Charge substitutions at the voltage-sensing module of domain III enhance actions of site-3 and site-4 toxins on an insect sodium channel

Abstract

Scorpion α-toxins bind at the pharmacologically-defined site-3 on the sodium channel and inhibit channel inactivation by preventing the outward movement of the voltage sensor in domain IV (IVS4), whereas scorpion β-toxins bind at site-4 on the sodium channel and enhance channel activation by trapping the voltage sensor of domain II (IIS4) in its outward position. However, limited information is available on the role of the voltage-sensing modules (VSM, comprising S1–S4) of domains I and III in toxin actions. We have previously shown that charge reversing substitutions of the innermost positively-charged residues in IIIS4 (R4E, R5E) increase the activity of an insect-selective site-4 scorpion toxin, Lqh-dprIT_3-c, on BgNa_v1–1a, a cockroach sodium channel. Here we show that substitutions R4E and R5E in IIIS4 also increase the activity of two site-3 toxins, LqhαIT from Leiurusquinquestriatus hebraeus and insect-selective Av3 from Anemonia viridis. Furthermore, charge reversal of either of two conserved negatively-charged residues, D1K and E2K, in IIIS2 also increase the action of the site-3 and site-4 toxins. Homology modeling suggests that S2-D1 and S2-E2 interact with S4-R4 and S4-R5 in the VSM of domain III (III-VSM), respectively, in the activated state of the channel. However, charge swapping between S2-D1 and S4-R4 had no compensatory effects on gating or toxin actions, suggesting that charged residue interactions are complex. Collectively, our results highlight the involvement of III-VSM in the actions of both site 3 and site 4 toxins, suggesting that charge reversing substitutions in III-VSM allosterically facilitate IIS4 or IVS4 voltage sensor trapping by these toxins.

1. Introduction

Voltage-gated sodium channels (Na_vs) initiate and propagate action potentials in most excitable cells. Na_vs activate rapidly upon membrane depolarization and then inactivate within milliseconds (Catterall, 2000; Hodgkin and Huxley, 1952). The pore-forming α-subunit of sodium channels consists of four homologous domains (I to IV), each containing six transmembrane segments (S1–S6). Segments S5, S6 and their interconnecting linker form the pore module (PM), while segments S1–S4 form the voltage-sensing module (VSM). The S4 segment in each domain contains repeated motifs of a positively charged Arg or Lys residue followed by two hydrophobic residues (Catterall, 2000). The S4 segments sense alterations in membrane potential and respond by outward movement during channel activation, initiating a conformational change that opens the channel pore (Armstrong, 1981; Catterall, 2010), which is then followed by fast inactivation.

Negative counter charges in S1–S3 of the VSM are highly conserved (Palovcak et al., 2014) and are proposed to interact with positively charged voltage sensor residues to coordinate the S4 translocation across a gating pore within the hydrophobic milieu of the membrane in response to membrane depolarization (reviewed by Groome and Bayless-Edwards, 2020). This ‘sliding helix’ model of ion channel gating has been supported by structural analysis coupled with disulfide cross-linking experiments of the VSM in prokaryotic sodium channels (DeCaen et al., 2011; DeCaen et al., 2009; DeCaen et al., 2008; Paldi and Gurevitz, 2010; Payandeh et al., 2012; Payandeh et al., 2011; Yarov-Yarovoy et al., 2012; Zhang et al., 2012a; reviewed by Groome, 2014) and with molecular dynamics simulations in eukaryotic channels (Delemotte et al., 2011; Gosselin-Badaroudine et al., 2012).

Due to their critical role in neuronal signaling and muscle fiber contractility, sodium channels are targeted by a variety of naturally-occurring neurotoxins, including venom toxins from scorpions and spiders. These toxins are thus valuable tools for exploring structure-function relations in sodium channels (Catterall, 1992; Cestele and Catterall, 2000; Dutertre and Lewis, 2010; Gordon, 1997; Nicholson, 2007; Terlau and Olivera, 2004). They are also potential leads for development of new selective insecticides because many of these venom toxins exhibit host specificity, being toxic to insects, but not to mammals (Krapcho et al., 1995; Strugatsky et al., 2005; Zlotkin et al., 2000; Zhu et al., 2020).

Scorpion α-toxins, several sea anemone toxins and some spider toxins inhibit channel fast inactivation upon binding to the pharmacologically defined site-3, which comprises residues in the VSM of domain IV and residues in the PM of domain I (Catterall et al., 2007; Gordon et al., 1992; Gur Barzilai et al., 2014; Gur et al., 2011; Gurevitz, 2012; Ma et al., 2013; Moran et al., 2007; Nicholson, 2007; Thomsen and Catterall, 1989; Wang et al., 2011). On the other hand, scorpion β-toxins enhance the activation of sodium channels upon binding to site-4, which is formed by residues in the VSM of domain II and the pore module of domain III (Gurevitz et al., 2007; Song et al., 2011; Zhang et al., 2011, 2012b).

The voltage sensor trapping model, in which toxins hinder the inward (deactivating) movement of the S4 segment from its outward activated state, illustrates the pivotal role of S4 in domains II and IV in the action of site-4 and site-3 toxins, respectively (Catterall et al., 2007; Cestele and Catterall, 2000; Zhang et al., 2011, b). However, the role of the VSMs of domains I and III in the action of these toxins is largely unclear. Song et al. (2011) previously reported that charge-reversing substitutions of the fourth and fifth gating charges in IIIS4, R4E and R5E, increased the sensitivity of the cockroach sodium channel, BgNa_v1–1a, to the insect-selective scorpion β-toxin, Lqh-dprIT₃, suggesting that IIIS4 modulates the voltage-sensor trapping of IIS4. In this study, we examined the role of the VSM of domain III (III-VSM) in the action of two site-3 toxins, LqhαIT from the scorpion Leiurusquinquestriatus hebraeus and the insect-selective Av3 from the sea anemone Anemonia viridis, on the BgNa_v1–1a channel. We systematically conducted charge-reversal mutational analysis of the five positively-charged residues in IIIS4 and of four negatively-charged residues in IIIS1 and IIIS2 in BgNa_v1–1a, and examined the effects of the two site-3 toxins and the site-4 toxin Lqh-dprIT₃ on these mutant channels. Our results show that alterations in III-VSM, which is not considered to constitute the binding site of site-3 or site-4 toxins, affect the action of both types of toxins, most likely by an allosteric mechanism.

2. Methods

2.1. Toxin production and functional analysis

Lqh-dprIT_3-c (hereafter referred to as Lqh-dprIT₃) and LqhαITof the scorpion Leiurusquinquestriatus hebraeus, and Av3 from the sea anemone Anemonia viridis, were produced in recombinant forms and analyzed as previously described (Moran et al., 2007; Strugatsky et al., 2005; Zilberberg et al., 1996).

2.2. Site-directed mutagenesis

Site-directed mutagenesis was performed via PCR using oligonucleotide primers and Pfu Turbo DNA polymerase (Stratagene, La Jolla, CA, USA). The primers used in site-directed mutagenesis are summarized in Table S1. The mutations were verified by DNA sequencing.

2.3. Expression of BgNa_v1–1a channels in Xenopus oocytes

Xenopus oocytes were purchased from Xenopus 1 (Dexter, Michigan). The procedures for oocyte preparation and cRNA injection were identical to those described previously (Tan et al., 2002). For robust expression of BgNa_v1–1a channels, cRNA (1 ng) was co-injected into oocytes with D. melanogaster tipE cRNA (1:1 ratio), to enhance channel expression (Feng et al., 1995; Warmke et al., 1997).

2.4. Electrophysiological recording and analysis

Voltage dependent activation and fast inactivation were measured and analyzed using the two-electrode voltage clamp technique as previously described (Tan et al., 2005). Sodium currents were measured with a Warner OC-725C oocyte clamp and a Digidata 1440A interface. Data were sampled at 50 kHz and filtered at 2 kHz. Leak currents were corrected by p/4 subtraction. pCLAMP 10.2 was used for data acquisition and analysis. The maximal sodium current peak amplitude was limited to ˂ 2.0 μA to achieve optimal voltage control by adjusting the concentration of cRNA injected and timing of recording. The protocols to determine gating properties and toxin effects were identical to those described previously (Gao et al., 2014; Song et al., 2011). To examine the effects of Lqh-dprIT₃, a 20 Hz train of 50, 5-ms depolarizing prepulses to 50 mV (conditioning pulses) were applied, followed by a 20 ms depolarizing test pulse between −80 and 65 mV from a holding potential of −120 mV. The voltage dependence of sodium channel conductance (G/V) was calculated by measuring peak current amplitude in response to test potentials ranging from −80 to 65 mV in 5 mV increments and divided by (V - V_rev), where V is the test potential and V_rev is the reversal potential for sodium ions. Peak conductance values were normalized to the maximal conductance (G_max) and fitted with a two-state Boltzmann equation G/G_max = [1 + exp (V–V_1/2)/k]⁻¹, where k is the slope factor, V is the command potential and V_½ is the voltage for half-maximal activation.

The extent of channel modification by Lqh-dprIT₃ was determined by the percentage of channels with the voltage dependence of activation shifted to negative membrane potentials, as derived from fractional amplitudes of double Boltzmann fits of the conductance-voltage relationship. The percentages of channel modification by LqhαIT and Av3 were measured by I₂₀/I_peak, where I_peak is the peak current elicited by a 20 ms step depolarization to −10 mV and I₂₀ is the non-inactivated current (due to toxin action) at the end of that depolarization. The ratio of I₂₀/I_peak reflects the potency of α-toxins affecting fast inactivation (Gao et al., 2014). Data are presented as means ± SEM. Statistical significance was determined by using one-way analysis of variance (ANOVA) with Scheffe’s post hoc analysis and significant value were set at p˂0.05.

2.5. Structural modeling

The secondary structure of Na_v channels was predicted using JPred4 (Drozdetskiy et al., 2015) to confirm the preservation of helices. Multiple sequence alignment for BgNa_v domains I-IV, Na_vAb from the bacterium Arcobacter butzleri (Payandeh et al., 2011), and NaChBac from Bacillus halodurans was carried out using MUSCLE (Edgar, 2004). The structure of Na_vAb was downloaded from the Protein Data Bank (4EKW. pdb) and used as template. Homology modelling was performed using Modeller 2.0 (Payandeh et al., 2012; Sali and Blundell, 1993) and the best structure was selected based on the GA341 output. Energy minimization (1500 steps) to detect possible charge interactions was carried out using AmberTools. The convpdb.pl utility in the MMTSB tool set (Feig et al., 2004) was used to prepare the input for the AmberTools. The resulting protein structures were visualized using VMD (Humphrey et al., 1996). Multi-sequence alignment was conducted with Multalign Viewer in UCSF Chimera package (Pettersen et al., 2004).

3. Results

3.1. Substitutions R4E and R5E in IIIS4 increase the inhibitory effect of Av3 and LqhαIT on channel inactivation

The inhibitory effect of Av3 and LqhαIT on fast inactivation of BgNa_v1–1a was quantified by determining the ratio of I₂₀/I_peak, as was previously described (Gao et al., 2014) (Fig. 1A and B). In addition, an increase in amplitude of the peak current was observed. The effect of charge reversal for each of the five positively-charged channel residues in IIIS4 was examined by measuring the activity of Av3 and LqhαIT on the channel mutants. Compared to WT BgNa_v1–1a, charge reversal of the 4th arginine (R4E) or 5th arginine (R5E) increased the inhibitory effect of either toxin (Fig. 1C and D; Fig. S1A), while charge reversals K1E or R3E had no effect on the action of either toxin at any dose tested. Substitution R2E revealed a significant effect on the action of LqhαIT at 10 nM, but not at 3 nM. Notably, in the absence of toxins, fast inactivation of the mutant channels was similar to that of the WT channel.

Fig. 1.

Substitutions R4E and R5E in IIIS4 increase the action of Av3 and LqhαIT. A-B, Representative sodium current traces in the absence or presence of Av3 (A) or LqhαIT (B). The protocols are presented above the current traces: the currents were elicited by a 20 ms test pulse to −10 mV, from a holding potential of −120 mV. C-D, Effects of charge reversal of the five positively-charged residues in IIIS4 on BgNa_v1–1a sensitivity to Av3 (C) or LqhαIT (D). The inhibitory effect on fast inactivation by the toxins was calculated by measuring the current that remained at 20 ms (I₂₀) divided by the peak current (I_peak). * denotes significant difference compared to WT BgNa_v1–1a channels (p˂ 0.05).

3.2. Charge reversal of two negative residues in IIIS2 facilitates the action of site-3 and site-4 toxins

Negative countercharges in S1–S3 of the VSMs presumably interact with positively-charged voltage sensor residues (DeCaen et al., 2008, 2009, 2011; Chakrapani et al., 2010; Paldi and Gurevitz, 2010; Gosselin-Badaroudine et al., 2012; Yarov-Yarovoy et al., 2012; Groome and Winston, 2013) and so stabilize intermediate channel states during S4 translocation in response to alteration in membrane potential. These interactions may affect toxin binding, as it has been previously shown that substitution of the outermost negative charge E1290 in IIIS1 increases the sensitivity of a BgNa_V1–1a splice variant to the site-4 toxin Lqh-dprIT₃ (Song et al., 2011). To examine whether other counter-charges in S1 or S2 play a role in the action of site-3 and site-4 toxins, we examined the effect of toxins on mutations for each of the four highly conserved negatively-charged residues in the VSM of channel domain III. In S1, intracellular E1268 and E1273 were substituted with lysine (S1-E1K and S1-E2K, respectively). In S2, the extracellular D1308 and intracellular E1318 were also substituted with lysine, (S2-D1K and S2-E2K, respectively). All four channel mutants generated sufficient sodium currents in Xenopus oocytes enabling functional analyses.

The effects of these charge reversing mutations on channel gating are shown in Fig. 2. Compared to WT channels, substitutions S1-E1K, S1-E2K and S2-E2K rendered a shift in the voltage-dependence of channel fast inactivation to more hyperpolarizing voltages with no significant effect on channel activation (Fig. 2B and C; Table 1). In contrast, substitution S2-D1K caused a shift in the voltage-dependence of channel activation and fast inactivation to more depolarizing voltages (Fig. 2B and C; Table 1), and increased the extent of inactivation at potentials more hyperpolarized than −80 mV (Fig. 2C).

Fig. 2.

Voltage dependence of activation and inactivation of four charge reversal channel mutants at IIIS1 and IIIS2. A, Schematic diagram of a sodium channel indicating the positions of the charge reversing substitutions of four highly conserved positively-charged residues and negatively-charged residues in domain III. B–C, Voltage dependence of activation (B) and inactivation (C). Data are presented as mean ± SEM for 10–20 oocytes. * indicates a significant difference compared to that of BgNav1–1a channels only using a one-way ANOVA with Scheffe’s post hoc analysis (p < 0.05).

Table 1.

Voltage dependence of activation and inactivation of BgNa_v1–1a and mutants.

Nav channel type	Activation		Inactivation		n
	V1/2 (mV)	k (mV)	V1/2 (mV)	k (mV)
BgNav1–1a	− 29.6 ± 1.3	4.6 ± 0.4	− 49.2±1.0	4.7 ± 0.1	11
IIIS1-E1K	− 31.3 ± 1.2	5.6 ± 0.2	−56.1 ± 0.9*	5.1 ± 0.1	26
IIIS1-E2K	− 30.8 ± 1.0	7.4 ± 0.4	− 61.9 ± 1.1*	5.2 ± 0.1	13
IIIS2-E2K	− 31.1 ± 1.6	6.6 ± 0.3	− 53.7 ± 1.2*	6.0 ± 0.2	13
IIIS2-D1K	−26.0 ± 0.9*	7.1 ± 0.2	− 43.7 ± 1.2*	3.7 ± 0.2	13
IIIR4E	− 36.2 ± 1.8*	4.6 ± 0.4	− 49.2 ± 1.0	4.7 ± 0.1	11
IIIS2-D1K/IIIR4E	− 41.2 ± 1.2*	8.1 ± 0.2	− 79.3 ± 1.5*	8.8 ± 0.5	15

The voltage dependence values of conductance and inactivation were fitted with a two-state Boltzmann equation to determine V_1/2, the voltage for half maximal conductance or inactivation, and k, the slope factor for conductance or inactivation. The values in the table represent the mean ± SEM, and n is the number of oocytes tested. The asterisks indicate significant differences as determined by one-way ANOVA (p < 0.05) with Scheffe’s post hoc analysis.

Charge reversing substitutions in S2, S2-D1K or S2-E2K, but not in S1, S1-E1K, S1-E2K, increased the effect of the site-3 toxins. Whereas both Av3 and LqhαIT induced greater non-inactivating currents in S2-D1K and S2-E2K channels compared to WT channels, their effect on S1-E1K and S1-E2K channel mutants was similar to that obtained on WT channels (Fig. 3; Fig. S1B). A similar profile was obtained when monitoring the effects of the charge reversing substitutions on the action of Lqh-dprIT₃ site-4 toxin. At 300 nM, Lqh-dprIT₃ enhanced BgNa_v1–1a activation, as indicated by a shift in the hyperpolarizing direction with 28% of the channels modified. This effect of toxin was increased in mutant IIIS2 channels (75% modified for S2-D1K channels and 67% for the S2-E2K channels (Fig. 4D–F; Table 2) but was unaltered by mutant IIIS1 channels (Fig. 4A–C, F; Table 2).

Fig. 3.

Charge reversal of two negatively-charged residues in IIIS2 increase the effect of site-3 toxins. A-B, Representative sodium current traces of channel mutants S2-D1K and S2-E2K in the absence or presence of Av3 (A) or LqhαIT (B). C-D, Effects of LqhαIT (C) and Av3 (D) on fast inactivation of the four charge reversal channel mutants. The toxin inhibitory effects on fast inactivation were calculated by measuring the current that remained after 20 ms (I₂₀) divided by the peak current (I_peak) for S1-E1K, S1-E2K and S2-E2K channel mutants. Since after 20 ms, the inactivation of S2-D1K was not complete, the remaining current of this channel mutant was determined only after 50 ms, and so the value measured was I₅₀/I_peak. *denotes significant difference compared to WT BgNa_v1–1a channels determined by one-way ANOVA with Scheffe’s post hoc analysis (p˂ 0.05).

Fig. 4.

Charge reversal of two negatively-charged residues in IIIS2 increase the effect of Lqh-dprIT₃. A-E, Conductance-voltage relations in WT BgNa_v1–1a channels (A) or the four channel mutants substituted in domain III (B-E) in the absence (●) or presence (○) of 300 nMLqh-dprIT₃. To measure the toxin effect, a 20 Hz train (50) of 5 ms depolarizing prepulses to 50 mV from a holding potential of −120 mV was followed by a series of 20 ms depolarizing test pulses between −80 and −65 mV. F, Percentage of modification of the WT and four mutant channels elicited by Lqh-dprIT₃. * denotes a significant difference compared to WT BgNa_v1–1a channels, as determined by one-way ANOVA with Scheffe’s post hoc analysis (p˂ 0.05).

Table 2.

Voltage dependence of activation of BgNa_v1–1a and of the four negatively-charged channel mutants before and after the application of 300 nM Lqh-dprIT₃.

Toxin-free			Lqh-dprIT3 (300 nM)
V0.5k			V0.5	k 1	A1 (%)		V0.5	k 2	A2 (%)
BgNav1–1a	−31.3 ± 2.2	4.9 ± 0.3	−32.0 ± 1.9		3.9 ± 0.3	72 ± 5	−68.0 ± 4.6		8.5 ± 1.0	28 ± 1
IIIS1-E1K	−31.2 ± 1.1	5.2 ± 0.2	−33.7 ± 1.2		4.4 ± 0.2	77 ± 4	−70.7 ± 2.6		9.2 ± 0.5	23 ± 4
IIIS1-E2K	−31.2 ± 0.5	5.7 ± 0.2	−34.2 ± 1.0		4.7 ± 0.2	69 ± 7	−75.0 ± 2.1		11.0 ± 0.8	31 ± 7
IIIS2-D1K	−25.9 ± 0.8	7.6 ± 0.2	−55.2 ± 5.8		4.3 ± 0.8	25 ± 6	−75.0 ± 3.5		5.4 ± 0.5	75 ± 6a
IIIS2-E2K	−29.4 ± 1.2	7.5 ± 0.3	−30.7 ± 1.8		4.8 ± 0.6	33 ± 6	−69.9 ± 1.9		7.7 ± 0.9	67 ± 6a

Each value represents the mean ± SEM for at least eight oocytes.

^aSignificant differences from other mutants were determined by one-way ANOVA analysis of variance (p ˂ 0.05).

3.3. The D1K/R4E channel is equally sensitive to the site-3 toxins as D1K or R4E channels

To further examine whether electrostatic interactions between R4 in IIIS4 and negatively-charged residues in IIIS2 affect the gating of BgNa_V1–1a and its interaction with site-3 and site-4 toxins, we constructed the double mutants S2-D1K/R4E and S2-E2K/R4E. Of these only the S2-D1K/R4E channels generated observable sodium currents in Xenopus oocytes. Compared to WT channels, S2-D1K/R4E channels exhibited a 12 mV hyperpolarizing shift in the voltage dependence of activation, and a 30 mV hyperpolarizing shift in the voltage dependence of inactivation (Table 1). The sensitivity of S2-D1K/R4E mutant channels to Av3 or LqhαIT was comparable, although not identical, to that of the single mutation channels, S2-D1K and R4E (Fig. 5). LqhαIT inhibited the fast inactivation of WT channels at 60%, S2-D1K channels at 85%, S4-R4E, and D1K/R4E channels at 100%. The sensitivity of the S2-D1K/ R4E mutant channels could not be evaluated in the presence of 300 nM Lqh-dprIT₃ due to a marked decrease in the peak current amplitude along a significant increase in leak currents, which prevented measurements of channel activation at more negative potentials.

Fig. 5.

The sensitivity of S2-D1K/R4E double mutant channels to Av3 and LqhαIT compared to channel mutants S2-D1K and R4E. A-B, Representative sodium current traces of R4E and IIIS2-D1K/R4E channels in the absence or presence of Av3 (A) and LqhαIT (B). C-D, Effect of S2-D1K/R4E double substitution on channel sensitivity to Av3 (C) and LqhαIT (D) in comparison to the single channel substitution. * denotes significant differences compared to WT BgNa_v1–1a channels as determined by one-way ANOVA with Scheffe’s post hoc analysis (p˂ 0.05).

3.4. Putative electrostatic interactions at the voltage-sensing module of domain III

The results on gating for charge reversal substitutions at III-VSM followed by those for the analyses of toxin effects show that interference of electrostatic interactions in this channel region not only affect the coordinated movement of the voltage sensor, but also influence the actions of the various toxins. To illustrate the critical electrostatic interactions in III-VSM, we aligned the corresponding sequences of various sodium channels (Fig. S2A) and constructed a three-dimensional model of that region in BgNa_v1–1a (Fig. S2B) using Na_vAbas template in the activated state (4EKW.pdb). The model shows that during channel activation R3, R4 and R5 are in close proximity and may interact with E1290 (IIIS1) (Song et al., 2011), D1308 (S2-D1) and E1318 (S2-E2), respectively, of the negatively-charged residue cluster in domain III. These putative interactions resemble those suggested in disulfide-locking studies between the analogous S4 and S2 residues of the bacterial channel (NaChBac; Fig. S2A).

4. Discussion

Binding site-3 is recognized by a variety of structurally distinct toxins, such as scorpion α-toxins (Gordon et al., 2007), sea anemone toxins (Moran et al., 2006, 2007), and spider toxins (Nicholson, 2007; Rash and Hodgson, 2002). Inhibition of sodium channel fast inactivation by site-3 toxins involves channel residues of the VSM in domain IV and/or of the pore module in domain I (Gordon et al., 1992; Gur et al., 2011; Gurevitz, 2012; Ma et al., 2013; Thomsen and Catterall, 1989; Wang et al., 2011). In domain-swapped channels, these regions are in proximity, suggesting overlapping but non-identical toxin binding sites in these two regions. For example, the sea anemone toxin Av3 competes with LqhαIT for binding to insect sodium channels and similarly inhibits the inactivation process. However, Av3 does not bind at linker S3–S4 in domain IV of site-3 in the Drosophila sodium channel, but rather at a cleft in S6 of domain I (Gur Barzilai et al., 2014). In contrast, the D1701R substitution in S3–S4 at domain IV of the DmNa_v1 channel, which abolishes the activity of a scorpion α-toxin, does not affect Av3 action (Moran et al., 2007).

Here we show that substitution of VSM residues in domain III (R4, R5 in IIIS4 and D1, E2 in IIIS2) increases the action of LqhαIT and Av3 site-3 toxins. Together with findings from a previous study of a β-toxin (Song et al., 2011), our results indicate that substitutions of the innermost positively-charged residues R4 and R5 in IIIS4 enhance the action of both site-3 and site-4 toxins on BgNa_v1–1a channels. Our functional data and interaction model suggest that R4 and R5 provide the essential interactions with III-VSM negative charges during gating that enhance the actions of site-3 and site-4 toxins. These residues in III-VSM are unlikely to be directly involved in toxin binding to the channel, as was previously suggested using site-directed antibodies (Gordon et al., 1992; Thomsen and Catterall, 1989) and mutagenic dissection (Gur et al., 2011; Ma et al., 2013), thus raising questions as to the mechanism involved. Despite their differences in binding, both Av3 and LqhαIT prolong the fast-inactivation of S2-D1K (D1308K), S2-E2K (E1318K), R4E and R5E BgNa_v1–1a channel mutants (Figs. 1 and 3). Thus, differences in toxin interaction with site-3 seem irrelevant to the effects mediated by substitutions in III-VSM. The increased action of the toxins upon charge reversals in III-VSM may therefore reflect an allosteric effect on the coordinated movements of the VSMs in domains III and IV during fast-inactivation, a possibility consistent with findings that show that movements of S4 in sodium channels are not independent, but coupled (Chanda et al., 2004).

Binding site-4 (for scorpion β-toxins) involves channel residues in the VSM of domain II and in the pore module of domain III (Zhang et al., 2011, 2012b). We have previously shown that L1285P in IIIS1 and E1290 A/K at the extracellular end of IIIS1 enhance β-toxin action, possibly by facilitating VSM trapping in domain II (Song et al., 2011). Similar to the action of site-3 toxins, the increased action of Lqh-dprIT₃ in the charge reversing channel mutants observed in this study could result from an allosteric effect between domains II and III.

In an attempt to visualize putative interactions among residues in III- VSM, we constructed a homology model of BgNa_v1–1a in the activated state, using as template the crystal structure of the analogous homotetrameric bacterial sodium channel Na_vAb (Payandeh et al., 2011). As shown in this model (Fig. S2B), the positive charge R3 likely interacts with E1290 during activation, consistent with results from a previous study on the role of E1290 in scorpion β-toxin action (Song et al., 2011). The model also suggests that R4 and R5 may interact with S2-D1 and S2-E2, respectively, consistent with our findings that substitutions D1K and E2K in IIIS2, but not E1K and E2K in IIIS1, markedly increased the activity of LqhαIT, Av3 and Lqh-dprIT₃. D1and E2 in IIIS2 of BgNa_v1–1a are analogous in position to E283 and E293 in S2 of the voltage-dependent Shaker potassium channel, in which electrostatic interactions exist between E283 in S2 and R368 and R371 in S4, and between E293 in S2 and K374 in S4 (Papazian et al., 1995; Tiwari-Woodruff et al., 1997). Studies using NaChBac have suggested that specific interactions of S4 positively-charged residues with negative countercharges in S1 and S2 regulate S4 outward movement upon membrane depolarization (Chakrapani et al., 2010; DeCaen et al., 2008, 2009, 2011; Paldi and Gurevitz, 2010; Yarov-Yarovoy et al., 2012). Negative countercharges may stabilize the S4 sensors at rest, and, upon membrane depolarization, facilitate sequential ion pair interactions controlling the movement of the S4 voltage sensors during gating (DeCaen et al., 2008, 2009, 2011; Paldi and Gurevitz, 2010). These countercharges are also analogous in position to those in BgNa_v1–1a in our study. Clearly, the extracellular IIIS1 residue E1290, extracellular IIIS2 residue D1308, and intracellular IIIS2 residue E1318 play a critical role in BgNa_v1–1a gating, and disruption of the putative interactions of these residues with IIIS4 partners by charge reversing substitutions in III-VSM likely accounts for the facilitation of toxin effects at sites 3 and 4.

Interestingly, our data from the D1K/R4E double mutant showed that charge swapping between S2-D1 and S4-R4 could not compensate each other for the effects on either gating or toxin actions, suggesting complexity in the interaction of charged residues in III-VSM. Specific interactions of negatively and positively charged residues have been demonstrated functionally in prokaryotic sodium channels (DeCaen et al., 2008, 2009, 2011; Paldi and Gurevitz, 2010; Yarov-Yarovoy et al., 2012) by molecular dynamics simulations in eukaryotic sodium channels (Gosselin Badaroudine et al., 2012; Moran et al., 2014; Groome and Bayless-Edwards, 2020) and by cryo-EM determination of channel structure (Payandeh et al., 2011, 2012; Zhang et al., 2012; Pan et al., 2018; Shen et al., 2018). These studies support our premise that proximity of S1 and S2 negatively charged residues with specific S4 positively charged residues in III-VSM, but also point out the need for future investigation of the specific nature of these interactions during gating and channel response to toxin exposure.

Recent cryo-EM studies of structures of sodium channels in complex with scorpion and spider toxins predicted that toxin-induced conformational changes in the VSMs of domains II and IV are transduced allosterically to the extracellular linkers in the VSM of domain III (Clairfeuille et al., 2019; Shen et al., 2019).In the case of the scorpion α-toxin AaH2 (Androctonus australis Hectorα-toxin 2) on a chimeric channel, Na_v1.7-Na_vPaS, which bears the pore module of a cockroach (Periplaneta americana) channel and the VSM of domain IV of the human channel Na_v1.7 (Clairfeuille et al., 2019), not only that toxin binding at the VSM of domain IV rendered a considerable shift (over 13 Å) of the VSM in the intracellular direction, but it also induced conspicuous alterations at the extracellular loops S3–S4 and S1–S2 in the VSM of domain III (Fig. S3A). Similarly, noticeable alterations in the extracellular loops in the VSM of domain III have been shown in another cryo-EM study upon binding of a site-4 toxin Dc1a (from the venom of the desert spider Diguetiacanities) at the VSM of domain II of a cockroach Na_vPaS channel (Fig. S3B) (Shen et al., 2018). Based on these observations and our results, we suggest that charge reversal of ionizable residues in III-VSM may alter closed-state transitions in III-VSM that allosterically modulate extracellular linkers of the VSMs of domains II and IV. These alterations may prolong toxin binding to sites-3 and 4 and increase their effects. Together, this study not only extends the proposed role of III-VSM in sodium channel gating, but also demonstrates the allosteric effects among the VSMs, as reflected by alterations in the activities of site-3 and site-4 toxins.