Metabolic changes associated with plasticity of studied phenotypes to N availability
Primary metabolism is associated with growth and levels of several primary metabolites are altered in response to changes in N availability (Sulpice et al., 2013). However, it is not clear if the plasticity of metabolites is associated with plasticity of the complex phenotypes studied here. To address this question, we first measured the levels of 66 primary metabolites in a subset of 43 accessions grown in the three N conditions (Table S6). To quantify plasticity, we calculated the CV of every metabolite across the three mean values for each of the 43 accessions (Table S7). The majority of primary metabolites are essential for plant growth and showed low plasticity across N conditions as expected (Figure 5a). The organic acids citrate and fumarate, and the amino acids ornithine, glutamine, lysine and arginine showed the highest median CV across the accessions (Figures 5a; Table S7). Of these, glutamine and glutamate are directly linked to N assimilation, whilst citrate is a precursor for 2-oxoglutarate needed in this process. Furthermore, malate and citrate valves are directly linked to carbon fixation and the tricarboxylic acid (TCA) cycle, providing the substrates for amino acids synthesis, and their conversion can favor the accumulation of other metabolites, such as fumarate (Eprintsev, Fedorin, Sazonova, & Igamberdiev, 2016).
To evaluate if metabolic plasticity relates to plasticity of the four phenotypes, we performed correlation analysis between the CVs of metabolites and CVs of phenotypes ERD, FRD, FT and YIE over the 43 accessions (Figure 5b). We found a significant positive correlation between the plasticity of ERD and plasticities of benzoate, dehydroascorbate (dimer), glutamine, and phosphorate levels, and a significant negative correlation with plasticity of glucose levels. Therefore, our findings suggest that stable levels of glucose over different N conditions, in contrast to dehydroascorbate, are associated with the higher plasticity of ERD. Further, plasticity of FRD showed significant negative correlation with plasticity of fumarate and glycerate levels, and positive correlation with plasticities in lactate and fucose levels (Figure 5b). The plasticity of FT showed positive correlation with homoserine levels, while that of YIE positively correlated with plasticity of fructose, glucose, rhamnose and shikimate levels (Figure 5b). The latter indicates that stable sugar contents stabilizes yield in respond to N availability.
Next, we asked if the significant correlation between the plasticity of a complex phenotype and the plasticity of metabolite levels was observed between the mean metabolite levels and the mean levels of ERD, FRD, FT and YIE in each N condition separately. Only two metabolites, benzoate and lactate, showed significant correlation both between the mean and plasticity values of the same trait; benzoate to ERD and lactate to FRD respectively (Figure 5, Figure S4). This analysis supports the hypothesis that the plasticities and the mean values of the studied phenotypes are independently regulated.