MODELING HUMAN METABOLISM: A DYNAMIC MULTI-TISSUE APPROACH

Despite significant advances in constraint-based modelling, a methodology for modelling
dynamic multi-tissue models of human metabolism is still missing. Additionally, prior to
analysing diseased models, it is important to develop a good methodology, as it would not only
enable us to capture the effects of metabolism-associated diseases, but it would also allow
us to recapitulate known physiological healthy properties of human metabolism. Therefore,
a dynamic multi-tissue model using a new methodology
was developed. The objective function comprises a set of complex functions that the
multi-tissue model needs to perform. To demonstrate the capabilities
of this new approach, different healthy, and unhealthy conditions were simulated. In a first
step, the effect of different healthy conditions was analysed (i.e. the fasting, the ingestion
of different meals, and exercising at various intensities, and conditions), demonstrating the
model’s capability to correctly predict metabolic changes occurring on energy-associated
pathways. In the second step, biomarkers for a range of inborn errors of metabolism were
predicted, and the predictions were shown to be in good agreement with previous data.
Finally, after verifying the capability of the dynamic multi-tissue model to review known
physiological aspects of human metabolism, this model was further integrated with a physiologically-
based pharmacokinetic model of glucose metabolism, previously developed by Schaller
et al. (2013). Contrasting conditions, such as healthy and diabetic, were simulated using the
multi-scale model during fasting and after an oral glucose tolerance test and candidate drugs
to treat type 2 diabetes mellitus were predicted. Five out of the 80 simulated drug targets
were predicted as candidate anti-diabetic targets, and the majority of drugs known to inhibit
the predicted drug targets, have already been shown to have anti-diabetic effects.
The developed approach can be applied to any metabolic disease and to any system
where homeostasis plays an important role, or where a simple biomass optimization function
is not applicable. Furthermore, the large amount of data collected for the multi-tissue model generation is of
significant value for tissue constraint-based metabolic modellers who need data to constrain
their models.