Inhibition of Sodium Ion Channel Function with Truncated Forms of Batrachotoxin

Abstract

A novel family of small molecule inhibitors of voltage-gated sodium channels (Na_vs) based on the structure of batrachotoxin (BTX) – a well-known channel agonist – is described. Protein mutagenesis and electrophysiology experiments reveal the binding site as the inner pore region of the channel, analogous to BTX, alkaloid toxins, and local anesthetics. Homology modeling of the eukaryotic channel based on recent crystallographic analyses of bacterial Na_Vs suggests a mechanism of action for ion conduction block.

Graphical Abstract

Voltage-gated sodium ion channels (Na_Vs) are responsible for the initiation and propagation of action potentials in excitable membranes.^[1] Eukaryotic Na_Vs are comprised of an α-subunit, which assembles as four homologous domains (D1–D4) to form the central ion conduction pore, and two auxiliary β-subunits. Transient changes in membrane voltage potential effect rapid alterations in channel conformation, which are critical for proper ion gating. Biophysical and computational studies to characterize Na_V dynamics have been aided through structural investigations that include protein X-ray crystallography,^[2] mutagenesis experiments,^[3] and ligand binding studies.^[4] The value of small molecules for interrogating Na_V structure-function is considerable, as SAR data in combination with site-directed mutagenesis have shaped current views of the molecular form of the eukaryotic channel structure.^[5] These insights notwithstanding, the ability of small molecule Na_V modulators to alter ion gating kinetics by influencing conformational states of the channel is poorly understood.^[6] Given the intensive interest in Na_Vs as targets for drug design,^[7] deeper insight into the complex molecular interactions between ligand and channel is sought.

Batrachotoxin (BTX), a natural product isolated from poison dart frogs of the genus Phyllobates,^[8] is a potent modulator of Na_Vs.^[9] Binding of this molecule to the inner pore of the channel (Site II) elicits a multitude of functional responses, including a hyperpolarized shift on the voltage-dependence of activation, inhibition of both fast and slow inactivation, and reduction in ion selectivity.^[10] The uniqueness of BTX as a Na_V agonist has motivated our efforts to understand the molecular details of its binding interactions with the channel and its structure-function properties. As part of this program, we have prepared simplified BTX-like structures 1 (Figure 1).^[11] Electrophysiology recordings with wild-type and mutant Na_V isoforms demonstrate that BTX analogues comprising the C, D, and E rings of the natural product block Na_V current – a stark contrast to the behavior of BTX itself. Protein mutagenesis data suggest that BTX analogues lodge in the inner pore lining of the channel, thus sharing a common receptor site with the parent compound. Surprisingly, both enantiomers of 1 display nearly identical potency as Na_V inhibitors.^[12] A homology model of the binding interactions of 1 with the channel pore is presented in an attempt to rationalize the lack of stereorecognition in the receptor site. Collectively, these results establish 1 as a novel chemotype for Na_V inhibition and as a chemical probe for understanding the molecular mechanism of Na_V function.

Figure 1.

C/D/E ring analogues of batrachotoxin (1) act as potent antagonists of voltage-gated sodium ion channels (Na_Vs).

Early structure-activity relationship studies on the effects of semi-synthetic BTX derivatives against Na_Vs highlight the importance of the ester and amine groups in addition to oxygen functional groups at C3, C9, and C11 for toxin function.^[13] Protein mutagenesis experiments,^[3] ligand binding studies,^[4] and competitive ligand displacement assays^[14] give substantive evidence that toxin interaction with Na_V is localized to Site II. Homology modeling and single-point mutagenesis data suggest that the C/D/E ring portion of BTX makes primary contacts with the inner pore lining.^[3f,15] The lack of availability of BTX from natural sources restricts access to new, modified BTX compounds and further examination of this model.^[9c] Investigations of structure-function with BTX analogues and protein mutants can give insight into the unique ability of BTX to influence voltage activation and inactivation through binding the central cavity of the channel. Accordingly, de novo chemical synthesis of the toxin and related structures is the only current available strategy for accessing such probes.^[16]

Truncated forms of BTX comprised of the C/D/E ring units and C20 ester moiety are available in two steps from silyl ether 2 (Scheme 1). Derivatives that contain either the acylpyrrole group found in BTX or those bearing benzoate, naphthoate, cyclohexylcarboxylate, and butyrate are generated in good yields (33–95%) using either acid chloride or mixed-anhydride reaction partners. Coupling of a mixed-anhydride form^[17] of BODIPY® FL to 3 gives a fluorescent conjugate 1f, which was prepared to examine the steric dimensions of the receptor site and for its potential as an imaging tool for ligand displacement assays.^[14] Each of these compounds has been synthesized from 2 in multi-milligram quantities.

Scheme 1. Synthesis of BTX analogue compounds.^a.

^a(a) benzoyl chloride, DMAP, CH₂Cl₂, 95%; (b) 1-naphthoyl chloride, DMAP, CH₂Cl₂, 89%; (c) (ethyl carbonic) 2,4-dimethyl-1H-pyrrole-3-carboxylic anhydride, i-Pr₂NEt, 1,4-dioxane, 90 °C, 72%; (d) cyclohexanecarboxylic(ethyl carbonic) anhydride, 1,4-dioxane, 90 °C, 65%; (e) butyric(ethyl carbonic) anhydride, 1,4-dioxane, 90 °C, 75%; (f) BODIPY^® FL propionic acid, 2-methyl-6-nitrobenzoic anhydride, Et₃N, DMAP, CH₂Cl₂, 33%.

Electrophysiology measurements were performed in a whole-cell voltage-clamp format against the α-subunit of the rat skeletal muscle Na_V (rNa_V1.4), heterologously expressed in Chinese hamster ovarian (CHO) cells. Currents were elicited by 10-ms step depolarizations from a holding potential of −100 to 0 mV and were recorded at steady state. In this assay, a channel modified by BTX, which binds preferentially to the open-state of the channel (use-dependence) exhibits a sustained current.^[3,10] By contrast, application of benzoate 1a or naphthoate 1b (100 μM) shows partial block of Na⁺ current in a use-dependent manner (Figure 2A, data shown for 1b). Measured IC₅₀ values for these two compounds are 81.4 ± 4.8 and 17.4 ± 0.4 μM, respectively (Figures 2B, S1). More surprisingly, experiments performed with either antipode of 1b (i.e., 1b*) demonstrate that inhibition of the channel by these agents is not stereoselective.

Figure 2.

Assay results of BTX analogues against rNav1.4. (A) Use-dependent block of sodium ion current of rNa_V1.4 by 1b (100 μM). Representative traces of superimposed Na⁺ currents are labeled with the number of repetitive pulses. (B) Structure-activity relationship studies on the ester group of the C/D/E core analogues 1a–1f and 3.

Other ester derivatives display similar inhibitory activities to 1a and 1b with the most potent having an IC50 value as low as 11.1 ± 0.3 μM (1c). Notably, block of Na⁺ current is also observed with the sterically large BODIPY® FL conjugate 1f (19.2 ± 2.4 μM). Assuming 1f shares a common receptor with BTX and other BTX analogues, this result suggests that a large open space may comprise the ligand binding site (vide infra). As with BTX, the absence of the ester group dramatically alters the affinity for channel binding.^[13,18] Electrophysiology recordings with alcohol 3 show < 5% block at 1 mM concentrations. Use-dependent block of sodium ion conduction by 1a–1f markedly contrasts the activity of BTX and establishes these molecules as a new class of Na_V inhibitors.

Subsequent investigations of BTX C/D/E analogues were performed with naphthoate 1b given the potency and ease of synthesis of this particular compound. Electrophysiology recordings with different NaV isoforms, which included rNaV1.2, hNaV1.5, and hNaV1.7, give measured IC₅₀ values that vary between 8–30 μM (Figures 3, S2). The similar potency of 1b against channel subtypes is perhaps unsurprising in light of the nearly identical primary structure of the amino acids that are believed to comprise the inner pore.

Figure 3.

Dose-response curve of 1b against selected Na_V isoforms. Normalized data were fit to the Hill equation.

Binding of BTX to Na_V hyperpolarizes the voltage-dependence of channel activation by −45 mV, allowing a greater population of channels to open at resting membrane potentials.^[3,9,10] In addition, BTX is known to inhibit fast inactivation. By contrast, electrophysiology experiments to test the influence of 1b on channel activation and inactivation reveal that this derivative has only a slight influence on the steady-state voltage dependence of activation and inactivation (Figure 4). Binding of 1b (25 μM) to rNa_V1.4 shifts V_0.5 of activation by +4.9 mV and the slope factor is increased by +2.0 mV. For steady-state inactivation, the midpoint voltage is shifted by −6.0 mV and the slope factor is not significantly changed (+0.7 mV).

Figure 4.

Effects on the voltage dependence of (A) activation and (B) inactivation of rNa_V1.4 by 1b (25 μM). The normalized data were fit to a Boltzmann equation.

The obvious discrepancy in activity between BTX and C/D/E ring analogues such as 1b has motivated additional experiments to examine ligand-receptor binding interactions. Amino acid residues that comprise the inner channel pore were selected for initial single-point protein mutagenesis experiments, following previous studies of Site II ligands such as BTX and veratridine^[20] and with the aid of a homology model comprising the S5, S6 and p-loop regions of the channel (Figures 5, S3). Ten amino acids positioned on the S6 helices of domains I–IV were changed to either lysine (K) or alanine (A). All mutant channels have been previously prepared, characterized and tested against BTX.^[3] In most cases, lysine modification (N434K, N784K, F1280K, F1579K, N1584K) results in an increase (~2–10 fold) in the IC₅₀ of naphthoate 1b, consistent with destabilization of ligand binding caused by Coulombic repulsion between the charged lysine and the protonated inhibitor. Alanine mutation at three sites (F1579A, N1584A, Y1586A) also influences the ability of 1b to block channel function, but the overall effect on IC₅₀ is much less pronounced.

Figure 5.

Inhibitory constants for 1b against single-point mutants of rNa_V1.4 (corresponding domains are labeled D1–D4). The effect of 5 μM BTX on wild-type and mutant Na_Vs is shown as s (sensitive) or r (resistant).

With the exception of the tyrosine residue at position 1586, all amino acid mutations shown in Figure 5 have been found to abrogate the effect of BTX on channel function. The marked influence of N434K, N784K, F1280K, N1579K, and N1584K on binding of both 1b and BTX suggests some commonality in the receptor site for these two ligands. Differences between BTX and 1b are noted, however, with two BTX-destabilizing mutations, F1236K and N1584A; against these mutant isoforms, 1b shows a modest increase in affinity from that of wt-Na_V1.4. Given the large volume size of the inner pore as estimated from Na_V homology models, it is possible for BTX and 1b to bind to the open state of the channel in different spatial orientations. The lack of stereoselectivity for inhibition by either antipode of 1b as well as the action of alkaloids such as aconitine and veratridine at Site II are consistent with an inner pore domain that can accommodate sterically large, topologically unique structures. This conclusion is further supported by docking experiments^[20] of 1b with a homology model of the channel pore^[21] constructed from recent X-ray structures of bacterial Na_Vs^[2] (Figures 6, S4, S5). In these poses, both enantiomers of 1b are positioned near residues N434, N784, L1280, F1579 and N1584 and the ammonium group faces the central pore, presumably blocking ion transport through charge repulsion. Future studies will examine these models in greater detail through additional mutagenesis and SAR experiments.

Figure 6.

Enantiomers of 1b docked to a homology model of rNa_v1.4. Top panels: views down the pore axis; Bottom panels: side views (Domain I residues removed for clarity). Domain I = green; Domain II = cyan; Domain III = magenta; Domain IV = yellow. Bottom panels show residues N434, N784, L1280 (not labeled), F1579; N1584.

In sum, we have established that analogue structures of BTX possessing the C/D/E ring elements function as low micromolar inhibitors of Na_V subtypes.^[22] This activity is in striking contrast to the behavior of BTX as a channel agonist. Despite the evident functional differences, it appears from protein mutagenesis experiments that 1b, and by inference related compounds, overlap the binding site of BTX in the inner pore cavity of Na_V. The availability of 1b and associated BTX structures is enabling subsequent investigations to understand how small molecules that bind the channel inner pore regulate ion gating.