Experimental validation of metabolic pathway modeling.

In the search for new drug targets in the human parasite Entamoeba histolytica, metabolic control analysis was applied to determine, experimentally, flux control distribution of amebal glycolysis. The first (hexokinase, hexose-6-phosphate isomerase, pyrophosphate-dependent phosphofructokinase (PP(i)-PFK), aldolase and triose-phosphate isomerase) and final (3-phosphoglycerate mutase, enolase and pyruvate phosphate dikinase) glycolytic segments were reconstituted in vitro with recombinant enzymes under near-physiological conditions of pH, temperature and enzyme proportion. Flux control was determined by titrating flux with each enzyme component. In parallel, both glycolytic segments were also modeled by using the rate equations and kinetic parameters previously determined. Because the flux control distribution predicted by modeling and that determined by reconstitution were not similar, kinetic interactions among all the reconstituted components were experimentally revised to unravel the causes of the discrepancy. For the final segment, it was found that 3-phosphoglycerate was a weakly competitive inhibitor of enolase, whereas PP(i) was a moderate inhibitor of 3-phosphoglycerate mutase and enolase. For the first segment, PP(i) was both a strong inhibitor of aldolase and a nonessential mixed-type activator of amebal hexokinase; in addition, lower V(max) values for hexose-6-phosphate isomerase, PP(i)-PFK and aldolase were induced by PP(i) or ATP inhibition. It should be noted that PP(i) and other metabolites were absent from the 3-phosphoglycerate mutase and enolase or aldolase and hexokinase kinetics experiments, but present in reconstitution experiments. Only by incorporating these modifications in the rate equations, modeling predicted values of flux control distribution, flux rate and metabolite concentrations similar to those experimentally determined. The experimentally validated segment models allowed 'in silico experimentation' to be carried out, which is not easy to achieve in in vivo or in vitro systems. The results predicted a nonsignificant effect on flux rate and flux control distribution by adding parallel routes (pyruvate kinase for the final segment and ATP-dependent PFK for the first segment), because of the much lower activity of these enzymes in the ameba. Furthermore, modeling predicted full flux-control by 3-phosphoglycerate mutase and hexokinase, in the presence of low physiological substrate and product concentrations. It is concluded that the combination of in vitro pathway reconstitution with modeling and enzyme kinetics experimentation permits a more comprehensive understanding of the pathway behavior and control properties.