AraGEM, a genome-scale reconstruction of the primary metabolic network in Arabidopsis

Abstract

Genome-scale metabolic network models have been successfully used to describe metabolism in a variety of microbial organisms as well as specific mammalian cell types and organelles. This systems-based framework enables the exploration of global phenotypic effects of gene knockouts, gene insertion, and up-regulation of gene expression. We have developed a genome-scale metabolic network model (AraGEM) covering primary metabolism for a compartmentalized plant cell based on the Arabidopsis (Arabidopsis thaliana) genome. AraGEM is a comprehensive literature-based, genome-scale metabolic reconstruction that accounts for the functions of 1,419 unique open reading frames, 1,748 metabolites, 5,253 gene-enzyme reaction-association entries, and 1,567 unique reactions compartmentalized into the cytoplasm, mitochondrion, plastid, peroxisome, and vacuole. The curation process identified 75 essential reactions with respective enzyme associations not assigned to any particular gene in the Kyoto Encyclopedia of Genes and Genomes or AraCyc. With the addition of these reactions, AraGEM describes a functional primary metabolism of Arabidopsis. The reconstructed network was transformed into an in silico metabolic flux model of plant metabolism and validated through the simulation of plant metabolic functions inferred from the literature. Using efficient resource utilization as the optimality criterion, AraGEM predicted the classical photorespiratory cycle as well as known key differences between redox metabolism in photosynthetic and nonphotosynthetic plant cells. AraGEM is a viable framework for in silico functional analysis and can be used to derive new, nontrivial hypotheses for exploring plant metabolism.

Figure 1.

A simplified scheme of the current textbook photorespiratory cycle. Numbers are as follows: 1, plastidic glycolate transporter; 2, peroxisomal glycolate transporter; 3, plastidic Glu-malate translocator; 4, plastidic 2-oxoglutarate-malate translocator; 5, peroxisomal Glu transporter; 6, peroxisomal 2-oxoglutarate transporter; 7, peroxisomal Gly transporter; 8, mitochondrial Gly transporter. GDC, Gly decarboxylase; Glt, glycolate; Gox, glyoxylate; GS, Gln synthetase; methylene-THF, methylene-tetrahydrofolate; 2-OG, 2-oxoglutarate; 3-PGA, 3-phosphoglycerate; 2-Pglt, 2-phosphoglycolate; P_i, inorganic phosphate; Ru 1,5-BP, ribulose 1,5-bisP.

Figure 2.

Flux map for the photorespiratory cycle. Comparison of fluxes between photorespiration and photosynthesis (no oxygenation of ribulose 1,5-bisP). In the diagram, solid lines represent fluxes and dashed lines represent the transporters. Green and red lines highlight fluxes that have increased and decreased, respectively, during photorespiration when compared with photosynthesis. Gray lines represent flux values that have not changed significantly. Steps 1 to 17 represent the order of events that complete the photorespiratory cycle. Transporters are described in the Figure 1 legend. Boxes 9, 15, and 16 represent ammonium and CO₂ being transported by diffusion. Box 17 represents CO₂ being fixed in the Calvin cycle.

Figure 3.

Flux map for respiration of a nonphotosynthetic plant cell. Flux comparison is shown between a nonphotosynthetic cell (respiration) and a photosynthetic cell (photosynthesis). Green and red lines represent increased and decreased flux, respectively, of a nonphotosynthetic cell compared with a photosynthetic cell. Gray lines represent flux values that have not changed significantly compared with photosynthesis. Boxes A and B represent the metabolic activity of some isoenzymes typically activated in nongreen tissues. Box A in plastidic glycolysis (glycolysis II) represents the plastid form of GAPDH [NAD(H)-dependent] activity. Box B represents the metabolic activity of NADH-GOGAT involved in ammonia assimilation in nongreen plastids.

Figure 4.

Information flow from a genome-scale model repository (Excel file) to a model document (SBML) to a flux map (Excel file). An initial list of components was compiled from the Arabidopsis genome, and the network was manually curated. The Java parsing program separates reactions into reactants and product stoichiometry and removes repeated reactions. Calculation output from MATLAB was used to update the flux map.