Multiscale metabolic modeling of C4 plants: connecting nonlinear genome-scale models to leaf-scale metabolism in developing maize leaves

TL;DR

This paper develops a nonlinear constraint-based metabolic model for maize C4 leaves, integrating spatial, biochemical, and gene expression data to predict metabolic states and responses to environmental changes.

Contribution

It introduces a novel nonlinear optimization framework for genome-scale C4 plant metabolism, linking leaf development, gene expression, and metabolic flux predictions.

Findings

High correlation between predicted fluxes and experimental data

Successful modeling of base-to-tip metabolic transition in maize leaves

Inclusion of nonzero growth rate in the model

Abstract

C4 plants, such as maize, concentrate carbon dioxide in a specialized compartment surrounding the veins of their leaves to improve the efficiency of carbon dioxide assimilation. Nonlinear relationships between carbon dioxide and oxygen levels and reaction rates are key to their physiology but cannot be handled with standard techniques of constraint-based metabolic modeling. We demonstrate that incorporating these relationships as constraints on reaction rates and solving the resulting nonlinear optimization problem yields realistic predictions of the response of C4 systems to environmental and biochemical perturbations. Using a new genome-scale reconstruction of maize metabolism, we build an 18000-reaction, nonlinearly constrained model describing mesophyll and bundle sheath cells in 15 segments of the developing maize leaf, interacting via metabolite exchange, and use RNA-seq and enzyme activity measurements to predict spatial variation in metabolic state by a novel method that optimizes correlation between fluxes and expression data. Though such correlations are known to be weak in general, here the predicted fluxes achieve high correlation with the data, successfully capture the experimentally observed base-to-tip transition between carbon-importing tissue and carbon-exporting tissue, and include a nonzero growth rate, in contrast to prior results from similar methods in other systems. We suggest that developmental gradients may be particularly suited to the inference of metabolic fluxes from expression data.