Application of Biochemical Systems Theory to Metabolism in Human Red Blood Cells: SIGNAL PROPAGATION AND ACCURACY OF REPRESENTATION

Abstract

Human erythrocytes are among the simplest of cells. Many of their enzymes have been characterized kinetically using steady-state methods in vitro, and several investigators have assembled this kinetic information into mathematical models of the integrated system. However, despite its relative simplicity, the integrated behavior of erythrocyte metabolism is still complex and not well understood. Errors will inevitably be encountered in any such model because of this complexity; thus, the construction of an integrative model must be considered an iterative process of assessment and refinement. In a previous study, we selected a recent model of erythrocyte metabolism as our starting point and took it through three stages of model assessment and refinement using systematic strategies provided by biochemical systems theory. At each stage deficiencies were diagnosed, putative remedies were identified, and modifications consistent with existing experimental evidence were incorporated into the working model. In this paper we address two issues: the propagation of biochemical signals within the metabolic network, and the accuracy of kinetic representation. The analysis of signal propagation reveals the importance of glutathione peroxidase, transaldolase, and the concentration of total glutathione in determining systemic behavior. It also reveals a highly amplified diversion of flux between the pathways of pentose phosphate and nucleotide metabolism. In determining the range of accurate representation based on alternative kinetic formalisms we discovered large discrepancies. These were identified with the behavior of the model represented within the Michaelis-Menten formalism. This model fails to exhibit a nominal steady state when the activity of glutathione peroxidase is decreased by as little as 9%. Our current understanding, as embodied in this working model, is in need of further refinement, and the results presented in this paper suggest areas of the model where such effort might profitably be concentrated.