Computational modelling of natural cell-to-cell heterogeneity reveals key parameters that control the diversity of human pancreatic islet β-cell excitability in response to glucose

Insulin-producing {beta}-cells demonstrate remarkable heterogeneity in their individual responsiveness to glucose, and that cellular heterogeneity contributes to coordinating islet activity and glucose homeostasis. Our current understanding of how variation in cell-intrinsic factors control cellular excitability and insulin secretion is informed by foundational experiments conducted on dispersed single {beta}-cells. Such studies are limited in their ability to link multiple electrical or metabolic properties within a single cell and preclude the ability to relate, post hoc, each parameters contribution to glucose responsiveness. Computational modelling represents a unique and underutilized tool to integrate and investigate the role of natural {beta}-cell heterogeneity in physiologic glucose responses. Herein, we utilize a high-volume single-cell electrophysiology "patch-seq" dataset to define the physiologically relevant sources of variability in human {beta}-cell electrophysiology and model their influence on single-cell glucose responses. Three thousand in silico human {beta}-cells were fitted to physiologically relevant variations in glucokinase activity, K+ current, Na+ current, Ca2+ current, and exocytotic function. Four dominant electrical phenotypes arose at low (2 mM) and high (20 mM) glucose: silent, bursting, spiking, and depolarized. Approximately 50% of uncoupled {beta}-cells remained electrically silent at high glucose. Furthermore, Na+ channel half-inactivation voltage was a major predictor of the silent and spiking phenotypes at each glucose concentration, and of cells that transition from silent to spiking when glucose increased. Indeed, experimentally observed variation in Na+ channel voltage dependence was second only to variation in ATP-sensitive potassium channel conductance in determining {beta}-cell excitability. Our data-driven computational modelling highlights the functional importance of electrical heterogeneity in human {beta}-cell glucose responses, and provides a useful tool for generating testable hypotheses.