specifically, thiol moieties on cysteine (Cys) residues are reactive at physiological pH and are thus sensitive to changes in the redox status of the cell. They are the primary targets of redox modifications (Jang et al., 2004; Zheng et al., 1998). Some of the protein targets of ROS, so‐called ROS‐sensing proteins, have been shown to have susceptible Cys residues, which can undergo reversible oxidation depending on the redox potential of their surroundings (Barford, 2004). Moreover, reduction/oxidation can profoundly change a protein’s function to suit the cell’s immediate needs (Jang et al., 2004). The identification of ROS‐sensing proteins in the phytopathogen F. graminearum forms the subject of this article. Fusarium graminearum causes Fusarium head blight on wheat and other grasses, and Fusarium ear rot on maize (Trail, 2009). The role of NOX in F. graminearum with respect to the production of superoxides, and in both development and pathogenicity, has been demonstrated (Fig. 1A; Wang et al., 2014).

Figure 1.

Biochemical characterization of NADPH oxidase (NOX) and workflow of two strategies used to identify redox‐modified proteins in Fusarium graminearum. (A) The production of superoxides is reduced in the ΔNoxA/B mutant strain. Wild‐type (WT) and ΔNoxAB strains were grown in 0.5 × potato dextrose agar for 6 days and the plates were flooded for 30 min with 2.5 mm nitroblue tetrazolium (NBT) in 5 mm (N‐morpholino)propane sulfonate–NaOH, pH 7.6. The blue precipitates, indicative of superoxide production, were photographed (Wang et al., 2014). (B) Identification of redox‐modified proteins by two‐dimensional electrophoresis (2‐DE). Proteins were labelled with monobromobimane (mBBR) (Bm) that binds covalently to reduced cysteines. Subsequent reduction and alkylation with iodoacetamide (IAM) labels oxidized cysteines. The proteins were separated by 2‐DE and affected cysteines were identified by mass spectrometry. DTT, dithiothreitol; LC‐MS/MS, liquid chromatography‐tandem mass spectrometry; CAM, carbamidomethyl. (C) A gel‐free approach to identify oxidized cysteines using a biotin switch method. Reduced cysteines were first blocked, and oxidized cysteines were then reduced and labelled with biotin. Tryptic peptides were enriched on streptavidin and peptides were recovered using β‐mercaptoethanol (β‐ME). Oxidized cysteines with a β‐ME adduct were identified by mass spectrometry. NEM, N‐ethylmaleimide. [Color figure can be viewed at wileyonlinelibrary.com]

We used two conditions and two strategies to identify proteins that are modified through Cys oxidation in wild‐type (WT) F. graminearum, but remain unmodified in the ΔNoxAB mutant strain (Fig. 1B,C). The two conditions were growth in a nutrient‐rich broth to identify redox‐modified proteins that are involved in growth and development, and growth in a nutrient‐limited broth to identify proteins potentially important for the pathogenicity of F. graminearum. The first strategy employed was a gel‐based method in which the soluble proteomes from the F. graminearum WT and ΔNoxAB strains were labelled with monobromobimane (mBBR) during the extraction process and then separated and compared by two‐dimensional electrophoresis (2‐DE) over time (Fig. 1B) (Yano et al., 2001). The second strategy used to identify targeted Cys residues was non‐gel based. In this gel‐free approach, all Cys residues within a proteome were specifically labelled with either N‐ethylmaleimide to block free sulfhydryls or with biotin to enrich oxidized Cys residues (Fig. 1C) (McDonagh et al., 2009).

The goal of this study was to use both strategies to compare the redox proteomes of WT F. graminearum and the ΔNoxAB strain and to develop a list of candidate redox‐sensing proteins. To confirm their roles as targets of NoxAB‐derived ROS, we made deletions of a few of the candidate redox genes and made point mutations in Cys residues in one candidate protein (FGSG_10089), and determined their phenotypes in vivo. The results indicated that certain proteins are specifically targeted by ROS originating from NoxAB and, furthermore, that genetic disruption of at least some of these target proteins results in mutant phenotypes similar to the ΔNoxAB strain.

Results

2‐DE reveals 19 candidate redox‐sensing proteins

The 2‐DE procedure labels reduced Cys residues whose levels are expected to be higher in the NoxAB mutant as it presents a less oxidizing environment than its WT counterpart. Protein spots with changing redox levels can be visualized by fluorescence imaging and proteins with active Cys residue(s) are identified by mass spectrometry (MS). The advantage of this approach is that relative levels of oxidized Cys residues are visibly rendered and can be easily quantified and compared (i.e. levels of redox relative to total protein). This gives a true measure of the changing oxidation states of targeted proteins. However, there are experimental limitations imposed by the 2‐DE method, principally in protein solubilization and in the limited dynamic range of separation (Rabillout et al., 2010). Co‐migration of proteins complicates correct identification; however, by analysing the same spot from proteins extracted from both WT and ΔNoxAB strains, it is possible to quantify peptides at the MS level using label‐free approaches, and hence to identify only targeted proteins (Bantscheff et al., 2007).

Oxidized Cys levels were expected to be higher in proteins in WT F. graminearum than in the ΔNoxAB mutant because the latter is expected to produce less O₂ ^–. This would present a slightly less oxidizing environment, and thus bimane labelling would be expected to be higher in proteins isolated from the ΔNoxAB mutant strain than from WT F. graminearum. Therefore, reduced Cys residues (Cys‐SH) will be labelled with bimane and oxidized Cys residues (e.g. Cys‐S‐OH, Cys‐S‐S‐Cys, etc.) will be labelled with carbamidomethyl (Fig. 1). To reveal differentially oxidized Cys residues, we ran 2‐DE polyacrylamide gel electrophoresis (PAGE) gels on proteins isolated from the WT and ΔNoxAB F. graminearum strains grown in nutrient‐rich medium for 24 h (0 h) and then switched to nutrient‐limiting medium for 5, 10 and 24 h. An example of a gel stained with Coomassie brilliant blue (CBB) and mBBR (and protein spots isolated from different time points for MS analysis) is shown in Fig. S1 (see Supporting Information).

To identify targeted Cys residues, we first calculated their fluorescence ratio (i.e. fluorescence signal/CBB signal), because this measurement takes into account both changes in protein abundance and changes in redox status. Proteins with targeted Cys residues will have significantly higher fluorescence ratios in the ΔNoxAB mutant than in WT. Proteins from spots in the 2‐DE gels from all four time points whose abundance had not changed between the two strains, yet were significantly more oxidized in the WT than mutant strain, were isolated and analysed by MS (Table S1, see Supporting Information). This is consistent with the hypothesis that NoxAB is a source of additional cytoplasmic hydrogen peroxide (H₂O₂), which oxidizes proteins with susceptible Cys residues—the redox‐sensing proteins. As all the protein spots contained more than one protein, it was important to determine which peptide(s) was responsible for the specific signal. This was achieved by comparing the precursor ion intensities of the Cys‐containing peptides from all spots in Table S2 (see Supporting Information). Nineteen of 28 redox‐susceptible proteins contained peptides in which the precursor ion intensity of the Cys peptide was at least two‐fold higher in the NoxAB mutant relative to the WT strain (Table S2).

Gel‐free approach yields 99 redox‐sensing proteins

Given the limitations of 2‐DE in protein solubilization and dynamic range, we also performed gel‐free, shotgun experiments on the same time points as used in the 2‐DE experiment. The advantage of this approach over 2‐DE is that it has an extended dynamic range, and is not limited to proteins of a given pI range, and more proteins are solubilized at the start of the experiment. As redox‐targeted peptides are enriched by affinity chromatography, the relative quantification of proteins is less accurate and, critically, quantitative changes in proteins with no relative change in redox between the two samples cannot be ruled out. In four independent biological replicates, the gel‐free approach revealed a total of 99 proteins that contained at least one modified Cys residue, indicating that this residue was oxidized at the start of the experiment [minimum of one significant peptide, P < 0.05, 0.8% decoy false discovery rate (FDR); Table S3, see Supporting Information]. As the gel‐free approach involved enrichment, peptides lacking Cys or with unmodified Cys at the outset were lost. The identities of the 99 proteins that were more oxidized in the ΔNoxAB strain are shown in Table S3.

Comparative and functional analysis of redox‐sensing proteins

From a total of 118 redox‐sensing proteins from all four time points, we noted that there was a limited overlap of proteins between the two approaches (Table S4, see Supporting Information). However, the three proteins that overlapped and returned the same peptides with identical targeted Cys residues were FGSG_08737, a Woronin body major protein precursor, FGSG_00872, a UDP‐N‐acetylglucosamine pyrophosphorylase, and FGSG_09321, an acetoacetyl thiolase. The latter is an enzyme involved in isoprenoid biosynthesis and is an important regulatory protein responsive to numerous abiotic stresses (Soto et al., 2001). The peroxisomal origin of the Woronin body protein (FGSG_08737) suggests that it could be subject to redox regulation. The homologue in the rice blast fungus Magnaporthe oryzea has shown to be critical for pathogenicity of the fungus (Soundararajan et al., 2004). Through targeted deletion of FGSG_08737, we confirmed the role of this protein in the pathogenicity of F. graminearum (Figs 2A and S2B, see Supporting Information). In addition, we showed that the FGSG_08737 mutant strain was unable to form perithecia, similar to the phenotype observed with the ΔNoxAB mutant strain (Fig. 2B) (Wang et al., 2014).

Figure 2.

Woronin body protein (FGSG_08737) is important for virulence and necessary for the formation of perithecia in Fusarium graminearum. (A) Each strain was assessed for the ability to infect susceptible wheat cultivar ‘Roblin’. Infection was scored by counting the number of visibly infected spikelets at days 7, 10, 14 and 21. (B) Perithecia were induced on carrot agar plates and photographed 5 days after Tween treatment. No perithecia were detected in either NoxA/B double mutant strains or the ΔFGSG_08737 deletion strain compared with the wild‐type (WT) strain. This experiment was repeated at least three times with similar results. [Color figure can be viewed at wileyonlinelibrary.com]

All the redox proteins were functionally categorized by the MIPS FunCat database (https://mips.helmholtz-muenchen.de/funcatDB/). A comparison between the two strategies revealed that proteins from the gel‐based method predominantly enriched for functions linked to energy and metabolism (Table S5, see Supporting Information). This group included proteins necessary for carbohydrate metabolism, such as phosphoglycerate kinase (PGK, FGSG_03992), a hexokinase (FGSG_00500), a glucokinase (FGSG_08774), a pyruvate decarboxylase (FGSG_09834) and a pyruvate kinase (FGSG_07528) (Table S5). Genetic analyses have shown that FGSG_00500 is important for virulence and mycotoxin production in F. graminearum (Zhang et al., 2016) and FGSG_09834 is dispensable for pathogenicity (Figs S2C,E). Interestingly, all of these metabolically associated genes were also shown to be temporally regulated and were identified in nutrient‐limiting conditions (Fig. 3; Table S6, see Supporting Information).

Figure 3.

Functional categories of 118 redox‐modified proteins in nutrient‐rich and nutrient‐limiting conditions The Venn diagram describes the proteins enriched in the nutrient‐rich condition (0 h), whereas nutrient‐limiting conditions are represented by the 5‐, 10‐ and 24‐h time points. The parentheses show the number of protein(s) identified by the gel‐based approach. Proteins from each time point were functionally categorized and showed that proteins in the nutrient‐rich condition were enriched for protein translation. The fungus growing in nutrient‐limiting conditions was enriched for proteins associated with primary metabolism at earlier time points (5 and 10 h) and proteins linked to secondary metabolism at the later time point (24 h). [Color figure can be viewed at wileyonlinelibrary.com]

Proteins unique to the gel‐free method were also enriched for proteins associated with metabolism (Table S5). However, additional functional categories, such as ‘Proteins with binding function or cofactor requirement’, ‘Cell type differentiation’, ‘Stress response’ and ‘Protein fate’, were also enriched (Table S5). Unlike the gel‐based method, the vast majority of the redox proteins in nutrient‐rich conditions (36 of 37) were identified by the gel‐free method (Fig. 3). In this environmental condition, proteins linked to protein translation, such as ribosomal proteins (FunCat 12: FGSG_07292, FGSG_ 10245, FGSG_10941) and translation initiation factor eIF5a (FGSG_01955), were specifically enriched. Translation initiation factor eIF5a has been shown to be modified post‐translationally by the unique amino acid hypusine, and interference with this process leads to substantial reduction in virulence in F. graminearum (Martinez‐Rocha et al., 2016). Interestingly, the genes that are involved in the hypusination process also regulate ROS production (Martinez‐Rocha et al., 2016). In addition, proteins involved in cell signalling and differentiation (FunCat 30 and 43: FGSG_08737, FGSG_02646) were also specifically enriched under this condition. FGSG_02646 was identified in a yeast two‐hybrid screen with TAP42 protein, a component of the target of rapamycin (TOR) complex in F. graminearum (Yu et al., 2014). The TOR complex is conserved in all eukaryotes, is involved in a myriad of cellular processes and is also subject to redox regulation (Yoshida et al., 2011).

The rationale for the use of two different culture conditions was to ensure that redox‐modified proteins contributing to fungal growth and development, and proteins that contribute to fungal pathogenicity, were identified. This was borne out by the identification of redox‐modified proteins associated with cell wall architecture (FGSG_02022, FGSG_10089) in nutrient‐rich conditions. Proteins that contribute to virulence are typically secreted and linked to the production of secondary metabolites. We identified these proteins in nutrient‐limiting conditions, including FGSG_02204, FGSG_04546, FGSG_06384, FGSG_06767, FGSG_04124, FGSG_04091, FGSG_03072, FGSG_00048, FGSG_00071, FGSG_00421, FGSG_01346 and FGSG_03638 (Fig. 3; Table S6) (Phalip et al., 2005). FGSG_00071 is annotated as a cytochrome P450 monooxygenase and is part of the cluster that regulates the deoxynivalenol (DON) mycotoxin pathway in F. graminearum. A mutation in this gene results in the accumulation of DON intermediates (McCormick et al., 2004). A network‐based analysis of virulence genes included FGSG_00071, associating this gene with the virulence pathway in F. graminearum (Lysenko et al., 2013). The secreted protein FGSG_03072 is related to acidic exopeptidase, was highly expressed during DON‐inducing conditions and may play an essential role in fungal cell wall remodelling to support mycelial growth at the infection front (Lowe et al., 2015). A homologue of another secreted protein, an enolase (FGSG_01346), has been shown to be under redox regulation in human neuroblastoma cells (Ishii and Uchida, 2004).

The cell wall‐associated proteins included proteins with glycosylphosphatidylinositol (GPI)‐anchored domains (FGSG_02022 and FGSG_10089) (Kinoshita and Fujita, 2016). GPI‐anchored proteins (GPI‐APs) have been associated with F. graminearum morphogenesis and pathogenicity (Rittenour and Harris, 2013). FGSG_02022 encodes β(1→3)glucanosyltransferase, part of a Gelp (glucan elongase) protein family, involved in the elongation of β(1→3)glucans, a main component of the fungal cell wall (Gastebois et al., 2010). Repeated attempts to obtain a deletion mutant of this gene in F. graminearum were unsuccessful. Similar attempts to delete a homologue in Aspergillus fumigatus were also unsuccessful; however, a heterokaryon rescue technique demonstrated conclusively that this is an essential gene (Gastebois et al., 2010). We were successful in obtaining the deletion of the other GPI anchor‐encoding gene FGSG_10089 (Fig. S3, see Supporting Information).

GPI‐anchored cell wall protein FGSG_10089 is required for virulence in F. graminearum

Analysis of FGSG_10089 identified domains conserved in many GPI‐APs, including a signal peptide at the N‐terminus (amino acids 1–20), an omega site (amino acid residue 374; P _value = 5.4 × 10⁻⁶) where GPI is added, and a hydrophobic tail at the C‐terminus (Fig. 4A) (Kinoshita and Fujita, 2016). FGSG_10089 has a homologue in yeast (ECM33; E = 1 × 10⁻²⁵) with proposed roles in sporulation and cell wall architecture (Chabane et al., 2006; Pardo et al., 2004). Similarly, ccg‐15, a Neurospora crassa homologue (E = 6 × 10⁻⁴⁴), is a clock‐regulated gene that governs the frequency of reproduction (i.e. sporulation) (Lombardi and Brody, 2005). A targeted deletion of FGSG_10089 showed no differences in spore production (unpublished, Subramaniam), but displayed a significant decrease in virulence compared with the WT strain (Table 1), which was mostly reversed in the complemented strain (ΔFGSG_10089:FGSG_10089) and in the strain that constitutively expressed FGSG_10089 (ΔFGSG_10089:FGSG_10089 _Oex) (Fig. S3). As FGSG_10089 was observed at 5 h in nutrient‐limiting medium, we were interested to know whether the deletion had any impact on the production of 15‐acetyldeoxynivalenol (15‐ADON) in culture. No significant changes in 15‐ADON production were measured (Table 2).

Figure 4.

Phenotypic characterizations of the glycosylphosphatidylinositol (GPI)‐anchored protein FGSG_10089. (A) The amino acid sequence of the GPI‐anchored protein FGSG_10089. SAA is the GPI anchor/cleavage site; the receptor L‐domain is underlined; the LRR domain is italicized and underlined. The two active cysteine peptides in the FGSG_10089 protein identified by mass spectrometry are shown in bold. The cysteine (C325, red) residue in the peptide (double underline) was mutated to serine and phenylalanine for functional analyses. (B) Point mutations in the cysteine residue of FGSG_10089 do not affect expression. Reverse transcription‐polymerase chain reaction (RT‐PCR) was used to monitor the expression of FGSG_10089 by the primer set P7/P8 in the wild‐type (lane 1), ΔFGSG_10089 (lane 2), ∆FGSG_10089:FGSG_10089 _Oex (lane 3), ∆FGSG_10089:FGSG_10089 _C→S,Oex (lane 4) and ∆FGSG_10089:FGSG_10089 _C→SF,Oex (lane 5) strains. β‐Tubulin (FGSG_09530) was used as an internal control. (C) Cellophane penetration assays to assess virulence. One thousand conidiophores from wild‐type (WT), NoxAB, ∆FGSG_10089, ∆FGSG_10089:FGSG_10089 _Oex, ∆FGSG_10089:FGSG_10089 _C→S,Oex and ∆FGSG_10089:FGSG_10089 _C→SF,Oex strains were spotted in the middle of minimal medium (MM) plates overlaid with cellophane membrane (+ Cellophane). After 5 days of growth, cellophane was peeled off (Post‐cellophane) and the growth of mycelia on MM plates was monitored after 3 days. [Color figure can be viewed at wileyonlinelibrary.com]

Table 1.

Pathogenicity tests of FGSG_10089 mutant strains.

Fusarium strains	Infected wheat heads*
Wild‐type	98 ± 2
ΔFGSG_10089	55 ± 9
ΔFGSG_10089:ΔFGSG_10089	80 ± 10
ΔFGSG_10089:ΔFGSG_10089 Oex	72 ± 8
ΔFGSG_10089:ΔFGSG_10089 C → S,Oex	57 ± 11
ΔFGSG_10089:ΔFGSG_10089 C → F,Oex	52 ± 8

*Percentage of infected heads ± standard deviation with P < 0.05.

Table 2.

Quantification of 15‐acetyl‐deoxynivalenol (15 ADON) in FGSG_10089 mutant strains.

Fusarium strains	15‐ADON (µg/mg)	Mycelial weight (mg)
Wild‐type	13 ± 2	41 ± 2
ΔFGSG_10089	11 ± 9	40 ± 2
ΔFGSG_10089:ΔFGSG_10089	13 ± 10	40 ± 2
ΔFGSG_10089:ΔFGSG_10089 Oex	12 ± 8	40 ± 2
ΔFGSG_10089:ΔFGSG_10089 C → S,Oex	15 ± 11	41 ± 2
ΔFGSG_10089:ΔFGSG_10089 C → F,Oex	14 ± 8	39 ± 2

Redox‐modified Cys residue is required for virulence and cell wall integrity in F. graminearum

MS identified a Cys residue at position 325 in the FGSG_10089 protein as one of the Cys residues (Cys³²⁵) targeted by NoxAB (Fig. 4A; Table S3). To determine the importance and function of Cys³²⁵, two point mutations were introduced to change Cys to either serine (S) or phenylalanine (F) in the FGSG_10089 protein, and the plasmid constructs were constitutively expressed in the ∆FGSG_10089 mutant strain (ΔFGSG_10089:FGSG_10089 _C→S,Oex, ΔFGSG_10089:FGSG_10089 _C→F,Oex) (Fig. S3). First, we confirmed that the point mutations did not affect the expression of the gene (Fig. 4B). Next, we performed a cellophane penetration assay, which has been used to assess virulence function in F. graminearum (Subramaniam et al., 2015). It was shown that the ΔFGSG_10089 strain was unable to breach the cellophane membrane, similar to the ΔNoxA/B strain demonstrated previously (Fig. 4C; Wang et al., 2014). The ΔFGSG_10089 mutant strain complemented with WT FGSG_10089 (ΔFGSG_10089:FGSG_10089 _Oex) regained its ability to penetrate cellophane. Similarly, the mutant strains complemented with both Cys mutants (ΔFGSG_10089:FGSG_10089 _C→S,Oex, ΔFGSG_10089:FGSG_10089 _C→F,Oex) were also able to penetrate the membrane. This suggests that neither Cys mutant affects FGSG_10089 protein function and is not necessary for the initiation of the infection process (Fig. 4C). We followed the cellophane experiments and performed pathological assays on wheat heads to assess the overall disease potential of FGSG_10089. In contrast with cellophane experiments, the constitutive expression of Cys‐modified strains reduced virulence to a comparable extent to the ΔFGSG_10089 mutant strain (Table 1). This strongly indicates that Cys³²⁵ is necessary to cause disease and contributes to the overall virulence function of the FGSG_10089 protein. Similar to the mutant strain, the Cys‐modified strains showed no changes in growth or in 15‐ADON production compared with the WT strain (Table 2).

Mutation of FGSG_10089 homologues in Saccharomyces cerevisiae and the human pathogen Candida albicans showed that the cell wall architecture was dramatically altered (Chabane et al., 2006; Martinez‐Lopez et al., 2004; Pardo et al., 2004). In view of this, we performed scanning electron microcopy on spores from WT, ∆FGSG_10089 and one of the Cys‐modified strains (ΔFGSG_10089:FGSG_10089 _C→F,Oex). Similar to the results observed in yeast, both ∆FGSG_10089 and ΔFGSG_10089:FGSG_10089 _C→F,Oex showed severe morphological defects compared with WT spores (Fig. 5B). Staining with calcofluor white (CFW), a compound that binds to chitin or glucan polymer of the cell wall, showed uniformly stained material in both WT and the ∆FGSG_10089 mutant spores (Fig. 5C). Together, these results suggest that FGSG_10089 plays a role in cell wall architecture.

Figure 5.

FGSG_10089 is involved in the cell wall architecture of Fusarium graminearum. (A) Scanning electron micrograph of wild‐type spores with a smooth cell wall. A wrinkled cell wall is visible in both the mutant ΔFGSG_10089 spores and in the cysteine‐modified mutant of FGSG_10089. The wrinkled phenotype was reversed in the complemented strain (ΔFGSG_10089:FGSG_10089). (B) The ΔFGSG_10089 mutant is not affected in chitin or glucan accumulation. Calcofluor white was used to stain both wild‐type and mutant spores, and there were no significant changes in glucan accumulation in the cell walls of spores. [Color figure can be viewed at wileyonlinelibrary.com]

Discussion

Superoxide anions generated by NADPH oxidases A and B are formed outside the cell and are rapidly converted to H₂O₂ by extracellular superoxide dismutases as a means of detoxification (Aguirre and Lambeth, 2010). As NOX influences both development and pathogenicity in many different fungi, we undertook this study to identify the downstream targets of this important enzyme. Real‐time imaging of H₂O₂ during the developmental stages of F. graminearum showed increased accumulation of H₂O₂ during hyphal septal formation and during the development of infection structures (Mentges and Bormann, 2015). The deletion of NoxA/B displayed severe defects in pathogenicity and in the development of sexual spore structures (Wang et al., 2014). Thus, a comparative analysis was undertaken between culture conditions in which proteins involved in fungal growth and development, i.e. nutrient‐rich medium (0‐h time point), and proteins related to virulence, i.e. nutrient‐limiting medium (5‐, 10‐ and 24‐h time points), were present. To maximize our output to identify redox proteins, we used two complementary strategies—gel‐based and non‐gel‐based proteomics approaches—and identified a total of 118 proteins that were modified by ROS produced by the action of NOX enzyme (Table S4).

Using the gel‐based approach, differences in bimane labelling were easily detected; however, assigning the difference in fluorescence signal to a particular Cys residue was more challenging because spots contained multiple proteins. To assign the difference in fluorescence signal to a particular Cys residue properly, we compared all the peptides that contained Cys residues in both WT and ΔNoxAB samples. As the peptide population contains both CAM‐ and bimane‐modified peptides, we first measured the precursor ion intensities of each type of peptide. This enabled us to determine the prevalence of modification (CAM or bimane) on each peptide, which was then used to calculate the relative abundance of the bimane peptides in ΔNoxAB and WT samples (Table S2). Occasionally, the peptide in question was not detected during the liquid chromatography‐mass spectrometry (LC‐MS) run in one or both samples, and therefore no comparison could be made. However, in other cases, the peptide was seen, but with only one modification. In these cases, there had been either no modification or near‐complete bimane modification.

The gel‐free ‘shotgun’ analysis yielded a far greater number of candidate redox‐sensing proteins and is not constrained by the experimental limitations of 2‐DE. In contrast with 2‐DE, the gel‐free approach yielded lower abundance proteins, membrane proteins and proteins with a pI falling outside the pH range of the 2‐DE analysis, exemplified by the identification of FGSG_10089, which is both membrane bound and is outside of the pH range of 2‐DE. One of the purported limitations of the gel‐free strategy is that it does not take into account changes in protein abundance through time. In other words, enriched, labelled peptides could simply be increased in abundance with no change in redox state. However, analysis from 2‐DE indicated that this was unlikely. There was very little measurable change in the overall proteome during the time course used, as measured by CBB. In all cases in which a change in the UV intensity was measured on the two‐dimensional gels, no significant differences in protein abundance were observed. Thus, 118 proteins that were identified by two distinct approaches could be considered as high‐confidence redox‐sensing proteins. This is substantiated by the number of proteins that were identified as regulated by redox status (Table 3).

Table 3.

Phenotype of redox‐sensing proteins identified in this study.

Gene name	Annotation	References
FGSG_00500	Probable hexokinase	Zhang et al. (2016)
FGSG_01346	Probable enolase	This study, ectopic
FGSG_01955	Translation initiation factor	Martinez‐Rocha et al. (2016)
FGSG_02022	β(1‐3)Glucanosyltransferase, gel3p	This study, ectopic
FGSG_02646	Related to hexamer binding protein	Yu et al. (2014)
FGSG_03072	Related tripeptidyl‐peptidase 1	Lowe et al. (2015)
FGSG_03992	Phosphoglycerate kinase	McDonagh B et al. (2009); Bosco et al. (2012)
FGSG_09834	Pyruvate decarboxylase	This study
FGSG_10089	Related sporulation‐specific gene SPS2	This study
FGSG_08737	Woronin body major protein precursor	This study, Soundararajan et al. (2004)
FGSG_09321	Acetoacetyl‐CoA thiolase	Soto et al. (2001)

A comparative analysis between the two approaches indicated that the 2‐DE approach identified proteins that were enriched in primary metabolic pathways. Enzymes of these pathways are a major target of post‐translational redox modifications, mainly as a result of the generation of NADH, NADPH and FADH which serve as reducing power, and as such, levels of these metabolites must be carefully balanced with the redox potential of the cell (McDonagh et al., 2009, 2011; Ralser et al., 2007). Rerouting of carbon flux from glycolysis to the pentose phosphate pathway is a major response to oxidative stress in Caenorhabditis elegans (Ralser et al., 2007), S. cerevisiae (McDonagh et al., 2011) and, probably in F. graminearum. At least six enzymes of primary carbon metabolism were targeted by NoxAB‐generated ROS, including glyceraldehyde dehydrogenase (GAPDH, FGSG_16627), which is modified in S. cerevisiae following H₂O₂ stress (McDonagh et al., 2011). In our study, GAPDH was identified in both the gel‐free and gel‐based experiments. Regulation of this enzyme allows metabolic flux to be redirected through the pentose phosphate pathway to generate more NADPH under oxidative stress conditions (Ralser et al., 2007). The GAPDH tryptic peptide with the same consensus Cys residues has also been shown to be oxidized in Arabidopsis thaliana and in human lung carcinoma cell lines treated with H₂O₂ (Brodie and Reed, 1987; Hancock et al., 2005). Another protein, methionine adenosyltransferase (FGSG_00421), with a known redox‐sensing function, was also identified in our study (McDonagh et al., 2009). Following treatment with H₂O₂, the yeast methionine adenosyltransferase has been shown to be modified at the same Cys residue as the protein from F. graminearum (McDonagh et al., 2009). The same enzyme in rat liver has been shown to be reversibly inactivated by covalent modification of Cys residues (Sánchez‐Góngora et al., 1997). The Cys residue at position 312 was found to affect enzyme activity and this same Cys residue was targeted in F. graminearum.

The importance of redox‐modified Cys residues has been demonstrated with the modification of C⁹⁸ in PGK in yeast, which affected its enzyme activity (McDonagh et al., 2009). Similarly, the activity of PGK was affected by the intracellular redox environment in the photosynthetic protist Phaeodactylum tricornutum (Bosco et al., 2012). The authors analysed all six Cys residues by molecular modelling and site‐directed mutagenesis, and identified three Cys residues that could be subjected to redox regulation (Bosco et al., 2012). Direct evidence for the role of disulfide bonds mediated by Cys residues in protein structure comes from OxyR characterized in S. cerevisiae (Jo et al., 2015). The transcription factor OxyR is active only under oxidized conditions to create a disulfide bridge, which changes its conformation, required for DNA binding (Jo et al., 2015). Another example comes from mammalian studies of the GPI‐AP Thy‐1 cell surface antigen of the immunoglobulin superfamily. Loss of oxidation leads to loss of epitope recognition by these antibodies (Bradley et al., 2013). In our study, the phenotypic changes as a result of deletion or substitution of Cys in FGSG_10089 are probably caused by a structural change in the cell wall or a change in the ability of FGSG_10089 to interact with other proteins involved in cell wall architecture. Moreover, similar changes in phenotypes were observed with both the deletion and Cys‐substituted mutant strains, leading us to conclude that Cys³²⁵ is required for the function of this GPI‐AP.

In conclusion, this study used two complementary proteomic approaches to identify redox‐sensing proteins in F. graminearum. A comparison between WT and the NADPH oxidase mutant (ΔNoxA/B) helped us to identify a number of proteins that have not been characterized previously as redox‐sensing proteins. Furthermore, the time course experiments also allowed us to separate redox‐sensing proteins involved in homeostasis from those involved in the active metabolic state. Genetic analysis also suggested that ROS from the NOX enzyme may act directly on these redox‐sensing proteins; however, recent studies of redox modification of transcription factors have suggested that ROS emerging from other sources, such as peroxidases, may work synergistically with NOX (Sikes, 2017).

Experimental Procedures

Biological materials

Fusarium graminearum WT strain DAOM 233423 originates from the Canadian Collection of Fungal Cultures (CCFC/DAOM), Agriculture and Agri‐Food Canada, Ottawa, ON, Canada. The NADPH oxidase mutant used in this study (ΔNoxAB) was constructed and the phenotype of this strain has been described previously (Wang et al., 2014). Cultures of the WT and ΔNoxAB mutant strains were grown separately in liquid medium as described previously (Walkowiak et al., 2015). Mycelia were harvested at the following times after transfer to the nutrient‐limiting medium: 0 (i.e. prior to transfer), 5, 10 and 24 h.

Protein extraction, labelling and 2‐DE

After filtering the mycelial mass through filter paper (Whatman #3 Sigma‐Aldrich, St. Louis, MO, USA), proteins were extracted and labelled with mBBR as described by Bykova et al. (2011). Briefly, mycelia were ground to a fine powder in liquid nitrogen, and 2 mL of 50 mm Tris‐HCl (pH 8.0), 2% (w/v) 3‐[(3‐Cholamidopropyl)dimethylammonio]‐1‐propanesulfonate hydrate, CHAPS, 2% (w/v) dodecyl β‐d‐maltose and 0.5 mm mBBR (Calbiochem, San Diego, CA, USA), with Complete™ ethylenediaminetetraacetic acid (EDTA)‐free protease inhibitor cocktail, were added to the powder. After 30 min in the dark, proteins were precipitated from acetone and finally redissolved in isoelectric focusing (IEF) solution [7 m urea, 2 m thiourea, 4% (w/v) CHAPS, 20 mm dithiothreitol (DTT) and 0.5% (v/v) ampholyte, pH 3–10 (BioRad Laboratories, Mississauga, ON, Canada)].

The 2‐DE was performed as described by Rampitsch and Bykova (2009). Briefly, samples were used to rehydrate 24‐cm IEF strips (IPG: GE Healthcare, Mississauga, ON, Canada), pH 4–7, in a Teflon reswelling tray at 22 ºC under a layer of mineral oil. Strips were then focused for a total of 58 kVh (Multiphor II: GE Healthcare). After focusing, the strips were reduced, alkylated and equilibrated as recommended by the manufacturer, and separated by sodium dodecylsulfate‐polyacrylamide gel electrophoresis (SDS‐PAGE) in 12% gels (Ettan Dalt‐6: GE Healthcare) at a constant power setting of 2.5 W per gel, 30 min, followed by 100 W (total). Gels were then fixed and photographed under UV‐light illumination at 365 nm (Alphaimager: Protein Simple, San Jose, CA, USA) before being stained with CBB as described by Bykova et al. (2011). Gels were analysed using PDQuest software (v8.0.1: BioRad Laboratories, Hercules, CA, USA) and scanned to produce TIF images.

Mass spectrometry

Proteins of interest were cut from the gels in cubes of approximately 1 mm³. In‐gel trypsin digestion and subsequent peptide extraction were carried out as described previously (Rampitsch and Bykova, 2009). Mass spectra of the resulting tryptic peptides were acquired in a hybrid quadrupole‐Orbitrap mass spectrometer (Q‐Exactive: ThermoFisher, Bremen, Germany). Tryptic peptides were separated through a C₁₈ column (12‐cm fused silica column, 75 µm ID, packed with Vydac C₁₈, 5 µm beads, 300 Å pores), coupled directly to the mass spectrometer via a nanoelectrospray ionization source. An acetonitrile gradient [2% (v/v) to 40% (v/v) in 0.1% (v/v) formic acid] was delivered at 300 nL/min over 40 min (Easy nLC1000: ThermoFisher, San Jose, CA, USA). A survey scan acquired over the range m/z 300–2000 was followed by 12 MS/MS scans of the most intense ions, with dynamic exclusion set to 15 s. Protein identification of the MS/MS spectra was performed using Mascot Server v2.4 (MatrixScience, London, UK). The following parameters were set: a monoisotopic mass accuracy of ±5 ppm; up to one missed cleavage; peptide charge up to +5; variable modifications of carbamidomethyl (Cys), mBBR (Cys with a mass shift of +190.074 atomic mass units) and oxidation (Met). Raw MS data files were converted to MGF and used to query the genomic sequence of F. graminearum containing 13 826 sequences downloaded from MIPS (Helmholz Zentrum, Munich, Germany: www.mips.helmholtz-muenchen.de/). Proteins were considered to be correctly identified if returns contained two or more peptides with a significant ion score, as defined by Perkins et al. (1999).

Data analysis

For 2‐DE, spot intensity values for both CBB‐stained and fluorescent spots were created by PDQuest for all three biological replicates. Values were reported as means. Quantitative changes were calculated as fluorescence ratios relative to protein abundance (Bykova et al., 2011). Precursor ion intensities were calculated for all peptides in these paired spots using Mascot Distiller (Quantitation Toolbox v2.5.1: MatrixScience).

Gel‐free enrichment and analysis of redox‐modified peptides

Peptides from putative redox‐sensing proteins were enriched using a biotin‐switch method (McDonagh et al., 2009). In brief, protein extracts from WT F. graminearum and the ΔNoxAB mutant strains were first blocked with Iodoacetamide (IAM) and then reduced with DTT. Proteins were then labelled with biotin‐HPDP (N‐[6‐(biotinamido)hexyl]‐3’‐(2’pyridylthio)propionamide)(Pierce, Rockford, IL, USA) through thiol groups of Cys, digested with trypsin and the peptides were enriched by affinity chromatography on streptavidin beads (Streptavidin–Agarose CL‐4B resin: Sigma‐Aldrich, St. Louis, MO, USA). Elution from the column with 5% (v/v) β‐mercaptoethanol yielded pools of enriched tryptic peptides whose Cys residues were oxidized prior to extraction and contained a mercaptoethanol adduct. These were analysed by LC‐MS/MS using a quadrupole time‐of‐flight (QTOF) mass spectrometer (TripleTOF 5600: ABI Sciex, Concord, ON, Canada) under a fee‐for‐service contract at the Manitoba Centre for Proteomics and Systems Biology (Winnipeg, MB, Canada), as described previously (Rampitsch et al., 2015). Data analysis using Mascot was performed as described above, except that the variable modifications were set to: carbamidomethyl (Cys), destreak (Cys; with a mass shift of +76 atomic mass units) and oxidation (Met). Scaffold (v4.4: Proteome Software Inc., Portland, OR, USA) was used to validate MS/MS‐based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 95.0% probability by the Peptide Prophet algorithm with Scaffold delta‐mass correction. Protein identifications were accepted if they could be established at greater than 95.0% probability and contained at least one identified peptide. Protein probabilities were assigned by the Protein Prophet algorithm. Spectral counts were assigned by Scaffold. The gel‐free enrichment was performed at the following time intervals post‐transfer into the nutrient‐limiting medium: 0, 5, 10 and 24 h. Four independent biological replicates were performed.

Generation of F. graminearum transgenic strains and fungal transformation

The deletion of FGSG_10089 was accomplished by polymerase chain reaction (PCR) amplification of the two homologous recombination sequences (HRSs), with primer set P1/P2, which amplified the 5′ flanking sequence of FGSG_10089 (HRS1), and primer set P3/P4, which amplified the 3′ flanking region of FGSG_10089 (HRS2) (Fig. S1A). The two HRS products were introduced into the pRF‐HU2 vector, with a hygromycin selection marker (Hyg) (Frandsen et al., 2008). The constitutive expression of the FGSG_10089 _Oex vector was constructed by PCR amplification of the FGSG_10089 coding sequence (CDS) with the primer set P5/P6, and cloned into the pRF‐GUE vector using the USER (Uracil‐Specific Excision Reagent) cloning procedure (New England Biolabs, Ipswich, MA, USA) (Frandsen et al., 2008). The conversion of Cys³²⁵ to phenylalanine and serine was achieved by first cloning the FGSG_10089 CDS amplified with the primer set P7/P8 into the pENTR/D‐TOPO Gateway vector (ThermoFisher, San Jose, CA, USA). The point mutation was achieved by PCR amplification of the FGSG_10089 CDS with P9/P10 and P11/P12, respectively, and digested with 1 μL of Dpn1 enzyme, and an aliquot was transformed into Escherichia coli DH5α cells. The mutations were confirmed by sequencing. The Cys mutants were cloned into the pRF‐GUE vector with the primer sets P5/P6. To complement the ΔFGSG_10089 deletion mutant strain, a construct with the native promoter fused to the CDS was cloned into the pRF‐GU vector. First, the ~1‐kb promoter region was amplified with the P1/P2 primer set. A second PCR product with a native CDS and the two CDSs with mutations at Cys³²⁵ was amplified with the primer set P19/P8. The fusion PCR product was created by combining the two PCR products with the primer set P13/P14, and then cloned into pRF‐GU using the USER reaction (New England Biolabs). All of the genetically altered strains were generated by Agrobacterium‐mediated transformation (Frandsen et al., 2008). The same strategy to delete FGSG_10089 was also used to delete FGSG_08737 and FGSG_09834. The two HRS PCR fragments for FGSG_08737 were amplified by the primer sets P20/P21 and P22/P23, and, similarly, the two HRS PCR fragments for FGSG_09834 were amplified by the primer sets P34/P35 and P36/P37. The HRS for each gene was cloned into a pRF‐HU2 vector. The deletion of each gene was verified by PCR using the P26/P27 primer set for the selection marker hygromycin and P28/P29 to detect the geneticin selection marker. The presence/absence of FGSG_10089 was detected by primer set P7/P8, of FGSG_08737 by primer set P24/P25, of FGSG_09834 by primer set P38/P39 and of Tri6 by primer set P30/P31.

Trichothecene analysis by high‐performance liquid chromatography (HPLC)

To induce trichothecene production in liquid culture, a modified two‐stage medium protocol was employed (Nasmith et al., 2011). A nylon net filter was added to the culture plates. This allowed for even growth and avoided mycelial clumps (Walkowiak et al., 2015). Briefly, 2 × 10⁴ spores of WT and mutant F. graminearum strains were inoculated into 4 mL of nutrient‐rich medium for 48 h and washed twice with water. The mycelial mass was suspended in 4 mL of nutrient‐limiting medium (1 g (NH₄)₂HPO₄, 3 g KH₂PO₄, 0.2 g MgSO₄.7H₂O, 5 g NaCl, 40 g sucrose, 10 g glycerol in 1 L, pH 4.0) and then transferred to six‐well culture trays. Trichothecenes were analysed by HPLC with direct injection of 100 μL of the culture filtrate into a 150 mm × 4.6 mm Hypersil ODS column (5 μm C₁₈: ThermoFisher, Mississauga, ON, Canada) using an isocratic methanol : water gradient at 15% over 25 min at a flow rate of 1 mL/min. Trichothecenes were monitored by UV light at 220 nm and the concentration of 15‐ADON was calculated based on a dilution curve derived from pure 15‐ADON standard (Nasmith et al., 2011).

DNA extraction, PCR, reverse transcription (RT)‐PCR and quantitative real‐time PCR

DNA from WT and mutant F. graminearum strains was isolated from mycelia grown on a fresh potato dextrose agar (PDA) plate with an E.Z.N.A. Fungal DNA Mini Kit (Omega Bio‐Tek, Norcoss, GA, USA). Briefly, mycelia were ground in a screw cap microcentrifuge tube using a Bertin Precellys 24 Homogenizer (Bertin Corp. Rockville, MD, USA). PCRs consisted of 0.2 μm forward primer, 0.2 μm reverse primer, 0.2 mm deoxyribonucleotide triphosphates (dNTPs), 1 × PCR buffer (Stratagene, San Diego, CA, USA) and 25 ng of DNA template and PfuTurbo Cx Hotstart polymerase (Stratagene). All of the qRT‐PCRs were performed in triplicate using an Applied Biosystems Power SYBR Green Kit (ThermoFisher, Canada) and an Applied Biosystems StepOne Plus Real‐Time PCR System, according to the manufacturer’s instructions (Applied Biosystems). To quantify the copy number of the hygromycin gene in the FGSG_08737 and FGSG_09834 transgenic lines, the Tri6 gene was used as control (Walkowiak et al., 2015). The data were imported and analysed in StepOne software (v2.1 ThermoFisher). The strains complemented with WT FGSG_10089 and the Cys mutants, FGSG_10089 _C→S and FGSG_10089 _C→F, were grown in potato dextrose broth (PDB) for 24 h, and RNA was extracted (Subramaniam et al., 2015). The expression of FGSG_10089 was monitored by RT‐PCR with the primer set P7/P8, and β‐tubulin was used as internal control (P42/P43). All the primers used in this study are listed in Table S7 (see Supporting Information).

Cellophane penetration, pathogenicity assays and analyses in Triticum aestivum

The cellophane penetration assays were performed in duplicate by the method described previously (Subramaniam et al., 2015). Briefly, 20 000 condiospores were spotted onto the centre of solid minimal medium overlaid with a cellophane sheet. After 5 days of growth at 28 ºC, the cellophane layer was removed and the plates were incubated for an additional 3 days to observe mycelial growth that had successfully penetrated the cellophane. Photographs were recorded with a Nikon D90 digital camera with an AF Micro‐Nikkor 60 mm f/2.8D lens (Nikon Canada Inc., Mississauga, ON, Canada).

All wheat pathogenicity assays were performed on Triticum aestivum L. ‘Roblin’, a susceptible wheat variety, and a resistant variety 'Sumai 3'. Plants were grown in growth chambers until anthesis was reached, as described previously (Schreiber et al., 2011). At mid‐anthesis, the plants were inoculated between the palea and lemma with approximately 1000 conidia. Inoculated plants were transferred to a contained misting facility and monitored for the development of disease symptoms, such as spikelet discoloration. Data were collected when heads in the control treatment group exhibited approximately >90% infection. Student’s t‐test was performed to assess any differences between the WT and transgenic strains.

Confocal microscopy for cell wall analysis

Confocal microscopy was conducted with the help of Denise Chabot (Agriculture and Agrifood Canada, Ottawa, ON, Canada). Fresh conidia from WT and ΔFGSG_10089 strains were suspended in CFW (0.07% in 0.1 m NaPO₄, pH 8) on a glass slide. A Zeiss LSM Duo 510 confocal microscope with ZEN 2009 software was used to image the mycelia and conidia. Epifluorescence was imaged with a Zeiss Filter #1 (full range of colours). CFW was excited with a laser at 405 nm and emission was recorded with BP 420–490 nm. A differential interference contrast channel was used for all observations (Objective‐Alpha Plan Apochromat 63X/1.46) Carl Zeiss Canada, Toronto, ON, Canada.

Scanning electron microscopy (SEM) of spores of WT and ΔFGSG_10089 F. graminearum strains

Spores isolated by carboxymethyl cellulose medium were fixed, dehydrated and gold coated before visualization by SEM. Briefly, silicon wafers were coated with 0.1% poly‐l‐Lysine (EMS cat#: 19320‐A) for 25 min and air dried. WT and ΔFGSG_10089 spores were spotted onto silicon wafers and incubated at room temperature for 1 h in a humid chamber. Subsequently, the silica wafers with spores were fixed in 4% paraformaldehyde and 2.5% glutaraldehyde in 0.1 m sodium cacodylate buffer, pH 7.2, for 1 h at room temperature, rinsed and dehydrated through a graded ethanol series, and critical point dried (Biodynamics Research Corporation, Rockville, MD, USA). The wafers were then mounted on stubs using carbon adhesive tabs, sputter coated with gold to a thickness of 8 nm using an Emitech K550V Gold Sputter Coater (EM Technologies Ltd., Ashford, Kent, UK) and imaged with a Quanta 600 SEM at 20 kV (FEI Company™, Brno, Czech Republic).

Supporting information

Fig. S1 Example of two‐dimensional electrophoresis (2‐DE). (A) Examples of 2‐DE gels of the soluble Fusarium graminearum proteome (pH 4–7) isolated at 10 h in nutrient‐limited medium from wild‐type and ΔNoxA/B mutant strains. The gels were stained with Coomassie brilliant blue (CBB) and monobromobimane (mBBR). Proteins whose intensity changed significantly only with the bimane label with a significant fluorescence ratio were processed for mass spectrometry analysis. (B) Examples of protein spots identified at different time points. All of the spots (boxed) that showed differential bimane labelling at different time points were isolated, characterized and quantified. Intensity measurements are shown in Tables S1 and S2 (see Supporting Information).

Fig. S2 Construction and characterization of FGSG_08737 and FGSG_09834 Fusarium graminearum deletion mutant strains. (A) Scheme used to construct both deletion mutants in the pRF‐HU2 vector with hygromycin (Hyg) as antibiotic selection marker in F. graminearum. The two homologous recombination sequences (HRSs) were amplified by polymerase chain reaction (PCR) for each gene and introduced into pRF‐HU2. All the primers are listed in Table S7 (see Supporting Information). (B) Characterization of the ΔFGSG_08737 F. graminearum strain. DNAs from the wild‐type (lane 2), ΔNoxA/B (lane 3) and ΔFGSG_08737 (lane 4) mutant strains were screened for Tri6 with the primer set P31/P32(Tri6), for FGSG_08737 with the primer set P25/P26 and for hygromycin (Hyg) with the primer set P27/P28. Lane 1 (dH₂O) represents the water control. (C) Characterization of the ΔFGSG_09834 F. graminearum strain. Genomic DNAs from the wild‐type (lane 2), ΔNoxA/B (lane 3) and ΔFGSG_09834 (lane 4) mutant strains were screened for Tri6 with the primer set P31/P32 (Tri6), for FGSG_09834 with the primer set P39/P40 and for hygromycin (Hyg) with the primer set P27/P28. Lane 1 (dH₂O) represents the water control. (D) Assessment of the copy number of the hygromycin marker gene in the deletion strains. Quantitative PCR (qPCR) was used to assess the copy numbers of hygromycin in ΔFGSG_09834 and ΔFGSG_08737 strains. The Tri6 deletion strain was used as control (Walkowiak et al., 2015). The qPCR primer set P33/P34 was used to detect Hyg. (E) The ΔFGSG_09834 F. graminearum strain shows no defects in virulence. Pathology tests were performed with spores from wild‐type, NoxA/B (ΔNoxA/B) and ΔFGSG_09834 mutant strains on a susceptible cultivar (’Roblin’) of wheat plants. The disease (% infected wheat heads) was monitored over time (days post‐inoculation, DPI). This was repeated twice with similar results.

Fig. S3 Construction and characterization of the FGSG_10089 mutant Fusarium graminearum strain. (A) Construction scheme of the F. graminearum FGSG_10089 deletion strain. The two homologous recombination sequences (HRS1 and HRS2) were amplified by the primer sets P1/P2 and P3/P4, respectively, cloned into the pRF‐HU2 vector with hygromycin (Hph) as the antibiotic selection marker, and transformed into the wild‐type F. graminearum strain. (B) Scheme of construction of the constitutive expression vectors for wild‐type FGSG_10089 (FGSG_10089 _Oex) and the two cysteine mutants, FGSG_10089 _C→S,Oex and FGSG_10089 _C→F,Oex, in F. graminearum. The coding sequences (CDSs) of wild‐type FGSG_10089 and the two cysteine mutants were polymerase chain reaction (PCR) amplified by the primer set P5/P6, cloned into the pRF‐GUE vector with geneticin (Gen) as the antibiotic selection marker, and transformed into the ΔFGSG_10089 deletion F. graminearum strain (∆FGSG_10089). (C) Scheme of construction of complementation vectors for wild‐type FGSG_10089 (FGSG_10089) and the two cysteine mutants, FGSG_10089 _C→S and FGSG_10089 _C→F, in F. graminearum. A 1‐kb promoter region (Pro) was PCR amplified with the primer set P13/P20. CDSs from the wild‐type and two cysteine mutants with an overhang at the 5′ end were PCR amplified with the primer set P19/P14. A fusion PCR product that contained both the promoter and CDSs was amplified with the primer set P13/P14, cloned into the pRF‐GU vector and transformed into the mutant ΔFGSG_10089 F. graminearum strain. (D) Characterization of deletion, complementation and constitutive expression of FGSG_10089 F. graminearum strains. Genomic DNA was isolated from wild‐type strain (lane 1), a deletion mutant strain (ΔFGSG_10089; lane 2), a strain that constitutively expresses the wild‐type FGSG_10089 in the deletion mutant background (∆FGSG_10089:FGSG_10089 _Oex; lane 3), a strain that constitutively expresses FGSG_10089 with Cys³²⁵ modified to serine in the deletion mutant background (∆FGSG_10089:FGSG_10089 _C→S,Oex; lane 4), a strain that constitutively expresses FGSG_10089 with Cys³²⁵ modified to phenylalanine in the deletion mutant background (∆FGSG_10089:FGSG_10089 _C→F,Oex; lane 5), a strain that complements ∆FGSG_10089 with a native promoter and the FGSG_10089 CDS (∆FGSG_10089:FGSG_10089 (lane 6), a strain that complements ∆FGSG_10089 with a native promoter and the modified cysteine residue to serine in FGSG_10089 (FGSG_10089:FGSG_10089 _C→S; lane 7) and a strain that complements ∆FGSG_10089 with a native promoter and the modified cysteine residue to phenylalanine in FGSG_10089 (∆FGSG_10089:FGSG_10089 _C→F; lane 8). Genomic DNAs from all the strains were screened for the presence/absence of FGSG_10089 with the primer set P7/P8, for Tri6 with the primer set P31/P32 and for the presence of the gpdA promoter and FGSG_10089 CDS with the primer set p41/P8.

Table S1 Proteins identified by a change in Coomassie brilliant blue (CBB) staining in two‐dimensional polyacrylamide gel electrophoresis (2D‐PAGE) between wild‐type (WT) and ΔNoxAB strain at 0 h. Proteins identified by a change in fluorescence signal between WT and ΔNoxAB at 0 h. The UV peptides of the cysteine‐labelled peptides at 0 h. Proteins identified by a change in CBB staining in 2D‐PAGE between WT and ΔNoxAB at 5 h. Proteins identified by a change in fluorescence signal between WT and ΔNoxAB at 5 h. The UV peptides of the cysteine‐labelled peptides at 5 h. Proteins identified by a change in CBB staining in 2D‐PAGE between WT and ΔNoxAB at 10 h. Proteins identified by a change in fluorescence signal between WT and ΔNoxAB at 10 h. The UV peptides of the cysteine‐labelled peptides at 10 h. Proteins identified by a change in CBB staining in 2D‐PAGE between WT and ΔNoxAB at 24 h. Proteins identified by a change in fluorescence signal between WT and ΔNoxAB at 24 h. The UV peptides of the bimane‐labelled peptides at 24 h.

Table S2 Precursor ion intensities of cysteine (Cys) peptides of proteins.

Table S3 Annotation of 99 proteins from the gel‐free approach.

Table S4 The 118 redox proteins identified by the gel‐based and gel‐free approaches.

Table S5 Functional categories of proteins identified by the gel‐based approach.

Table S6 Functional categories of all redox‐sensing proteins at time 0 h. Functional categories of all redox‐sensing proteins at time 5 h. Functional categories of all redox‐sensing proteins at time 10 h. Functional categories of all redox‐sensing proteins at time 24 h.

Table S7 Sequences of primers used in this study.