Dissection of the structure–function relationship of Na_v channels

Significance

Na_vchannels are of tremendous physiological, pathophysiological, and pharmacological significance for their primary function in initiating and transmitting electrical signals. Structural determination of Na_vchannels in different working states is critical for an advanced understanding of their working mechanism and instrumental to rational drug development. In this study, we continue our efforts in dissecting the structure–function relationship of Na_vchannels. We report the structures of three human Na_v1.7 mutants that contain different sets of point mutations. Comparative electrophysiological and structural analyses reveal a correlation of the right shift of the activation and static inactivation I–V curves with the contraction of the pore domain, which is featured with the π form helical turn in the S6 segments.

Abstract

Voltage-gated sodium channels (Na_v) undergo conformational shifts in response to membrane potential changes, a mechanism known as the electromechanical coupling. To delineate the structure–function relationship of human Na_v channels, we have performed systematic structural analysis using human Na_v1.7 as a prototype. Guided by the structural differences between wild-type (WT) Na_v1.7 and an eleven mutation-containing variant, designated Na_v1.7-M11, we generated three additional intermediate mutants and solved their structures at overall resolutions of 2.9–3.4 Å. The mutant with nine-point mutations in the pore domain (PD), named Na_v1.7-M9, has a reduced cavity volume and a sealed gate, with all voltage-sensing domains (VSDs) remaining up. Structural comparison of WT and Na_v1.7-M9 pinpoints two residues that may be critical to the tightening of the PD. However, the variant containing these two mutations, Na_v1.7-M2, or even in combination with two additional mutations in the VSDs, named Na_v1.7-M4, failed to tighten the PD. Our structural analysis reveals a tendency of PD contraction correlated with the right shift of the static inactivation I–V curves. We predict that the channel in the resting state should have a “tight” PD with down VSDs.

Eukaryotic voltage-gated sodium (Na_v) channels govern membrane excitability by initiating and propagating electrical signals in the form of action potentials (1–4). At resting state, the membrane of excitable cells, like neurons and myocytes, is polarized with a negative potential. Na_v channels in the resting state are nonconductive. They activate in response to membrane depolarization (5). A critical mechanism to ensure repetitive firing of action potentials is the rapid inactivation of Na_v channels, which occurs within one or several milliseconds after activation (6, 7). Inactivated channels return to the resting state after membrane hyperpolarization to prepare for the next working cycle (5, 8).

Because of their foundational significance in physiology, aberrant function or regulation of Na_v channels is associated with a myriad of pathogenic conditions, such as epileptic seizure, arrhythmia, and pain disorders (9–12). Consequently, Na_v channels represent a major class of drug targets (13–15). Structural resolution of Na_v channels in different functional states is not only critical for an advanced understanding of the fundamental physiological process of electrical signaling but also instrumental to the development of Na_v-targeting drugs (16, 17).

Na_v channels, like other voltage-gated cation channels, comprise two modules, the central pore domain (PD), which is responsible for selective and gated ion permeation, and the voltage-sensing domains (VSDs) that detect the changes of membrane potential and undergo conformational shifts accordingly (18). It is generally accepted that a Na_v channel in the resting state has a closed PD with four “down” VSDs, in which the gating charge residues on the S4 segment face the intracellular side (4, 13). Activation of a Na_v channel refers to pore opening induced by the upward movement of VSDs (19, 20). The concerted movements of VSDs with pore gating are known as the electromechanical coupling (21). From a physical perspective, a Na_v channel is in essence an ensemble of multiple conformers, each of which may be specifically recognized by a modulator.

Up to date, the structures of eight subtypes of human Na_v channels, Na_v1.1–Na_v1.8 have been reported (22–29). The only one whose structure is still at large is Na_v1.9. However, most of the reported structures represent a similarly inactivated conformation, characteristic of non-conductive PD and four up VSDs. The major challenge in the structural biology of Na_v channels is to unveil conformations associated with distinct states during the electromechanical coupling. Toward this goal, we have been using human Na_v1.7 as a scaffold and exploited multiple strategies, such as introduction of point mutations and inclusion of state-dependent toxins or drugs (28, 30–33).

One of our recent attempts was to rationally substitute some of the Na_v1.7 residues with the counterparts from Na_vPaS, which is the first eukaryotic Na_v channel whose atomic structure was determined (34). Compared to human Na_v channels, the structure of Na_vPaS, an isoform from American cockroach, is distinct from the human ones, with a more contracted PD and the VSDs in intermediate states (34). When nine loci on the PD of Na_v1.7 were replaced by residues from Na_vPaS, the resulting variant, named Na_v1.7-M9 (short as M9), showed right-shifted I–V curves for both activation and static state inactivation (SI Appendix, Fig. S1 and Table S1) (31). When combined with two additional mutations in VSD_I and VSD_II, the right shifts of these I–V curves of the variant, Na_v1.7-M11, were more dramatic (Fig. 1 A and B) (31). Based on these I–V curves, we predicted that the structure of purified M11, in the absence of any membrane potential, might represent a closed-state inactivated (CSI) conformation. Indeed, the structure of M11-class I deviates substantially from that of the wild-type channel (Na_v1.7-WT), characteristic of a tightly closed PD and a completely down VSD in the first repeat (VSD_I) (Fig. 1C). In addition, VSD_II moves toward the intracellular side by one gating charge (31). For simplicity, unless otherwise specified, Na_v1.7-M11 or M11 will refer to the structure of M11-class I.

Fig. 1.

An overview of our rational design of Na_v1.7 variants for structure–function relationship study. (A) A diagram illustrating the distribution of 11-point mutations on human Na_v1.7 (M11). The nine loci on the pore-forming segments are substituted with corresponding residues from Na_vPaS. The mutant that contains these nine mutations is named M9. (B) Right shift of the activation and static inactivation I–V curves of Na_v1.7 variants. This panel is adapted from Huang et al. (31). (C) Substantial conformational changes in the PD, VSD_I, and VSD_II between M11 and wild type (WT). Structures of M11 and WT are superimposed relative to the α subunit. Red arrows indicate the shifts of VSD_I and the amino terminal domain (NTD) from WT to M11. PDB codes for M11 and WT are 7XVE and 7W9K, respectively. Insets: Structural mapping of the 11 mutations on Na_v1.7-M11. The Cα atoms of the 11 loci are highlighted as spheres. (D) Na_v1.7-M9 exhibits an intermediate conformation between M11 and WT. Left: The PD of M9 is similar to that of M11. Right: VSD_I in M9 resembles that in WT. M9 is domain colored, and M11 (Left) and WT (Right) are colored green and gray, respectively. (E) Contracted PD in M9 and M11. The pore radii of M9 (orange), M11 (green) and WT (gray) were calculated in HOLE (35) and tabulated. Structural figures were prepared in ChimeraX (36) or PyMol (37).

In this study, we attempted to dissect the contributions of the eleven point mutations to the structural changes between WT and M11. In addition to M9, we generated two additional variants, Na_v1.7-M2 that contains two PD mutations and its derivative Na_v1.7-M4 that has the two additional VSD mutations as in M11. Our systematic structural examination of these Na_v1.7 mutants affords an updated view of the structure–function relationship of Na_v channels. These structures highlight the conformational complexity of the PD associated with non-conductive states.

Results

Cryo-EM Analysis Na_v1.7-M9.

Following our established protocols (31), we obtained a three-dimensional (3D) electron microscopy (EM) reconstruction of Na_v1.7-M9 in complex with the β1 subunit at overall resolution of 2.9 Å, using 483,417 particles (SI Appendix, Figs. S2–S4 and Table S2). Structural comparison of M9 with WT (PDB code: 7W9K) and M11 (PDB code: 7XVE) reveals that the extracellular loops (ECLs) of PD, VSD_III, and VSD_IV align consistently across all three structures (30, 31). Yet, the conformation of M9 lies in an intermediate state between WT and M11. While the PD in M9 is nearly identical to that in M11, VSD_I and VSD_II of M9 resemble those in WT (Fig. 1D). In this regard, the overall conformation of M9 is mostly similar to the class II conformation of M11, except that its VSD_II is fully up (SI Appendix, Fig. S5).

Permeation path calculations in HOLE (35) reveal that the radius of the constriction point along the intracellular gate of M9 is nearly identical to that of M11, approximately 1.5 Å narrower than that of WT (Fig. 1E). The contracted PD in M9 is consistent with our previous analysis that introduction of bulkier hydrophobic residues in fenestrations increases the energy barrier to open the pore as it is more difficult to disrupt the interactions between adjacent pore-forming segments (31). This observation is consistent with the right shift of the I–V curve for M9 activation (Fig. 1B).

Structural Variations between M9 and WT.

As we have previously elaborated the detailed structural differences between M11 and WT (31), we will mainly focus on the unique structural variations between M9 and WT to avoid redundancy.

Despite overall structural similarities of the VSDs between M9 and WT, there are slight conformational deviations in VSD_I and VSD_II. VSD_I in M9 appears to be up, as in contrast to the completely inactivated conformation of that in M11. However, the S4_I segment in M9 shifts toward the intracellular side by less than a half turn compared to WT (Fig. 2A and SI Appendix, Fig. S6). VSD_II in M9 is completely up; the S4_II segment is even slightly above that in WT (Fig. 2A).

Fig. 2.

Structural differences between Na_v1.7-M9 and Na_v1.7-WT. (A) Minor structural shifts of VSD_I and VSD_II between M9 and WT. Despite the marked structural distinctions in the PD, the VSDs remain largely the same between M9 and WT. (B) Structural deviations of the PD between M9 and WT. Top: The S6_II and S6_IV segments undergo α to π transition from WT to M9. Red arrows indicate the coupled motions of S4-5_II, S6_II, and S6_III. Bottom: Concerted movements of S4-5_I with adjacent S4-5_II and S5_II. These conformational changes are similar to those between M11 and WT (31).

The tight PD and the up VSDs in M9 place the S4-5 segments in repeats I and II in conformations different from those in WT or M11 (Fig. 2B). The S4-5_I segment moves closer to the central axis in M9, accompanying the centripetal movement of the ensuing S5_I segment. The most dramatic conformational changes occur in S4-5_II and the lower half of S5_II (Fig. 2). In WT, S4-5_II and S5_II are connected by an unwound sharp turn (the elbow). In M9, the elbow transforms to a curved helix, resulting in a flattened S4-5_II segment (the forearm).

Similar to what has been observed between M11 and WT (31), the movement of S4-5_II toward the central axis, combined with the substitution of L866F on S5_II, results in concerted conformational shifts in several PD segments to avoid steric clashes. The S4-5_I and S4-5_II undergo coupled movements, as the distance between the amino termini of S4-5_I and S5_II in M9 remains nearly unchanged compared to that in WT. These structural changes result in the α to π helix transition in the middle of S6_II (Fig. 2B).

Examination of Key Residues on the Conformational Shifts of the PD.

Structural comparison between M9 and WT confirms that the nine mutations of Na_v1.7 are sufficient to lock the PD in a tight conformation. Scrutiny of these loci show that mutations of residues on the upper half of the PD do not change the backbone positions, while those on the lower region, exemplified by L866F, T870M, and G1454C, undergo marked shifts (Fig. 3A). We then examined the effect of these mutations on the biophysical properties of Na_v1.7.

Fig. 3.

The conformation of two Na_v1.7 variants remains the same as WT. (A) Rationale to reduce the number of mutations in the PD. Among the nine mutations in M9, the backbone atoms of L866F, T870M, and G1454C exhibit structural variations, which are labeled in magenta. (B) Electrophysiological characterizations of additional Na_v1.7 variants. The activation and steady-state inactivation curves of indicated Na_v1.7 variants are shown. Data points are presented as mean ± SEM and n is the number of recorded cells. Please refer to Materials and Methods and SI Appendix, Table S1 for details. Na_v1.7-M2 contains two-point mutations, L866F and G1454C. When two additional mutations, E156K on S2_I and G779R on S2_II, were introduced to M2, the resulting mutant, named Na_v1.7-M4 could not be recorded despite our extensive efforts. (C) Structures of M2 and M4 are similar to that of WT. The three structures are superimposed relative to the α subunit. Insets: The backbone atoms of the four substituted residues in M4 remain unchanged to those in WT.

Single-point mutations L866F or G1454C both led to right shift in the voltage dependence of activation, with the V_1/2 of activation curves shifting from −25.59 ± 0.22 mV for WT to −14.62 ± 0.43 mV, and −13.55 ± 0.37 mV, respectively (Fig. 3B and SI Appendix, Table S1). In contrast, T870M induced a left shift, resulting in the V_1/2 of activation curve to −32.52 ± 0.31 mV (Fig. 3B and SI Appendix, Table S1). When L866F and G1454C were combined, named Na_v1.7-M2, the activation curve was further shifted rightward with the V_1/2 of activation reaching −3.30 ± 0.51 mV, which is close to 0.55 ± 0.43 mV for Na_v1.7-M9, but with a less steep slope (Fig. 3B and SI Appendix, Table S1). The voltage dependence of the static inactivation curve of M2 is nearly identical with that of WT (Fig. 3B and SI Appendix, Table S1). Being curious whether the conformation of M2 resembles that of M9 or WT, we set out to solve the structure of Na_v1.7-M2.

A 3D EM reconstruction of M2 was obtained with the overall resolution at 3.4 Å, using merely 77,511 particles (SI Appendix, Fig. S3 and Table S2). The structure of M2 is nearly identical with that of WT, with the RMSD of 0.5 Å over 1,181 Cα atoms of α subunit (Fig. 3C).

We then examined whether inclusion of E156K and G779R to M2, the resulting mutant named Na_v1.7-M4, would alter the conformation, as these two VSD mutations induce marked structural shifts of VSD_I and VSD_II from M9 to M11 (SI Appendix, Fig. S6). However, M4 could not be recorded using whole-cell patch clamp in HEK293T cells despite extensive trials. Nevertheless, we proceeded with cryo-EM analysis of M4 and determined its structure to an overall resolution of 3.4 Å, using 181,009 particles (SI Appendix, Fig. S3 and Table S2). The structure of M4 is also nearly identical to WT, or M2 (Fig. 3C). Of note, the backbone of the four substituted loci, E156K, G779R, L866F, and G1454C in M4 completely overlap with those in WT (Fig. 3C).

Structural Heterogeneity in Na_v1.7-Variants.

The structural similarity of M2 and M4 with WT suggests that the other seven loci on the PD, although sharing identical backbones in all the Na_v1.7 variants, should have contributed to the structural malleability of the PD. If that is the case, we reasoned that the altered structural plasticity should be reflected from the homo- or heterogeneity of the particles in the cryo-samples.

It is noted that we recombinantly expressed, purified, and prepared cryo-EM samples for Na_v1.7-WT, M2, M4, M9, and M11, following an identical protocol. The particles for WT are quite homogeneous. The majority of the particles was selected from the 2D classification to yield a 2.2 Å 3D reconstruction (30). In contrast, two major classes were obtained for M11, representing only 18.7% and 29.6% of the total 2D classified particles (31). M9 was also quite heterogeneous, with only 18.3% particles were selected from the 2D classification for one major 3D reconstruction. We did not collect a large number of micrographs for M2 or M4, as the preliminary assessment suggested they were also rather homogeneous, with 48.5% and 54.9% particles selected for final 3D reconstruction, respectively (Fig. 4A). These observations support our speculation that the seven point mutations confer enhanced structural flexibility in the absence of any membrane potential.

Fig. 4.

Increased conformational heterogeneity of Na_v1.7-M9 and Na_v1.7-M11. (A) The percentiles of selected particles in the final 3D reconstruction may indicate the conformational heterogeneity. M9 and M11 appear to be more heterogeneous than M2 and M4. The EM maps were contoured at 5~7 σ in ChimeraX. (B) Growing propensity for α to π transition in S6_IV with increasing number of mutations. Potential conformational heterogeneity, reflected by compromised local resolutions, are indicated with short red arrows. The curved arrows in M9 and M11 indicate the rotation of S6_IV as a result of α to π transition in the middle. The densities were prepared in ChimeraX and contoured at 6 σ.

Furthermore, scrutiny of the 3D EM map shows that increasing number of mutations is correlated with the α → π transition of the S6 segments, or in another word, the tightening of the PD (Fig. 4B). In WT, both S6_II and S6_IV are made of α helical turns only. In M9 and M11, both have the α → π transition in the middle helical turn. When it comes to M2 and M4, although we modeled S6_IV in the form of α helical turns, the ambiguity of the local density suggests the existence of another conformer, as seen in the 3D reconstruction of Na_v1.7 in the presence of the gating modifier toxins Huwentoxin IV (HWTX-IV) or Protoxin II (30) (Fig. 4B).

Discussion

In this study, we systematically compared the electrophysiological properties and the cryo-EM analysis of five Na_v1.7 variants, WT, M2, M4, M9, and M11. In the WT channel, the PD has four fenestrations and a sterol-like density penetrating the intracellular gate. We assigned a glycol-diosgenin (GDN) molecule into this density, as it is the primary detergent used for protein purification. Although the gate radius at the constriction point is between 2.5 and 3 Å, it is not wide enough to permeate hydrated Na⁺ ions. We describe this conformation as the relaxed PD (Fig. 5 A and B). In M9 and M11, S6_II and S6_IV undergo α to π helical transition, leading to reduced volume of the central cavity, diminished or shrunk fenestration, and sealed intracellular gate that can no longer accommodate a lipid. We call it a tight PD. With regard to M2 and M4, despite similar conformation to WT, assigning GDN was challenging due to the shorter density (SI Appendix, Fig. S4B). Although the dominant density for S6_IV prefers the assignment of the α helix, a small portion may exist as the π form, a conformer that was resolved for Na_v1.7 bound with HWTX-IV and saxitoxin, referred to as Na_v1.7-HS-π (30). Notably, the structure of Na_v1.7-HS-π has the π helical turn in S6_IV but α helix in S6_II. Its PD represents an intermediate state between the relaxed and the tight forms (Fig. 5B). In fact, the conformational complexity for non-conductive PD has also been discussed for voltage-gated Ca²⁺ (Ca_v) channels. In addition to the relaxed and tight states, which are similar to the corresponding ones discussed here, a loose state with even more dilated PD is observed in human Ca_v2.1 bound to a blocker pinaverium (38).

Fig. 5.

Structure–function relationship of the PD of Na_v channels. (A) A diagram illustrating three resolved states of PD: relaxed, intermediate, and tight. There is a correlation of the cavity volume and the number of π-form S6 segments, as seen in panel B. S6_II and S6_IV are α helices in relaxed PD, and a sterol-like molecule penetrates the intracellular gate. There are four fenestrations in the relaxed PD. S6_IV in the intermediate PD undergoes α to π helix transition, leading to the closure of I:IV fenestration. In the tight PD, both S6_II and S6_IV exhibit α to π helix transition, resulting in a completely sealed intracellular gate. (B) Non-conductive PD can correspond to multiple conformations. The corresponding pore radii of Na_v1.7-M9 (orange), M11 (green), WT (gray), and HS-π (Na_v1.7 in complex with HWTX-IV and saxitoxin, PDB code: 7W9P, purple) are compared, suggesting multiple conformations of PD. (C) Predicted electrophysiological properties for Na_v variants to be structurally resolved in the resting state using standard protocols. For an Na_v channel to stably stay in the resting state in the absence of membrane potential, its activation and steady-state inactivation thresholds should be both beyond 0 mV, as indicated by the dashed lines.

A challenge in the structural characterizations of voltage-gated ion channels (VGICs) is to assign the functional states to the structures. Before the complete working cycle of a VGIC is structurally resolved, it is difficult to unambiguously establish the structure– function relationship of a particular conformer. Our systematic electrophysiological and cryo-EM characterizations of rationally designed Na_v1.7 mutants shed light on the structural interpretation of the I–V curves for voltage-dependent activation and static-inactivation.

The activation curve reflects the voltage-dependent equilibrium between the resting and activated states, while the static inactivation curve indicates that between the resting and the inactivated states. Taking Na_v1.7 as an example, the WT channel can fully activate upon membrane depolarization to 0 mV; however, after prolonged duration, all the channels are trapped in the inactivated state (Figs. 1B and 5C). The structure obtained under such condition is expected to represent an open-state inactivated (OSI) conformation, featured with the relaxed PD and all up VSDs. In the case of M11, a depolarization to 0 mV is not sufficient to open the channel, yet the majority can be trapped in the inactivated state. Therefore, the structure, which is characteristic of a tight PD with at least two fully up VSDs, may represent the CSI conformation.

Along this line of thinking, additional mutations may help capture the structure of a resting-state Na_v, although the activated state is always transient due to the fast inactivation mechanism. The variant to be trapped in the resting state in the absence of membrane potential should have the right-shifted activation curve as in M11 and, more importantly, a further right-shifted static inactivation curve that has 100% open capacity at 0 mV (Fig. 5C). As the tight PD cannot possibly further contract, we predict that the resting-state structure should have a tight PD conformation as seen in M9 and M11. The S4 segments in the VSDs should have extensive interactions with the pore-forming segments. The interface between the fully deactivated VSD_I and the segments in the tight PD in M11 may shed light on the modeling of a bona fide resting state structure.

Materials and Methods

Whole-Cell Electrophysiology.

The codon-optimized cDNAs of wild-type Na_v1.7 (Na_v1.7-WT, UniProt: Q15858) and Na_v1.7-M9 (L866F, T870M, A874F, V947F, M952F, V953F, V1438I, V1439F, and G1454C) were synthesized (BGI Geneland Scientific Co., Ltd) and subcloned into the pCAG vector with twin Strep-tag and FLAG tag in tandem at the amino terminus (39). Na_v1.7-M11 was generated by introducing two additional VSD mutations (E156K and G779R) to Na_v1.7-M9. Na_v1.7-M2 was generated by introducing two mutations (L866F, and G1454C) to Na_v1.7-WT. Na_v1.7-M4 was generated by introducing two additional VSD mutations (E156K and G779R) to Na_v1.7-M2. All of these variants were generated using a standard One-step Quick Change cloning. All plasmids were validated by DNA sequencing.

For whole-cell patch-clamp recordings, the overall methods were similar to our previously reported protocols (31). The HEK293T cells (Invitrogen) were cultured in Dulbecco’s Modified Eagle Medium (DMEM, BI) supplemented with 4.5 mg/mL glucose and 10% (v/v) fetal bovine serum (FBS, BI). Cells for recordings were plated onto glass coverslips and transiently co-transfected with the expression plasmid for Na_v1.7-variants and an eGFP-encoding plasmid using lipofectamine 2000 (Invitrogen). Cells with green fluorescence were selected for patch-clamp recording 18 to 36 h after transfection. All experiments were performed at room temperature. No further authentication was performed for the commercially available cell line. Mycoplasma contamination was not tested.

The whole-cell Na⁺ currents were recorded using an EPC10-USB amplifier with Patchmaster software v2*90.2 (HEKA Elektronic), filtered at 3 kHz (low-pass Bessel filter) and sampled at 50 kHz. The borosilicate pipettes (Sutter Instrument) had a resistance of 2 to 4 MΩ and the electrodes were filled with the internal solution composed of (in mM) 105 CsF, 40 CsCl, 10 NaCl, 10 EGTA, 10 HEPES, pH 7.4 with CsOH. The bath solutions contained (in mM) 140 NaCl, 4 KCl, 10 HEPES, 10 D-glucose, 1 MgCl₂, 1.5 CaCl₂, pH 7.4 with NaOH.

The voltage dependence activation was recorded by using a protocol consisting of steps from a holding potential of −120 mV (for 200 ms) to voltages ranging from −90 mV to +80 mV for 50 ms in 5 mV increment. In the activation property analysis, we used the equation, G = I/(V − Vr), where Vr (the reversal potential) represents the voltage at which the current is zero. For the activation curves, conductance (G) was normalized and plotted against the voltage from −90 mV to +20 mV ~ +60 mV as appropriate. For voltage dependence inactivation, cells were clamped at a holding potential of −90 mV and were applied to step pre-pulses from −130 mV or −120 mV to 0 mV for 50 ms with an increment of 5 mV. Then, the Na⁺ currents were recorded at the test pulse of 0 mV for 50 ms. The peak currents under the test pulses were normalized and plotted against the pre-pulse voltage. The steady-state inactivation curves were fitted with a Boltzmann function to obtain V_1/2 and slope values. The linear component of leaky currents and capacitive transients were subtracted using the P/4 procedure. Data were analyzed using Origin (OriginLab) and GraphPad Prism (GraphPad Software).

Recombinant Co-Expression and Purification of the Human Na_v1.7-Variants.

For protein expression, purification of human Na_v1.7-variants, as well as the subsequent data processing, was carried out following our optimized protocols (31). The Na_v1.7-variants were coexpressed with the auxiliary subunits β1 and/or β2 following an identical protocol as reported (28). HEK293F cells (Invitrogen) were cultured in SMM 293 T-II medium (Sino Biological Inc.) under 5% CO₂ in a Multitron-Pro shaker (Infors, 130 r.p.m.) at 37 °C and transfected with plasmids when the cell density reached 2.0 × 10⁶ cells per mL. For 1 L cell culture, 1.5 mg for Na_v1.7-variants plus 0.5 mg for β1 and/or 0.5 mg for β2 were preincubated with 4 mg 25-kDa linear polyethylenimines (PEIs, Polysciences) in 25 mL fresh medium for 20 min before adding into cell culture. Transfected cells were cultured for 48 h before harvesting.

Protein purification was carried out following our optimized protocol (28). Twenty L cells co-transfected with Na_v1.7-variants plus β subunits were used for one batch of sample preparation. They were harvested by centrifugation at 800×g and resuspended in the lysis buffer containing 25 mM Tris-HCl (pH 7.5) and 150 mM NaCl. The suspension was supplemented with 1% (w/v) n-dodecyl-β-D-maltopyranoside (DDM, Anatrace), 0.1% (w/v) cholesteryl hemisuccinate Tris salt (CHS, Anatrace), and protease inhibitor cocktail containing 2 mM phenylmethylsulfonyl fluoride (PMSF), aprotinin (6.5 μg/mL), pepstatin (3.5 μg/mL), and leupeptin (25 μg/mL). After incubation at 4 °C for 2 h, the cell lysate was ultra-centrifuged at 20,000×g for 45 min, and the supernatant was applied to anti-Flag M2 affinity gel (Sigma) by gravity at 4 °C. The resin was rinsed four times with the wash buffer (buffer W) containing 25 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.05% glycol-diosgenin (GDN, Anatrace), and the protease inhibitor cocktail. Proteins were eluted with buffer W plus 200 μg/mL FLAG peptide (Sigma). The eluent was then applied to Strep-Tactin Sepharose (IBA) and the purification protocol was similar to the previous steps except that the elution buffer was buffer W plus 2.5 mM D-Desthiobiotin (IBA). It was then concentrated using a 100-kDa cut-off Centricon (Millipore) and further purified through Superose-6 increase column (GE Healthcare) in buffer W. The peak fractions were pooled and concentrated to ~1 mg/mL, and then conducted cryo-EM sample preparation.

Cryo-EM Data Acquisition.

Aliquots of 3.5 μL freshly purified Na_v1.7-variants were placed on glow-discharged holey carbon grids (Quantifoil Au 300 mesh, R1.2/1.3). Grids were blotted for 3.0 s and plunge frozen in liquid ethane cooled by liquid nitrogen with Vitrobot Mark IV (Thermo Fisher).

For Na_v1.7-M2, electron micrographs were acquired on a Titan Krios electron microscope (Thermo Fisher) operating at 300 kV and equipped with a Gatan K3 Summit detector and GIF Quantum energy filter. A total of 5,106 movie stacks were automatically collected using AutoEMation (40) with a slit width of 20 eV on the energy filter and a preset defocus ranging from -1.8 µm to -1.5 µm in super-resolution mode at a nominal magnification of 105,000 X. Each stack was exposed for 2.56 s with 0.08 s per frame, resulting in 32 frames per stack.

For Na_v1.7-M4 and Na_v1.7-M9, complex, electron micrographs were acquired on a Titan Krios electron microscope (Thermo Fisher) operating at 300 kV and equipped with Gatan K3 Summit detector and GIF Quantum energy filter. 7468/5167 movie stacks were automatically collected using AutoEMation (40) with a slit width of 20 eV on the energy filter and a preset defocus ranging from −1.8 µm to −1.5 µm in super-resolution mode at a nominal magnification of 64,000 X. Each stack was exposed for 2.56 s with 0.08 s per frame, resulting in 32 frames per stack.

The total dose was 50 e⁻/Ų for all the datasets. The stacks were motion-corrected with MotionCor2 (41) or Warp (42) and binned twofold, resulting in 0.8374 Å/pixel for Na_v1.7-M2, and 1.0979 Å/pixel for Na_v1.7-M4 and Na_v1.7-M9. Dose weighting was performed as described (43). Defocus values were estimated using Gctf (44) or cryoSPARC (45).

Image Processing.

For simplicity, the data processing of Na_v1.7-M9 was selected as representative with a diagram showed in SI Appendix Fig. S2. A total of 4,400,497 particles were automatically picked using template picking in cryoSPARC. After 2D classification, a total of 2,634,558 good particles were selected, these particles were subject to one round of Non-uniform Refinement, resulting in 3D reconstructions at 2.89 Å. The resulting particles were subject to one additional round of Heterogeneous Refinement in cryoSPARC, with the number of classes set to 10. A total of 483,417 particles were selected and applied to another round of Non-uniform Refinement and CTF refinement, resulting in final reconstructions at overall resolutions of 2.89 Å (SI Appendix, Fig. S3 and Table S2). Other datasets were processed with the same workflow. Briefly, a total of 6,398,495 and 11,394,867 particles of Na_v1.7-M2 and Na_v1.7-M4 were automatically picked, respectively. After 2D classification, 159,730 and 329,563 good particles were selected for Na_v1.7-M2 and Na_v1.7-M4. After a round of Non-uniform Refinement, Heterogenous Refinement and additional Non-uniform Refinement, the final reconstructions of Na_v1.7-M2 and Na_v1.7-M4 were reconstituted at overall resolution of 3.37 and 3.39 Å using 77,511 and 181,009 particles, respectively.

Model Building and Structure Refinement.

Model building was performed based on the map acquired in the data processing. The starting model of Na_v1.7 (PDB code: 7W9K) (30), was fitted into the EM map by Chimera (46). All Na_v1.7 residues, lipids, and sugar moieties were manually checked in COOT (47). The chemical properties of amino acids were taken into consideration during model building.

Structural refinement was performed using phenix.real_space_refine application in PHENIX (48) with secondary structure and geometry restraints. Over-fitting of the overall model was monitored by refining the model in one of the two independent maps from the gold-standard refinement approach and testing the refined model against the other map (49). Statistics of the map reconstruction and model refinement can be found in SI Appendix Table S2.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

The atomic coordinates of Na_v1.7-variants have been deposited in the Protein Data Bank (http://www.rcsb.org) under the accession codes 8XMN (Na_v1.7-M2) (50), 8XMO (Na_v1.7-M4) (51), and 8XMM (Na_v1.7-M9) (52). The corresponding electron microscopy maps have been deposited in the Electron Microscopy Data Bank (http://www.ebi.ac.uk/pdbe/emdb/) with the accession codes EMD-38483 (53), EMD-38484 (54), and EMD-38482 (55), respectively. All other data are included in the manuscript and/or SI Appendix.