Modeling in Large-scale Metabolic Networks
Diseases such as hepatosteatosis (NASH), non-alcoholic fatty liver disease (NAFLD), diabetes and states of fasting and overnutrition are reflected in hepatocyte's metabolism. To investigate the specific mechanisms involved is a precondition for rational design of strategies in the treatment of hepatic diseases. Some main factors of e.g. NAFLD development are understood but for a specific pharmacologic intervention a much greater level of detail is required.
An important precondition to understand the complex interplay in the realization of a metabolic function is the estimation of intracellular flux rates of specific biochemical conversion and transport processes. However, its measurement is difficult. The straightforward method to estimate the reaction fluxes directly is to to measure the fate of marked atoms (for instance 13C) in given substrate molecules. However, this process is experimentally challenging. From marked glucose feeding, finally only 15-30 reaction fluxes are estimated. This estimate is also an average of a relatively long time period - it can not be used for the investigation of short-term effects. Thus, it is necessary to infer the fluxes and flux changes indirectly from other measurements.
Flux changes can be estimated from the change of intracellular concentrations but their measurement is also difficult, typically 50 different metabolites are measured in an experimental protocol. To infer flux rates, accurate kinetic models are necessary.
The disadvantage of these low-scale methods is that only hypotheses can be confirmed, but processes in unsuspected parts of the metabolism are not found. Currently high-thruput measurements are available for protein and RNA abundance. The latter is cheaper, needs less effort, and more complete. For the human and mouse hepatocyte, relatively many RNA profiles for different conditions are available and the basis of this work.
Infer metabolic objective
The challenge is that there is a relatively long cascade of processes involved from messaenger-RNA to metabolic fluxes. In this cascade, regulation processes or even basic biophysical constraints interrupt or modify the flow. Thus, it can be expected that the image of the metabolic function reflected by RNA expression is blurred.
The central requirement to the modelling method is that even though the image is blurred, semi-quantitative information can still be obtained. This information can then be verified by a more accurate targertd metabolomic experimental techniques.