Kinetic and data-driven modeling of pancreatic β-cell central carbon metabolism and insulin secretion

• Published in: PLoS computational biology 2022-10, Vol.18 (10), p.e1010555-e1010555
• Main Authors: Gelbach, Patrick E., Zheng, Dongqing, Fraser, Scott E., White, Kate L., Graham, Nicholas A., Finley, Stacey D.
• Format: Article
• Language: English
• Subjects: Beta cells; Biology and Life Sciences; Carbon; Computer and Information Sciences; Diabetes; Diabetes mellitus; Enzymes; Glucose; Glucose metabolism; Insulin; Insulin secretion; Least squares method; Medicine and Health Sciences; Metabolic flux; Metabolic networks; Metabolic pathways; Metabolism; Metabolites; Metabolomics; Mitochondria; Modelling; Pancreas; Pathways; Physical Sciences; Regression models; Secretion
• ISSN: 1553-7358; 1553-734X; 1553-7358
• DOI: 10.1371/journal.pcbi.1010555

Pancreatic β-cells respond to increased extracellular glucose levels by initiating a metabolic shift. That change in metabolism is part of the process of glucose-stimulated insulin secretion and is of particular interest in the context of diabetes. However, we do not fully understand how the coordinated changes in metabolic pathways and metabolite products influence insulin secretion. In this work, we apply systems biology approaches to develop a detailed kinetic model of the intracellular central carbon metabolic pathways in pancreatic β-cells upon stimulation with high levels of glucose. The model is calibrated to published metabolomics datasets for the INS1 823/13 cell line, accurately capturing the measured metabolite fold-changes. We first employed the calibrated mechanistic model to estimate the stimulated cell’s fluxome. We then used the predicted network fluxes in a data-driven approach to build a partial least squares regression model. By developing the combined kinetic and data-driven modeling framework, we gain insights into the link between β-cell metabolism and glucose-stimulated insulin secretion. The combined modeling framework was used to predict the effects of common anti-diabetic pharmacological interventions on metabolite levels, flux through the metabolic network, and insulin secretion. Our simulations reveal targets that can be modulated to enhance insulin secretion. The model is a promising tool to contextualize and extend the usefulness of metabolomics data and to predict dynamics and metabolite levels that are difficult to measure in vitro . In addition, the modeling framework can be applied to identify, explain, and assess novel and clinically-relevant interventions that may be particularly valuable in diabetes treatment.