With this study we have identified for the first time a cardiac target for BPA: the compound blocks currents through hNav1.5 in a tonic- and use-dependent manner. The modulation profiles of hNav1.5 by BPA and LAs share striking similarities. Both ligand types have a higher affinity for the inactivated state (Kr
), display a shift of steady-state fast inactivation to more hyperpolarized potentials and, in addition to a strong tonic component, both show a strengthening of use-dependent current decline. The effect on steady-state fast inactivation of co-applied BPA and mexiletine () was not additive; the observed shift instead indicated competitive binding. To test the hypothesis that the compounds exert their modulation via interaction with a common receptor site, BPA and mexiletine were tested with the F1760A mutant. This single substitution greatly attenuated both use-dependent and tonic block by BPA (), thus marking another commonality with LAs 
and supporting the prediction of an overlapping binding site in the pore.
F1760 is a critical binding determinant for LAs and engages pore-blockers in two distinct interactions depending on the functional state of Nav
. For use-dependent block of the open-inactivated pore, an aromatic side-chain is required at the F1760 position to satisfy a putative cation-π interaction with the tertiary amine of a LA 
. However, BPA lacks a charge and its interaction with the open-inactivated pore may therefore share similarities with LAs that are neutral at physiological pH e.g. benzocaine, phenytoin or carbamazepine 
. For tonic block, the physicochemical requisite is hydrophobicity – not aromaticity – for interaction at the F1760 position 
, which suggests an alteration in ligand contacts between the open- and closed-state pore in accordance with the modulated receptor hypothesis 
The Kis for mexiletine and BPA are almost the same although mexiletine displays a larger apparent use-dependent block. This may be due to the higher solubility of mexiletine in water, which would allow quicker access to the binding site from the intracellular side when the channel is opening repetitively. Ki, on the other hand, was determined after the cells were held on an inactivating potential for several seconds, rendering time-dependent processes less relevant. 30 µM BPA and 30 µM mexiletine both induce significant use-dependent block (), whereas tonic block with 30 µM mexiletine is almost absent. This also suggests that BPA accesses the binding site in the closed-state channel with comparatively greater ease than mexiletine.
The availability of a bacterial sodium channel crystal structure 
has enabled us to examine the structural basis of tonic block using a hNav1.5 homology model. Automated docking predictions identified a common binding site for both BPA and mexiletine in a hydrophobic cavity atop the F1760 side chain on S6 of domain IV (). This pronounced cavity was not evident in a previous closed-state Nav
model based on the KcsA potassium channel crystal structure 
. F1760 forms significant van der Waals contact with both docked ligands and, in agreement with our electrophysiology data, its mutation to the small side-chain alanine residue would diminish a key underpinning contact. It is the hydrophobic section and not the amine group of mexiletine that contacts F1760, which precludes a cation-π-interaction and is consistent with the hydrophobic but not aromatic requirement at this position for tonic block 
. Therefore this tonic block receptor is distinct from the use-dependent-block binding site for which residue Y1767 was shown to be important 
. Y1767 does not appear to form part of the BPA binding site. Significant structural rearrangements would be necessary to reposition a LA to form a cation-π interaction with F1760 and simultaneously contact Y1767 (e.g. a change in the F1760 side-chain rotamer, 
). In the structural model Y1767 is closer to the cytoplasmic pore opening and may represent the first major binding determinant for substances entering the pore from the intracellular side (use-dependent block). F1760 on the other hand is closer to the side fenestrations, offering a possible binding site for molecules potentially entering via this pathway (tonic block).
BPA is a potent tonic blocker (Kr
58.6±8 µM) and its affinity for the closed-state pore may derive from its considerable hydrophobicity (logP 4.04; Table S1
). The docking predictions identify I1756 as a key hydrophobic contributor to the tonic-block receptor (, located behind the ligands). Mutation of I1756 to alanine produces a modified tonic block, not necessarily due to altered binding, but it seems to ease access to the pore for extracellularly-applied LAs 
. Permanently-charged derivatives of lidocaine are unable to partition completely in the membrane but can still access the closed-state pore of the I1756A mutant. Studies with µ-conotoxin rule out the selectivity filter as the route for this novel entry pathway 
. I1756 is located in the model in the vicinity of the predicted interface between the lipid bilayer, extracellular solution and the DIII–DIV side fenestration. It is possible that substitution to the small side chain alanine enlarges the DIII–DIV side fenestration, such that a charged LA molecule which is semi-partitioned in the lipid bilayer and orientated with the tertiary amine contacting negatively-charged lipid head group(s) may diffuse laterally into the pore, passing I1756A.
We employed the hNav1.5 model to qualitative assess potential routes for BPA or mexiletine ingress using SMD simulations. The force exerted on the mobile ligand during the simulations was used as a descriptor of how well the ligand would fit through the access pathway (). The highly-constricted selectivity filter comprised the path of greatest resistance for both ligands while the pore fenestrations were comparably more accessible. The four fenestrations hindered the crossing of BPA to a similar extent (peaks of ~700–900 pN in the force profiles; ) but in the case of the more flexible mexiletine the fenestration at the DIII–DIV interface proved most accessible (peak of ~425 pN), whereas DI–DII and DII–III showed a higher resistance to mexiletine than to BPA entry. A ligand ingressing the pore via the DIII–DIV fenestration would directly contact F1760 and therefore be in favorable position to occupy the tonic-block receptor. The compatibility of the DIII–DIV fenestration with closed-state ligand ingress when compared with the selectivity filter correlates with results from an analogous study by Bruhova et al. 
. However, this latter study used the KcsA potassium channel structure as template and consequently the pore fenestrations of the Nav
homology model were significantly smaller.
BPA exposure has recently been linked to heart disease 
, raising the question of a cardiac target. The highest reported blood level of BPA is 290 nM 
, which is still lower than 1 µM, the lowest concentration to have a significant effect on hNav1.5 in our study (). The lowest concentration tested at which hNav1.5 currents were not affected was 100 nM (no effect level, NOEL). The mean human blood levels are reported to be ~10 nM 
, just within the tolerable range, according to safety calculations based on these results. There might be a very slight risk that BPA may have adverse effects on cardiac function, especially if patients are preconditioned (e.g. treated with class 1 antiarrhythics). Still, arrhythmias induced by BPA via its effects on Nav1.5 alone do not seem very likely. Our recordings were performed in a heterologous expression system at room temperature, and many factors may influence potential cardiac BPA effects in physiological conditions, such as additional binding proteins and body temperature. We have determined a short term effect of BPA, and potential long term effects of BPA exposure to cardiac function remain to be investigated.
BPA was shown to affect voltage-gated sodium currents in dissociated dorsal root ganglion neurons 
. However, these studies did not investigate BPA action in a defined heterologous expression system or identify a specific binding site.
In this study, we have identified the first cardiac-specific target for BPA and further localized its binding site to the LA receptor in the hNav1.5 pore. Modeling and computational methods provide a basis for understanding BPA-binding interactions in the closed-state channel and assess routes for entry into the pore lumen. Our data may contribute to the on-going debate on BPA safety limits at a time of heightened public concern about BPA exposure.