The Pore-Lipid Interface: Role of Amino-Acid Determinants of Lipophilic Access by Ivabradine to the hERG1 Pore Domain

Abstract

Abnormal cardiac electrical activity is a common side effect caused by unintended block of the promiscuous drug target human ether-à-go-go-related gene (hERG1), the pore-forming domain of the delayed rectifier K 1 channel in the heart. hERG1 block leads to a prolongation of the QT interval, a phase of the cardiac cycle that underlies myocyte repolarization detectable on the electrocardiogram. Even newly released drugs such as heart-rate lowering agent ivabradine block the rapid delayed rectifier current I Kr , prolong action potential duration, and induce potentially lethal arrhythmia known as torsades de pointes. In this study, we describe a critical drug-binding pocket located at the lateral pore surface facing the cellular membrane. Mutations of the conserved M651 residue alter ivabradine-induced block but not by the common hERG1 blocker dofetilide. As revealed by molecular dynamics simulations, binding of ivabradine to a lipophilic pore access site is coupled to a state-dependent reorientation of aromatic residues F557 and F656 in the S5 and S6 helices. We show that the M651 mutation impedes state-dependent dynamics of F557 and F656 aromatic cassettes at the protein-lipid interface, which has a potential to disrupt drug-induced block of the channel. This fundamentally new mechanism coupling the channel dynamics and small-molecule access from the membrane into the hERG1 intracavitary site provides a simple rationale for the well established state-dependence of drug blockade. SIGNIFICANCE STATEMENT The drug interference with the function of the cardiac hERG channels represents one of the major sources of drug-induced heart disturbances. We found a novel and a critical drug-binding pocket adjacent to a lipid-facing surface of the hERG1 channel, which furthers our molecular understanding of drug-induced QT syndrome.

An Update on the Structure of hERG

“…However evidence remains equivocal about direct interactions between drugs and F557; mutation of other residues that line the hydrophobic pockets (F619 and L622; see Figure 4C) have negligible (cisapride, haloperidol) or limited (dofetilide) effects on block by these drugs (Saxena et al, 2016), and this is surprising since computational docking indicates that drugs that bind deep within a hERG pore hydrophobic pocket are constrained within a compact binding site; as discussed below, mutations of F619 and L622 strongly attenuate the effects of a hERG activator, indicating that drug binding in this part of the channel is expected to be susceptible to mutation of these side chains. Likewise, for all drugs so far tested except cisapride, the attenuation of block in F557L is similar to the effect of the hERG Y652A mutation (Saxena et al, 2016;Helliwell et al, 2018;Cheng et al, 2019;Perissinotti et al, 2019) suggesting that the contributions of these side chains to hERG block may be linked. This interpretation is reinforced by the finding that the voltage-dependence of hERG block by Cavalli-2 is lost in hERG F557L (Helliwell et al, 2018), similar to the loss of voltagedependence of block by this and other high-affinity blockers in hERG Y652A (Sanchez-Chapula et al, 2003).…”

“…The new hERG structure also provides a context for exploring the structural basis for a potential alternative access of hERG blockers to the central pore via the lipid bilayer, as recently proposed for the bradycardic agent ivabradine (Lees-Miller et al, 2015;Perissinotti et al, 2019). Many channel blockers, including many hERG blockers, are lipophilic and are expected to partition into the lipid bilayer phase; direct drug access from the membrane to the pore is well characterized for a number of Na v (Payandeh et al, 2012) and twin-pore (K2P) channels (Dong et al, 2015), although the fenestrations required for pore access via the lipid are not well established in K v channels (e.g.…”

“…In neither the hERG nor EAG structures is there a pathway that connects the central pore to the membrane that is large enough for drug molecules to enter via the membrane, and access via a lipid-facing fenestration would seem to require either a conformational state of hERG not represented in the new structures, considerable conformational flexibility or conformational transitions associated with gating. On the basis of a suppression of ivabradine block in hERG M651T and modulation of F557 and F656 interactions by M651 in MD simulations (see Figure 4C for the location of these residues) it is proposed that these three residues act as a dynamic gate to facilitate ivabradine access from the membrane to the inner pore (Perissinotti et al, 2019); however, direct evidence from experiment or simulation for access to the pore [the likely site of ivabradine channel block (Melgari et al, 2015b;Perissinotti et al, 2019)] from the membrane remains to be established.…”

Atomistic modeling towards predictive cardiotoxicity

“…Indeed, our ~2.5 s long unbiased MD simulations demonstrate that the neutral form of the drug quickly (in <200ns) embeds into the membrane, while DOFC remains predominantly in aqueous solution (see SI Figure S11A). These results indicated that DOFN could potentially access hERG through a hydrophobic pathway, but that the charged form will only enter through the aqueous pathway (52). Moreover, due to a substantially smaller energetic barrier (Figure 2C), we expect much faster membrane translocation of DOFN.…”

Allosteric Coupling Between Drug Binding and the Aromatic Cassette in the Pore Domain of the hERG1 Channel: Implications for a State-Dependent Blockade

“…The focus on the neutral state of two drugs is also in agreement with recent studies by Clancy and co-workers where the neutral states of two hERG blockers (dofetilide and moxifloxacin) were found to be more stable than the charged ones in the narrow intra-cavity space in the study employing exhaustive Umbrella-Sampling simulations (Yang et al, 2020). It is important to stress that computational and experimental studies showed that various charged-states of the blockers exist in a highly dynamical equilibrium, where preference for a specific charge state of the drug can be shifted by the environment (membrane, binding pocket, or aqueous phase) (Shagufta et al, 2009;Carvalho et al, 2013;Demarco et al, 2018;Perissinotti et al, 2019;Yue et al, 2019).…”