MS-DIAL features were clustered by applying a Pearson correlation, with a minimum correlation of 0.8 and maximum p value of 0.05, retaining two features per cluster according to most intense and the most connected peak filters . Selected peaks were imported into MS-FINDER for annotation . Mass tolerance for MS1 and MS2 were set to 5 and 15 ppm respectively, the relative abundance cut off set at 1% and the formula finder was configured to use C, H, O, N, P, and S atoms. FooDB, PlantCyc, ChEBI, NPA, NANPDB, COCONUT, KNApSAcK, PubChem, and UNPD were used as local databases. During the final merge step in MS-CleanR, the best annotation for each peak was based on MS-FINDER scores. The normalized annotated peaks list produced by MS-CleanR was used for the final statistical analyses in R. A partial least squares supervised model of the complete log-transformed and Pareto-scaled dataset was done using the ropls package , with the three treatment groups as the response variables. Ellipses were drawn around treatments using stat_ellipsebased on a 95% confidence level. This distance type considers the correlation between variables and the ellipses are created around the centroid data point. Heatmaps were created using the log-transformed data within the ComplexHeatmap package in R , with hierarchical clustering according to the complete-linkage method and Euclidean distance measure across columns and rows, displayed as dendrograms.Pairwise multivariate analysis was performed across all time points between ANE and H2O, and between AA and H2O, using an orthogonal projections to latent structures discriminant analysis . OPLS-DA models were generated using the ropls package, macetas plastico cuadradas with the predictive components set to 1 and orthogonal components to 7. S-plots were generated following sample sum normalization and Pareto scaling via calculation of p1 and pcorr1 of the OPLS-DA scores using the muma package source code within R.
Chemical class enrichment analysis was achieved using ChemRICH for each two-treatment comparison at each time point within R using the source code . A student’s t-test of the signal was conducted to generate p values and effect size. Previous transcriptomic work revealed a high level of congruency in differentially expressed genes in AA- and ANE-treated tomato seedlings compared to H2O-treated controls . To further this line of investigation, the metabolomic profiles of AA- and ANEroot-treated plants were compared to H2O after 24-, 48-, 72-, and 96-hours exposure to their respective treatments. Locally-treated roots and distal leaves were harvested, flash frozen, extracted for metabolites, and subsequently analyzed via LC-MS, which with underivatized samples primarily captures the nonvolatile metabolome . Partial least squares score plots of tomato root tissue revealed distinct clustering by treatment irrespective of time point . No overlap was observed in the 95% confidence ellipses for any treatment group. Likewise, heatmap visualization of the log10 signal of metabolites showed clear clustering of metabolomic profiles by treatment . Features displayed in the heat map were filtered from the total dataset with a p-value < 10−6 and absolute fold change > 5 in roots. Less defined clustering was observed in PLS score plots of distal leaf tissue across sampled timepoints . Ellipses representing the 95% confidence interval for both H2O and AA treatments both partially overlap with the ANE treatment group. Similarly, a heatmap depicting metabolite log10 signal showed more diffuse clustering by treatment . Visualized metabolites from leaves displayed in the heatmap used a p-value < 0.001 and an absolute fold-change > 2. These findings are refilective of distal tissue not directly treated with either elicitor. Changes in the distal leaves were not as robust as in the directly treated roots, likely due to diminution of systemic signals that effect metabolic changes throughout the plant. An assessment of the total annotated features across the metabolomic analysis revealed shared and unique annotated features between roots and leaves .
Roots and leaves share 44 features with leaves displaying the largest number of unique metabolic features . There were 330 unique identified features in leaves compared to 223 features unique to root tissue . A comparison of differential metabolic features for AA- and ANE-treated plants compared to the H2O control revealed robust overlap for both roots and leaves . AA- and ANE-treated roots shared 68 differential metabolic features, with AA and ANE treatments possessing 37 and 29 differential features unique to each, respectively . Less overlap was observed in leaves with 39 shared differential metabolites, with AA- and ANE-root treated plants displaying 34 and 19 uniquely differential metabolites, respectively .Chemical enrichment analyses were conducted to identify classes of metabolites whose accumulation was locally or systemically altered in AA- and ANE-root-treated tomato seedlings. Enrichment analyses of metabolites whose mean signal was significantly changed in AA- or ANE-treated plants compared to H2O identified numerous affected chemical classes . These changes were most robust in directly treated roots compared to distal leaves. Treatment of tomato seedlings with AA showed strong modulation of metabolomic features classified as triterpenoids and linoleic acid and derivatives in roots. AA-treated roots also showed increases in hydroxycinnamic acids and derivatives and fatty acyl glycosides of mono- and disaccharides. ANE-treated roots showed modulation in the accumulation of triterpenoids, steroidal glycosides, and hydroxycinnamic acids and derivatives. Similar to AA-treated plants, the roots of plants treated with ANE also showed increases in metabolites classified as fatty acyl glycosides of mono- and di-saccharides. Although less striking than the chemical enrichment analysis of roots, leaf tissue of root treated plants did reveal an altered metabolome . These changes in metabolite accumulation occurred most prominently at 96 hours, the latest tested time point. Increases in sesquiter penoids and steroidal saponins were seen in leaves of AA-treated plants at 96 hours.
A mix of accumulation and suppression of terpenoids and an increase of methoxyphenols was observed in the leaves of ANE-root-treated plants. Chemical enrichment analyses broadly revealed classes of metabolites that were induced or suppressed in AA- or ANE-treated plants. To examine which specific variables provide the strongest discriminatory power between the two treatment groups, a two-group comparative supervised multivariate analysis, orthogonal projections to latent structures discriminant analysis , was utilized. OPLS-DA score plots show strong between group variability discrimination between AA and ANE treatment groups compared to the H2O control across all tested time points with the x-axis describing the inter-treatment variability,macetas 30l and the y-axis showing the intra-treatment variability . S-plots derived from OPLS-DA were examined for both AA and ANE treatments in pairwise comparison with H2O control. S-plots of OPLS-DA revealed that treatment with AA or ANE induced shared changes in the levels of several defense-related metabolites in roots . Variables with the most negative and positive correlation and covariance values are the most influential in the model. These metabolites are located on either tail of the S-plot and contribute most greatly to the separation between treatment groups Bar charts depicting mean LC-MS signals for top OPLS-DA S-plot metabolites visualized across all time points illustrate that AA and ANE have similar effects on plant metabolic response . Treatment of tomato roots with AA and ANE resulted in a sharp increase in metabolic intermediates in ligno-suberin biosynthesis. This includes AA-induced accumulation of moupinamide and significant increases in coniferyl alcohol in the roots of ANE-treated plants across all tested time points. In roots,AA and ANE treatments also induced increased levels of N-ethyl phydroxycinnamide and N1-trans-feruloylagmatine compared to H2O treatment, refilecting strong upregulation of the shikimate pathway and phenolic compound synthesis. Reduced levels of tomatine and dehydrotomatine were observed in the roots of AA- and ANE-treated plants indicating suppression of steroid glycoalkaloid biosynthesis. Treated plants also showed lower levels of lyso-phosphatidyl ethanolamine that could refilect enhanced membrane lipid turnover. AA and ANE can induce disease resistance locally and systemically, alter the accumulation of key phytohormones, and change the transcriptional profile of tomato with a striking level of overlap between the two treatments . The current study examined and characterized the AA- and ANE-induced metabolomes of tomato. AA and ANE locally and systemically induce metabolome remodeling toward defense-associated metabolic features. Early studies investigating transcriptional and metabolic changes in potato revealed selective partitioning and shifting of terpenoid biosynthesis from steroidal glycoalkaloids to sesquiterpenes following treatment with AA or EPA or challenged with P. infestans . Similarly, our work here with AA- and ANE-treated tomato seedlings has shown a marked decrease in the levels of two abundant glycoalkaloids, tomatine and dehydrotomatine . Our data also show strong enrichment of sesquiterpenes in leaves of AA-treated plants at 96 hours post treatment, although the identity of these sesquiterpenes is unresolved .
This work further supports evidence for differential regulation and sub-functionalization of sterol/ glycoalkaloid and sesquiterpene biosynthetic pathways in solanaceous plants in different stress contexts . AA and EPA are strong elicitors that are abundant in structural and storage lipids of oomycete pathogens, but absent from higher plants. Although their initial perception by the plant is likely different from that of canonical MAMPs , there is some convergence in downstream defenses induced by these various MAMPs. Work to characterize the effect of canonical MAMP treatment on the metabolomes of various plant species has implicated common metabolic changes that prime for defense. Cells and leaf tissue of A. thaliana treated with lipopolysaccharide showed enrichment of phenylpropanoid pathway metabolites, including cinnamic acid derivatives and glycosides . In the same study, SA and JA were also positively correlated with LPS treatment, as we also observed in tomato following treatment with AA . Recent work in A. thaliana wild-type and receptor mutants treated with two chemotypes of LPS showed increases in hydroxycinnamic acid and derivatives and enrichment of the associated phenylpropanoid pathway . Work in tobacco similarly found treatment with LPS, chitosan, and flg22 all induced accumulation of hydroxycinnamic acid and derivatives, and that defense responses elicited by these MAMPs were modulated by both SA and JA . More recent work in the cells of Sorghum bicolor treated with LPS showed enrichment of hydroxycinnamic acids and other phenylpropanoids in coordination with accumulation of both SA and JA . Treatment of tomato with flg22 and flgII-28 also enriched hydroxycinnamic acids, and tomato treatment with cps22 revealed a metabolic shift toward the phenylpropanoid pathway with hydroxycinnamic acid, conjugates and derivatives as key biomarkers . Similar to traditional MAMPs, AA and the AA/EPA-containing complex mixture, ANE, both induce enrichment of cinnamic acid and derivatives in tomato seedlings . This supports the hypothesis that MAMPs broadly induce similar metabolic changes to enrich pools of specialized secondary metabolites that contribute to plant immunity. AA- and ANE-treated roots showed strong enrichment of metabolic features classified as fatty acyl glycosides of mono- and disaccharides . Fatty acyl glycosides have been studied in several plant families and are most extensively characterized in members of Solanaceae . Investigations into the function of fatty acyl glycosides in plants suggest they may act to protect against insect herbivory through various mechanisms and provide protection against fungal pathogens . A recent study isolated and identified fatty acyl glycosides from strawberry capable of inducing immune responses in A. thaliana, including ROS burst, callose deposition, increased expression of defense-related genes, and induced resistance to bacterial and fungal challenge . This same work also demonstrated that the strawberry-derived fatty acyl glycosides induced resistance in soybean and, due to their antimicrobial activity, also protected lemon fruits post harvest from fungal infection . AA- and ANE-root treatments locally elicit accumulation of the same class of defense associated metabolites that Grellet et al. illustrated to have direct antimicrobial activity and protect against disease . Cell wall fortification is an important plant defense often initiated upon pathogen infection. Cell wall lignification is a well-studied mechanism with localized accumulation of phenolic intermediates and lignin at attempted penetration sites . Lignification reinforces and rigidifies the cell wall to create an impervious barrier to microbial ingress . In our study, AA treatment of tomato roots induced accumulation of a phenylcoumaran intermediate in lignin biosynthesis, while ANE treatment induced accumulation of coniferyl alcohol, an important monomer unit of lignin. Interestingly, coniferyl alcohol has recently been shown to act in a signaling capacity in a regulatory feedback mechanism to intricately control lignin biosynthesis, an irreversible process that is energetically costly .