Pool size measurements facilitate the determination of fluxes at branching points in non-stationary metabolic flux analysis: the case of Arabidopsis thaliana

Abstract

Pool size measurements are important for the estimation of absolute intracellular fluxes in particular scenarios based on data from heavy carbon isotope experiments. Recently, steady-state fluxes estimates were obtained for central carbon metabolism in an intact illuminated rosette of Arabidopsis thaliana grown photoautotrophically (Szecowka et al., 2013; Heise et al., 2014). Fluxes were estimated therein by integrating mass-spectrometric data of the dynamics of the unlabeled metabolic fraction, data on metabolic pool sizes, partitioning of metabolic pools between cellular compartments and estimates of photosynthetically inactive pools, with a simplified model of plant central carbon metabolism. However, the fluxes were determined by treating the pool sizes as fixed parameters. Here we investigated whether and, if so, to what extent the treatment of pool sizes as parameters to be optimized in three scenarios may affect the flux estimates. The results are discussed in terms of benchmark values for canonical pathways and reactions, including starch and sucrose synthesis as well as the ribulose-1,5-bisphosphate carboxylation and oxygenation reactions. In addition, we discuss pathways emerging from a divergent branch point for which pool sizes are required for flux estimation, irrespective of the computational approach used for the simulation of the observable labeling pattern. Therefore, our findings indicate the necessity for development of techniques for accurate pool size measurements to improve the quality of flux estimates from non-stationary flux estimates in intact plant cells in the absence of alternative flux measurements.

Keywords: Arabidopsis thaliana; carbon metabolism; flux profiling; isotopic labeling; isotopically non-stationary; metabolic flux analysis; metabolite pool sizes; photoautotrophic growth.

Figure 1

Changes of optimized metabolic content in Scenario B. The differences (errors) between the optimized metabolic content c_{sim, n} (red circle) as well as their lower and upper 95%-confidence limits (error bars) to the measured values c_{obs, n} were scaled by the measured standard deviation by $c_{obs.,n}^{*}=\frac{c_{obs,n}-c_{sim,n}}{{\sigma }_{c_{obs,n}}}$. All changes of the values by the optimization can be explained by measurements errors. Note that for the assumed normal distribution of the measurement errors, the 95% confidence intervals corresponds to 2 σ_{cobs., n}.

Figure 2

Sensitivities of flux estimates and time error to perturbation in metabolic content for scenario B. The heatmap shows the relative changes in flux estimates and time error upon relative changes in metabolic content in Scenarios B. The content of a selected metabolic pool was changed and fixed and the model was re-optimized for the remaining parameters (see Material and Methods).