The Role of Pool Size Measurements in Improving Flux Estimations in Non-Stationary Metabolic Flux Analysis

Abstract

Over the past several decades, Metabolic Flux Analysis (MFA) has gained importance in the biomedical and pharmaceutical sciences. MFA is based on a computational procedure used to determine metabolic reaction rates (fluxes) and in some cases metabolite concentrations (pool sizes) within metabolic pathways where direct observation and measurement of intracellular function is not feasible. Applying 13C MFA to biological data, such as that in liver metabolism, helps to determine, for example, the fate of glucose in regulatory metabolic pathways under physiological and pathophysiological steady state conditions of a cell. However, designing models with reliable flux estimations remains a challenge. A debate that still persists is whether metabolite pool size measurements added as constraints in optimization algorithms improve metabolic flux predictions. We studied the role of pool sizes using simple model networks with total of 8 and 9 metabolites. To investigate the effect of pool size constraints on flux estimations, we set up a series of scenarios varying inputs into the model for several metabolites (such as information on pool size measurements and noise levels in the mass isotopomer distribution (MID) labeling data). Our findings indicate that in models when a full set of MID labeling data (as oppose to just M0 labeling data) is available, pool sizes become dispensable in flux estimations. Moreover, in biological networks where a full set of labeling data for certain metabolites is accessible but the exact model network is unknown, flux estimations are more reliable if pool sizes are not included as model constraints.