Cell-Free Expression of Sodium Channel Domains for Pharmacology Studies. Noncanonical Spider Toxin Binding Site in the Second Voltage-Sensing Domain of Human Nav1.4 Channel

Abstract

Voltage-gated sodium (Na_V) channels are essential for the normal functioning of cardiovascular, muscular, and nervous systems. These channels have modular organization; the central pore domain allows current flow and provides ion selectivity, whereas four peripherally located voltage-sensing domains (VSDs-I/IV) are needed for voltage-dependent gating. Mutations in the S4 voltage-sensing segments of VSDs in the skeletal muscle channel Na_V1.4 trigger leak (gating pore) currents and cause hypokalemic and normokalemic periodic paralyses. Previously, we have shown that the gating modifier toxin Hm-3 from the crab spider Heriaeus melloteei binds to the S3-S4 extracellular loop in VSD-I of Na_V1.4 channel and inhibits gating pore currents through the channel with mutations in VSD-I. Here, we report that Hm-3 also inhibits gating pore currents through the same channel with the R675G mutation in VSD-II. To investigate the molecular basis of Hm-3 interaction with VSD-II, we produced the corresponding 554-696 fragment of Na_V1.4 in a continuous exchange cell-free expression system based on the Escherichia coli S30 extract. We then performed a combined nuclear magnetic resonance (NMR) and electron paramagnetic resonance spectroscopy study of isolated VSD-II in zwitterionic dodecylphosphocholine/lauryldimethylamine-N-oxide or dodecylphosphocholine micelles. To speed up the assignment of backbone resonances, five selectively ¹³C,¹⁵N-labeled VSD-II samples were produced in accordance with specially calculated combinatorial scheme. This labeling approach provides assignment for ∼50% of the backbone. Obtained NMR and electron paramagnetic resonance data revealed correct secondary structure, quasi-native VSD-II fold, and enhanced ps-ns timescale dynamics in the micelle-solubilized domain. We modeled the structure of the VSD-II/Hm-3 complex by protein-protein docking involving binding surfaces mapped by NMR. Hm-3 binds to VSDs I and II using different modes. In VSD-II, the protruding ß-hairpin of Hm-3 interacts with the S1-S2 extracellular loop, and the complex is stabilized by ionic interactions between the positively charged toxin residue K24 and the negatively charged channel residues E604 or D607. We suggest that Hm-3 binding to these charged groups inhibits voltage sensor transition to the activated state and blocks the depolarization-activated gating pore currents. Our results indicate that spider toxins represent a useful hit for periodic paralyses therapy development and may have multiple structurally different binding sites within one Na_V molecule.

Keywords: NMR spectroscopy; cell-free expression; channelopathies; combinatorial selective labeling; gating modifier; ligand-receptor interactions; sodium channel.

Figure 1

(A) Transmembrane topology of Na_V channels. The S1-S4 helices are in blue and S5-S6 in gray. Conserved Arg/Lys residues are marked by the plus sign (+). Sites of the VSD mutations associated with different diseases (Simkin, 2011; Nicole and Fontaine, 2015) are marked (HypoPP, NormoPP, HyperPP, hypokalemic, normokalemic, or hyperkalemic periodic paralysis; CMS, congenital myasthenic syndrome). (B) Spatial organization of Na_V channels with one pore domain and four VSDs. Approximate positions of the ligand-binding sites are shown. (C) Alignment of VSD-II and VSD-I of human Na_V1.4 channel with VSDs of other Na_V and K_V channels. Conserved aromatic/hydrophobic, charged, and polar residues are color-coded. TM segments are highlighted in gray. The gating charge transfer center is marked by green asterisks. The sites of conserved charged residues in the S4 helix are numbered. Mutations of R222 (Na_V1.4-DI) and R675 (Na_V1.4-DII) (red and green diamonds) are associated with HypoPP and NormoPP, respectively. The binding sites of spider GMTs are boxed. There are two types of sites with the primary binding interface on the S3b helix (blue) and on the S1-S2 loop (orange). Sites are shown for: Na_V1.4-DII/Hm-3 (present work); Na_V1.4-DI/Hm-3 (Männikkö et al., 2018); Na_V1.7-DII/ProTx2 (Xu et al., 2019); Na_VPaS-DII/Dc1a (Shen et al., 2018); K_VAP/VsTx1 (Lau et al., 2016); and K_V1.2-2.1/hanatoxin (Swartz and MacKinnon, 1997). Two numbering schemes are given: residue numbers in the expressed VSD-II construct starting from the N-terminal Met1 and in the full-length Na_V1.4 channel (in parentheses).

Figure 2

Hm-3 inhibits gating pore currents through Na_V1.4-R675G. The voltage dependence of gating pore currents in absence (black symbols) and presence (red symbols) of 10-µM Hm-3, n = 4. The currents in each cell were normalized to peak negative current in control condition. The insert shows the last 200 ms of the currents in response to 300-ms pulses to voltages ranging from –140 to 50 mV in 5-mV increments in control condition (black traces) and in the presence of 10-µM Hm-3 (red traces).

Figure 3

Characterization of cell-free-produced isolated voltage-sensing domain II of Na_V1.4 channel. (A) 12% SDS-PAGE of purified VSD-II. Lane 1, molecular mass markers; lane 2, ¹³C,¹⁵N-VSD-II after purification by Ni²⁺-chromatography; lane 3, double M25C/S115C VSD-II mutant after labeling with MTSL and Ni²⁺ purification. (B) SDS-PAGE of purified A140C mutant of VSD-K_VAP. Lane 1, molecular mass markers; lanes 2 and 3, protein samples after and before treatment with DTT. Bands corresponding to monomeric and dimeric VSDs are labeled. (C) SEC analysis of partially aggregated VSD-II M25/S115-SL, ¹³C,¹⁵N-VSD-II, empty DPC micelles, VSD-K_VAP A140C, and VSD-K_VAP A140C after treatment with DTT.

Figure 4

Nuclear magnetic resonance spectra and resonance assignment of voltage-sensing domain II. (A) Overlay of ¹H,¹⁵N-TROSY spectra of 30-µM ¹⁵N-labeled VSD-II in DPC/LDAO (11/11 mM, pH 5.5, 45°C, 800 MHz) before (black) and after (red) addition of 150-µM unlabeled Hm-3. Final concentrations: 23-µM VSD-II, 114-µM Hm-3, and detergent to Hm-3 molar ratio of 190:1. The insert on the left panel shows ¹H¹⁵N^ε1 signals of Trp side chains (not assigned). The dashed frame highlights the region expanded in the right panel. The residue numbering scheme corresponds to the expressed VSD-II construct with the N-terminal Met1. The observed ¹H-¹⁵N cross peak of the Met1 residue indicates the presence of unprocessed N-terminal formyl, which is typical for proteins synthesized in CF systems based on bacterial extracts. (B) Labeling pattern for five VSD-II samples (rows in the table) used for backbone resonance assignment. Histidine was unlabeled, and cysteine was not used for the CF synthesis. (C) An example of ¹H-¹⁵N cross peaks assignment. The overlays of 2D TROSY, TROSY-HNCO, and TROSY-HNCA spectra are shown. Absent peaks are marked by rectangles. The degree of ambiguity in the assignment for each of the cross peaks is shown in the insert. The remaining ambiguities were resolved by 3D HNCA and HN(CO)CA spectra measured for uniformly ¹³C,¹⁵N-labeled VSD-II.

Figure 5

Nuclear magnetic resonance data define the secondary structure and backbone dynamics of voltage-sensing domain II in n-dodecylphosphocholine/n-dodecyl-N,N-dimethylamine-N-oxide micelles. Unassigned residues are underlined. Artificially introduced and mutated residues are in gray. The secondary structure in the published cryo-EM structure of Na_V1.4 channel (Pan et al., 2018) is shown by bars and vertical shading. The distorted S12 helix and a fragment of 3₁₀-helix in the S4 segment are shown in light gray. The conserved Arg residues that are responsible for voltage gating and two Lys residues are marked by “+”. (A) Positive values of the Δδ¹³C^α and Δδ¹³C’ secondary chemical shifts (the direction is shown by arrow) indicate backbone helical conformation. (B) Difference in the VSD ¹³C^α, ¹³C’, and ¹H^N chemical shifts between DPC/LDAO and LPPG micelles. Positive and negative values of Δδ¹³C^α _DL and Δδ¹³C’_DL indicate an increase and decrease of helicity, respectively, upon protein transfer from LPPG to DPC/LDAO. (C) Intensity (log values) of peaks in the 3D HNCO spectrum. The level corresponding to the average intensity is shown by a dashed line. Intensities larger than twice the average (blue) reveal residues with high conformational mobility in the ps–ns time scale. Intensities smaller than half the average (red) reveal residues either belonging to the less mobile TM helices or subjected to conformational exchange in the µs–ms time scale. Residues displaying ¹⁵N-{¹H} NOE <0.6 are subjected to high-amplitude motions in the ps–ns time scale. For comparison, ¹⁵N-{¹H} NOE values for VSD-II in LPPG micelles are shown in green.

Figure 6

Electron paramagnetic resonance data reveal partial unfolding of spin-labeled voltage-sensing domain II variants in DPC micelles. (A) X-band CW EPR spectra of the A45/S131-SL and M25/S115-SL VSD-II variants in 20% d ₃₈-DPC and 15% LPPG at 300 K. The dashed line shows a simulated spectrum obtained assuming a mixture of the free spin-label (8%) with the isotropic motion with τ_c = 0.4 ns and of the spin-label attached to the protein (92%) having anisotropic motion with τ_c = 1 ns. (B) Distance measurements in single- and double-spin labeled VSD-II variants in different membrane-mimicking media (see legend above the figure). Background-corrected four-pulse Q-band DEER traces are shown with normalized intensity. Solid lines show the best fits obtained using DeerAnalysis2013. (C). Interspin distance distributions obtained from the analysis of the DEER traces. In all cases, the regularization parameter L is set to 1,000. (D) X-band CW EPR spectra of the A45/S131-SL and A45-SL variants of VSD-II at 140 K. The spectra of the M25/S115-SL variant are identical and are not shown in the figure. (E) Cartoon depicting folded and partially unfolded VSD-II in complex with a DPC micelle.

Figure 7

Nuclear magnetic resonance data define the interface of the Hm-3 interaction with voltage-sensing domain II. (A) Overlay of ¹H,¹⁵N-HSQC spectra of 30-µM ¹⁵N-labeled Hm-3 in 57/57-mM DPC/LDAO (pH 5.5, 45°C, 800 MHz) before (black) and after (red) addition of unlabeled VSD-II. Final concentrations: 20-µM Hm-3, 40-µM VSD-II, and detergent to Hm-3 molar ratio of 5,700:1. The signals demonstrating the biggest changes in the intensity or chemical shifts are marked by ellipses. (B) Overlay of ¹H,¹⁵N-HSQC spectra of 30-µM ¹⁵N-labeled VsTx1 in 45/45-mM DPC/LDAO (pH 5.5, 45°C, 800 MHz) before (black) and after (red) addition of unlabeled VSD-II. Final concentrations: 20-µM VsTx1, 40-µM VSD-II, and detergent to VsTx1 molar ratio of 4,500:1. (C) Relative changes of the ¹H-¹⁵N chemical shifts and intensities of the Hm-3 resonances upon addition of VSD-II (VSD-II/Hm-3 molar ratio of 2:1). Dilution of the Hm-3 sample was accounted for during calculation of the intensity ratio. The threshold levels of 0.015 ppm and 0.6 are shown by dotted lines. Hm-3 residues involved in the interaction with DPC/LDAO micelle are highlighted in gray. (D) The interfaces of the Hm-3 interaction with the micelle (blue) and VSD-II (orange) are mapped on the Hm-3 structure. Gray mesh shows approximate micelle surface with a radius of ∼24 Å.

Figure 8

Nuclear magnetic resonance data define the interfaces of voltage-sensing domain II interaction with Hm-3. (A) Relative changes in the ¹H-¹⁵N chemical shift and in the intensity of VSD-II resonances upon addition of Hm-3 (pH 5.5, 45°C, 800 MHz). Final concentrations: 23-µM VSD-II, 114-µM Hm-3, and 11/11-mM DPC/LDAO. Dilution of the VSD-II sample was accounted for during calculation of the intensity ratio. The threshold levels of 0.08 ppm and 0.6 are shown by dotted lines. (B) The interface of VSD-II interaction with Hm-3 is mapped on the structure of VSD-II. The residues forming the extracellular and cytoplasmic Hm-3 binding sites are shown in red and magenta, respectively. The side chains forming the extracellular interaction interface are annotated. The conserved Arg/Lys residues of the S4 helix, negatively charged Asp/Glu residues, and conserved F70(621) residue are also shown. The residue numbering scheme corresponds to the expressed VSD-II construct with the N-terminal Met1. (C) Fragment of the ¹H,¹⁵N-TROSY spectra of ¹⁵N-labeled VSD-II at different Hm-3 concentrations. (D) The binding curves of Hm-3 to VSD-II in DPC/LDAO micelles measured for F28(579) and N57(608) residues (cytoplasmic and extracellular binding sites, respectively).

Figure 9

Molecular modeling of Hm-3 in complex with voltage-sensing domain II. (A) RMSF of VSD-II calculated over the “stable” part of the MD trajectory. Secondary structure of the Na_V1.4 cryo-EM structure is shown (Pan et al., 2018). Distorted helices are in light gray. (B) RMSF of Hm-3 calculated over the “stable” parts of the MD trajectories (shown is the spread for three independent runs). The toxin secondary structure is shown above; the dotted lines indicate the positions of Cys residues. (C) Representative docking solutions of the VSD-II/Hm-3 complex after filtration. The backbone of the fragments, which according to the NMR data form the interaction interfaces, is colored in red (VSD-II) and orange (Hm-3). The charged side chains at the interaction interfaces and in the TM part of VSD-II are shown. The VSD gating charge transfer center (residue F70) and aromatic side chains at the toxin membrane-binding face are also shown. VSD-II residues are in italics type. The residues participating in intermolecular ionic interactions are underlined. Dashed arches in the lower panels show an approximate micelle surface with the radius of ∼24 Å. The upper panels show the position of Hm-3 relative to the full-length α-subunit of Na_V1.4 (reconstruction based on docking data).