Plant primary metabolism is a highly coordinated, central, and complex network of biochemical processes regulated at both the genetic and post-translational levels. The genetic basis of this network can be explored by analyzing the metabolic composition of genetically diverse genotypes in a given plant species. Here, we report an integrative strategy combining quantitative genetic mapping and metabolite‒transcript correlation networks to identify functional associations between genes and primary metabolites in Arabidopsis thaliana. Genome-wide association study (GWAS) was used to identify metabolic quantitative trait loci (mQTL). Correlation networks built using metabolite and transcript data derived from a previously published time-course stress study yielded metabolite‒transcript correlations identified by covariation. Finally, results obtained in this study were compared with mQTL previously described. We applied a statistical framework to test and compare the performance of different single methods (network approach and quantitative genetics methods, representing the two orthogonal approaches combined in our strategy) with that of the combined strategy. We show that the combined strategy has improved performance manifested by increased sensitivity and accuracy. This combined strategy allowed the identification of 92 candidate associations between structural genes and primary metabolites, which not only included previously well-characterized gene‒metabolite associations, but also revealed novel associations. Using loss-of-function mutants, we validated two of the novel associations with genes involved in tyrosine degradation and in β-alanine metabolism. In conclusion, we demonstrate that applying our integrative strategy to the largely untapped resource of metabolite–transcript associations can facilitate the discovery of novel metabolite-related genes. This integrative strategy is not limited to A. thaliana, but generally applicable to other plant species. Primary metabolites are key elements in plant growth and development. Our partial understanding of their biosynthesis and regulation derives mostly from biochemical and genetic modification experiments. The recent generation of large-scale genome-wide data, along with the advances in mass-spectrometry techniques, allows us to treat metabolite levels as quantitative traits and to link them to genomic information, resulting in the identification of so-called metabolic quantitative trait loci (mQTL). These mQTL contribute to the discovery of new biosynthetic and regulatory elements that control the plant's metabolic landscape. Low mapping resolution, however, normally limits discovery to one causal gene per locus. Here, we utilize a complementary strategy to support the identification of casual genes by genetic mapping. We measured metabolite levels in 314 A. thaliana accessions, then used genome-wide association mapping to identify mQTL. We next used previously published results from a time-course stress study to construct metabolite‒transcript correlation networks. Integrating data from both approaches enabled us to select candidate genes linked to specific metabolites. We finally validated two of the novel gene‒metabolite associations using knockout lines. We demonstrated that by using an integrative strategy, we can validate previously characterized gene–metabolite associations, and most importantly, identify novel associations between metabolites and genes. The combined quantitative genetics and metabolite‒transcript networks that we present here can be applied to other organisms and fields of research.

Dark-induced senescence provokes profound metabolic shifts to recycle nutrients and to guarantee plant survival. To date, research on these processes has largely focused on characterizing mutants deficient in individual pathways. Here, we adopted a time-resolved genome-wide association-based approach to characterize dark-induced senescence by evaluating the photochemical efficiency and content of primary and lipid metabolites at the beginning, or after 3 or 6 days in darkness. We discovered six patterns of metabolic shifts and identified 215 associations with 81 candidate genes being involved in this process. Among these associations, we validated the roles of four genes associated with glycine, galactinol, threonine, and ornithine levels. We also demonstrated the function of threonine and galactinol catabolism during dark-induced senescence. Intriguingly, we determined that the association between tyrosine contents and TYROSINE AMINOTRANSFERASE 1 influences enzyme activity of the encoded protein and transcriptional activity of the gene under normal and dark conditions, respectively. Moreover, the single-nucleotide polymorphisms affecting the expression of THREONINE ALDOLASE 1 and the amino acid transporter gene AVT1B, respectively, only underlie the variation in threonine and glycine levels in the dark. Taken together, these results allow us to present a very detailed model of the metabolic aspects of dark-induced senescence, as well as the process itself.

A growing body of research points to a positive interplay between viruses and plants. Tomato yellow curl virus (TYLCV) is able to protect tomato host plants against extreme drought. To envisage the use of virus protective capacity in agriculture, TYLCV-resistant tomato lines have to be infected first with the virus before planting. Such virus-resistant tomato plants contain virus amounts that do not cause disease symptoms, growth inhibition, or yield loss, but are sufficient to modify the metabolism of the plant, resulting in improved tolerance to drought. This phenomenon is based on the TYLCV-dependent stabilization of amounts of key osmoprotectants induced by drought (soluble sugars, amino acids, and proteins). Although in infected TYLCV-susceptible tomatoes, stress markers also show an enhanced stability, in infected TYLCV-resistant plants, water balance and osmolyte homeostasis reach particularly high levels. These tomato plants survive long periods of time during water withholding. However, after recovery to normal irrigation, they produce fruits which are not exposed to drought, similarly to the control plants. Using these features, it might be possible to cultivate TYLCV-resistant plants during seasons characterized by water scarcity.