Optimization of computational hERG models for prediction and translation of cardiovascular risk in patients

The human ether-à-go-go (hERG) channel plays an essential role in the repolarization of the cardiac action potential (AP) as it conducts the rapid component of the delayed rectifier potassium current. The inhibition of hERG channels by research compounds and drugs constitutes a relevant cardiovascular risk factor manifesting as AP broadening, early afterdepolarization phases, drug-induced QT interval prolongation and increased risk of Torsade de Pointes, a potentially lethal ventricular arrhythmia. Therefore, assessment of hERG blockade by new drug candidates is one of the mandatory tests to evaluate cardiac safety. To study the drug-induced proarrhythmic risk, a combination of both in vitro and in silico assays (in combination with pre-clinical in vivo ECG testing) has been proposed. However, the computer models currently available have important limitations that reduce its applicability: Hodking-Huxley models do not consider the transition kinetics among the different ion channel states, being inefficient to predict the cardiovascular risk of pharmacologically complex compounds. In turn, Markov formalisms cover transition kinetics and multiple channel conformations, but are both mathematically and technically demanding, and they are based on a set of assumptions that reduce their predictive power. Hence, the aim of this project is to improve available Markov hERG models by establishing optimized recording conditions that allow for replacement of set model assumptions by recorded parameters to predict more precisely pro-arrhythmic risk of compounds, independently of their pharmacological complexity. For that purpose, more informative and physiologically relevant electrophysiological protocols were developed and tested at controlled near-to-physiological temperature (35–36 °C) using the patch-clamp technique in HEK-293 cells overexpressing hERG channels. The most promising and optimized model consistently reproduced the maximal conductance, activation, deactivation, inactivation, and recovery parameters when a protocol consisting of a pre-pulse followed by a train of action potential waveforms was applied. The simulated traces were comparable to the experimentally recorded and the parameters were consistently fitted even after ten different simulations. However, although the model has been proved to fit the parameters under basal conditions, further pharmacological experiments using reference compounds reported to be hERG channel blockers, such as E-4031, are currently being performed to calibrate and validate the model.