2016 Conference on Computational Modelling with COPASI
Manchester Institute of Biotechnology, 12th – 13th May, 2016
1 - University of Surrey, UK; 2 - University of Reading, UK; 3 - University of Leeds, UK
Keywords: dynamic flux variability analysis, sugars, fatty liver, insulin signalling
Very high doses of fructose alter human hepatic insulin sensitivity and increase lipogenesis. However, the relevance of these data to population consumption is unclear. The objective of this work is to develop a predictive, multi-scale model of human hepatic monosaccharide transport, signalling and metabolism. This computational model will be used to predict the regulatory and metabolic outcomes to physiological levels of glucose and fructose in healthy and fatty liver.
Utilising quasi steady state Petri nets (QSSPN; Fisher, et al. 2013 Bioinformatics), the aim of this work is to build a multi-scale model composed of: (i) gene regulation and signalling relevant to lipid and sugar metabolism; (ii) a kinetic model of insulin signalling created by integration of published ODE models; and (iii) the HepatoNet1 liver-specific genome-scale metabolic network constrained by in vitro flux measurements. In these simulations, we propose a novel analysis approach ‘dynamic flux variability analysis’ (dFVA). Here, the exchange flux of interest was set as the objective function and used the minimal and maximal objective function values to calculate upper and lower bound time courses of metabolite concentration in the medium or extracellular space. These bounds were consistent with stoichiometric and thermodynamic constraints of the model, whilst also satisfying the demands of a ‘healthy hepatocyte’ biomass function. Alongside this, an immortalised hepatocyte cell line, HepG2, was used to provide in vitro data to experimentally validate in silico predictions. Insulin sensitivity by western blot analysis (n=3-4) and sugar uptake with and without insulin stimulation (n=3-5) were measured.
To date, we have reconstructed a dynamic regulatory network of hepatic glucose and fructose transport using the Petri net formalism and integrated this with HepatoNet1 constrained by in vitro flux data (Jain, et al. 2012 Science). Together with our newly proposed dFVA method, simulations have predicted minimum and maximum transport rates allowing the calculation of extracellular glucose and fructose concentrations over time. Insulin sensitivity was confirmed in HepG2 cells with a 1.7-fold increase of pAKT/AKT expression in response to postprandial levels of insulin. HepG2 medium glucose and fructose concentrations were found to be within our predicted dFVA solution space. In addition, a significant increase of sugar uptake was seen in insulin-treated versus untreated cells. Preliminary simulations of a published kinetic model of hepatic insulin signalling (Kubota, et al. 2012 Molecular Cell) have been implemented in COPASI and successfully replicated by QSSPN.
In conclusion, we are able to reproduce hepatic monosaccharide uptake in vitro in our in silico model. Future work will integrate the regulatory insulin signalling network to the metabolic network to predict the outcomes of insulin regulation on sugar and lipid metabolism in response to physiological levels of glucose and fructose in healthy and fatty liver.