Multiscale and modular analysis of cardiac energy metabolism: repairing the broken interfaces of isolated system components.

Computational models of large molecular systems can be assembled from modules representing biological function emerging from interactions among a small subset of molecules. Experimental information on isolated molecules can be integrated with the response of the network as a whole to estimate crucial missing parameters. As an example, a "skeleton" model is analyzed for the module regulating dynamic adaptation of myocardial oxidative phosphorylation (OxPhos) to fluctuating cardiac energy demand. The module contains adenine nucleotides, creatine, and phosphate groups. Enzyme kinetic equations for two creatine kinase (CK) isoforms were combined with the response time of OxPhos (t mito; generalized time constant) to steps in the cardiac pacing rate to identify all module parameters. To obtain t mito, the time course of O2 uptake was measured for the whole heart. An O2 transport model was used to deconvolute the whole-heart response to the mitochondrial level. By optimizing mitochondrial outer membrane permeability to 21 microm/s the experimental t mito = 3.7 s was reproduced. This in vivo value is about four times larger, or smaller, respectively, than conflicting values obtained from two different in vitro studies. This demonstrates an important rule for multiscale analysis: experimental responses and modeling of the system at the larger scale allow one to estimate essential parameters for the interfaces of components which may have been altered during physical isolation. The model correctly predicts a smaller t mito when CK activity is reduced. The model further predicts a slower response if the muscle CK isoform is overexpressed and a faster response if mitochondrial CK is overexpressed. The CK system is very effective in decreasing maximum levels of ADP during systole and reducing average Pi levels over the whole cardiac cycle.