Model-guided concurrent data assimilation for calibrating cardiac ion-channel kinetics

Abstract

Potassium channels (Kv) are responsible for repolarizing the action potential in cardiomyocytes. There is a variety of Kv isoforms and corresponding currents (e.g. IKto, IKslow1, IKslow2) that contribute to different phases of repolarization. Because only the sum of their activities can be measured in the form of currents (IKsum), there is a need to delineate individual K+ currents. Most existing studies make inference of Kv activities via curve-fitting procedures but encounter certain limitations as follows: (1) curve-fitting decomposition only relies on the shape of K+ current traces, which does not discern the underlying kinetics; (2) IKsum traces can only be fitted for one clamp voltage at each time, and then analyzed in a population-averaged way later. This paper presents a novel concurrent data assimilation method to calibrate biophysics-based models and delineate kinetics of Kv isoforms with multiple voltage-clamp responses simultaneously. The proposed method is evaluated and validated with whole-cell IKsum recordings from wild-type and chronically glycosylation-deficient cardiomyocytes. Experimental results show that the proposed method effectively handles multiple-response data and describes glycosylation-conferred perturbations to Kv isoforms. Further, we develop a graphical-user-interface (GUI) application that provides an enabling tool to biomedical scientists for data-driven modeling and analysis of Kv kinetics in various heart diseases.