Combined Metabonomic and Quantitative RT-PCR Analyses Revealed Metabolic Reprogramming Associated with Fusarium graminearum Resistance in Transgenic Arabidopsis thaliana

Fusarium head blight disease resulting from Fusarium graminearum (FG) infection causes huge losses in global production of cereals and development of FG-resistant plants is urgently needed. To understand biochemistry mechanisms for FG resistance, here, we have systematically investigated the plant metabolomic phenotypes associated with FG resistance for transgenic Arabidopsis thaliana expressing a class-I chitinase (Chi), a Fusarium-specific recombinant antibody gene (CWP2) and fused Chi-CWP2. Plant disease indices, mycotoxin levels, metabonomic characteristics, and expression levels of several key genes were measured together with their correlations. We found that A. thaliana expressing Chi-CWP2 showed higher FG resistance with much lower disease indices and mycotoxin levels than the wild-type and the plants expressing Chi or CWP2 alone. The combined metabonomic and quantitative RT-PCR analyses revealed that such FG-resistance was closely associated with the promoted biosynthesis of secondary metabolites (phenylpropanoids, alkanoids) and organic osmolytes (proline, betaine, glucose, myo-inositol) together with enhanced TCA cycle and GABA shunt. These suggest that the concurrently enhanced biosyntheses of the shikimate-mediated secondary metabolites and organic osmolytes be an important strategy for A. thaliana to develop and improve FG resistance. These findings provide essential biochemical information related to FG resistance which is important for developing FG-resistant cereals.

Fusarium head blight disease resulting from Fusarium graminearum (FG) infection causes huge losses in global production of cereals and development of FG-resistant plants is urgently needed. To understand biochemistry mechanisms for FG resistance, here, we have systematically investigated the plant metabolomic phenotypes associated with FG resistance for transgenic Arabidopsis thaliana expressing a class-I chitinase (Chi), a Fusarium-specific recombinant antibody gene (CWP2) and fused Chi-CWP2. Plant disease indices, mycotoxin levels, metabonomic characteristics, and expression levels of several key genes were measured together with their correlations. We found that A. thaliana expressing Chi-CWP2 showed higher FG resistance with much lower disease indices and mycotoxin levels than the wild-type and the plants expressing Chi or CWP2 alone. The combined metabonomic and quantitative RT-PCR analyses revealed that such FG-resistance was closely associated with the promoted biosynthesis of secondary metabolites (phenylpropanoids, alkanoids) and organic osmolytes (proline, betaine, glucose, myo-inositol) together with enhanced TCA cycle and GABA shunt. These suggest that the concurrently enhanced biosyntheses of the shikimate-mediated secondary metabolites and organic osmolytes be an important strategy for A. thaliana to develop and improve FG resistance. These findings provide essential biochemical information related to FG resistance which is important for developing FG-resistant cereals.
Keywords: Arabidopsis thaliana, Fusarium graminearum, metabolic responses, mycotoxins, Chi-CWP2, NMR INTRODUCTION Fusarium head blight (FHB) resulting from Fusarium graminearum infection is a serious plant disease causing losses of several millions of tons cereals in worldwide production of wheat, barley and maize (Xu et al., 2005;Xu and Nicholson, 2009;Kazan et al., 2012;Zhang et al., 2013). Such losses include both FHB-caused reductions in crop yields and in usability of cereal grains contaminated by potent trichothecene mycotoxins of FG (Li et al., 2002;Kazan et al., 2012). FG-resistant cultivars is thus urgently required to overcome the problems although natural germplasm resources with Fusarium resistance are unavailable at present (Liu, 2001). Currently, FG control relies largely on the use of chemical fungicides. This often causes undesirable environmental consequences and the development of fungicide-resistant Fusarium populations (D'Mello et al., 2000;Leonard and Bushnell, 2003;Zhang et al., 2009). Therefore, the development of resistant plants using transgenic technology becomes a promising and environmentally friendly way for FHB control (Collinge et al., 2008(Collinge et al., , 2010. Such technology will also be beneficial to reduce or eliminate mycotoxins in cereals (Chen et al., 1999;Anand et al., 2003;Makandar et al., 2006).
In fact, progresses have been made in transgenic plants with FG resistance already. There are many studies aiming to enhance plant resistance against FG using transgenic approaches (Koch et al., 2013;Collinge et al., 2016;Majumdar et al., 2017). For example, expressing dsRNA to silence CYP51 genes of fungi could render susceptible plants highly resistant to FG in Arabidopsis and barley (Koch et al., 2013). In addition, overexpressing the defense response genes α-1-purothionin, thaumatin-like protein 1 tlp-1, and β-1, 3-glucanase (Mackintosh et al., 2007) or expressing the maize b-32 antifungal gene (Balconi et al., 2007) in wheat could enhance resistance to FG. Furthermore, numerous antibodies derived from animals recognizing pathogen-specific antigens have been successfully transformed into plants (Safarnejad et al., 2011) to alter immune regulations and to enhance disease resistance of the resultant plants (Li et al., 2008;Safarnejad et al., 2009;Cervera et al., 2010). Some functional antibodies have also been derived from plants and already employed in commercial production (Schillberg et al., 2001). For instance, some genes encoding antifungal peptides from both plants and animals were transformed into Bobwhite, a model wheat cultivar, to improve FHB resistance (Chen et al., 1999;Anand et al., 2003;Makandar et al., 2006). Transgenic wheat with a gene encoding an antimicrobial peptide RsAFP2 showed improved resistance to fungal pathogens (Jha and Chattoo, 2010;Li et al., 2011).
Recently, the use of a chicken-derived Fusarium-specific recombinant antibody (CWP2) recognizing a surface antigen of F. graminearum attracted attention for developing the FGresistant plants. When transforming CWP2 antibody that was fused to antifungal peptides into A. thaliana and wheat, both transgenic plants showed increased resistance to Fusarium pathogens and mycotoxins (Peschen et al., 2004;Li et al., 2008;Cheng et al., 2015). Such fusion proteins not only enhanced the stability of scFvs (Wörn et al., 2000), but also increased the activity of antimicrobial peptides and inhibitory effects on pathogens even with low level expression (Peschen et al., 2004). Such disease resistance plants improved both crop yields and quality. However, the biochemistry details of plants related to the disease-resistant functions of these fusion proteins remain to be fully understood and plant metabolic reprogramming is probably important.
In this work, we investigated the FG resistance properties of A. thaliana lines expressing a wheat class I chitinase (Chi), a chicken-derived Fusarium-specific recombinant antibody (CWP2), a fusion protein (Chi-CWP2), and the wild type, respectively. We analyzed plant morphological phenotypes for disease indices, production of mycotoxins, metabonomic phenotypes together with the expressions of relevant genes in these A. thaliana plants with and without FG infection. The aim of this study was to reveal the metabolomic details of A. thaliana related to its FG resistance.

Plant Materials
Four lines of Arabidopsis thaliana used in this study included wild-type (Columbia ecotype) and three transgenic lines expressing Chi, CWP2, and Chi-CWP2, all of which were generated via Agrobacterium tumefaciens-mediated transformation (Peschen et al., 2004). All A. thaliana seeds were surface-sterilized with 70% ethanol for 20 min, followed by 3 washes with 95% ethanol for 1 min and air-dried on sterilized filter paper. These seeds were then sown in the sterilized 0.5 × Murashige and Skoog medium containing 3% (w/v) sucrose and 7.5% agar (w/v). After germination, seedlings were transplanted to sterilized soil and grew for 2 weeks with a 16-h photoperiod at 22 • C, 8,000 lux and 75% humidity. Nutrient solution was applied twice a week as previously described (Chen et al., 2000).

Fungus Culture
F. graminearum strain 5035 isolated from a scabby wheat spike in Wuhan, China, is a DON producer and highly pathogenic to wheat (Xu et al., 2010). This Fusarium strain was inoculated on sterilized glass-membrane paper over potato-dextrose agar at 28 • C for 2 days and then cultured in CMC broth at 28 • C (200 rpm) for 5 days. Conidiaspores were collected to prepare a suspension with concentration of 1 × 10 5 spores per milliliter and stored at −20 • C until used in subsequent inoculations.

F. graminearum Inoculation
The flowers of A. thaliana plants (∼6-week-old) were carefully sprayed with a spore suspension containing 1 × 10 5 spores per mL as done previously (Urban et al., 2002). Control plants were inoculated in the same manner but with deionized water instead of the spore suspension. The inoculated plants were kept in a large plastic propagator at 100% relative humidity for 7 days. During the first 2 days, lights were switched off to maintain darkness and turned on the third day. A numerical scoring method was applied to quantify the disease index of A. thaliana plants from flower infection, as previously described (Urban et al., 2002). After inoculation for 4 days, aerial materials of A. thaliana were respectively collected, frozen with liquid nitrogen, ground into powder and then stored at −80 • C. Each sample was divided into two parts for metabonomic analysis and RNA extraction, respectively. Six biological replicates were sampled for each line. Matured seeds were also collected from new and old siliques for observation. The disease index was calculated from aerial mycelia development in flower, new siliques, and old siliques.

Measurements of Mycotoxins
The infected A. thaliana floral tissues from each plant line were harvested on day 7 post inoculation. All samples were stored at −80 • C prior to analysis. The samples were then ground into fine powders with liquid nitrogen and dried in an electric blast drying oven, respectively. About 1 g such powder for each sample was added with 4 mL of acetonitrile-water solution (84:16, v/v) in a flask followed with shaking for 1 h. The resultant extract for each sample was individually obtained after removal of residues through vacuum filtration using a Buchner funnel. The crude extract solution was then purified with Mycosep 227 @ columns. Mycotoxins were collected by eluating the columns with 5 mL of acetonitrile-water solution (84:16, v/v) and then evaporated to dryness. Group B trichothecenes (DON, 3Ac-DON, 15Ac-DON) were derivatized using 100 µL of trimethylsilyl imidazole (TMSI) containing 1% trimethylchlorosilane (TMCS) (v/v) for 15 min. Then 1 mL of isooctane (with 4 mg/L Mirex) was added and the reaction was then quenched with 1 mL of double distilled water. Upper layer was employed for analysis of mycotoxins using gas chromatography equipped with a mass spectrometer (QP2010, SHIMADZU) using a DB-5MS (30 m × 0.25 mm × 0.25 µm) capillary column as previously described . Each sample was analyzed twice with one in full scan mode (m/z 100-600) for the identification and the other in selected ion-monitoring (SIM) mode for quantifying the targeted analytes using known mycotoxin standards. Three independent replicates were analyzed for each plant line.

Extraction of Plant Metabolites
Each freeze-dried plant powder sample (ca. 25 mg) from the aforementioned four plant lines in a microtube was added with 1 mL of pre-cooled 50% aqueous methanol (−40 • C). These samples were then treated with intermittent sonication (i.e., 30 s of sonication with intermittent 30 s break) for 10 min in an ice bath. Following centrifugation (16,000 g, 4 • C) for 10 min, the supernatant was transferred into a new microtube (5 mL); the remaining solid residues were extracted two more times using the same procedure and three resultant supernatants were combined. After removal of methanol under vacuum using a Speed-Vac Concentrator (Thermo SAVANT, SC110A-230), the supernatants were lyophilized in a freeze-drier for 24 h. The dried extracts were re-dissolved into 550 µL of phosphate buffer (0.1 M, pH 7.4) containing 10% D 2 O (v/v) and 0.02 mM TSP. After another centrifugation for 10 min, 500 µL of supernatant for each sample was transferred into a 5 mm NMR tube for NMR analysis. In the experiment, toxins and organic solvents were used in strict accordance with the local biological experimental safety management regulations, waste generated during the experiment were disposed by specialized agencies.

NMR Measurements
All NMR spectra were recorded at 298 K on a Bruker AV III 600 NMR spectrometer with a cryogenic TXI probe (Bruker Biospin, Germany) operating at 600.13 MHz for 1 H. 1 H NMR spectra were acquired using a standard onedimensional NOESY-based pulse sequence (RD-90 • -t 1 -90 • -t m -90 • -acquisition) with water signal suppressed during the recycle delay (RD, 2 s) and mixing time (t m , 100 ms). The 90 • pulse length was about 9.5 µs and t 1 was 3 µs. Sixty-four transients were collected into 32 k data points for each spectrum with a spectral width of 12 kHz. An exponential window function with line broadening factor of 0.5 Hz was applied to all free induction decays (FIDs) prior to Fourier transformation (FT). The chemical shifts were referenced to TSP at δ 0.00. For metabolite signal assignments, a set of 2D NMR (including 1 H-1 H COSY, 1 H-1 H TOCSY, 1 H-JRES, 1 H-13 C HSQC, and 1 H-13 C HMBC) spectra were recorded and processed as previously reported (Teague et al., 2004;Xiao et al., 2008).

Spectral Processing and Data Analysis
After phase-and baseline-corrections using TOPSPIN (v3.1, Bruker Biospin GmbH, Germany), the region δ 0.50-9.50 of all 1 H NMR spectra was divided into bins with a width of 0.003 ppm (1.8 Hz) using the AMIX software (v 3.8.3, Bruker Biospin GmbH, Germany). The region δ 4.40-5.15 was discarded to eliminate the effects of imperfect water pre-saturation. The spectral integrals of all buckets were then normalized to the dry weight of A. thaliana plants to give dataset in the form of signal area (metabolite quantity) per gram dry plant tissue. All NMR spectral bins were then subjected to the Student's test or nonparametric tests where appropriate with MUDA (multiple univariate data analysis) method as previously described (Duan et al., 2013). The results for inter-group differences can be displayed as a differential metabogram plot where p-values are color-coded for all variables with those metabolites having significant inter-group differences (p < 0.05) colored in red (Duan et al., 2013).
Concentration of some metabolites were calculated from the integrals of their NMR signals (non-overlapping ones) against that of an internal reference (TSP) with known concentration ( Table 1). This is done by taking relaxation times (T 1 ) into consideration (Table S1) as previously described (Dai et al., 2010a,b). The resultant data were subjected to statistical analyses (one way-ANOVA) using SAS software (V8, Statistics Analysis System, USA).

Quantitative Real-Time PCR Analysis
Total RNAs were isolated from aerial materials challenged by FG using Trizol reagent (Invitrogen) and treated with DNase I (Thermo) to remove contaminated genomic DNA. The first strand cDNA was prepared using superscript III reversetranscriptase (Invitrogen). qRT-PCR analysis was performed on an iQ5 Cycler (Bio-Rad) under the previous condition  with three biological repeats. Two-tailed t-test (confidence interval, 95%) was performed using GraphPad Prism 5 software for the statistical analysis of the data. Gene-specific primers were used for the qRT-PCRs as listed in Table S2.

Identification of the Insertion Sites of Transgenic A. thaliana Plants
Three pairs of nested primers (Table S3) were designed, based on the left arm sequence of the transformation vector pTRAKC, for use in Tail-PCR amplification according to the previous method (Liu and Chen, 2007). The insertion sites for Chi, CWP2 and Chi-CWP2 were chromosome 1 BAC F9K20, chromosome 5 BAC clone F2I11 and chromosome 2 clone F1O11 map g6825, respectively.

RESULTS
Phenotypes of the F. graminearum Infected A. thaliana FG conidia were sprayed only into flowers to localize the infection, in this study, and all plants were maintained in the same growth chamber after inoculation until maturation. Therefore, FG-induced changes in plant morphological phenotypes were readily observable for flower but not any other tissues. For wild type A. thaliana, aerial mycelia were developed around anther and new siliques on day-3 post inoculation (3-DPI) of F. graminearum conidia. For A. thaliana expressing either Chi or CWP2, aerial mycelia were developed around anther and new siliques on 4-DPI but around flower and older siliques only after 7-DPI. A. thaliana expressing Chi-CWP2, in contrast, had significantly alleviation in such aerial mycelia development even after 7-DPI ( Figure 1A).
On 7-DPI, the disease indices (DI) for flower and old siliques of all three transgenic plants expressing Chi, CWP2, and Chi-CWP2 respectively were significantly lower than that of wild type ( Figure 1B). For new siliques, however, only Chi-CWP2 plants had significantly lower DI values than wild type ( Figure 1B). Harvested ripen seeds from diseased plants were shriveled and dark brown in color whereas the healthy ones were plump and yellow in color ( Figure S1). Compared to these from the uninfected plants, seeds from the FG-inoculated wild-type plants were markedly affected whereas about 50% seeds from Chi or CWP2 plants were affected. In contrast, most of the seeds from Chi-CWP2 plants were healthy after FG inoculation ( Figure S1). All these data indicated that transgenic A. thaliana expressing a single gene Chi or CWP2 showed moderate FG resistance whereas, in contrast, the plant expressing fusion protein Chi-CWP2 displayed remarkably higher FG resistance (Figures 1A,B).

Mycotoxins in Flowers of the F. graminearum Infected A. thaliana
It is well known that F. graminearum (FG) produces type-B trichothecene mycotoxins which typically include deoxynivalenol (DON) and its two derivatives 15-acetyldeoxynivalenol (15Ac-DON) and 3-acetyl-deoxynivalenol (3Ac-DON); 15Ac-DON is much more toxic to plants than DON (Desjardins et al., 1993;Kimura et al., 2007). The levels of these mycotoxins positively associate with severity of FG-caused FHB in wheat (Gautam and Dill-Macky, 2011). Therefore, the levels of these mycotoxins in plant tissues generally have inverse correlation with FG resistance of the given plants. In this study, mycotoxins were measured for the FG-invaded flower tissues on 7-DPI ( Figure 1C). DON levels in Chi and CWP2 plants (about 4.04 and 3.77 nmol per gram dry tissue, respectively) were much lower than in wild type (about 7.23 nmol/g); 15Ac-DON levels in Chi and CWP2 plants (0.18 and 0.37 nmol/g, respectively) were also significantly lower than that in wild type (1.14 nmol/g). In Chi-CWP2 plants, remarkably, DON level (1.18 nmol/g) was <50% of that in the other two transgenic lines;15Ac-DON level in Chi-CWP2 plants was close to detection limit and <10% of that in the other two transgenic lines ( Figure 1C).
After challenging with FG, however, the aforementioned inter-group metabolomic differences became much greater and metabolically more widespread. After FG challenge, Chi plants contained significantly more Phe, malate, and citrate but less GABA than wild type; CWP2 plants had significantly more glucose, myo-inositol, betaine, malate, α-ketoglutarate, citrate, Phe, Pro, and Thr but less GABA and Ala than wild type (Table 2, Figure S3). In contrast, Chi-CWP2 plants differed significantly from the other lines in their metabolic responses toward FG infection (Table S5). This transgenic line contained significantly more glucose, myo-inositol, Pro, Phe, Trp, Tyr, His, Asn, pyruvate, succinate, malate, α-ketoglutarate, citrate, fumarate, betaine, and uridine but less Ala, GABA, and choline than wild type (Table 2). Furthermore, Chi-CWP2 plants had significantly higher levels in glucose, betaine, Phe, Tyr, Trp, His, Arg, pyruvate, malate, α-ketoglutarate, and citrate but lower choline level than Chi plants (Table S5, Figure S3); Chi-CWP2 plants also had significantly higher levels in betaine, Phe, Tyr, Trp, His, pyruvate, malate, α-ketoglutarate, and citrate but lower levels in Ala, Thr, and choline than CWP2 plants (Table S5, Figure  S3). These results were broadly consistent with the quantitative results ( Table 1) though more comprehensive.

DISCUSSION
Transgenic plants expressing anti-Fusarium peptides and proteins have shown Fusarium resistance though natural germplasm has limited resistance. Chi has shown anti-fungal activity and FHB controlling potentials in birch and wheat (Punja, 2001;Pasonen et al., 2004;Shin et al., 2008) probably due to Chi's hydrolytic activity for chitin, which is a main cell wall polysaccharide in fungi but not in plants (Roncero, 2002;Ruiz-Herrera et al., 2002;Kim et al., 2009;Xu et al., 2010). Transgenic wheat expressing CWP2, a Fusarium-specific recombinant antibody, also showed some Fusarium resistance (Peschen et al., 2004;Cheng et al., 2015) whilst the CWP2 antibody fused to Chi showed excellent Fusarium hyphae inhibitory activity in vitro (Peschen et al., 2004). However, the roles of transgene-facilitated plant metabolic reprogramming in fungal resistance development remain to be fully understood. Therefore, this study investigated the FG resistance of A. thaliana expressing Chi, CWP2, and fused Chi-CWP2, in comparison with the wild-type line. Metabonomic responses of A. thaliana toward transgenesis and all these plants toward FG infection were systematically analyzed.
First, results from both disease index and mycotoxin production indicated that Arabidopsis plants expressing Chi, CWP2, and Chi-CWP2 had significantly better FG resistance than the wild-type (Figure 1). Whilst no significant differences in FG resistance were observed between Chi and CWP2 plants, Chi-CWP2 plants displayed remarkably higher resistance against FG than both Chi and CWP2 lines (Figure 1). The infected Arabidopsis flowers contained much more DON than 15Ac-DON (about 10% of total toxins) in all A. thaliana lines. Nevertheless, reduction of 15Ac-DON level is important for FG resistance since 15Ac-DON is more toxic to plants than DON (Kimura et al., 2006). After FG inoculation, noticeably, Chi-CWP2 plants contained significantly less DON than Chi and CWP2 plants with almost no 15Ac-DON in Chi-CWP2 plants ( Figure 1C).
Metabolomic analysis showed that, under normal circumstances, plants expressing Chi and CWP2 had limited impacts to plant metabolism with elevation of fumurate (TCA cycle intermediate) and enhanced biosynthesis of osmolytes (betaine or glucose) compared with wild type (Tables 1, 2). Expression of Chi-CWP2 led to more metabolic alterations in A. thaliana than wild-type which was highlighted by significant elevation of aromatic amino acids (Phe, Trp, and His), citrate and Asn (Tables 1, 2). This probably indicates that expressing Chi-CWP2 mainly promotes the shikimate pathway-mediated secondary metabolism especially biosynthesis of phenylalaninemediated phenylpropanoids and tryptophan-mediated alkanoids in addition to TCA cycle.
When challenged with FG, however, more metabolic differences were observed between three transgenic plant lines and wild type as well as between transgenic plants having different FG resistance (Tables 1, 2, Figure 5). Compared to wild type, Chi transgenic plants having limited FG resistance showed elevation of Phe, malate, and citrate but decline of GABA (Tables 1, 2) suggesting enhanced shikimate-mediated secondary metabolism, TCA cycle, and probably GABA-shunt ( Figure 5). CWP2 plants having moderate FG resistance (slightly better than Chi plants) showed elevation of Phe, malate, αketoglutarate, citrate, betaine, glucose, myo-inositol, and proline but decline of GABA and alanine (Tables 1, 2) indicating enhanced shikimate-mediated secondary metabolism, TCA cycle, biosynthesis of osmolytes, and GABA-shunt (Figure 5). This notion is further supported by moderate up-regulations of BADH for choline-mediated betaine biosynthesis, SSADH for GABA-to-succinate conversion in GABA shunt, OAT for ornithine-mediated proline biosynthesis and SS for alkaloid biosynthesis (Figures 1C, 5).
In contrast, however, plants expressing Chi-CWP2 with strong FG resistance had much more widespread and significant metabolic responses to FG challenge than wild type, Chi and CWP2 plants (Tables 1, 2, Figure 5). FG-induced elevation of phenylalanine, tyrosine and tryptophan (Tables 1, 2) together with up-regulation of TAT, IDO, SS, G6PD, and PAL (Figure 3) indicates promotion of the shikimate-mediated biosynthesis of phenylpropanoids and indole-related alkanoids (Figure 5). Several studies also supported the notion that both PAL and the phenylpropanoid pathway were involved in the resistance of wheat to FG and DON (Paranidharan et al., 2008;Walter et al., 2010). Elevations of succinate, malate, α-ketoglutarate, citrate, and fumarate (Tables 1, 2) together with up-regulations of α-KGDH and IDH (Figure 3) indicate enhancement of TCA cycle (Figure 5). Elevation of glucose, myo-inositol, pyruvate and concurrent decline of alanine suggest promotion of gluconeogenesis to generate glucose and myo-inositol as osmolytes (Liu et al., 2010;Kumar et al., 2016). Significant decline of GABA and up-regulation of SSADH suggest the promotion of GABA shunt to yield succinate feeding into TCA (Figure 5). Significant elevation of betaine, reduction of choline and upregulation of BADH indicate promoted conversion of choline into betaine, which is a strong organic osmolyte (Figure 5). FG-induced elevation of proline and up-regulation of OAT probably suggests enhanced proline biosynthesis since ornithine aminotransferase catalyzes the conversion of ornithine into glutamate-5-semialdehyde and cyclization into 1-pyrroline-5carboxylic acid followed with its reduction into proline. In A. thaliana, nevertheless, ornithine can also be turned into GABA via Arg-mediated putrescine-spermidine pathway, one cannot rule out such possibility here with the up-regulation of OAT especially with no changes in Arg and ornithine. The resultant GABA can be converted into succinate feeding into TCA cycle FIGURE 5 | FG-induced metabolic changes in wild-type (WT) and transgenic A. thaliana expressing Chi, CWP2, and Chi-CWP2 highlighted by changes in metabolites and gene expression levels. Identified metabolites and measured genes were in bold letters. Red and green symbols denoted significant increases and decreases (p < 0.05), respectively. Suc, sucrose; Ala, alanine; Arg, arginine; Asn, asparagine; α-KG, α-ketoglutarate; Cho, choline; Cit, citrate; E4P, erythrose-4-phosphate; Fum, fumarate; GABA, γ-aminobutyrate; G6P, glucose-6-phosphate; Glc, glucose; Glu, glutamate; Icit, isocitrate; Mal, malate; mIno, myo-inositol; OAA, oxalacetate; Orn, ornithine; PEP, phosphoenolpyruvate; 3PG, 3-phosphoglycerate; Pyr, pyruvate; SSA, succinate semialdehyde; Ser, serine; Shik, shikimate; Succ, succinate; Trp, tryptophan; Tyr, tyrosine; Phe, phenylalanine; Uri, uridine.
as discussed earlier and gluconeogenesis (Figure 5). This notion is further supported by elevation of Asn (via oxaloacetate-Asp route) and greatly enhanced ACS regulation probably to convert acetyl-CoA generated from TCA into acetate and then into pyruvate for gluconeogenesis (Figure 5).
The above propositions were further confirmed with strong correlations between these changed metabolites, gene expressions and production of mycotoxins (Figure 4). Inverse correlations between the levels of both mycotoxins and the expression levels of TAT, ACS, IDH, α-KGDH, BADH, SSADH, OAT, PAL, IDO, XS, G6PD, and SS (Figures 4E,F, Figure S4) suggest the above discussed metabolic reprogramming be positively associated with FG resistance. Inverse correlations between the production of mycotoxins and the levels of aromatic amino acids, TCA intermediates, osmolytes but positive correlation with the levels of choline and GABA (Figures 4A-D Figure  S4) suggest positive correlation of FG resistance with promoted biosynthesis of phenylpropanoids, indole-related alkanoids, osymolytes, and GABA shunt. This is conceivable since Phe, Tyr, and Trp are precursors for biosynthesis of phenylpropanoids and tryptophan-related alkanoids which have antibacterial activity in plant resistance to biotic stressors (Abdel-Farid et al., 2009). Biosynthesis of osmolytes such as glucose, myo-inositol, betaine, and proline are important metabolic reprogramming strategy for plant resistance to both abiotic and biotic stressors (Choi et al., 2004;Browne and Brindle, 2007;Paranidharan et al., 2008;Bollina et al., 2010;Dai et al., 2010b;Liu et al., 2010Liu et al., , 2017Zhang et al., 2011;Kumar et al., 2016).

CONCLUSION
Transgenic A. thaliana expressing Chi-CWP2 had much stronger FG resistance than wild-type and transgenic plants expressing Chi and CWP2 alone, both of which showed moderate FG resistance. Such FG resistance was highlighted in both reduced disease indices and production of mycotoxins with Chi-CWP2 plants producing much less mycotoxins after FG inoculation. The combined metabonomic and quantitative RT-PCR analyses revealed that enhanced biosynthesis of phenylpropanoids, tryptophan-related alkaloids, and organic osmolytes played important roles in FHB resistance for these transgenic A. thaliana lines especially Chi-CWP2 plants. Nevertheless, introduction of these genes had only limited metabolic effects on plants under normal circumstances thus limited effects on plant physiological biochemistry. These findings suggest that expressing Chi-CWP2 is an attractive approach for developing transgenic plants with intrinsic FG resistance. Concurrent enhancement of biosynthesis of shikimate-mediated secondary metabolites and organic osmolytes is a potentially important strategy worth exploring in development of the FG resistant plants in sustainable agriculture.

AUTHOR CONTRIBUTIONS
Y-CL and HT: designed the project; FC, CL, and JZ: performed the sampling; FC and CL: measured the contents of mycotoxins and performed quantitative real-Time PCR analysis; FC, CL, and JZ: performed NMR measurements, spectral processing and data analysis; HL and H-PL: involved in the manuscript refinement; FC, CL, Y-CL, and HT: wrote the article.

FUNDING
This project was supported by the National Natural Science Foundation of China (31600240, 31200208, 21175149, and 31272004).