GPU-Enabled stochastic Spatiotemporal Model of Rat Ventricular Myocyte Calcium Dynamics

Abstract

The dysfunction of the normal calcium dynamics is a major factor in certain types of cardiac arrhythmias. These cardiac arrhythmias are thought to result from Ca2+ waves which occur when Ca2+ release propagates from one release site to another outside of the normal time during systole resulting in depolarization of the cell's outer membrane. Experimental results suggest that the elementary event underlying calcium release at these sites is the Ca2+ spark and the summation of these Ca2+ sparks result in the global [Ca2+]i transient that causes contraction. We have developed a model of the cardiac myocyte that includes the spatial organization and microsecond level resolution of clusters of ryanodine receptor (RyR) that are Ca2+ release channels responsible for the generation of Ca2+ sparks. We use this model to explore how Ca2+ overload, RyR Ca2+ sensitivity, RyR coupling, and other factors that affect the propagation of Ca2+ release between release sites. We will utilize our newly developed Ultrafast Markov chain Monte Carlo method which allows the rapid simulation of a whole-cell model containing 20,000 release sites, each containing 7 L-type Ca2+ channels and 50 RyRs. This algorithm greatly reduces computation time by using adaptive time step approach and a compact representation of the Markov chain state space. Hence, this novel method provides a powerful tool for performing stochastic cellular simulation with realistic Ca2+ dynamics. Also, with the availability of the next generation graphics processing units (GPU) computing architecture - codename Fermi from NVIDIA - model solution is greatly accelerated allowing the implementation of such a detailed model for the first time.