Metabolic Modeling of Inborn Errors of Metabolism: Carnitine Palmitoyltransferase II Deficiency and Respiratory Chain Complex I Deficiency

Abstract

The research goal was to assess the current capabilities of a metabolic modeling environment to support exploration of inborn errors of metabolism (IEMs); and to assess whether, drawing on evidence from published studies of EMs, the current capabilities of this modeling environment correlate with clinical measures of energy production, fatty acid oxidation, accumulation of toxic by-products of defective metabolism, and mitigation via therapeutic agents. IEMs comprise several hundred disorders of energy production, often with significant impact on morbidity and mortality. Despite advances in genomic medicine, currently the majority of therapeutic options for IEMs are supportive only, and most only weakly evidenced. Metabolic modeling could potentially offer an in silico alternative for exploring therapeutic possibilities. This research established models of two inborn errors of metabolism (IEMs), carnitine palmitoyltransferase (CPT) II deficiency and respiratory chain complex I deficiency, allowing exploration of combinations of IEMs at different degrees of enzyme deficiency. It utilized a modified version of the human metabolic network reconstruction, Recon 2, which includes known metabolic reactions and metabolites in human cells, and which allows constraint-based modeling within a computational and mathematical representation of human metabolism. It utilized the Matlab-based COBRA (Constraint-based Reconstruction and Analysis) Toolbox 2.0, and a customized suite of functions, to model ATP production, long-chain fatty acid oxidation (LCFA), and acylcarnitine accumulation in response to varying defect levels, inputs and a simulated candidate therapy. Following significant curation of the metabolic network reconstruction and customization of COBRA/Matlab functions, this study demonstrated that ATP production and LCFA oxidation were within expected ranges, and correlated with clinical data for enzyme deficiencies, while acylcarnitine accumulation inversely correlated with the degree of enzyme deficiency; and that it was possible to simulate upregulation of enzyme activity with a therapeutic agent. Results of the curation effort contributed to development of an updated version of the metabolic reconstruction Recon 2. Customization of modeling approaches resulted in a suite of re-usable Matlab functions and scripts usable with COBRA Toolbox methods available for further exploration of IEMs. While this research points to potentially greater suitability of kinetic modeling for some aspects of metabolic modeling of IEMs, it helps to demonstrate potential viability of constraint-based steady state modeling as a means to explore some clinically relevant measures of metabolic function for single and combined inborn errors of metabolism.