Dynamics of drug binding to the hERG channel: an in silico cardiac action potential model for early safety assessment

Pharmaceutical companies are under-pressure to produce potential safe drugs (i.e., no effect on cardiac action potential (AP) and QT interval). Early testing strategies to reduce this risk rely on concentration-effect (C-E) curve data obtained from a panel of heterologously-expressed key ion channels (ICs) using standard IonWorks? voltage protocols. We had previously presented an in silico canine midmyocardial AP model that uses C-E data to predict changes in AP duration (APD; Hussein L et al., 2010 SPS meeting). We have now investigated whether incorporating the dynamics of drug binding to the hERG channel might improve the accuracy of the model. New IonWorks? protocols were used to extract binding kinetics of dofetilide, almokalant and cisapride at ~20�C. Data were fitted with a newly developed hERG model to obtain rate constants for drug binding and unbinding. Rate constants were then incorporated into a new canine IKr model developed at ~37�C for simulating APs at various concentrations and pacing frequencies. As seen experimentally, AP simulations predict that all three drugs evoked concentration-dependent increase in APD. If instead a traditional C-E curve was employed, results show that APD increases could be significantly under-predicted for cisapride (31% vs. 44% at 1�M) and almokalant (36% vs. 54% at 3�M) at 1.5Hz. Thus, predictivity of the current AP model can be improved with new alternative IonWorks? protocols. This has the potential to enhance upstream understanding of direct IC modulation of the AP via a rapid cost-effective in silico evaluation.