Mitochondrial phosphate transporter and methyltransferase genes contribute to Fusarium head blight Type II disease resistance and grain development in wheat

Fusarium head blight (FHB) is an economically important disease of wheat that results in yield loss and grain contaminated with fungal mycotoxins that are harmful to human and animal health. Herein we characterised two wheat genes involved in the FHB response in wheat: a wheat mitochondrial phosphate transporter (TaMPT) and a methyltransferase (TaSAM). Wheat has three sub-genomes (A, B, and D) and gene expression studies demonstrated that TaMPT and TaSAM homoeologs were differentially expressed in response to FHB infection and the mycotoxigenic Fusarium virulence factor deoxynivalenol (DON) in FHB resistant wheat cv. CM82036 and susceptible cv. Remus. Virus-induced gene silencing (VIGS) of either TaMPT or TaSAM enhanced the susceptibility of cv. CM82036 to FHB disease, reducing disease spread (Type II disease resistance). VIGS of TaMPT and TaSAM significantly reduced grain number and grain weight. This indicates TaSAM and TaMPT genes also contribute to grain development in wheat and adds to the increasing body of evidence linking FHB resistance genes to grain development. Hence, Fusarium responsive genes TaSAM and TaMPT warrant further study to determine their potential to enhance both disease resistance and grain development in wheat.


Introduction
Fusarium head blight (FHB) is an economically important disease of wheat caused by Fusarium fungi. It reduces yield and contaminates grain with mycotoxins harmful to human and animal health, most commonly deoxynivalenol (DON) [1]. DON is also a virulence factor, facilitating the spread of Fusarium within wheat heads [2,3]. Many components of FHB resistance have been described, the most common being resistance to initial infection (Type I resistance) and resistance to disease spread (Type II resistance) and studies have shown that resistance to the  [3]). Several studies have shown that specific processes and pathways are activated in wheat in response to . UDP-glycosyltransferases (UGTs) gene involved in DON detoxification pathway have been shown to convert DON to less toxic DON-3-O-glucoside and overexpression of wheat UGT (TaUGT3) and barley UGT (HvUGT13248) increased DON tolerance in transgenic plants [4,8]. Genes involved in the classic detoxification pathway (drug transporters and cytochrome P450) have been shown to contribute to DON resistance in wheat, as well as the evolutionary divergent orphan gene TaFROG and the wheat sucrose non-fermenting-1 (SNF1)-related protein kinase 1 catalytic subunit α (SnRK1α) [9][10][11][12]. The diversity of pathways activated in response to DON in wheat [13] and barley [14] reaffirm the important role of this toxin in facilitating disease development.
Wheat genes encoding a mitochondrial phosphate transporter (TaMPT) and S-adenosyl methionine (SAM)-dependent methyltransferase (TaSAM) were identified as being responsive to DON based on a microarray analysis and were differentially expressed in cv. CM82036 x cv. Remus double haploid lines segregating for the FHB resistance QTL Fhb1 [13]. But the role of these genes and their associated pathways in the wheat response to DON or to FHB is unknown. MPTs belong to the phosphate transporter 3 (PHT3) gene family. They are located in the inner membrane of mitochondria and are responsible for transporting inorganic phosphate (Pi) into the mitochondrial matrix, wherein the Pi is utilised for the oxidative phosphorylation of ADP to ATP [15][16][17]. MPT genes have been identified and characterised in many plant species, but studies on wheat MPT genes are very limited [18]. There is one report of the down-regulation of a MPT in wheat heads (resistant to FHB) in response to F. graminearum [19]. Other pathosystems also provide evidence for the involvement of MPTs in wheat disease responses. Yu et al. [20] identified two wheat MPT genes responsive to wheat stripe rust (Puccinia striiformis). Using microarray analysis, Xin et al. [21] showed that a MPT gene was differentially expressed in wheat in response to the causal agent of powdery mildew disease, Blumeria graminis f. sp. tritici, with higher expression in a resistant wheat line than in a susceptible line. Recently, genome-wide association studies (GWAS) and fine-mapping studies identified a MPT gene that co-segregated with the Pch1 locus that confers resistance to eyespot disease caused by Pseudocercosporella herpotrichoides [22]. SAM-dependent methyltransferases enzymes catalyse the transfer of methyl groups from SAM to a large variety of acceptor substrates, ranging from small metabolites to bio-macromolecules [23]. These enzymes contain a cofactor (SAM) binding site and a substrate binding site and share little sequence identity [24]. Several studies have reported the responsive of wheat methyltransferase genes to F. graminearum. Gunnaiah et al. [25] demonstrated that the phenylpropanoid pathway genes encoding caffeic acid-O-methyltransferase, caffeoyl-CoA-Omethyltransferase, and flavonoid-O-methyltransferase were up-regulated in resistant wheat near-isogenic lines containing the FHB resistance QTL Fhb1. Schweiger et al. [26] finemapped and sequenced a 1Mb contig containing the Fhb1 region from the FHB resistant cv. CM82036 and identified 28 candidate genes including a methyltransferase domain containing protein. However, the methyltransferase gene is unlikely an exclusive determinant of Fhb1 resistance, since a deletion mutation in a histidine-rich calcium binding protein has been shown to confer Fhb1 resistance [27]. Cho et al. [28] showed that a methyltransferase gene was differentially expressed in both the FHB resistant cultivar (cv.) Dahongmil and the susceptible cv. Urimil after inoculation with F. graminearum. Long et al. [29] found that a SAM-methyltransferase domain-encoding gene was one of eight candidates whose expression correlated with the FHB resistance QTL on chromosome 2D. Recently, AlTaweel et al. [30] highlighted a methyltransferase gene that was up-regulated in the presence of F. graminearum infection in FHB resistance cv. Sumai 3 and susceptible cv. Caledonia.
Herein, we characterised the mycotoxin-responsive TaSAM and the TaMPT genes first identified by Brennan et al. [13]. Homoeologous genes were identified on all wheat subgenomes and gene expression studies were conducted to determine their responsiveness to DON and DON-producing F. graminearum in the FHB resistant cv. CM82036 and susceptible wheat cv. Remus. Using VIGS, we determined the contribution of both TaSAM and the TaMPT genes to FHB disease resistance in wheat. Furthermore, the VIGS experiment assessed the contribution of TaSAM and TaMPT genes to grain development in wheat.

Adult plant DON and FHB time course experiment
An adult plant experiment was conducted to analyse the temporal response of TaMPT and TaSAM genes homoeologs to both DON and FHB disease in the wheat cvs. CM82036 and Remus. At mid-anthesis (growth stage (GS) 65) [36], two central spikelets of heads from secondary tillers (and of similar size/number of spikelets) were inoculated with 20μl (40μl per head) of either deoxynivalenol (DON) (Santa Cruz, Texas, USA) (5mg ml -1 in 0.02% Tween-20) or 10 6 conidia of F. graminearum strain GZ3639 [37] or 0.02% Tween-20 (mock treatment). After treatment, the heads were covered with a plastic bag for 48 hours to maintain high humidity. Treated spikelets were harvested at either 0, 12, 24, 48, 72, or 96 hours postinoculation (hpi), flash-frozen in liquid nitrogen (N 2 ) and stored at -70˚C prior to RNA extraction. The experiment comprised three replicate trials (each conducted independently at different times), each including eight heads from four individual plants (two heads per plant) per treatment combination (therefore, across the independent trials, there was a total of 12 plants/ 24 heads per treatment combination). See S1 Table for

DNA, RNA extraction and cDNA synthesis
DNA was extracted from wheat leaves using the HP plant DNA mini kit (OMEGA) following the manufacturer's instructions. RNA was extracted from wheat heads as previously described [38] and was DNase-treated using the TURBO DNA-freeTM kit (Ambion Inc., USA). The quality, yield and integrity of the RNA was analysed using both the ND-1000 spectrophotometer (NanoDrop, Thermo Fisher Scientific, USA) and electrophoresis. Reverse transcription of total RNA and the quality check of synthesized cDNA for DNA contamination was conducted as previously described [13].

Cloning of TaSAM-D and TaMPT-A genes and promoters
Homoeolog-specific primers for the TaSAM from chromosome 2D (hereafter referred to as TaSAM-D) and TaMPT from chromosome 5A (hereafter referred to as TaMPT-A) genes and promoters were designed based on the cv. Chinese spring wheat reference sequence (IWGSC RefSeq V1.1) using genome-specific primers (GSPs), a web-based platform for designing genome-specific primers in polyploids [39]. For TaMPT-A promoter and gene cloning, primers were designed to amplify around 1500 bp upstream of the start codon and another set to amplify the coding region of the gene homoeologs (S2 Table). TaSAM-D primers was designed to amplify both the promoter and gene together as one product (S2 Table). Due to high sequence similarities between the 2A and 2B homoeologs of TaSAM gene it was not possible to design homoeolog-specific primers for these two genes. Homoeolog-specific primers for TaSAM-D gene and promoter were used to amplify targets from DNA of wheat cvs. CM82036 and Remus. PCR reactions contained 100 ng genomic DNA, 1.25 U of Takara Ex Taq™, 1X Ex Taq buffer (Mg 2+ plus) and 2.5 mM of each dNTP in 50 μl reaction. Reaction conditions were as follows: 94˚C for 5 min, 35 cycles of 94˚C for 30 s, 55˚C for 30 s, 68˚C for 2 minutes and a final extension step at 72˚C for 10 min and conducted in a ProFlex PCR System (Applied Biosystems by Life Technologies, USA). The amplified PCR products were cloned into the pCR 1 -XL-TOPO 1 vector using the TOPO 1 XL cloning kit (Invitrogen, UK) and sequenced using M13 forward and reverse primers. Results were validated for at least two independent PCR amplicons per target sequence. Sequences were aligned to IWGSC wheat genome database (https://wheat-urgi.versailles.inra.fr/Seq-Repository/BLAST) via BLASTn analysis.

Sequence and phylogenetic analysis
TaMPT-A and TaSAM-D genes cloned from cvs. CM82036 and Remus were used to identify their homoeologs in wheat cv. Chinese spring via BLASTn analysis against Ensembl Genomes (http://plants.ensembl.org) and the wheat genome (IWGSC Refseq V1.1). Multiple sequence alignments of TaMPT-A and TaSAM-D genes and their homoeologs was constructed using MultAlin (http://multalin.toulouse.inra.fr/). TaMPT-A and TaSAM-D sequences from cvs. CM82036 and Remus were used to extract homologous sequences from other Poaceae via BLASTp against the EnsemblPlants database (http://plants.ensembl.org) [40]. Phylogenetic relationships of TaMPT and TaSAM homologs were deduced using the Neighbour-joining method with bootstrapping (1000 replicates) using the MEGA 7.0 program [41]. The sub-cellular localisation of the TaSAM and TaMPT genes was predicted using Multiloc2 [42]. For domain analysis, the amino acid sequence of TaMPT and TaSAM were used for BLASTp against the EnsemblPlants database and were further scanned using the TMHMMv 2.0 and the Interpro protein database [43].

Virus-induced gene silencing (VIGS) constructs
The barley stripe mosaic virus (BSMV)-derived VIGS vectors used in this study consisted of the wild type BSMV ND18 α, β and γ tripartite genome [44,45]. For transient gene silencing, two independent, non-overlapping fragments were amplified for each gene: 194 and 177 bp for TaSAM and 171 and 145 bp for TaMPT (see S2 Table for PCR primers). The fragments were amplified from cv. CM82036 cDNA. Gene silencing fragments were selected to target all the homoeologs of either TaMPT (chromosome 5A, 5B and 5D genes) or TaSAM (chromosome 2A, 2B and 2D genes) (S3 Fig). The homology and gene silencing specificity of fragments for TaMPT homoeologs and TaSAM homoeologs was assessed via BLASTn analysis against the wheat genome and qRT-PCR using homoeolog-specific primers (S8 Table). For cloning the VIGS fragments, PCR reactions were performed with 30ng plasmid DNA, 1μM each of forward and reverse fragment-specific primers (S2 Table) in a 10μl reaction containing 0.5U Taq DNA polymerase and 1x PCR buffer (Invitrogen, UK), 1.5mM MgCl2, and 125μM of each dNTP. PCR reactions were conducted in a ProFlex PCR System (Applied Biosystems by Life Technologies, USA) and the PCR program consisted of an initial denaturation step at 94˚C for 2 min, 35 cycles of 94˚C for 30 s and 60˚C for 30 s and a final extension step at 72˚C for 5 min. The amplified silencing fragments were cloned into the pGEM-T vector (pGEM-T Easy cloning kit; Promega, UK). The pGEM-T vectors carrying the silencing fragments were digested with NotI and PacI sites (New England Biolabs, MA, USA), thereby generating NotI and PacI ends in DNA fragment. Inserts was purified by gel extraction and then the fragment was subsequently ligated to NotI-digested γ RNA vector pSL038-1 [45]. Plasmids containing the silencing fragments were sequenced by Macrogen (korea) using the vector-specific pGamma-F/R primers (S2 Table). A BSMV γ RNA vector carrying 185 bp fragment of barley phytoene desaturase (PDS) gene served as positive control [45]. The plasmids BSMV α and γ genome as well as γ RNA constructs with silencing fragments for PDS, TaSAM and TaMPT were linearized with Mlu1 restriction enzyme whereas BSMV β was linearized using the Spe1 enzyme. Linearized plasmids were converted into capped in vitro transcripts using mMessage Machine™ T7 in-vitro transcription kit (Ambion, Inc., Austin, TX, USA), following the manufacturer's protocol. Once cloned into BSMV, these constructs were respectively named BSMV:SAM1 and BSMV:SAM2 for TaSAM and BSMV:MPT1 and BSMV:MPT2 for TaMPT.

Virus-induced gene silencing of TaMPT and TaSAM genes
FHB resistant wheat cv. CM82036 plants were used for the VIGS experiment. The experiment consisted of three randomly designed independent trials (conducted at different times); each trial included 20 heads per treatment combination (10 plants, two heads per plant) (i.e. across the three independent trials there was a total of 30 plants/60 heads per treatment combination). See S1 Table for experimental design. Plants were grown as described above. At growth stage 47 [36] just before the emergence of the first wheat head, the flag leaves of secondary tillers (of similar size) were rub-inoculated with VIGS buffer FES or buffer containing a 1:1:1 mixture of the in vitro transcripts of BSMV α, β and γ RNA (BSMV:00) or derivatives γ RNA that contained plant fragments (BSMV:PDS, BSMV:SAM1, BSMV:SAM2, BSMV:MPT1, or BSMV:MPT2) [45]. At mid-anthesis (growth stage 65) [36], the two central spikelets of heads on VIGS-treated tillers were treated with either 10 6 conidia of F. graminearum strain GZ3639 or 0.02% Tween-20 (mock treatment). Treated heads were covered with plastic bags for 2 days to maintain high humidity. The third spikelet above the treated spikelet was harvested 24h post-treatment, flash frozen in liquid N 2 and stored at -70˚C prior to RNA extraction (for gene expression studies). RNA was extracted from individual spikelets and equivalent amounts of RNA for the four treated heads per pot were bulked to give a total of five bulk RNA samples per treatment combination per trial (i.e. a total of 15 bulked RNA samples per treatment combination across all three trials). For all treated heads, the number of diseased (discoloured and necrotic) spikelets (including treated spikelets) was assessed at 14 and 21 days post-inoculation (dpi). At growth stage (GS 90), the treated heads were harvested and the average dry grain weight and number of seeds per head was determined.

Quantitative Reverse Transcriptase PCR analysis
Quantitative real-time PCR (qRT-PCR) analyses were conducted using the Stratagene Mx3000TM Real Time PCR (Stratagene, Germany). Homoeolog-specific PCR primers used in this study (S2 Table) were designed using genome-specific primer (GSP) [46]. The specificity of the primers targeting the TaMPT and TaSAM homoeologs was checked via PCR of DNA extracts from nullisomic-tetrasomic lines of cv. Chinese Spring (obtained from Germplasm Resources Unit, JIC, Norwich http://www.jic.ac.uk/germplasm/). An off-target TaMPT gene variant located on chromosome 2 was also analysed to confirm VIGS specificity (see S2 Table  for primers; this gene was chosen because it was next closest homolog and matches 20bp sequence identity with construct BSMV:MPT2). The efficiency of the qRT-PCR primers was checked via qPCR of a dilution series of samples. Each reaction contained 1.25 μl of a 1:5 (v/v) dilution of cDNA (1000-1.6 ng/μl), 0.2 μM of each primer and 1X SYBR 1 Premix Ex Taq™ (Tli RNase H plus, RR420A, Takara) in a total reaction volume of 12.5 μl. PCR conditions were: 1 cycle of 1 min at 95˚C; 40 cycles of 5 s at 95˚C and 20 s at 60˚C; and a final cycle of 1 min at 95˚C, 30 s at 55˚C and 30 s at 95˚C for the dissociation curve. The threshold cycle (CT) values obtained were used to calculate the standard curve. The specificity of PCR amplification was confirmed by the presence of a single peak in melting temperature curve analysis of amplified fragments. Housekeeping genes used for wheat gene expression studies were α-tubulin (GenBank No. U76558.1) [47], Yellow-leaf specific gene 8 (YLS8, TraesCS1D02G332500) and protein phosphatase 2A subunit A3 (TaPP2AA3, TraesCS5B02G165200). These genes were verified not to be differentially expressed in the experiments or in publicly available RNA-seq studies for FHB experiments (results not shown). CT values obtained by real-time RT-PCR were used to calculate the relative gene expression using the formula 2^-(CT target gene − CT housekeeping gene) as described previously [48]. For validation of virus-induced gene silencing, the same qRT-PCR conditions and homoeolog-specific primers were used, (the primers did not overlap with the VIGS construct sequences).

Statistical analysis
All data were analysed using MINITAB 16 (Inc, 2010) (Minitab Ltd., Coventry, UK). Non-normally distributed data sets were transformed to fit a normal distribution using the Johnson transformation (Ryan & Joiner, 1976) and the statistical significance of difference was analysed using one-way analysis of variance incorporating Tukey's test (P = 0.05). The data which could not be transformed using the Johnson transformation (Ryan & Joiner, 1976) was analysed using the non-parametric Mann-Whitney test in MINITAB.

Analysis of TaMPT and TaSAM sequences and phylogeny
Previous studies within our laboratory identified novel TaMPT and TaSAM genes that were responsive to the toxigenic Fusarium mycotoxin DON (Walter et al., 2008). Based on homology, we deduced that the genes of interest were on chromosomes 5A and 2D, respectively, with homoeologs on the other wheat sub-genomes. The TaMPT gene on wheat chromosome 5A and TaSAM on 2D were cloned and sequenced from wheat cvs. CM82036 and Remus; sequences were then compared with homoeologs from the sequenced genome of cv. Chinese spring (IWGSC Ref seq v1.1). The TaMPT-A DNA sequence from cvs. CM82036 and Remus showed 100% identity with a sequence from cv. Chinese spring (CS) (S3 and S4 Tables). Two wheat homoeologs of TaMPT-A were identified by BLASTn analysis and are located on cv. Chinese spring chromosomes 5B and 5D, and are hereafter respectively referred to as TaMPT-B and TaMPT-D. Both homoeologs shared a high homology (> 98% nucleotide and amino acid identity) with the TaMPT-A (98.1 and 98.2% respectively; S3 and S4 Tables). In silico analysis of protein sequences predicted that all TaMPT homoeologs contain the mitochondrial carrier domain (IPR018108, S1 Fig and S5 Table). Multiloc2 predicted that TaMPT homoeologs have a high probability of localising to the mitochondria (S5 Table). The DNA sequence of TaSAM-D cloned from cvs. CM82036 and Remus showed 100% identity with a sequence on chromosome 2D of wheat cv. Chinese spring (S6 Table). Two homoeologs of TaSAM-D was located on cv. Chinese spring chromosomes 2A and 2B, and are hereafter referred to as TaSAM-A and TaSAM-B. Both share > 96% nucleotide and amino acid homology with TaSAM-D (S6 and S7 Tables). In silico sequence analysis predicted that all TaSAM homoeologs contain S-adenosyl-L-methionine-dependent methyltransferase and methyltransferase type 11 domains (IPR013216, S2 Fig and S8 Table). Multiloc2 predicted that TaSAM homoeologs have a high probability of localising to the cytoplasmic region (S8 Table). Phylogenetic analysis showed that wheat TaMPT-A and TaSAM-D proteins represent conserved gene families and these variants cluster with proteins from Poaceae plants (Fig 1A and 1B).

TaMPT and TaSAM homoeologs are up-regulated in wheat heads in response to both DON and FHB disease
Gene expression studies analysed the response of TaMPT and TaSAM gene homoeologs to both DON and FHB in two wheat cultivars: one that is FHB and DON-resistant (cv CM82036), and one that is susceptible to both the toxin and the pathogen (cv. Remus). Transcripts of all homoeologous genes of TaMPT (5A, 5B, 5D) and TaSAM (2A, 2B, and 2D) were detected in heads of both cultivars.
At the earlier time points, wheat TaMPT genes were generally activated by both DON and FHB in both genotypes, but at later time points, they were activated to a greater extent in the susceptible cv. Remus (Fig 2). Also of note is that all homoeologs were not responsive to FHB in cv. CM82036 until 24h, unlike cv. Remus where they were FHB responsive at 12 hpi. The basal expression of 5A, 5B and 5D homoeologs of TaMPT in control (mock) treated tissue was near the detectable limit in both cultivars. In cv. CM82036, DON induction of all the TaMPT homoeologs (5A, 5B, and 5D) peaked at 12-24hpi and diminished thereafter, whereas the DON induction of the three variants of the gene in cv. Remus was more constant and generally increased over time, up to 96 hpi (Fig 2A-2C). Peak expression in DON-treated samples was 2-16-fold higher in cv. Remus (at 96hpi) than in cv. CM82036 (at 12 hpi). In terms of their response to F. graminearum expression of all three variants peaked in cv. CM82036 by 48hpi, with expression slightly, but not significantly (P > 0.05), declining thereafter (Fig 2D-2F). In cv. Remus, the expression in response to the pathogen of the 5A variant peaked at 72 hpi whereas that of the 5B and 5D variants peaked at 96 hpi. Peak expression in pathogen-treated tissue was 3-7.5-fold higher in FHB-treated cv. Remus (at 96hpi) as compared to CM82036 (at 48hpi). While all TaMPT homoeologs (5A, 5B, and 5D) showed a similar expression pattern in response to DON and wild type F. graminearum (Fig 2), in both wheat cultivars TaMPT-A was more responsive than the other variants to both the pathogen and DON treatment. In the DON-treated samples, at peak expression (12 hpi in cv. CM82036 and 96 hpi in cv. Remus), TaMPT-A transcript levels were 1.5-6-fold higher than those of either 5B or 5D variants (Fig  2A-2C). Similarly, in F. graminearum-treated samples, peak expression of the 5A variant in cvs. CM82036 (48 hpi) and Remus (96 hpi) was 2-9-fold higher than that of the other two variants (Fig 2D-2F).
TaSAM homoeologs were activated earlier in the resistant cv. CM82036 in response to DON, as compared to the susceptible cv. Remus, but the opposite was true for Fusarium induction of gene expression. All homoeologs of TaSAM (2A, 2B, and 2D) were induced as an early response to DON in both genotypes, peaking at 12-24h in both cultivars, decreasing  [50] or Tween-20 (mock treatment). harvested at 0, 12, 24, 48, 72, and 96 hours post inoculation, all as previously described [9]. Gene expression was quantified relative to wheat α-tubulin, YLS8 and TaPP2AA3 housekeeping genes (average of [2^-(CT target-CT α-tubulin) ], [2^-(CT target-CT YLS8) ] and [2^-(CT target-CT PP2AA3) ] and is presented on a log 10 scale on y-axis. Results represent the mean from three biological replicates (n = 6; each sample representing an RNA sample pooled from 8 heads) and bars indicate SEM. Columns with the same letter are, statistically, not significantly different (P < 0.05).
https://doi.org/10.1371/journal.pone.0258726.g002 thereafter in cv. CM82036 but not in cv. Remus (Fig 3A-3C). Peak expression of all three variants in DON-treated samples was 1.5-2-fold higher in cv. Remus (at 96 hpi) than in cv. CM82036 (at 12 hpi). Expression of all three variants was induced by the pathogen at 12 hpi in cv. Remus but not until 24 hpi in cv. CM82036 (Fig 3D-3F). Peak expression in pathogentreated samples of cv. Remus (at 12 hpi) was 2-6-fold higher than in cv. CM82036 (at 48hpi). TaSAM-D was the homoeolog most responsive to the pathogen and DON treatment in both wheat cultivars. Peak expression of the 2D variant in DON-treated samples of cvs. CM82036 (12 hpi) and Remus (96 hpi) was 3-5.5-fold higher than that of either the 2A or 2B homoeologs (Fig 3A-3C). Similarly, in the Fusarium-treated samples, peak expression of the 2D variant in cvs CM82036 (48 hpi) and Remus (12 hpi) was 1.7-5-fold higher than that of either the 2A or 2B variants (Fig 3D-3F).

TaMPT and TaSAM genes contribute to Type II FHB resistance in wheat
A VIGS experiment was conducted to determine if TaMPT and TaSAM genes contribute to Type II FHB resistance in wheat cv. CM82036 (resistance to disease spread from centrally inoculated spikelets). Silencing was independently achieved using two constructs for TaMPT (BSMV:MPT1 and BSMV:MPT2) and for TaSAM (BSMV:SAM1 and BSMV:SAM2). The constructs specifically targeted all wheat homoeologs located on chromosome 5A, 5B and 5D for TaMPT (S3A Fig and S9 Table) and 2A, 2B, and 2D for TaSAM (S3B Fig and S9 Table). For the chromosome 5D homoeolog (TaMPT-D) the expression was below detectable limits in both control (BSMV:00) and VIGS treated samples (result not shown). In the absence of gene silencing (FES buffer treatment or empty virus BSMV:00 treatment), TaMPT-A and TaMPT-B were significantly upregulated in response to the fungal pathogen F. graminearum (P < 0.05;  For TaSAM, in the absence of gene silencing (FES buffer treatment or empty virus BSMV:00 treatment), all three homoeologs were significantly upregulated in response to Fusarium (P < 0.05; Fig 5A-5C). In the absence of Fusarium treatment, neither BSMV:SAM1 nor BSMV:SAM2 reduced basal TaSAM transcript levels, relative to BSMV:00. But, in the presence of Fusarium, VIGS reduced the expression of all three homoeologs by 55-89%, as

TaMPT and TaSAM genes positively influence grain number
For plants treated with FES (the VIGS buffer), point inoculation of spikelets with F. graminearum resulted in a small (8%, relative to Tween-20) but insignificant (P > 0.05) reduction in grain number, and it did not affect average grain weight (Figs 6 and 7). The effects of Fusarium on grain number were greater for plants in which the flag leaf was treated with empty virus BSMV:00 (P < 0.05; 29% relative to Tween-20), suggesting that the virus exacerbated the effects of FHB on grain development. For both TaMPT and TaSAM, the effect of VIGS on grain development were independent of Fusarium treatment, as similar effects were observed in both the mock and FHB-treated tissue (Figs 6 and 7). Silencing of TaMPT using BSMV: MPT1 in mock and Fusarium-inoculated heads resulted in respective reductions of 28 and 24% in grain number. The second construct (BSMV:MPT2) had similar effects to BSMV: MPT1 on grain number (28 and 27% reductions for mock and Fusarium-treated heads, relative to BSMV:00; P < 0.05) (Fig 6A). In terms of grain weight, BSMV:MPT1 treatment results in a 27% reduction in grain weight in mock and Fusarium treated samples (P < 0.05), as compared to BSMV:00 treatment (Fig 6B). But construct BSMV:MPT2 treatment did not significantly reduce grain weight in mock and Fusarium-treated heads (P > 0.05).
VIGS of TaSAM using BSMV:SAM1 in mock and Fusarium-inoculated heads resulted in respective reductions of 21 and 29% in seed number (P < 0.05), as compared to the BSMV:00 treatment (Fig 7A). The second construct (BSMV:SAM2) had similar effects to BSMV:SAM1 on grain number, resulting in a reduction in total grain number of 20 and 26% in mock and Fusarium-inoculated heads, respectively (P < 0.05). For grain weight, in Fusarium-inoculated heads construct BSMV:SAM1 treatment results in a 34% reduction, as compared to BSMV:00 treatment (P < 0.05; Fig 7B). In mock-treated samples, the 10% reduction in grain weight in BSMV:SAM1 versus BSMV:00 treated samples was not statistically significant (P > 0.05). The second construct (BSMV:SAM2) did not significantly reduce grain weight in mock inoculated heads but resulted in a 29% reduction of grain weight in Fusarium-inoculated heads, relative to BSMV:00 (P < 0.05) (Fig 7B).

Discussion
Resistance to FHB is complex and governed by several QTL on wheat chromosomes, indicating that multiple genes affect the resistance [51]. This study highlighted TaMPT and TaSAM as positive contributors to the wheat defense response against DON mycotoxin and FHB, inhibiting the spread of the disease. In comparing with previous studies and by searching the physical position of QTL-associated markers [52,53] in the cv. Chinese Spring wheat genome (IWGSC v.1.1), we deduced that the physical position of TaSAM-D gene is outside (150 kb distal to) the FHB QTL on chromosome 2DS, suggesting that TaSAM gene is not directly associated with 2DS QTL [54]. The TaMPT-A gene was located within the FHB 5A (Qfhs.ifa-5A) QTL interval [32]. But in our study, we showed that TaMPT contributed to resistance to disease spread rather than the Type I resistance associated with the 5A QTL (resistance to infection). Furthermore, QTL mapping using a DH population of CM82036x Remus [32,55] deduced that TaMPT-A was not co-localised within the fine-mapped 5A QTL (Qfhs.ifa-5A) region (results not shown), suggesting that TaMPT-A is not the causal gene within 5A QTL region derived from cv. CM82036 [56].
Gene expression studies showed that DON induced expression of both genes earlier in cv. CM82036, but to a greater extent in cv. Remus. But pathogen up-regulation of both TaMPT and TaSAM variants was generally as quick, if not quicker, and greater in the susceptible as compared to the resistant cultivar. This is likely reflective of faster DON accumulation in the susceptible as compared to the resistant cultivar. DON production in wheat spikes typically starts at around 36h after fungal inoculation, and high amounts of DON are accumulated between 36-96 hpi [57,58]. Interestingly FHB response of TaSAM genes peaked as early as 12hpi; hence it is either responsive to very low toxin levels or to other plant/fungal metabolites that form part of the initial response to FHB disease. There is precedence for the early response of methyltransferases to FHB diseases: the accumulation of methyltransferase gene was earlier and higher at 12 hpi in susceptible cv. Calendonia than FHB resistant cv. Sumai 3 [30].
We hypothesized that genomes would not contribute equally in response to FHB and DON, and consistent with this, we found some variation in the relative expression of the homoeologs with higher expression of TaMPT-A and TaSAM-D, as compared to their homoeologous counterparts, in both cultivars. In polyploid genomes, homoeologous genes may contribute in an additive manner or may have different expression patterns, giving rise to expression dominance from one or two sub-genomes [59]. Furthermore, homoeologous genes may alternatively been subjected to sub-functionalization. Nussbaumer et al. [59] identified that D subgenome was more abundant and responsive to F. graminearum than either the A or B subgenome of wheat. RNA-seq studies conducted by Powell et al. [60] identified that B and D homoeologs were more responsive than A homoeolog genes during infection by the fungal pathogen Fusarium pseudograminearum, indicating a homoeolog expression bias in hexaploid wheat.
The TaMPT genes belong to the phosphate transporter 3 (PHT3) gene family localised in the inner mitochondrial membrane [61]. TaMPT is the first MPT3 gene implicated in FHB resistance. This was validated via VIGS in the FHB resistant wheat cultivar CM82036 wherein it reduced disease spread. It also affected grain development in that gene silencing reduced grain number, while effects of grain weight were observed for one of the two silencing constructs. The closest Arabidopsis ortholog (MPT3, AT5G14040; 77.1% homology) was shown to be responsive to both salt and drought stress. Overexpression of AtMPT3 increased plant sensitivity to salt stress compared to wild-type plants, suggesting that ATP-dependent pathway are more activated by high AtMPT3 expression levels under salt stress [62]. Knowledge regarding the biological functions and molecular mechanisms of mitochondrial phosphate transporter in plants is still limited [61]. MPTs, located in the mitochondrial inner membrane, catalyse the transport of phosphate from the cytosol into the mitochondrial matrix where cellular ATP is generated from ADP and phosphate through oxidative phosphorylation [15][16][17]. Stress alters central metabolic pathways, including protein turnover, reactive oxygen species (ROS) production and changes in redox ratios. These metabolic changes enhances the demand for the fast turnover of ADP to ATP cycle that is mediated by respiratory oxidative phosphorylation [63]. Although ROS is an important signalling molecule in diverse biological processes, excessive amounts are toxic [64]. The Fusarium mycotoxin DON stimulates the overproduction of ROS [65]. High amount of H 2 O 2 accumulation leads to programmed cell death (PCD) [65], which could help the infection process of F. graminearum. ATP synthesis in mitochondria and higher amounts of phosphate generation is important for the reduction of excessive amounts of ROS. ATP molecules are exported from the mitochondria to the cytoplasm and help to minimise excessive amounts of ROS production [66,67]. Thus, MPT activity might further delay the oxidative burst and reduce FHB disease symptoms and PCD during the fungal infection process.
Silencing of TaMPT genes in wheat might result in enhanced damage to mitochondrial proteins, which in turn might have facilitated the accumulation of DON and the overproduction of ROS, thus enabling fungal spread in wheat heads. Many studies have shown that the modulation of mitochondrion-associated proteins disturbed mitochondrial functions that affect both plant growth and development [68]. Overexpressing the AtMPT3 gene in Arabidopsis disturbed the cellular redox homeostasis; transgenic plants accumulated excess amount of H 2 O 2 and O 2 levels which lead to PCD and hampered growth and development, suggesting that finely tuned mitochondrion activities are necessary for plant normal growth and development [68]. Tiwari et al. [69] demonstrated that reduced levels of ATP in the mitochondrial cells, due to oxidative stress, resulted in damage to the mitochondrial respiratory chain. It is likely that VIGS of the TaMPT gene leads to mitochondrial dysfunction during FHB treatment in the silenced plants. In this study, the observed phenotypes in silencing plants could be explained by the inability of mitochondria to meet the energy demands (ATP synthesis) in cells as a result of reduced Pi transport into mitochondria, which in turn would reduce the electron flux in the mitochondrial membrane and increase ROS formation, leading to more diseased symptoms in the spikelets of silenced plants. Jia et al. [68] reported the induced expression of genes involved in mitochondrial respiratory chain such as ATP synthase and alternative oxidases (AOX) in Arabidopsis MPT3 overexpressing lines using microarray studies. More research is needed to understand if silencing TaMPT has effect on other genes involved in mitochondrial respiration in wheat. It has been suggested that phosphate homeostasis and energy production plays important roles in grain development [18,70]. Wheat mitochondrial phosphate transporter genes (PHT3) were involved in grain development with high expression of TaPHT3;1 in embryo and rachis, and TaPHT3;2 in aleurone, suggesting its role in phosphate related homeostasis [18]. Recently, Yu et al. [71] showed that overexpressing tomato mitochondrial phosphate transporter gene (SIMPT3;1) in transgenic rice significantly promoted the uptake of phosphate and increased grain yield. In this study, the reduction in the total number of seeds (due to both VIGS silencing constructs) and weight (due to one of the two VIGS silencing constructs) in TaMPTsilenced plants suggests that these genes may have role to play in grain development. Biochemical and genetic studies are needed to decipher the relationship between TaMPT and oxidative phosphorylation, ATP synthesis and ROS accumulation and grain development. This will further help us to understand the role of plant mitochondrial phosphate transporters in FHB disease and DON in wheat and other cereals.
TaSAM are the first methyltransferase genes functionally characterised for their role in resistance to the spread of FHB disease. But, various microarray and gene expression studies have shown the upregulation of methyltransferase genes in response to F. graminearum in wheat [28,30]. Cho et al.
[28] highlighted a SAM-dependent methyltransferase from a microarray study and the gene was differentially expressed in the resistant cv. Dahongmil and the susceptible cv. Urimil after inoculation with F. graminearum. Using microarray expression profiling, Long et al. [29] demonstrated that a SAM-methyltransferase gene was upregulated in response to F. graminearum in a wheat Near isogenic line (NIL) that segregated for a FHB resistance QTL on chromosome 2DL. Recently, AlTaweel et al. [30] found the upregulation of a methyltransferase gene in the presence of F. graminearum infection in the FHB resistance cv. Sumai 3 and the susceptible cv. Caledonia, and they suggested that the methyltransferase may be involved in the response to oxidative stress. In this study, a DON and FHB-responsive TaSAM gene from the SAM-methyltransferase superfamily was shown to contribute to FHB resistance in wheat. The Arabidopsis ortholog (AT2G41380), which shares 53% homology with TaSAM-2D, is responsive to ROS and acts against oxidative stress [72][73][74]. Further experiments are needed to determine the role, if any, of TaSAM gene against oxidative stress in wheat. In rice, knockdown of OsSAMS1, 2 and 3 using RNA interference resulted in late flowering, dwarfism, and reduced fertility in transgenic plants, suggesting a putative role of this gene in histone H3K4me3 and DNA methylation. Moreover, they proposed that SAM deficiency or transport reduces SAM-dependent methyltransferase activities, leading to hypo methylation in plants [75]. Despite the role of methyltransferase in methylation and biosynthesis, to our knowledge, SAM-methyltransferase are rarely reported to be involved in grain yield and development in wheat. Previously, a study has shown that disruption of a methyltransferase gene was associated with reduced grain yield in rice. Hong et al. [76] demonstrated that disruption of the rice methyltransferase gene OsMTS1 resulted in premature leaf senescence, a low rate of photosynthesis, accumulation of ROS and grain yield reduction. The reduced grain number and weight in TaSAM silenced plants in our study suggested that it may have role in wheat yield components.
In conclusion, VIGS studies showed that TaMPT and TaSAM genes positively contributed to FHB resistance in wheat, most likely indirectly rather than through direct effects on the pathogen or DON. The results of VIGS study suggests that both TaMPT and TaSAM genes have potential for enhancing FHB resistance and augmenting grain development in wheat. Thus, TaMPT and TaSAM add to the relatively short list of FHB resistance genes that can be used to engineer crops with improved FHB resistance and yield performance. More detailed investigations of TaMPT and TaSAM genes and their pathways in wheat will extend our understanding of FHB resistance, and the functional roles and mechanisms of TaMPT and TaSAM.