Broadband Tuning the Voltage Dependence of a Sodium Channel

Voltage-gated sodium channels initiate action potentials in prokaryotes and in many eukaryotic cells, including vertebrate nerve and muscle. Their activation is steeply voltage-dependent, but it is unclear how the voltage sensitivity is set or whether it can be broadly shifted to positive voltages. Here we show that the voltage dependence of activation (VA) of the ancestral bacterial sodium channel NaVAb can be progressively shifted from −118 mV to +35 mV in chimeras with increasing numbers of amino acid residues from the extracellular half of the voltage sensor of human NaV1.7 channels. In a minimal chimera in which only 32 residues were transferred, we analyzed the effects of six additional mutations of conserved amino acid residues singly, in pairs, and as triple mutations. The resulting chimeric mutants exhibited a broad range of voltage sensitivity from VA=−118 mV to VA=+120 mV. Three mutations (N48K, L112A, and M119V) shifted VA to +61 mV when substituted in NaVAb itself, and substitution of two additional Cys residues in the Cys-free background of NaVAb further shifted VA to +105 mV. In these mutants, measurement of gating currents revealed that the voltage dependence of gating charge movement (VQ) shifted to positive membrane potentials as much or more than VA, confirming that the gating charges are trapped in their resting positions by these VA-shifting mutations. Our results demonstrate broadband shifting of VA and VQ of a sodium channel across a range of 240 mV and provide a toolbox of methods and constructs to analyze sodium channel structure and function in the resting state at 0 mV and in activated states at positive membrane potentials. GRAPHICAL ABSTRACT The complete range of broadband tuning of voltage-dependent activation of a sodium channel.


INTRODUCTION
Animal life depends on electrical signaling via action potentials. The swift conduction of action potentials along the cell surface of neurons, myocytes, and other excitable cells confers the ability to coordinate rapid events -allowing animals to breathe, move, and think (Hille, 2001;Hodgkin and Huxley, 1952). Action potentials are initiated by voltage-gated sodium channels, which open in response to membrane depolarization and then inactivate within 1-5 msec (Hodgkin and Huxley, 1952). Depolarization of the cell membrane also activates voltage-gated potassium channels, which open more slowly and conduct potassium out of the cell to repolarize the membrane potential (Hodgkin and Huxley, 1952). In order to explain the voltage dependence of activation of sodium channels, Hodgkin and Huxley proposed that membrane-bound charges associated with sodium channels respond to the change in electric field, move across the membrane upon depolarization, and return to their resting position upon repolarization. Based on the steepness of voltage-dependent activation, they proposed that 3 singly charged 'gating particles' control activation and one gating particle controls inactivation (Hodgkin and Huxley, 1952). Their prescient prediction was confirmed when the movement of these gating particles was directly measured in the form of gating currents, which are nonlinear capacitive currents that arise from the movement of the gating particles Bezanilla, 1973, 1977;Bezanilla and Armstrong, 1977). Fast, outward movement of positive gating charge was triggered by depolarization (ON gating currents), and a corresponding inward OFF gating current was observed upon hyperpolarization of the membrane. These gating currents are the electrical signature of the voltage-dependent activation process.
The protein components of sodium channels were initially identified by photolabeling studies using scorpion toxins, which revealed a complex of an a subunit and a smaller b subunit (Beneski and Catterall, 1980). Sodium channels were purified from electric eel electroplax (Barchi, 1983;Miller et al., 1983), skeletal muscle (Barchi, 1983), and brain Catterall, 1981, 1984;Hartshorne et al., 1982). The brain sodium channel was reconstituted into phospholipid bilayers and vesicles, and measurement of ion flux and single-channel currents verified that these purified proteins were in fact voltage-gated sodium channels (Hartshorne et al., 1985;Talvenheimo et al., 1982;Tamkun et al., 1984). Cloning and sequencing cDNA encoding sodium channel a subunits revealed a polypeptide of ~2000 amino acid residues having 24 transmembrane segments organized in six homologous domains (Noda et al., 1986;Noda et al., 1984). Extensive structure-function studies showed that the first four helices (S1-S4) of each subunit comprise the voltage-sensing domain (VS), while the fifth and sixth (S5-S6) transmembrane helices from all four subunits come together to form the pore module (PM) (Catterall, 2000). The S4 segment contains a repeated motif of a positively charged residue flanked on each side by hydrophobic ones in a three-residue pattern that is repeated 4-8 times in sodium channel VS. Mutagenesis studies revealed that these charged residues on the S4 segment serve as the gating charges that enable the VS to sense fluctuations in membrane potential and respond by opening the pore (Bezanilla, 2000;Catterall, 2000;Kontis et al., 1997;Stuhmer et al., 1989). Chemical labeling, voltage-clamp fluorometry, and disulfide-crosslinking studies revealed that changes in membrane potential drive the S4 segments of sodium channels across the membrane through a narrow hydrophobic seal, catalyzed by interactions with negative charges on the intracellular and extracellular sides (Chanda and Bezanilla, 2002;DeCaen et al., 2011;DeCaen et al., 2009;DeCaen et al., 2008;Yang et al., 1996;Yang and Horn, 1995;Yarov-Yarovoy et al., 2012). A "sliding-helix" model in which the S4 segment moves outward through the VS upon depolarization and exchanges ion pair partners has emerged as a consensus mechanism for voltage sensing (Catterall, 1986;Catterall et al., 2017;Vargas et al., 2012;Wisedchaisri et al., 2019;Yarov-Yarovoy et al., 2012).
The unexpected discovery of bacterial sodium channels showed that they are homotetrameric complexes of four identical subunits with six transmembrane segments (Ren et al., 2001). This simple organization facilitated structural studies using X-ray crystallography, which defined the structure of the bacterial sodium channel NaVAb at 2.7 Å resolution (Payandeh et al., 2011). Other bacterial sodium channels have similar structures (Ahern et al., 2016;Catterall et al., 2017;McCusker et al., 2012;Zhang et al., 2012). Moreover, eukaryotic sodium channels analyzed by cryo-electron microscopy (cryo-EM) have a similar core structure, with RMSD in the range of 3 Å for the transmembrane regions of mammalian sodium channels from skeletal muscle, nerve, and heart (Jiang et al., 2020;Pan et al., 2019;Pan et al., 2018;Shen et al., 2019). These structural results enabled analysis of the molecular basis of voltage-dependent gating at the atomic level, but the range of functional states observed in these structural studies was limited because they were all carried out in the absence of a membrane potential where partially activated and steadystate inactivated states of sodium channels are most stable. Here we used a combination of structure-guided mutagenesis and biophysical analysis to tune the voltage dependence of sodium channels over a broad range and determine whether they can function normally in the positive membrane potential range. Our results reveal that broadband tuning of the voltage dependence of a sodium channel over a range up to 240 mV is possible (see Graphical Abstract), and they provide a novel approach to analysis of the structure and function of sodium channels in multiple states that normally would not be accessible for study by structural biology methods at 0 mV. As proof of this concept, we have used one of the constructs described here to determine the structure of the resting state of a sodium channel at high resolution (Wisedchaisri et al., 2019).

Construction of NaVAb/NaV1.7-VS2 chimera and mutants
NaVAb/NaV1.7-VS2 cDNAs with a wide range of voltage-shifting mutations and combinations were constructed in order to explore the range of positive shifts in the voltage dependence of activation that are possible. These NaVAb/NaV1.7-VS2 chimeras were generated by grafting nucleotide sequences of S1-S2 and S3-S4 segments of NaV1.7 VS2 into homologous regions in NaVAb. The cDNAs encoding NaVAb/NaV1.7-VS2 chimeras were generated by overlap extension PCR (Heckman and Pease, 2007) for NaVAb 7DII, and by gene synthesis for NaVAb 7DII-4 and 7DII-5. These cDNAs were then amplified by PCR and cloned into the modified pIZT vector described previously (Wisedchaisri et al. 2019). Briefly, the pIZT vector (Life Technologies) was modified to include GFP followed by a self-cleaving P2A sequence under an OpIE2 promoter to facilitate identification of transfected Sf9 cells for electrophysiological recording, and NaVAb and mutants were expressed as GFP-P2A-NaVAb. NaVAb 7DII-2 and 7DII-3 chimeras were derived from NaVAb 7DII-4 by site-directed mutagenesis. Single, double, and triple residue substitutions in NaVAb 7DII chimeras and NaVAb were carried out by site-directed mutagenesis. All constructs were characterized by electrophysiology using transfected Sf9 cells as described previously The measured liquid junction potential was -7 mV for ionic current recordings, and the reported voltages were not adjusted. Ionic current families were generated by holding cells at -150 mV and stimulating with a series of 50-ms pulses at 10-mV intervals. All scale bars denote 1 nA/10 msec.
To determine the reversal potential of outward current, peak current was elicited by pulsing to an appropriate potential for each construct and decreasing the voltage in 10 mV steps until substantial inward current was seen in the tail currents. An agar bridge consisting of extracellular solution with 3% Agar was placed between the reference electrode and the bath in experiments utilizing perfusion of extracellular solution containing reducing agents to isolate the electrode from the reducing agents.

Data analysis
Data were analyzed using IGOR Pro (6.37). The peak current at each voltage of the current family was averaged and plotted as a function of the stimulus voltage to visualize the current vs voltage (I/V) relationship. When the I/V curves contained only outward current, the reversal potential (Vrev) was determined from the tail currents at peak activation. The instantaneous current of the tail current family was plotted as a function of voltage, fit with a line near 0 mV, and the X intercept was used as Vrev to generate the conductance vs voltage (G/V) curve from the I/V curve by calculating G=I/(Vm-Vrev). Normalized G/V curves were fit with a simple two-state Boltzmann equation 1/(1+exp(V1/2-Vm)/k) in which Vm is the stimulus potential, VA is the half-activation voltage, and k is a slope factor. I/V plots of inward currents were fit with (Vm-Vrev)*(Imin/(1+exp((Vm-VA)/k)) to generate Boltzmann relationships fit to normalized G/V curves. Q/V curves were generated by integrating the P/N subtracted curve either during the test pulse or after the return to the prepulse voltage.

RESULTS
We took a stepwise approach to determining whether broadband tuning of the voltage dependence of activation (VA) of a sodium channel is feasible. First, we asked whether formation of chimeric channels with amino acid residues from a mammalian sodium channel is able to induce positive shifts in the VA of NaVAb. Second, we surveyed the VA-shifting effects of single amino acid substitutions selected from the literature, which were known to induce positive shifts in VA in other channel types. Third, we assessed the effects of pairwise and triple combinations of these mutations in chimeric mammalian NaV1.7/NaVAb channels. Fourth, we determined if a

Shifting Voltage Dependence of a Chimeric Sodium Channel
The bacterial sodium channel NaVAb activates at very negative membrane potentials, with VA = -98 mV (Gamal El-Din et al., 2013a). Mammalian NaV channels activate at much more depolarized membrane potentials. For example, the NaV1.7 channel activates with a VA of -22 mV (Wu et al., 2013). To test whether substitution of amino acid residues from mammalian NaV1.7 into the VS of NaVAb could shift its voltage dependence, we constructed and analyzed chimeras of NaVAb and NaV1.7 ( Figure 1; Table 1). The basic chimera 7DII contained the extracellular residues of the human NaV1.7 DII VSD, starting near the extracellular end of each transmembrane helix ( Figure 1A and C, left, Table 1). This construction shortened the S1-S2 liker by one residue and lengthened the S3-S4 linker by four residues, resulting in a net gain of three residues for the channel (Table 1). We measured the sodium current observed when this chimera was depolarized to a range of membrane potentials, and we calculated the conductance as a function of stimulus potential. The resulting conductance/voltage (G/V) curves for chimera 7DII revealed a similar voltage dependence of activation to NaVAb (VA = -118 ± 0.9 mV; Figure 1B, red; Table   2). In contrast to this small effect on voltage dependence, substitution of additional amino acid residues in chimera 7DII caused large positive voltage shifts. VA shifted to -7.5 ± 1.7 mV for chimera 7DII-2 ( Figure 1B, orange) and further shifted to +34.6 ± 1.4 mV for chimera 7DII-3 ( Figure 1B, blue), a total voltage shift of +153 mV (Table 2). These large voltage shifts were caused by human NaV1.7 residues substituted for NaVAb residues over the extracellular halves of the S1-S4 segments ( Figure 1C). This dramatic positive shift of VA suggested the possibility that the voltage dependence of chimera 7DII could be tuned over a broad voltage range by substitutions of individual amino acid residues and by pairwise and triple combinations of individual mutations.

Shifting VA with Single Mutations
NaVAb has three distinct kinetic phases of slow inactivation: two that are easily reversible by hyperpolarization, and one "late use-dependent phase" that is induced by repetitive depolarizing pulses and reverses very slowly (Blanchet and Chahine, 2007;Gamal El-Din et al., 2013b;Pavlov et al., 2005). The late phase of inactivation makes stable recordings of voltage-dependent properties of NaVAb channels impractical. Previous work showed that the mutation N49K (numbered N48K in the NaVAb/NaV1.7 7DII chimera) caused a positive shift in VA and blocked slow, use-dependent inactivation of NaVAb (Gamal El-Din et al., 2013a). As expected from this prior work, the N48K mutation in chimera 7DII shifted 73 mV from VA = -118 ± 0.9 mV to VA = -45.4 ± 0.9 mV (Figures 2 and 3, red), and it eliminated the late use-dependent inactivation (not shown). This residue is located in a conserved site on the S2 segment containing an amino acid residue in the Extracellular Negative Cluster (ENC) that acts as a counter charge to the positive gating charges ( Figure 2C). In NaVAb, the native residue is Asn, which is hydrophilic and has a partial net negative charge. Placing a positive charge in this position may alter gating charge interactions and oppose both activation and slow, use-dependent inactivation.
To test whether other substitutions of single non-gating-charge residues would have comparable effects, we searched the literature for previously described, highly conserved residues whose mutation caused large positive shifts in VA. Mutagenesis studies of Shaker B potassium channels highlighted the residue equivalent to N25, located in the center of the VS, as a potential voltage-shifting amino acid substitution ( (Lacroix et al., 2014); Figure 2C, Table 1).
Mutation of N25 to Phe shifted the VA by +57 mV to VA=-61.0 ± 1.4 mV (Figure 2A and B, orange; Table 3). Thus, this mutation has comparable effects on VA in an insect potassium channel and a bacterial sodium channel.
Continuing this line of investigation, we sought well-characterized mutations of conserved amino acid residues located near the intracellular end of S4 that cause positive shifts in VA. We first focused on NaVAb/L109 (Table 1), whose counterpart in Shaker B potassium channels and NaV1.2 sodium channels yielded large positive shifts in VA upon mutation (Kontis et al., 1997;Lopez et al., 1991). Mutation of the corresponding residue in 7DII (L112, Table 1, Figure 2C) to Ala caused a positive shift of VA to -55.0 ± 4.9 mV in 7DII (Figure 2A and B, green; Table 3).
Taking a similar approach, we found that mutation M119V of 7DII positively shifted VA to -70.0 ± 1.2 mV (Figure 2A and B, purple; Table 3). L112A is located directly after R4, and M119 begins the S4-S5 linker ( Figure 2C). Thus, these two amino acid residues flank the intracellular end of the S4 segment and may impede its movement during voltage sensing and gating.
These results suggested that the intracellular end of the S4 segment might be a hotspot for VA-shifting mutations. To follow up that idea, we analyzed the effects of a mutation of the R4 gating charge that preserves its positive charge (R111K) and a mutation of T114 (T114K; Tables 1 and 3) that introduces an additional positive charge in register with the R1-R4 gating charges.
Substitution of these residues also resulted in positive shifts of VA, with R111K causing a +58-mV shift of VA to -59.8 ± 6.6 mV (Figure 3, blue) and T114K causing a +16-mV shift of VA to -102 ± 2.9 mV (Figure 3, yellow). Altogether, these experiments identify a set of conserved amino acid residues whose mutation individually causes substantial positive shifts in VA.

Pairwise Combinations of VA Shifting Mutations
Because these individual mutations gave promising results showing substantial positive shifts in VA, we analyzed the effects of pairwise combinations of selected VA-shifting mutations (Figure 4).
The effects of the combined mutations R111K and T114K were greater than additive, as the total shift in VA of the double mutant (+102 mV) was greater than the sum of the two individual shifts in VA (+74 mV, Figure 4A and B, brown, Table 4). The combined shift of VA of the N48K/R111K double mutant was +156 mV, which exceeds the sum of the individual shifts of +131 mV ( Figure   4C, blue; Table 4). The combinatorial effects of mutation N48K with R111S or R111C on VA were smaller than R111K, but also showed cumulative positive shifts in VA ( Figure 4C, yellow and green; Table 4). These results encouraged us to construct triple mutations to search for combinations that gave even larger positive shifts in VA.

Triple Combinations of VA Shifting Mutations
We studied substitutions of residue L112 with Ala and Phe in combination with N48K and R111K ( Figure 5). When combined with N48K, L112A and L112F had similar VA values of -9.4 ± 4.3 mV and -6.2 ± 1.1 mV, respectively, whereas the VA for N48K alone was -45.4 ± 0.9 mV ( Figure 5B, yellow and green; Table 3). With the further addition of R111K, the two triple mutants again had similar VA values at +51.2 ± 5.7 mV for N48K/R111K/L112A and +61.3 ± 1.9 mV for N48K/R111K/L112F ( Figure 5B, purple and orange). The shift between N48K and N48K/R111K was +84 mV ( Figure 5A, Table 4). When R111K was added to N48K/L112A and N48K/L112F, the shifts were much smaller at +61 mV and +68 mV, respectively, suggesting that there is an opposing interaction between R111K and the two L112 substitutions ( Figure 5B, purple and orange, Table 4). The triple mutant N48K/R111K/M119V produced the largest shift in VA of all constructs tested, +240 mV, to give a VA value of +120 ± 1.8 mV ( Figure 5A, green, Table 4).
Remarkably, the triple mutation N48K/L112F/M119V shifted VA by +170 mV (Figure 6, brown, Table 4) and the triple mutation N48K/L112A/M119V shifted VA by +200 mV, all the way to VA = +80.8 ± 4.0 mV, without mutation of the R4 gating charge (Figure 6, orange, Table 4). These results demonstrate that the electrostatic force exerted on the gating charges by the transmembrane electrical field can be overcome by only three amino acid substitutions, which do not change the charge of the affected amino acid residues yet substantially alter the energetic landscape due to the chemical energy gained from new protein interactions.

Positive Shifts in the Voltage Dependence of Gating Charge Movement
Our extensive studies of broadband tuning of the voltage dependence of activation of the bacterial/human chimera 7DII showed that its voltage sensitivity could be shifted cumulatively by up to +240 mV ( Figure 5A, green). However, VA is an indirect measure of movement of the S4 gating charges. In order to compare the effects of our VA-shifting mutations on both VA and the voltage dependence of gating charge movement, VQ, we used the NaVAbD28 construct that we developed for high-level expression for structural studies (Lenaeus et al., 2017), in which 28 nonconserved amino acid residues are deleted from the C-terminus. This construct allowed introduction of combined VA-shifting mutations with high expression. We focused on combination mutations that did not include the R4 gating charge. Introduction of the combination of N48K, L112A, and M119V equivalent mutations (KAV) into NaVAb caused a substantial positive shift in the VA of NaVAb ( Figure 7A, green; Table 5). The N48K equivalent mutation alone caused a +75 mV shift of VA in NaVAb ( Figure 7A, blue). The combined KAV mutation further shifted activation positively, to a VA of +61.3 ± 2.1 mV ( Figure 7A, green). Importantly, there was no detectable sodium current activated at 0 mV ( Figure 7A, green). Thus, the NaVAbD28/KAV channel functions entirely in the positive voltage domain.
NaVAb is Cys-free, and its relative, NaChBac, has been extensively studied by insertion of Cys residues and subsequent crosslinking by disulfide bond formation (Yarov-Yarovoy et al., 2012). We reasoned that introduction of carefully selected Cys residues might further shift VA and VQ. As described previously (Wisedchaisri et al., 2019), we identified constructs with pairs of substituted Cys residues that expressed well, retained functional activity, and formed disulfide bonds in the resting state in a voltage-dependent manner. Surprisingly, we found even more positive shifts in VA when single Cys residues were substituted for G94 and Q150 (Table 5; (Wisedchaisri et al., 2019)). G94C and Q150C shifted the voltage dependence of activation of NaVAb/KAV positively to VA = +88.4 ± 1.1 mV and +76.7 ± 3.1 mV, respectively. The double mutant, analyzed under in reducing conditions to prevent disulfide bond formation, exhibited a VA of +105 ± 0.8 mV (Table 5; (Wisedchaisri et al., 2019)). Importantly, these channels do not open at 0 mV, but the question of whether their voltage sensors remain in their functionally resting position at that voltage remained to be addressed.
The outward movement of the gating charges can be measured as a capacitative current as shown in classical voltage-clamp studies of the squid giant axon (Armstrong and Bezanilla, 1973).
By measuring the outward (ON) movement of the S4 gating charges in response to incremental voltage increases from a very negative membrane potential ( Figure 7B), we found that VQ was strongly shifted to positive membrane potentials by these VA-shifting mutations ( Figure 7C). The VQ values for these constructs were -83.1 ± 0.7 mV for NaVAb/N48K, -21.5 ± 3.7 mV for NaVAb/KAV, and +77.2 ± 0.4 mV for NaVAb/KAV/G94C. These results represent a shift of +62 mV with the addition of the 'AV' mutations to N49K, and +99 mV for the subsequent addition of G94C, for a total Q/V shift of +160 mV ( Figure 7C, Table 6). For comparison, the same substitutions shifted the G/V curve by only +127 mV in total (Table 6) The OFF gating charge movement occurred at more negative potentials for all constructs, as expected, because of charge immobilization during inactivation that must be reversed by hyperpolarization. The VQ for OFF gating charge movement for NaVAb/N49K was -152 ± 0.6 mV, and the addition of 'AV' in the KAV mutant increased this by +77 mV to a VQ of -74.7 ± 6.0 mV ( Figure 7D, Table 6). Addition of G94C resulted in a further +96 mV shift of VQ to +21.4 ± 3.3 mV, representing a total shift of +173 mV ( Figure 7D, Table 6). No gating charge movement was observed for KAV/G94C/Q150C in the absence of dithiothreitol to reduce disulfide bonds (n=4, data not shown), which supports our conclusion that disulfide-locking of G94C to Q150C prevents S4 translocation.

DISCUSSION
Our aim in this project was to explore whether broadband tuning of the voltage dependence of a voltage-gated sodium channel is possible. Sodium channels are in the resting state at negative membrane potentials and are activated by depolarization of the membrane to more positive membrane potentials. Although previous work had identified mutations that shift VA significantly, those experiments had not tested the extent of voltage shifting that is possible by combinations of mutations. Here we combined previously described mutations in multiple voltage-gated ion channels in order to test the range of positive shifting of VA that is possible. We were surprised to find that combination of a small number of mutations was able to shift VA up to 240 mV and yet retain normal sodium channel expression and function. Our findings show that broadband shifting of VA of a sodium channel is possible and provide a toolbox of methods and constructs that enable exploration of sodium channel states, functions, and structure in the positive membrane potential domain that was previously inaccessible to detailed biophysical and structural studies. The utility of this approach has been proven by use of one of these constructs to lock NaVAb in its resting state and determine its structure at high resolution (Wisedchaisri et al., 2019). In addition, our results lead to several important insights into the voltage sensing and voltage-dependent gating of sodium channels as discussed below.

Voltage dependence of gating of mammalian NaV channels can be transferred to NaVAb
Sodium channels in bacteria have VA's in the range of -130 mV, consistent with the negative membrane potential of bacteria in the range of -160 mV. In contrast, mammalian sodium channels have VA's in the range of -50 mV to -20 mV, consistent with the resting membrane potentials of nerve and muscle cells of -90 mV to -70 mV. Despite this large difference in voltage sensitivity, the VS of bacterial and mammalian sodium channels are very similar in structure, with RMSD values of <3 Å in their core transmembrane domains (eg., (Jiang et al., 2020)). Our results show that changes of a small number of amino acid sidechains by mutation is sufficient to give this large difference on voltage-dependent function. Thus, in evolution, it was possible to tune VS function over a broad range without significant structural change.

NaVAb
The similarity of structure of bacterial and mammalian sodium channel VS suggests that mutations that alter conserved amino acid residues across the broad range of ion channels might have similar effects on voltage dependence. Our results show that mutations initially described in Drosophila potassium channels, mammalian sodium channels, and other distantly related ion channels have similar effects when transferred into the bacterial sodium channel NaVAb by sitedirected mutations. Evidently, the structural basis for VS function is conserved across phylogeny, and only a small number of molecular changes are required to elicit dramatic changes in voltage dependence.

Combined mutations have additive effects on voltage-dependent activation
Although individual mutations have been frequently shown to shift the VA of sodium channels, an exhaustive study of additive effects of combined mutations that shift VA has not previously been carried out. Many combination mutations we studied showed increases in VA that were greater than either individual mutation alone, combined increases that were approximately additive, or increases in VA that were even more than additive when considered simply as the sum of VA.
Thus, a majority of these mutations worked together to enhance positive shifts in VA rather than to oppose those shifts, as judged by this empirical procedure.

Combined mutations can shift sodium channel function entirely into the positive membrane potential domain
We found that multiple triple combinations of mutations are able to shift VA entirely into the positive membrane potential domain. For these triple mutations, essentially all individual sodium channels are in the resting state at 0 mV. These results show that a negative transmembrane potential exerting an electrostatic force on the positive gating charges to pull them inward is not necessary for formation of the resting state at 0 mV. Moreover, application of stimuli in the positive voltage range lead to voltage sensor activation and pore opening. Evidently, the chemical energy of altered protein interactions caused by our mutations is sufficient to pull the gating charges into their resting conformation without a negative resting membrane potential, from which they can be activated by imposing positive voltage stimuli.

VA-shifting mutations also shift the voltage dependence of gating charge movement
In principle, mutations might shift VA by impeding the outward movement of the gating charges in response to depolarization, preventing the coupling of gating charge movement to pore opening, or a combination of both effects. Importantly, for the mutations we studied most completely, VQ measured from gating currents was shifted to a comparable or even greater extent than VA. These studies support the conclusion that the gating charges are locked in place in their resting state by the chemical energy of protein interactions contributed by the mutated amino acid residues in our constructs until the membrane potential is depolarized to 0 mV or beyond.

Stabilizing sodium channel functional states
All structures of native wild-type bacterial and eukaryotic sodium channels have activated or partially activated voltage sensors because there is no membrane potential in a protein crystal or on a cryo-EM grid. Determination of the structures of other sodium channel states will require stabilizing them at 0 mV. Our results provide a toolbox of methods and mutant cDNAs that can stabilize specific functional states at 0 mV for structural studies. We have already used the KAV triple mutant and disulfide locking of substituted Cys residues to stabilize the resting state of Wisedchaisri, G., Tonggu, L., McCord, E., Gamal El-Din, T.M., Wang, L., Zheng, N., and Catterall, W.A. (2019). Resting-state structure and gating mechanism of a voltage-gated sodium channel.
Zhang, X., Ren, W., DeCaen, P., Yan, C., Tao, X., Tang, L., Wang, J., Hasegawa, K., Kumasaka, T., He, J., et al. (2012). Crystal structure of an orthologue of the NaChBac voltage-gated sodium channel. Nature 486, 130-134. Figure 1. Voltage dependence of activation of NaVAb and NaVAb/NaV1.7 chimeras. A. Voltage sensor model highlighting key functional components: ENC, extracellular negative cluster (red); HCS, hydrophobic constriction site (gold); INC, intracellular negative cluster (red); gating charge Arg residues (blue). B. Conductance-voltage (G/V) relationships of three chimeras containing increasing numbers of human NaV1.7 residues. C. Homology models (Swiss-Model, 5X0M template (Shen et al., 2017), Domain II) were made to highlight the extent of substitution of human residues in five chimeric human-bacterial sodium channels created by replacing NaVAb residues with those of human NaV1.7 DII. The S1-S4 segments are colored from light to dark gray, with the S4-S5 linker in black. The colored portion indicates human residues, while native NaVAb residues are in varying shades of gray. The colors correspond to the different chimeric channels. The channels are arranged in order from least to most humanized: red, 7DII; orange, 7DII-2; blue, 7DII-3; yellow, 7DII-4; and pink, 7DII-5. The latter two constructs did not conduct sodium current. D. Current records of the three functional chimeras colored as indicated in B and C. Individual G/V curves were normalized to peak current values and fit with a standard Boltzmann equation. The average VA and k values were used to draw the solid line fits. Datapoints for each construct represent average values at the indicated voltages, and error bars denote mean ± SEM.

Figure 7. Comparison of the effects of the KAV mutations on G/V and Q/V relationships. A.
G/V relationships for the indicated constructs: 7DII (black dashes); NaVAb (gray dashes); 7DII/N48K (red dashes); 7DIIΔ28/N48K (purple diamonds); NaVAb/N49K (blue circles); NaVAb N49K/L112A/M1119V (KAV, light green circles); NaVAbΔ28/KAV (green diamonds), and 7DII/KAV (orange dashes). B. Voltage command diagrams and representative gating current traces of NaVAbD28/KAV for ON (left) and OFF (right) gating current measurements. Calibration bars indicate 1 nA and 10 msec. C. ON gating charge measurements in response to depolarization in media that contained NMDG-Cl substituted for NaCl and CaCl2. Cells were briefly hyperpolarized, and the subsequent current step was increased in 10 mV increments. D. OFF gating charge movement in response to hyperpolarization following brief depolarizating prepulses. A P/5 leak subtraction was used at voltages that remained well outside the range of charge movement in either the positive or negative direction. The area under the curve was calculated and plotted as a function of voltage. When necessary, traces were baseline corrected to the current remaining at the end of the test pulse. Individual Q/V curves were fit with a standard Boltzmann equation. The average VQ and k values were used to draw the solid line fits. Datapoints for each construct represent average values at the indicated voltage. Error bars denote mean ± SEM. KAV d28 denotes NaVAb/KAV/D28. Table 1. Sequence alignment of NaV1.7-DII, NaVAb WT, and chimeras. Yellow highlighting indicates transmembrane segments. Green highlighting indicates intracellular helices. Cyan highlighting indicates human residues transplanted into NaVAb to create the indicated chimeras. Intracellular and extracellular negative cluster residues in S1-S3 segments are shown in red. The arginine gating charges in S4 segment are shown in blue. As a landmark, mutations characterized in this study are highlighted in black background with residue numbers for NaVAb and 7DII chimera (in parenthesis) shown beneath.  VA, half maximal activation voltage; k, slope factor; zapp, apparent gating charge movement. Mean ± SEM. N.D., not determined. -61.0 ± 1.4 8.6 ± 0.5 2.9 ± 0.2 5 N48K -45.4 ± 0.9 6.0 ± 0.2 4.2 ± 0.2 8 N48K/D28 -47.8 ± 1.8 6.9 ± 0.3 3.6 ± 0.2 3 R111K -59.8 ± 6.6 8.4 ± 0.6 3.0 ± 0.02 2 L112A -55.0 ± 4.9 10.5 ± 1.6 2.4 ± 0.4 2 T114K -102.0 ± 2.9 8.9 ± 1.0 2.9 ± 0.4 3 M119V -70.0 ± 1.2 7.4 ± 0.2 3.3 ± 0.1 5 VA, half maximal activation voltage; k, slope factor; zapp, apparent gating charge movement. Mean ± SEM.