Pled in this system, which motivates the development of a new method, based on an objective function that rewards consistency between the pattern of expression change along the developing leaf and the pattern of flux change along the leaf for each reaction. With this approach, we predict reaction rates in a model of mesophyll and LOR-253 web bundle sheath tissue in fifteen segments of the leaf, interacting through vascular transport of sucrose, glycine, and glutathione. We compare our predictions to results from radiolabeling experiments.Results Metabolic reconstruction of Zea maysA novel genome-scale metabolic model was generated from version 4.0 of the CornCyc metabolic pathway database [26] and is presented in two forms. The comprehensive reconstruction involves 2720 reactions among 2725 chemical species, and incorporates CornCyc predictions for the function of 5204 maize genes, with 2064 reactions MK-8742 web associated with at least one gene. A high-confidence subset of the model, excluding many reactions not associated with manually curated pathways or lacking computationally predicted gene assignments as well as all reactions which could not achieve nonzero flux in FBA calculations, involves 635 reactions among 603 species, with 469 reactions associated with a total of 2140 genes. Both the comprehensive and high-confidence models can simulate the production of all major maize biomass constituents (including amino acids, nucleic acids, fatty acids and lipids, cellulose and hemicellulose, starch, other carbohydrates, and lignins, as well as chlorophyll) under either heterotrophic or photoautotrophic conditions and include chloroplast,PLOS ONE | DOI:10.1371/journal.pone.0151722 March 18,4 /Multiscale Metabolic Modeling of C4 Plantsmitochondrion, and peroxisome compartments, with key reactions SART.S23506 of photosynthesis (including a detailed representation of the light reactions), photorespiration, the NADP-ME C4 cycle, and mitochondrial respiration localized appropriately. Gene associations for reactions present in more than one subcellular compartment have been refined based on the results of subcellular proteomics experiments and computational predictions (as collected by the Plant Proteomics Database [27]) to assign genes to reactions in appropriate compartments. Two alternative sets of biomass production reactions are incorporated in the model. One system (based closely on iRS1563 [22]) allows the production of biomass components only in a fixed ratio (as is j.jebo.2013.04.005 appropriate in FBA calculations that maximize biomass production.) The other set of reactions allows individual biomass components to be produced without any constraint on their rates, and is used in some calculations below to allow shifts in biomass composition along the leaf developmental gradient to be predicted based on experimental data. A model for interacting mesophyll and bundle sheath tissue in the leaf was created by combining two copies of the high-confidence model, with transport reactions to represent oxygen and CO2 diffusion and metabolite transport through the plasmodesmata, and restricting exchange reactions appropriately (nutrient uptake from the vascular system to the bundle sheath, and gas exchange with the intercellular airspace to the mesophyll). A schematic of the two-cell model is shown in Fig 1b. Both single-cell versions of the model and the two-cell model, designated iEB5204, iEB2140, and iEB2140x2 respectively (based on the primary author’s initials and number of genes included, a.Pled in this system, which motivates the development of a new method, based on an objective function that rewards consistency between the pattern of expression change along the developing leaf and the pattern of flux change along the leaf for each reaction. With this approach, we predict reaction rates in a model of mesophyll and bundle sheath tissue in fifteen segments of the leaf, interacting through vascular transport of sucrose, glycine, and glutathione. We compare our predictions to results from radiolabeling experiments.Results Metabolic reconstruction of Zea maysA novel genome-scale metabolic model was generated from version 4.0 of the CornCyc metabolic pathway database [26] and is presented in two forms. The comprehensive reconstruction involves 2720 reactions among 2725 chemical species, and incorporates CornCyc predictions for the function of 5204 maize genes, with 2064 reactions associated with at least one gene. A high-confidence subset of the model, excluding many reactions not associated with manually curated pathways or lacking computationally predicted gene assignments as well as all reactions which could not achieve nonzero flux in FBA calculations, involves 635 reactions among 603 species, with 469 reactions associated with a total of 2140 genes. Both the comprehensive and high-confidence models can simulate the production of all major maize biomass constituents (including amino acids, nucleic acids, fatty acids and lipids, cellulose and hemicellulose, starch, other carbohydrates, and lignins, as well as chlorophyll) under either heterotrophic or photoautotrophic conditions and include chloroplast,PLOS ONE | DOI:10.1371/journal.pone.0151722 March 18,4 /Multiscale Metabolic Modeling of C4 Plantsmitochondrion, and peroxisome compartments, with key reactions SART.S23506 of photosynthesis (including a detailed representation of the light reactions), photorespiration, the NADP-ME C4 cycle, and mitochondrial respiration localized appropriately. Gene associations for reactions present in more than one subcellular compartment have been refined based on the results of subcellular proteomics experiments and computational predictions (as collected by the Plant Proteomics Database [27]) to assign genes to reactions in appropriate compartments. Two alternative sets of biomass production reactions are incorporated in the model. One system (based closely on iRS1563 [22]) allows the production of biomass components only in a fixed ratio (as is j.jebo.2013.04.005 appropriate in FBA calculations that maximize biomass production.) The other set of reactions allows individual biomass components to be produced without any constraint on their rates, and is used in some calculations below to allow shifts in biomass composition along the leaf developmental gradient to be predicted based on experimental data. A model for interacting mesophyll and bundle sheath tissue in the leaf was created by combining two copies of the high-confidence model, with transport reactions to represent oxygen and CO2 diffusion and metabolite transport through the plasmodesmata, and restricting exchange reactions appropriately (nutrient uptake from the vascular system to the bundle sheath, and gas exchange with the intercellular airspace to the mesophyll). A schematic of the two-cell model is shown in Fig 1b. Both single-cell versions of the model and the two-cell model, designated iEB5204, iEB2140, and iEB2140x2 respectively (based on the primary author’s initials and number of genes included, a.