On the Analysis of Metabolic Networks

Abstract

In the past quarter of a century, it has become evident that the major pathways of intermediary metabolism are regulated by a remarkable variety of interlocking control mechanisms. These include covalent modification of enzymes: acetylation, methylation, adenylylation, phosphorylation by cAMP-sensitive and -insensitive protein kinases, changes in SH and S-S redox states, among others. Regulation by noncovalent modification is virtually ubiquitous, and includes sensitivity to various ions, especially Ca²⁺, to substrates and to end products, to ratios of modifier concentrations such as [ATP]/[ADP] · [P_i], [NADH]/ [NAD⁺], [acyl CoA]/[CoA], etc. While elucidation of regulatory mechanisms is clearly prerequisite to an understanding of the controls acting in vivo, it is not in itself sufficient to allow prediction of the flux of metabolite through any particular metabolic step, even if one were to perform a “complete” analysis of all the known regulatory effectors in the cell. This lack of certainty comes about not only because of the ever-present possibility that compounds which were not measured are important regulatory substances, but also because frequently the enzymes and/or putative regulatory metabolites are in different subcellular environments, whether these be readily indentifiable organelles (such as mitochondria, peroxisomes, glycogen particles) or microenvironments formed by multienzyme complexes (Srere and Mosbach, 1974). The situation is further complicated by much evidence suggesting that proteins interact with one another in the relatively concentrated intracellular or intraorganellar environments where they function, while most of the in vitro studies are performed on partially purified enzymes and at unphysiologically low protein concentrations. The development of techniques for rapid separation of mitochondria from the rest of the cytosol (Siess et al.,1977; Tischler et al.,1977) considerably reduces the ambiguity inherent in measurements of cell metabolite content, but even if it were possible to separate the peroxisomes and other organelles at the same time, it would remain very difficult to assess all the factors determining the metabolite flux through any particular step in vivo. Metabolic network analysis addresses a much simpler question than those we have just discussed—namely, what is the flux of metabolite through the steps of a metabolic pathway in the living cell? Even assuming that the question may be answered by the methods to be described below, it is clear that such information by itself contributes very little to an understanding of how metabolism is regulated. Flux values obtained by network analysis may be compared with the enzyme activity for each step measured in vitro under “optimal” conditions. Frequently the in vitro activity exceeds the in vivo flux, and this information may reinforce accepted views on regulated steps or may indicate new steps where control may be sought. In some cases, however, the apparent in vivo flux exceeds the enzyme activity assayed under “optimal” conditions. This reemphasizes the difficulty of extrapolating from in vitro measurements of enzyme activity to in vivo behavior. When flux values from network analysis are interpreted in the light of information on the in vitro behavior of the enzymes involved and on the concentrations of ligands in the relevant subcellular compartments, one may expect to acquire a deeper understanding of the regulation of metabolism in the living cell.