Multivariate analysis of metabolome in leaves
To further study the effect of H2S and rhizobia on water deficiency, a metabolomic analysis was performed in the control, Q8 and Q8+NaHS -treated soybean seedlings under NW and SW conditions. The OPLS-DA results (Fig. 11A, C and E, Fig. S2A, B, Fig. 12A, B) showed different responses in levels of metabolites by leaves of the three treatments to water stress. The importance of metabolites in sample discrimination associated with difference treatments were easily visible in the volcano plot which combined univariate and multivariate analyses (−log (P ) obtained in the two-way ANOVA was plotted against the loading (pcorr) along axis (Fig. 11B, D, F). As can be seen from the volcano plot, in the control, SW condition resulted in more obvious metabolic response than NW condition (Fig. 11B), while the differential metabolites were less changed under the Q8 treatment (Fig. 11D). In addition, Q8+NaHS treatment seemed to have more up-regulation metabolites under SW conditions than under NW condition (Fig. 11F). The metabolites in the leaves were changed depending on the Q8 and Q8+NaHS treatments under water deficiency (Table 1). Our results showed that fatty acyl metabolites including methyl furfuracrylate and 2-Octenedioic acid synthesis were appreciably regulated by control treatment under SW condition. In addition, under SW condition, fatty acyl metabolites and isoflavone metabolites (Biochanin A) were sharply influenced by Q8 treatment. Furthermore, Q8+NaHS treatment regulated more metabolites than control and Q8 treatments under water deficiency. We focused on separate analysis of lipid metabolism changes between different components (Table S2), and based on P ≤0.05, Q8+NaHS treatment significantly increased up-regulation metabolites compared with control.
Furthermore, the results showed that the addition of rhizobia appreciably affected the changes of metabolites under NW condition (Fig. S2C, D, Table S4). In SW condition, control/Q8 treatment led to a more distinct metabolic response than treatments with Q8/Q8+NaHS (Table 2). We could visually observe the change of metabolites under different treatments by the volcano plot (Fig. 12C, D). Under SW condition, there were 37 different metabolites based on P ≤0.001 in the comparison between control and Q8 treatments (Table 2). Among them, we separately analyzed the composition of lipids among different treatments (Table S5). The number of up-regulated lipid metabolites were remarkably increased between control and Q8 treatments under SW condition. Under SW condition, the up-regulated lipid metabolites were similar to the NW condition between Q8 and Q8+NaHS treatments. It is worth noting that metabolomics that was indicated H2S and rhizobia synergistically regulated lipid metabolites under SW condition, including PE, PG, agavoside A, and dephospho-CoA in leaves, showing an effective H2S and rhizobia synergistically regulation and improved tolerance to water deficiency.