4.4. Metabolite Analysis of Soybean Leaves and Root Extracts

The metabolites of soybean leaves and roots were analyzed via LC–MS. The preparation of sample, analysis of LC–MS, and analysis of multivariate are detailed in this section. For the sample preparation, the soybean leaves and roots under 5 and 100 µM Mn stress were used for metabolome analysis. Each treatment was sampled after cultivation for 15 days, and four biological repetitions were employed for each treatment. A proper amount of each sample was weighed precisely before moving it into a centrifuge tube (2 mL cubage). Then, 600 µL MeOH, kept at a temperature of −20 °C, containing 4 ppm 2-Amino-3-(2-chloro-phenyl)-propionic acid, was supplemented and vortex-oscillated for 30 s. Moreover, 100 mg glass beads were supplemented and moved into a tissue grinder at 60 Hz for 90 s. The ultrasonic treatment (at 25 °C) was performed for 15 min, followed by centrifuging treatment at 4 °C for 10 min at the speed of 12,000 rpm. The supernatant liquid was filtered by a 0.22 µm filtering membrane and moved into the test bottle for LC–MS (liquid chromatography–mass spectrometry) testing. The LC judgment was implemented on a system of ultra-high-performance liquid chromatography (UHPLC) (Thermo Fisher Scientific, USA). Mass spectrometric testing of metabolic products was implemented on Q Exactive (Thermo FisherScientific, USA) with an ESI ion source.

For the analyses of LC–MS and multivariate, the original data were firstly transformed into mzXML format via the MS Convert in ProteoWizard software package (v3.0.8789) [67]. They were further processed with XCMS (various forms (X) of chromatography mass spectrometry) [68] for property detection, keeping time correction, and alignment. The metabolic products were authenticated by accuracy mass (<30 ppm), and the data of MS/MS were further matched with the HMDB (human metabolome database) (http://www.hmdb.ca, accessed on 27 August 2023) [69], the massbank (http://www.massbank.jp/, accessed on 27 August 2023) [70], the lipidMaps (http://www.lipidmaps.org, accessed on 27 August 2023) [71], the mzclou (https://www.mzcloud.org, accessed on 27 August 2023) [72], and the KEGG (http://www.genome.jp/kegg/, accessed on 27 August 2023) [73]. The robust LOESS signal correction (QC-RLSC) [74] was used for normalization of data to remedy systematic bias. After normalization processing, only peaks of ion with relativity SD (standard deviations) of below 30% in quality control (QC) were reserved to guarantee the correct identification of metabolic products.

Ropls software (version 3.0.2) was applied to all analyses of multivariate data and models [75]. After scaling data, the models were built based on the PCA (principal component analysis), the PLS-DA (partial least-square discriminant analysis), and the OPLS-DA (orthogonal partial least-square discriminant analysis). The metabolic profiles were visualized as score plots, in which each point represented a sample. The relevant loading plot and S-plot were created to offer information on the metabolic products that affected sample clustering. All of the models evaluated were verified for overfitting by permutation inspection. The descriptive performance of the models was tested by R2X (cumulative) (perfect model: R2X (cum) = 1) and R2Y (cumulative) (perfect model: R2Y (cum) = 1) values. Their prediction performance was tested by Q2 (cumulative) (perfect model: Q2 (cum) = 1) and a permutation inspection. The permuted model might not be capable of predicting classes. The R2 and Q2 values at the Y-axis intercept should be lower than those of Q2 and the R2 of the model of nonpermutation. OPLS-DA permitted the testing of the discriminating metabolites by the VIP. The p value, VIP produced by OPLS-DA, and FC were adopted to find the contributable variable for classification. In the end, p value < 0.05 and VIP values > 1 were recognized as metabolic products with statistical significance. DMs were applied to analysis of pathway with the help of MetaboAnalyst [76], which integrated the findings from strong analysis of pathway enrichment with the analysis of pathway topology. Then, the identified metabolic products in metabolomics were reflected in the pathway of KEGG (Kyoto Encyclopedia of Genes and Genomes) for biological analysis of advanced systemic functions. The metabolic products and relevant pathways were directly visualized by the KEGG Mapper tool.