Structure of human Na_v1.5 reveals the fast inactivation-related segments as a mutational hotspot for the long QT syndrome

Significance

Dysfunction of Na_v1.5, the primary cardiac Na_v channel, is associated with multiple arrhythmia syndromes, exemplified by type 3 long QT syndrome (LQT3) and Brugada syndrome (BrS). Establishment of the structure-function relationship and mechanistic understanding of the disease variants will facilitate the development of antiarrhythmic drugs. Here we report the cryo-EM structure of human Na_v1.5-E1784K, the most common variant shared by LQT3 and BrS. Structural mapping of 91 LQT3-associated mutations reveal a hotspot that involves the fast inactivation segments. The high density of LQT3 mutation sites in this region can be reasonably interpreted by the “door wedge” model for fast inactivation, which was derived from our previous structural observations and is supported by a wealth of functional characterizations.

Abstract

Na_v1.5 is the primary voltage-gated Na⁺ (Na_v) channel in the heart. Mutations of Na_v1.5 are associated with various cardiac disorders exemplified by the type 3 long QT syndrome (LQT3) and Brugada syndrome (BrS). E1784K is a common mutation that has been found in both LQT3 and BrS patients. Here we present the cryo-EM structure of the human Na_v1.5-E1784K variant at an overall resolution of 3.3 Å. The structure is nearly identical to that of the wild-type human Na_v1.5 bound to quinidine. Structural mapping of 91- and 178-point mutations that are respectively associated with LQT3 and BrS reveals a unique distribution pattern for LQT3 mutations. Whereas the BrS mutations spread evenly on the structure, LQT3 mutations are clustered mainly to the segments in repeats III and IV that are involved in gating, voltage-sensing, and particularly inactivation. A mutational hotspot involving the fast inactivation segments is identified and can be mechanistically interpreted by our “door wedge” model for fast inactivation. The structural analysis presented here, with a focus on the impact of mutations on inactivation and late sodium current, establishes a structure-function relationship for the mechanistic understanding of Na_v1.5 channelopathies.

Cardiac arrhythmia affects 1 to 2% of the world’s population and accounts for 230,000 to 350,000 sudden deaths each year in the United States (1, 2). A leading cause for arrythmia is aberrant firing of electrical signals, mediated by the voltage-gated Na channel Na_v1.5 encoded by SCN5A. Dysfunction of Na_v1.5 is related to various arrhythmia syndromes, including Brugada syndrome (BrS), long QT syndrome (LQTS), and sick sinus syndrome (SSS) (3–5) (SI Appendix, Tables S1 to S3). Congenital LQTS and BrS are the two most prevalent genetic disorders for cardiac channelopathies, accounting for approximately one-half of sudden arrhythmic deaths (5-7).

LQTS was named for the phenotype of prolonged QT intervals on electrocardiography that was first reported more than 6 decades ago (8). Ensuing studies revealed that the functional alteration of channels that control repolarization underlies many hereditary LQT cases. LQT3, with arrhythmia during sleep or on waking, is the most harmful LQT type. Inherited LQT3 is mainly associated with mutations in Na_v1.5, many of which result in increased late Na⁺ current (I_Na), a form of gain of function (GOF) (9, 10). In contrast, BrS-associated Na_v1.5 mutations are mostly loss of function (LOF), manifested by reduced peak I_Na or a leftward shift of inactivation curves (11–13).

Since the identification of the LQTS-related SCN5A mutation in 1995, more than 400 missense mutations in the primary sequence of Na_v1.5 have been reported related to various cardiac disorders (9, 14). Among these, 154 and 230 mutations have been identified in patients with LQT3 and BrS, respectively (SI Appendix, Tables S1 and S2). Intriguingly, ∼30 mutations were found in both LQT3 and BrS (SI Appendix, Tables S1 and S2).

To further the molecular understanding of Na_v1.5 channelopathies, we sought to solve structures of human Na_v1.5 with disease mutations in addition to our previous structural resolution of wild-type (WT) human Na_v1.5 bound to pore blockers (15). We started with Na_v1.5-E1784K, which is the most common mutation shared by LQT3 and BrS. This mutation enhances the risk of sudden death resulting from ventricular tachyarrhythmias in LQT3 patients or idiopathic ventricular fibrillation in BrS patients (16, 17). Recombinantly expressed Na_v1.5-E1784K exhibited a leftward shift (∼15 mV) of the steady-state inactivation, reduced peak I_Na, and increased late I_Na (16, 18, 19). How one single point mutation leads to both GOF and LOF at the same time remains enigmatic.

In this paper, we report the structural determination of Na_v1.5-E1784K using single-particle cryo-EM. Mapping of disease mutations shows distinct distribution patterns for LQT3 and BrS mutations on the three-dimensional (3D) structure. The fast inactivation-related structural elements stand out as a hotspot for LQT3 mutations. In the accompanying paper (20), we present a comparative structural analysis of human Na_v1.1 and Na_v1.5 that reveals several mutational hotspots in different sodium channelopathies.

Results

Cryo-EM Analysis of Human Na_v1.5-E1784K.

Consistent with previous report, substitution of Glu1784 with Lys led to a leftward shift (∼20 mV) of the steady-state inactivation, increased late I_Na at room temperature, and significantly decreased the conductance of Na_v1.5 compared to the WT channel (16, 18, 19) (Fig. 1A and SI Appendix, Fig. S1 and Table S4).

Fig. 1.

Structure of full-length human Na_v1.5-E1784K. (A) Electrophysiological properties of Na_v1.5-E1784K heterogeneously expressed in HEK293T cells. (Left) Voltage-dependent activation and inactivation curves. Parameters are presented in SI Appendix, Table S4. (Right) Persistent sodium current at different voltages. The working temperature is 22 °C. (B) Overall structure of human Na_v1.5-E1784K. A side view and a cytoplasmic view are shown. The structure is color-coded for distinct repeats. The III-IV linker is in orange, and the fast inactivation motif, IFM, is shown as ball and sticks. The sugar moieties are shown as black and blue sticks. Eα1/3, extracellular α helix in repeat I or III. All structural figures are prepared in PyMol. (C) The structure of Na_v1.5-E1784K is nearly identical to that of Na_v1.5-quinidine (PDB ID code 6LQA). The complex structures of Na_v1.5-E1784K and Na_v1.5-quinidine can be superimposed with an rmsd of 0.708 Å over 1,075 Cα atoms.

Four auxiliary β subunits, β1 to β4, modulate the localization and channel properties of the α core subunit of Na_v channels (21). We coexpressed β1 with Na_v1.5-E1784K to improve the expression level, although there was no corresponding density for any auxiliary subunit in the 3D EM map. Details of protein preparation, image acquisition, and data processing are presented in Methods. A 3D EM reconstruction was attained at an overall resolution of 3.3 Å out of 147,600 selected particles (SI Appendix, Figs. S2 and S3 and Table S5).

The final structural model of Na_v1.5-E1784K contains 1,151 residues, including the complete transmembrane domain, the extracellular segments, and the III-IV linker. Nine glycosylation sites were observed and assigned to each site (Fig. 1B). Despite the change of the channel properties, the structure of Na_v1.5-E1784K is nearly identical to that of quinidine-bound WT Na_v1.5 (Fig. 1C) (15). More importantly, the last resolved residue is Glu1781, which marks the end of S6_IV. The invisibility of Glu(Lys)1784 prevents a direct structural interpretation of the pathogenic mechanism of this mutation. Therefore, we shifted our focus to systematic structure-based analysis of resolved mutation sites, in the hope of facilitating the establishment of a structure-function relationship of Na_v1.5 disease variants.

Distinct Structural Distributions of LQT3 and BrS Mutations.

The high-resolution structure of human Na_v1.5-E1784K allows for mapping of a total of 86 mutations associated with LQT3 and 174 mutations associated with BrS (SI Appendix, Tables S1 and S2). In the comparative structural analysis reported in the accompanying paper (20), we divided disease-associated mutations on Na_v channels into “structural mutations” and “functional mutations.” The former may impair protein folding and structural stability, and the latter alters channel properties. Our analysis suggests that many of the mutations mapped to the extracellular loops (ECLs) above the pore domain (PD) and the selectivity filter (SF)-supporting pore helices P1 and P2 (the P1-SF-P2 region) are structural mutations, while those on voltage-sensing domians (VSDs) and segments related to pore gating and fast inactivation are mostly functional mutations (20).

Structural mapping of the disease mutations reveals distinctive distribution patterns of BrS and LQT3 mutations. The BrS mutations span the entire structure, including 39 mutations affecting 32 residues in the ECLs and 30 mutations of 24 residues in the P1-SF-P2 region (the SF zone hereinafter) (Fig. 2A and SI Appendix, Tables S1 and S2). In contrast, only six and three LQT3 mutations are identified in the ECL and SF zone regions, respectively (Fig. 2B). In fact, the structural distribution of LQT3 mutations appears to be highly polarized, with 33 residues (37 mutations) clustered on the III-IV linker and the nearby S4-S5, S5, and S6 segments in repeats III and IV (Fig. 3 B and C).

Fig. 2.

Structural mapping of Na_v1.5 mutations associated with BrS and LQT3. (A and B) Mapping of the BrS (A) and LQT3 (B) disease mutations on the structure. The Cα atoms of the disease-related residues are shown as spheres. Diagonal repeats are shown in the top and bottom rows. Whereas BrS mutations are distributed throughout the structure, LQT3 mutations are found mainly in the VSDs and the segments related to fast inactivation. (C) The IFM motif-carrying III-IV linker and its receptor site represent a hotspot for LQT3 mutations. The cluster of mutations is indicated by the semitransparent pink oval. The III-IV linker is in yellow.

Fig. 3.

Mapping of LQT3 mutations on VSDs. VSD_III and VSD_IV harbor more LQT3 mutations than the other two VSDs. The LQT3-related residues that are mapped to VSDs, including the S4-S5 linker, and their respective interface with the PD are shown as blue sticks.

Analysis of representative BrS mutations is presented in the accompanying paper (20). In what follows, we focus on a structural overview of LQT3 mutations.

LQT3 Mutations in VSDs.

An asynchronous activation model of the four VSDs has been established for the Na_v channel, in which the first three VSDs activate in advance of VSD_IV. Whereas activation of the first three VSDs leads to pore opening, activation of VSD_IV elicits fast inactivation of the channel (22–25). Consistent with their distinct functions, distribution of LQT3-related residues is uneven among the four VSDs, with four on VSD_I, five on VSD_II, nine on VSD_III, and ten on VSD_IV (Fig. 3 and SI Appendix, Table S1). Of note, here the affected loci are counted, although multiple substitutions to the same residue can occur. None of the mutations in VSD_II and VSD_III involves the well-defined functional residues, such as the gating charge (GC) residues, whereas two of the four affected residues are GC residues R2 and R3 (R222Q and R225Q/W) in VSD_I (26, 27). Similarly, mutations of two GCs, R1 and R2, in VSD_IV are found in LQT3 patients (R1623L/Q and R1626H/P) (10, 28, 29) (SI Appendix, Table S1).

Mechanistic dissection of the LQT3 mutations in VSDs, which undergo pronounced conformational changes during the working cycle of Na_v channels for each firing, requires structural determination of the channel in multiple functioning states. Before that, comparison of the structures of Na_v1.5-E1784K and Na_vPaS, which exhibits a conformation distinct from all mammalian Na_v channels of known structures, provides important clues to understanding the primary mutational hotspot on Na_v1.5 for LQT3.

A LQT3 Mutational Hotspot Mapped to the Inactivation Segments.

An important discovery derived from our comprehensive structural mapping of disease mutations is that ∼20% of all LQT3 mutations, or approximately one-third of structurally resolved mutations, are clustered to one corner of the PD, which is close to the intracellular gate along the permeation path in repeats III and IV (Fig. 4A). In particular, the Ile/Phe/Met (IFM) motif and the ensuing α helix on the III-IV linker host an exceptionally high density of LQT3 mutations (Fig. 4B).

Fig. 4.

Structural analysis of fast inactivation-related LQT3 mutations. (A) Asymmetric distribution of LQT3 mutations on the PD. The LQT3-related residues on the PD are shown as blue sticks. An intracellular view (Left) and a side view (Right) are presented. These mutations are clustered to the gating and fast inactivation-related segments in repeats III and IV. (B) The interface of the III-IV linker and S4-S5_IV is enriched of LQT3 mutations. (C) Mapping of the Na_v1.5 LQT3 mutations to Na_vPaS. The structure of Na_vPaS, a Na_v channel from American cockroach, exhibits a distinct conformation from that of Na_v1.5, featured with less depolarized VSDs, sealed PD, and the CTD-sequestered III-IV linker (PDB ID code 5X0M). (Left) Structural superimposition of Na_v1.5-E1784K and Na_vPaS. (Right) Many LQT3-related residues may be mapped to the interface of the III-IV linker and the CTD. The Cα atoms of the Na_vPaS residues that correspond to the LQT3 mutations in Na_v1.5 are shown as spheres. (D) Mapping of LQT3 mutations in a Na_vPaS-derived Na_v1.5 model, in which the conformation of the PD and the III-IV linker may represent a potentially resting state. (Upper) A Na_vPaS-derived structural model for Na_v1.5. The LQT3-related residues that map to the interface of the III-IV linker and the CTD are shown as blue sticks. (Lower) Lys1493 on the III-IV helix may be sandwiched by the invariant Glu1780 and Glu1784 on the end of S6_IV in a potentially resting state. Mutation K1493R may strengthen these interactions, thereby impeding the conformational shift toward the inactivation conformation captured by the structure, as seen in B. Consistently, the LQT3 mutation K1493R led to a right shift of the steady-state inactivation curve (41). Glu1784 may also be involved in the interaction with CTD and the III-IV linker. Mutation E1784K may disrupt this interaction, resulting in accelerated fast inactivation and a leftward shift of the inactivation curve, but the molecular basis for the increased late I_Na awaits further characterization (Fig. 1A) (17, 18).

Decades of research have confirmed that the III-IV linker plays a fundamental role in the fast inactivation of Na_v channels, among which the hydrophobic cluster IFM motif is defined as the fast inactivation particle (30, 31). In contrast to the conventional “ball and chain” model (31), our structural comparison of Na_vPaS with other Na_v channels suggested an “allosteric blocking” mechanism (32, 33), or “door wedge” model, for fast inactivating Na_v channels. In Na_vPaS, the III-IV linker is sequestered by the carboxy terminal domain (CTD) (33, 34), leaving the IFM-corresponding residues away from the PD, a conformation that is also observed in the voltage-gated Ca²⁺ channel Ca_v1.1 (35) (Fig. 4C). In the structures of EeNa_v1.4 and all resolved human Na_v channels (15, 20, 32, 36–38), the IFM motif undergoes a pronounced displacement to plug into a cavity composed of the S6 helices and the S4-S5 segments in repeats III and IV (Fig. 4C). Insertion of the IFM motif into this receptor site outside the intracellular gate is expected to push the S6 segments to close at the intracellular gate (33), reminiscent of a door stopper wedge. Therefore, these structures, characteristic of “up” VSDs and a plugged IFM motif, may represent the inactivated state of Na_v channels. For simplicity, we refer to the IFM motif and its receptor site as the “fast inactivation corner.”

Based on our “door wedge” model, we hypothesize that mutations that impair interactions between the IFM motif-carrying III-IV linker and its receptor site relax the S4-S5 restriction ring surrounding the intracellular gate or disrupt the intracellular gate may hinder fast inactivation and increase the probability of gate opening during prolonged depolarization. These molecular events would be detected in the electrophysiological characterizations as a right shift of the steady-state inactivation curve and increased late I_Na, which are just the commonly observed channel alterations associated with LQT3 mutations.

Our analysis is supported by documented characterizations of Na_v1.5 variants containing LQT3 mutations. For instance, mutation of the Phe in the IFM motif, F1486L, or alternation of a Phe in its receptor cavity, F1473C, may decrease the affinity between IFM and the receptor site. Both mutations led to increased late I_Na (39, 40). Met1652 on the S4-S5_IV segment interacts with Tyr1495 and Met1498 on the III-IV linker (Fig. 4B). Single point mutation M1652R, which may disrupt the interaction with the III-IV linker, resulted in slowed inactivation and increased late I_Na (10).

Along the same line, mutations that stabilize the resting state and impede conformational shifts toward the inactivated structure also may lead to the LQT3 phenotype. Before the structure of any eukaryotic Na_v channel in the resting state is captured, that of the Na_vPaS structure provides important clues (Fig. 4 C and D) (33, 34). To facilitate structural mapping, a 3D model of Na_v1.5 was generated based on sequence homology with Na_vPaS (Protein Data Bank [PDB] ID code 5X0M).

In this Na_vPaS-derived Na_v1.5 model, the III-IV linker interacts with the CTD. Although several LQT3 mutations are mapped to the CTD (Fig. 4 D, Upper), we refrain from discussing them because the resolution of CTD does not support accurate side chain modeling in Na_vPaS (33, 34). We focus on the residues whose counterparts in Na_vPaS have been reliably resolved. Na_v1.5-Lys1493, which is on the III-IV linker helix and away from the interface with other structural segments in the cryo-EM structure of Na_v1.5-E1784K (Fig. 4B), would be sandwiched by Glu1780 and Glu1784, which are on the C-terminal end of S6_IV, in the conformation of Na_vPaS (Fig. 4 D, Lower). Replacement of Lys1493 by Arg may even strengthen these interactions, hence impeding dislocation of the III-IV linker toward an inactivated conformation. Indeed, the LQT3 mutation K1493R resulted in a right shift of the steady-state inactivation curve (41).

Na_v1.5-Glu1784, although invisible in the current structure, is predicted to be positioned at the cytosolic tip of S6_IV. Based on the Na_vPaS-derived model, replacement of the acidic residue with Lys may weaken the sequestration of the S6_IV segment by the CTD, resulting in accelerated fast inactivation and a left shift of the inactivation curve, which may account for the BrS phenotype. However, the overall impact of this single point mutation may be more complex, as it also led to increased late I_Na underlying the LQT3 phenotype, which awaits a mechanistic explanation (16, 18).

Discussion

In this paper, we report the structure of human Na_v1.5-E1784K and focus on the systematic structural analysis of LQT3-associated mutations. BrS-related mutations, which are distributed rather uniformly on the entire structure and lack a clear preference for specific structural entities, are analyzed in the accompanying paper (20).

Instead of characterizing specific disease mutations, we attempted to summarize the distributing pattern of the resolved LQT3 mutations to acquire mechanistic insight. The structural analysis, together with reported functional characterizations, helps establish a structure-function relationship of the pathogenic mutations and in turn illuminates the mechanistic understanding of the Na_v channels.

An advanced molecular dissection of these pathogenic mutations necessitates structural determination of Na_v channels in multiple functional states. Among all the structures of eukaryotic Na_v channels, only two major conformations were captured. The structure of Na_vPaS is featured with a closed PD without fenestration, a CTD-sequestered III-IV linker, and VSDs in less polarized conformations (33, 34), while all the human, electric eel, and rat Na_v channels exhibit similar conformations, characterized by insertion of the IFM motif into a cavity outside the S6 tetrahelical bundle (15, 20, 32, 36–38, 42). Although the functional state of the Na_vPaS structure cannot be defined because of its resistance to electrophysiological recording, the conformations of the PD, the CTD, and the III-IV linker are consistent with those expected in a resting or a preopened channel. Thus, the pronounced structural differences between Na_vPaS and other Na_v channels provide an opportunity for understanding the function and disease mechanism of Na_v channels.

Structural mapping reveals the fast inactivation-related segments to be a hotspot for LQT3 mutations. Our “door wedge” model for fast inactivation, originally derived from structural comparison of Na_vPaS and Na_v1.4 channels (32, 36) and supported by functional data reported over the past 3 decades, affords a plausible mechanistic interpretation for the exceptionally high density of LQT3 mutations in the III-IV linker and the surrounding segments.

Taken together, the structural analyses presented here and in the accompanying paper (20) provide a comprehensive overview of dozens of Na_v mutations in several major channelopathies, lay the foundation for a precise structure-function relationship understanding of their pathogenic mechanisms, and reveal the molecular basis for a wealth of experimental and clinical observations accumulated over the past several decades.

Materials and Methods

Transient Coexpression of Human Na_v1.5-E1784K and β1.

A single point mutation, E1784K, was introduced to full-length human Na_v1.5 (UniProt accession no. Q14524) by QuikChange site-directed mutagenesis and confirmed by PCR sequencing. Then the cDNA was cloned into the pEG BacMam vector with a twin Strep tag and FLAG tag in tandem at the amino terminus, and β1 (UniProt accession no. Q07699) was cloned into the pCAG vector without any affinity tag (43, 44). The cDNAs of Na_v1.5-E1784K and β1 were optimized into the mammalian cell expression system and coexpressed in HEK293F cells (Invitrogen), which were cultured in SMM 293T-II medium (Sino Biological) under 5% CO₂ in a Multitron Pro shaker (Infors; 130 rpm) at 37 °C. When the cell density reached ∼2.0 × 10⁶ cells/mL, ∼2.0 mg plasmids (1.4 mg for Na_v1.5-E1784K and 0.6 mg for β1) and 4 mg of 25-kDa linear polyethyleneimines (Polysciences) were mixed into 30 mL of fresh medium and preincubated for 15 to 30 min before being added into 1 L of cell culture. Transfected cells were harvested after 48 h of cultivation.

Protein Purification of Na_v1.5-E1784K and β1.

Protein purification was conducted following a protocol identical to that for the WT human Na_v1.5 channel (15), which was derived from the experiments for human Na_v1.4 (36). Here 24 L of transfected cells were harvested by centrifugation at 800 × g for 12 min, after which the cell pellet was resuspended in the lysis buffer containing 25 mM Tris⋅HCl pH 7.5 and 150 mM NaCl. The cell suspension was supplemented with 1% (wt/vol) n-dodecyl-β-d-maltopyranoside (Anatrace), 0.1% (wt/vol) cholesteryl hemisuccinate Tris salt (Anatrace), and protease inhibitor mixture containing 2 mM phenylmethylsulfonyl fluoride, aprotinin (3.9 μg/mL), pepstatin (2.1 μg/mL), and leupeptin (15 μg/mL) and then incubated at 4 °C for 3 h. The cell lysate was subsequently ultra-centrifuged at 20,000 × g for 45 min, and the supernatant was applied to anti-FLAG M2 affinity gel (Sigma-Aldrich) by gravity at 4 °C. The resin was washed with 10 column volumes of wash buffer containing 25 mM Tris⋅HCl pH 7.5, 150 mM NaCl, 0.06% glycol-diosgenin (Anatrace), and the protease inhibitor mixture. Target proteins were eluted with five column volumes of wash buffer plus 200 μg/mL FLAG peptide (Sigma-Aldrich). The eluent was then applied to Strep-Tactin Sepharose (IBA Lifesciences). The purification protocol was similar to the previous steps except for the elution buffer, which was wash buffer plus 2.5 mM D-Desthiobiotin (IBA Lifesciences). The eluent was then concentrated using a 100-kDa cutoff Centricon centrifugal filter (Millipore Sigma) and further purified with size exclusion chromatography (Superose-6 Increase 10/300 column; GE Healthcare) in the wash buffer. The presence of target proteins was confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis and mass spectrometry. The purified proteins were pooled and concentrated to ∼2 mg/mL for cryo-EM sample preparation.

Whole-Cell Electrophysiology.

HEK293T cells (Invitrogen) cultured in Dulbecco’s Modified Eagle’s Medium (Biological Industries) supplemented with 4.5 mg/mL glucose and 10% fetal bovine serum (Biological Industries) were plated onto glass coverslips for subsequent patch clamp recordings. Cells were transiently cotransfected using Lipofectamine 2000 (Invitrogen) with the expression plasmids for the indicated Na_v1.5 or Na_v1.5-E1784K and an eGFP-encoding plasmid. Cells with green fluorescence were selected for patch clamp recording at 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 in HEK293T cells 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) used had a resistance of 2 to 4 MΩ, and the electrodes were filled with the internal solution composed of 105 mM CsF, 40 mM CsCl, 10 mM NaCl, 10 mM EGTA, and 10 mM Hepes, pH 7.4 with CsOH. The bath solutions contained 140 mM NaCl, 4 mM KCl, 10 mM Hepes, 10 mM d-glucose, 1 mM MgCl₂, and 1.5 mM CaCl₂, pH 7.4 with NaOH. Data were analyzed using Origin (OriginLab) and GraphPad Prism (GraphPad Software).

The voltage dependence of ion current (I-V) was analyzed using a protocol consisting of steps from a holding potential of −120 mV (for 100 ms) to voltages ranging from −90 to +80 mV for 50 ms in 5-mV increments. The linear component of leaky currents and capacitive transients were subtracted using the P/4 procedure. In the activation and conductance density calculations, 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. To obtain the conductance density curves, G was divided by the capacitance (C) and plotted against the voltage from −90 mV to +20 mV. For voltage dependence of inactivation, cells were clamped at a holding potential of −90 mV and were applied to step prepulses from −120 mV to +20 mV for 100 ms in increments of 5 mV. Then the Na⁺ current was recorded at the test pulse of 0 mV for 50 ms. The peak currents under the test pulses were normalized and plotted against the prepulse voltage. Activation and inactivation curves were fit to a Boltzmann function to obtain V_1/2 and slope values. The time course of inactivation data from the peak current at 0 mV was fitted to a single exponential equation, y = A1 exp(−x/τ_inac) + y0, where A1 is the relative fraction of current inactivation, τ_inac is the time constant, x is the time, and y0 is the amplitude of the steady-state component. The late sodium current was measured as the mean inward current between 40 ms and 50 ms at the end of a 100-ms depolarization to voltage between −30 mV and +30 mV. Then the current was divided by the peak inward current at the same potential to show the late sodium current percentage.

All data points are presented as mean ± SEM, and n is the number of experimental cells from which recordings were obtained. Statistical significance was assessed using an unpaired t test with Welch’s correction, two-way ANOVA, and an extra sum-of-squares F test.

Cryo-EM Data Acquisition.

For the cryo-EM sample preparation and data collection, the same machine and software were used with similar protocols as described previously (36). Aliquots of 3.5 μL of freshly purified Na_v1.5-E1784K were placed on glow-discharged holey carbon grids (Quantifoil Au 300 mesh, R1.2/1.3). Grids were blotted for 3.0 s and then plunge-frozen in liquid ethane cooled by liquid nitrogen with the Vitrobot Mark IV system (Thermo Fisher Scientific). Electron micrographs were acquired on a Titan Krios electron microscope (Thermo Fisher Scientific) operating at 300 kV, a Gatan K3 Summit detector, and a GIF Quantum energy filter. A total of 4,849 movie stacks were collected automatically using AutoEMation (45) with a slit width of 20 eV on the energy filter and a preset defocus range from −1.8 μm to −1.3 μm in superresolution mode at a nominal magnification of 81,000×. Each stack was exposed for 2.56 s with 0.08 s per frame, resulting in 32 frames per stack. The total dose rate was 50 e⁻/Ų for each stack. The stacks were motion-corrected with MotionCor2 (46) and binned twofold, resulting in 1.0825 Å/pixel. Meanwhile, dose weighting was performed (47). The defocus values were estimated with Gctf (48).

Image Processing.

A diagram of the data processing workflow is presented in SI Appendix, Fig. S2. A total of 1,471,342 particles were automatically picked from 4,633 manually selected micrographs using Gautomatch. Subsequent two-dimensional (2D) and 3D classifications and refinement were performed using cryoSPARC (49). After 2D classification, 276,241 good particles were selected and applied to 3D homogeneous refinement. After that one additional round of heterogeneous refinement was performed, during which those particles were classified into four classes, and the good class was selected, resulting in a dataset of 147,600 particles. The selected particles were applied to the uniform refinement and nonuniform refinement procedure, finally yielding a 3D EM map with an overall resolution of 3.3 Å. The resolution was estimated with the gold standard Fourier shell correlation 0.143 criterion (50) with high-resolution noise substitution (51).

Model Building and Structure Refinement.

The initial model of Na_v1.5-E1784K was based on the coordinate of human Na_v1.5-quinidine (PDB ID code 6LQA), and the structure refinement process was as described previously (15, 36), fitted into the EM map by Chimera (52) and manually adjusted in Coot (53). The chemical properties of amino acids were taken into consideration during model building. There was no density corresponding to β1. A total of 1,151 residues in Na_v1.5-E1784K were assigned with side chains, and nine sugar moieties were built. The N-terminal 118 residues, intracellular I-II linker (residues 430 to 698), II-III linker (residues 945 to 1187), and C-terminal sequences after Ser1782 were not modeled due to the lack of corresponding densities.

Structure refinement was performed using phenix.real_space_refine application in Phenix (54) real space with secondary structure and geometry restraints. Overfitting 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 (55). Statistics for the map reconstruction and model refinement are provided in SI Appendix, Table S5.

Data Availability

Density map and model data have been deposited in the Electron Microscopy Data Bank (accession code 30850) and the PDB (ID code 7DTC).