Metabolomics analysis of grains of wheat infected and noninfected with Tilletia controversa Kühn

Dwarf bunt caused by the pathogen Tilletia controversa Kühn is one of the most serious quarantine diseases of winter wheat. Metabolomics studies provide detailed information about the biochemical changes at the cell and tissue levels of plants. In the present study, a liquid chromatography/mass spectrometry (LC/MS) metabolomics approach was used to investigate the changes in the grain metabolomics of infected and noninfected with T. controversa samples. PCA suggested that T. controversa-infected and noninfected samples were separated during the interaction. LC/MS analysis showed that 62 different metabolites were recorded in the grains, among which a total of 34 metabolites were upregulated and 28 metabolites were downregulated. Prostaglandins (PGs) and 9-hydroxyoctadecadienoic acids (9-HODEs) are fungal toxin-related substances, and their expression significantly increased in T. controversa-infected grains. Additionally, the concentrations of cucurbic acid and octadecatrienoic acid changed significantly after pathogen infection, which play a large role in plant defense. The eight different metabolic pathways activated during T. controversa and wheat plant interactions included phenylalanine metabolism, isoquinoline alkaloid biosynthesis, starch and sucrose metabolism, tyrosine metabolism, sphingolipid metabolism, arginine and proline metabolism, alanine, aspartate, and glutamate metabolism, and tryptophan metabolism. In conclusion, we found differences in the metabolic profiles of wheat grains after T. controversa infection. To our knowledge, this is the first study to evaluate the metabolites in wheat grains after T. controversa infection.

Sample processing. Metabolites were extracted from 50 mg of crushed grains using ultrapure water (Watsons, China). The choice of extraction method is an important factor in any metabolomics study. The grains were crushed by following the method developed in our laboratory. Briefly, two to three grains were dipped into 2 ml Eppendorf tubes containing an appropriate amount of sterilized steel balls (1.5 mm) ground in a grinder machine (FastPrep 24 5G, MP Biomedicals, USA) for 1 min at 70 Hz for crushing. Every sample was ground 3 times for better results. Powdered tissue samples (50 mg) were first mixed with 500 µl of ultrapure water (50% methanol) and sonicated for 10 min. After this time, 50 µl of a homogenate mixture was added to 450 µl of precipitant containing internal standard (50 ng/ml propranolol for positive ion mode and 50 ng/ml tolbutamide for negative ion mode). The samples were vortexed for 1 min and centrifuged at 13,000 rpm for 10 min, and then transferred to a new sterilized Eppendorf tube and concentrated in a speed vacuum (Songyuan, Beijing, China) at 30 ℃ and the dried extract was re-dissolved of 100 µl in 50% methanol (v/v) for LC-MS analysis.
The parameters were as follows. Data processing. All LC-MS data were further filtered by using the R platform loaded with the xcms tool kit, including peak matching, retention time correction, variable integration (integrating the overall contribution of each variable) and data standardization (nontarget metabolite data processing). The raw data were preprocessed by noise reduction, baseline correction, peak alignment, standardization, and scaling and then analyzed by multivariate analysis, including principal component analysis (PCA), orthogonal partial least squares discriminant analysis (OPLS-DA) and hierarchical cluster analysis (HCA) with MetaboAnalyst

Results
Metabolite separated by LC-MS. The spike traits of normal and infected by T. controversa were showed in Fig. S1. The metabolites separated by chromatography continuously entered the mass spectrometer for data collection. The total ion strength was obtained by adding all the ionic strengths in each mass spectrum. The peak area of endogenous substances was obtained for each sample. The chromatograms of the positive and negative ions of wheat grains are shown in Fig. 1. The total ion chromatograms of positive ions (Fig. 1a,c) and negative ions (Fig. 1b,d) of control and infected wheat were obtained by using high-resolution mass spectrometry. The significant differences based on the VIP values and the presented upregulated and downregulated metabolites were shown in Table 1, there were 62 metabolites were got totally.
PCA of T. controversa-infected and control samples. PCA score plot exhibited good clustering between T. controversa-infected and noninfected samples (Fig. 2a,b). The results showed that the distribution of T. controversa-infected and noninfected metabolites was significantly different. The samples of T. controversainfected were near each other, and the noninfected samples were located together, indicating that there was good reproducibility among the biological replicates of the same treatments but differences between the treatments (Fig. 2a), with R2 (0.998) and Q2 (0.997), and the first 2PCs contribution was 77.6% and 4.7%, respectively. The OPLS-DA plot (R2, 0.989; Q2, 0.968) also revealed that the samples of T. controversa were mainly distributed on the right side and noninfected samples were distributed on the left side, indicating that there was good reproducibility among the biological replicates of the same treatments but differences between the treatments (Figs. 2b, S2). A receiver operating characteristic (ROC) curve was generated by using Monte Carlo cross validation (MCCV) under balanced subsampling. In each MCCV, two-thirds of the samples were used to assess the importance of features. Our results showed that the differential metabolites with variable importance in the projection (VIP) values > 1 were successfully detected under the ROC curve (Fig. S3). The list of metabolites and their significant differences based on the VIP values are shown in Table 1. According to the results presented in Table 1 and in Fig. 2b, 62 differential metabolites were found. Out of these 62 metabolites, 34 were upregulated and 28 were downregulated. Out of the 34 upregulated metabolites, 12 were related to amino acids, 10 were organic acids, 3 were alcohols, 1 was a nucleoside and 8 other classes were identified. Similarly, out of 25 downregulated metabolites, 5 were related to amino acids, 10 were organic acids, 2 were fatty acids, 1 was an alcohol, 3 were sugars, 1 was a nucleoside and there were 6 other metabolites. The upregulated and downregulated metabolites are listed in Table 1. Some of the above up-and downregulated metabolites play comprehensive roles in plant pathogen interactions. For example, caffeic acid (upregulated by 457-fold) has a strong ability to scavenge peroxy free radicals after plant pathogen infection 25 . Phenylalanine metabolism (upregulated by 22-fold) will result in an increase in lignin monomers, thereby solidifying the cell wall of plants when exposed to pathogens 26 . Pyro-  www.nature.com/scientificreports/ glutamic acid (upregulated by 14-fold) suppresses certain pathogen interactions 27 . Tetracosanoic acid (upregulated by 33-fold) is also related to antifungal activity 28 .
Screening and statistical analysis of differential metabolites. The correlation between the T. controversa-infected and non-infected samples was calculated to gain a better understanding. Red indicates a positive correlation, while green indicates a negative correlation. The results showed that metabolites in the same branch along the horizontal axis had the strongest positive correlation (Fig. 3a). The HCA results showed that 9-HODE (9) and Prostaglandin D3 (56) were positively correlated, indicating which involved in the pathogenic process of T. controversa. Besides, 9-HODE (9) were negatively correlated with many metabolites, such as malic acid (6), Fumaric acid (33) and Azelaic acid (57). Similarly, Prostaglandin D3 (56) were also negatively correlated with many metabolites, such as malic acid (6), Fumaric acid (33) and Azelaic acid (57). Additionally, the metabolites between infected and control samples were investigated by heatmap analysis, in which the horizontal and vertical axes represent the sample and variable information, respectively. Red indicates that the concentration of metabolites increased, green indicates that the concentration of metabolites decreased, and dark indicates that the change in metabolite concentration increased (Fig. 3b) As we saw from Fig. 3b, 9-HODE (9) and Prostaglandin D3 (56) was increased and malic acid (6), L-Proline (19) and Fumaric acid (33) were decreased in all infected groups, suggesting the invasion of pathogen will affect the normal metabolism of plants. Besides, the upregulated of antifungal substance, like caffeic acid (18), Pyroglutamic acid (24) and Tetracosanoic acid (44), were increased in infected groups, indicating they play a role in responding to the invasion of T. controversa.
Metabolic pathways analysis. By investigating the influence coefficient of the metabolic pathways, the pathways with impact values greater than 0.1 was found to be a potential target pathway. As a result, phenyla-  www.nature.com/scientificreports/ lanine metabolism, isoquinoline alkaloid biosynthesis, starch and sucrose metabolism, tyrosine metabolism, sphingolipid metabolism, arginine and proline metabolism, alanine, aspartate and glutamate metabolism and tryptophan metabolism were closely related pathways in this study ( Table 2, Fig. 4).

Discussion
Wheat is a major staple food crop worldwide. Wheat crops have evolved efficient mechanisms to inhibit the negative effects of pathogen attack 29 . The critical difference between plants that become diseased or remain healthy after pathogen infection is the recognition of elicitor molecules released by pathogens 30,31 . The earliest response in plants after pathogen infection is due to the oxidative burst that can trigger hypersensitive cell death. This is called the hypersensitive response (HR) and is a basic cellular response following successful pathogen recognition in plants and a major element of plant disease resistance 32 . Therefore, any change in the primary and secondary metabolites of plants in response to pathogen infection may thus be a key difference in successful defense against pathogens. Proline is a multifunctional amino acid that confers resistance against plant pathogens 33,34 and abiotic factors 35,36 . Resistant plants increase proline accumulation against pathogen infections and vice versa in susceptible plants 33 . Similarly, cucurbic acid, a compound similar to jasmonic acid (JA), is actively involved in defense mechanisms and tuberization in different crops 37 . Our results showed that the concentrations of proline, hydroxyproline, hydroxymethylproline and cucurbic acid decreased after T. controversa infection (Table 1). Decreasing proline content is a common reaction, and energy demand increases for cross-signaling in plant pathogen interactions. It is also possible that elicitor molecules that are not recognized by susceptible plants cause  www.nature.com/scientificreports/ decreases in proline and cucurbic levels, which can facilitate pathogen infection. Prostaglandin molecules act in a multidimensional way in pathogens and are used as signaling molecules to induce pathogen infection 38 . The concentration of prostaglandin increased during pathogen infection, which resembled our results (Table 1). A previous study revealed that silencing of the triple-ppo (this gene has a role in prostaglandin production) mutant of Aspergillus fumigatus was more virulent than wild-type A. fumigatus in an infected murine test model, and its mutant showed higher resistance than wild-type A. fumigatus against reactive oxygen species (ROS) produced by plants for a strong defense system after A. fumigatus infection 39 . Additionally, prostaglandin has a role in increasing other fungal metabolites, including tyrosine, phenylalanine and other secondary metabolites 40,41 . 9-Hydroxyoctadecaenoic acid (9-HODE) molecules are oxidation products of linoleic acid molecules that play a role in the defense mechanism, increased plant growth and promote aflatoxin B1 in A. parasiticus and sterigmatocystin (ST) from A. nidulans [42][43][44] . 9-HODEs also act as quorum sensing signal molecules to regulate the development and growth of A. ochraceus 45 . Our results showed that during T. controversa and wheat interactions, the concentration of 9-HODE molecules significantly increased, suggesting that T. controversa affects wheat grains. In this experiment, l-phenylalanine was upregulated 1.95-fold (Table 1). In the metabolic pathways activated by T. controversa (Table 2), phenylalanine metabolism affects the formation and deposition of lignin, which will help plants improve immunity to fungal pathogens 26,46 . Phenylalanine ammonia lyase (PAL) is an enzyme related to wheat lignification that participates in phenylalanine metabolism by decomposing l-phenylalanine and provides precursors for lignin biosynthesis. It has been reported that the enzymatic activity of PAL in wheat resistant lines to Karnal Bunt (caused by Tilletia indica) was significantly higher than that of susceptible lines, which is considered to be a marker for identifying KB resistance 46 . We also found changes in starch and sucrose metabolism, which will be influenced when plants are exposed to pathogen infection, and the activation of starch and sucrose metabolism pathways can reduce the toxic substances secreted by pathogens on plants [47][48][49][50] . The activity of isoquinoline alkaloids is due to the response of plants to the invasion of pathogenic fungi, and upregulated expression of tyrosine is closely related to tyrosine metabolism, indicating that the plant signal transduction pathways are active 51,52 , as the tyrosine metabolism pathway is related to the host immune response 53 .
An important and preliminary characteristic of metabolomics data is that it can measure targeted and untargeted analysis and allows the discovery of correlations between pairs of metabolites, even when the biological connection between them is not clear. In our results, for the metabolites from the grains, a small strong correlation was noted. However, for some metabolites, there was a positive or negative correlation in the heatmap analysis (Fig. 3a,b). Additionally, the relative abundances of pathogen-infected and control groups were scattered separately, which may suggest that grain metabolites were infected by T. controversa. For example, metabolites that were decreased or increased after pathogen infection are involved in energy metabolism. These changes in plant metabolites may be the result of T. controversa infection.  24 . 1 phenylalanine metabolism; 2 isoquinoline alkaloid biosynthesis; 3 starch and sucrose metabolism; 4 tyrosine metabolisms; 5 sphingolipid metabolisms; 6 arginine and proline metabolism; 7 alanine, aspartate and glutamate metabolism; and 8 tryptophan metabolisms. A pathway with an impact value greater than 0.1 was considered a potential target.