Scorpion toxin inhibits the voltage‐gated proton channel using a Zn²⁺‐like long‐range conformational coupling mechanism

Abstract

Background and Purpose

Blocking the voltage‐gated proton channel H_V1 is a promising strategy for the treatment of diseases like ischaemia stroke and cancer. However, few H_V1 channel antagonists have been reported. Here, we have identified a novel H_V1 channel antagonist from scorpion venom and have elucidated its action mechanism.

Experimental Approach

H_V1 and NaV channels were heterologously expressed in mammalian cell lines and their currents recorded using whole‐cell patch clamp. Site‐directed mutagenesis was used to generate mutants. Toxins were recombinantly produced in Escherichia coli. AGAP/W38F‐H_V1 interaction was modelled by molecular dynamics simulations.

Key Results

The scorpion toxin AGAP (anti‐tumour analgesic peptide) potently inhibited H_V1 currents. One AGAP mutant has reduced Na_V channel activity but intact H_V1 activity (AGAP/W38F). AGAP/W38F inhibited H_V1 channel activation by trapping its S4 voltage sensor in a deactivated state and inhibited H_V1 currents with less pH dependence than Zn²⁺. Mutation analysis showed that the binding pockets of AGAP/W38F and Zn²⁺ in H_V1 channel partly overlapped (common sites are His140 and His193). The E153A mutation at the intracellular Coulombic network (ICN) in H_V1 channel markedly reduced AGAP/W38F inhibition, as observed for Zn²⁺. Experimental data and MD simulations suggested that AGAP/W38F inhibited H_V1 channel using a Zn²⁺‐like long‐range conformational coupling mechanism.

Conclusion and Implications

Our results suggest that the Zn²⁺ binding pocket in H_V1 channel might be a hotspot for modulators and valuable for designing H_V1 channel ligands. Moreover, AGAP/W38F is a useful molecular probe to study H_V1 channel and a lead compound for drug development.

Abbreviations

2GBI

2‐guanidinobenzimidazole

AGAP

antitumour analgesic peptide

ESI‐MS

electrospray ionization mass spectrometry

ICN

intracellular Coulombic network

IPTG

isopropyl β‐d‐thiogalactoside

MALDI‐TOF MS

matrix‐assisted laser desorption/ionization time‐of‐flight mass spectrometry

MD

molecular dynamics

NOX

NADPH oxidase

NPT

the constant temperature and pressure

PLL

poly‐l‐lysine

POPC

1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phosphocholine

RP‐HPLC

reverse‐phase HPLC

What is already known

• Blockade of H_V1 channel could be a promising strategy for the treatment of H_V1‐related diseases.

• Zn²⁺ inhibits H_V1 channel currents by binding to an extracellular pocket on the channel.

What this study adds

• The scorpion toxin mutant AGAP/W38F is a novel H_V1 channel antagonist with improved selectivity.

• AGAP/W38F blocks H_V1 channel using a Zn²⁺‐like long range conformational coupling mechanism.

What is the clinical significance

• The Zn²⁺ binding pocket on H_V1 might be a promising docking site for drug design.

• AGAP/W38F is a valuable probe to study H_V1 and a lead molecule for drug development.

1. INTRODUCTION

In 2006, the voltage‐gated proton channel, https://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=124, underlying the voltage‐gated proton currents was identified by two independent groups (Ramsey, Moran, Chong, & Clapham, 2006; Sasaki, Takagi, & Okamura, 2006). This channel is a functional dimer, with each subunit constructed by four transmembrane segments and containing a proton‐conducting pore (Koch et al., 2008; Lee, Letts, & Mackinnon, 2008; Tombola, Ulbrich, & Isacoff, 2008). These features make it unique in the voltage‐gated ion channel (VGIC) family as other members are usually fourfold symmetry and contain one central pore. As a VGIC member, the gating of the H_V1 channel is primarily operated by transmembrane voltage changes. The S4 segment which carries the gating charges moves outward in response to membrane depolarization (Carmona et al., 2018; De La Rosa & Ramsey, 2018; Fujiwara et al., 2012; Gonzalez, Rebolledo, Perez, & Larsson, 2013), and the concerted conformation change within the voltage sensor opens the pore (Mony, Berger, & Isacoff, 2015; Qiu, Rebolledo, Gonzalez, & Larsson, 2013; Villalba‐Galea, 2014). Another important feature of the gating of H_V1 channel is that ΔpH across the membrane profoundly modulates voltage‐dependent activation (Cherny, Markin, & DeCoursey, 1995). The underlying mechanism remains not fully understood, and some titratable residues in the channel might be involved in stabilizing the channel in its closed or open conformation, depending on their protonation states (DeCoursey, 2018).

H_V1 channel is widely expressed in immune cells, microglia, sperm, airway epithelial cells and in some cancer cells, where it plays critical roles in physiological and pathophysiological conditions. The best known physiological role of H_V1 channel is that it maintains sustained https://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=993 activity in the immune cells by compensating charge and relieving intracellular acidification (DeCoursey, Morgan, & Cherny, 2003; El Chemaly et al., 2010; Henderson, Chappell, & Jones, 1987; Henderson, Chappell, & Jones, 1988; Morgan et al., 2009; Okochi, Sasaki, Iwasaki, & Okamura, 2009; Ramsey, Ruchti, Kaczmarek, & Clapham, 2009; Schrenzel et al., 1998). Hence, pharmacological inhibition of the H_V1 channel could be promising in the treatment of ischaemic stroke, inflammatory, and allergic diseases (Seredenina, Demaurex, & Krause, 2015), where H_V1‐coupled ROS production is an important pathogenic factor. This has been supported by study showing that H_V1‐deficient mice are protected from brain damage in the pMCAO ischaemic stroke model (Wu et al., 2012). Furthermore, several studies suggested H_V1 channel might play a role in cancer, being involved in proliferation and metastasis of tumour cells (Wang et al., 2011; Wang, Li, Wu, Che, & Li, 2012; Wang, Wu, Li, Zhang, & Li, 2013; Wang, Zhang, & Li, 2013). It was proposed that blocking the H_V1 channel could be a promising approach in cancer therapy by modulating the immunosuppressive tumour micro‐environment and inducing tumour cell acidification (Fernandez, Pupo, Mena‐Ulecia, & Gonzalez, 2016). An earlier study showed that H_V1 channel inhibition promoted apoptosis of leukaemic Jurkat cells (Asuaje et al., 2017). However, compared with the voltage‐gated Na⁺, K⁺, and Ca²⁺ channels, for which an abundance of modulators have been discovered, few modulators have been described for H_V1 channel. Zn²⁺ (Cherny & DeCoursey, 1999; Ramsey et al., 2006; Takeshita et al., 2014), guanidine derivatives (Hong, Kim, & Tombola, 2014; Hong, Pathak, Kim, Ta, & Tombola, 2013), hanatoxin (Alabi, Bahamonde, Jung, Kim, & Swartz, 2007), and Corza6 peptide (Zhao et al., 2018) were reported to effectively inhibit H_V1 currents by directly interacting with the channel. For Zn²⁺ and guanidine derivatives, their mechanisms of inhibiting H_V1 channel have been extensively studied, but shortcomings in their affinity and selectivity as H_V1 channel antagonists have also been observed. Recently, De La Rosa et al demonstrated that Zn²⁺ inhibits H_V1 channel using a long‐range conformational coupling mechanism (De La Rosa, Bennett, & Ramsey, 2018). However, it remains unknown whether other extracellularly bound H_V1 channel antagonists use similar mechanism.

In summary, identifying novel H_V1 channel antagonists with high affinity and selectivity and clarifying their action mechanism will certainly advance our understanding of the this proton channel and promote development of H_V1 channel ligands. Here, we have characterized a scorpion toxin as a novel H_V1 channel antagonist and shown that it acts through a Zn²⁺‐like long‐range conformational coupling mechanism.

2. METHODS

2.1. Venom and toxin purification

Scorpion Buthus martensii venom was collected by an electrical stimulation method, lyophilized, and stored at −80°C. The venom was dissolved in ddH₂O to a final concentration of 5 mg·ml⁻¹ and subjected to the first round of RP‐HPLC purification in a C18 column (10 × 250 mm, 5 μm, Welch Materials Inc., Shanghai, China) using a 50‐min linear acetonitrile gradient from 5% to 55%, at a flow rate of 3 ml·min⁻¹. The antitumour analgesic peptide (AGAP) fraction was further purified to homogeneity by the second round of RP‐HPLC purification in an analytical C18 column (4.6 × 250 mm, 5 μm, Welch Materials Inc., Shanghai, China) using a 16‐min linear acetonitrile gradient from 22% to 30% at a flow rate of 1 ml·min⁻¹. The AGAP toxin was identified by combining de novo MS/MS (tandem mass spectrometry) sequencing in a MALDI‐TOF/TOF mass spectrometry platform (AB SCIEX TOF/TOF™ 5800 system, Applied Biosystems, Foster City, CA, USA) and ESI‐MS determining the accurate molecular weight. For de novo sequencing, 30‐μg purified toxin was reduced, alkylated, digested by trypsin, and subjected to MS/MS analysis, searching the MS/MS data against the Swiss‐Prot database matched several scorpion toxins including this AGAP. AGAP was unequivocally identified by its accurate molecular weight.

2.2. Expression of recombinant toxins

AGAP cDNA was optimized for expression in Escherichia coli BL21(DE3), synthesized by Genscript (Genscript Corp., Nanjing, China), and cloned between KpnI and XhoI sites in the pET‐32a(+) plasmid (pET‐32a‐AGAP). AGAP mutant vectors were constructed by site‐directed mutations based on pET‐32a‐AGAP (for more details, please refer to Section 2.3). Wild‐type and mutant toxins were produced as fusion proteins containing 6×His and thioredoxin tags at their N‐termini (thioredoxin‐6×His‐toxin). Briefly, BL21(DE3)‐competent cells were transformed, a single clone in the plate was cultured overnight in 5‐ml Luria–Bertani broth (LB medium) and then scaled up in 500‐ml LB medium, and IPTG was added to a final concentration of 0.1 mM to induce protein expression at 37°C when the OD600 of the culture reached 0.6. Ten hours later, bacterial cells were collected, washed, resuspended with PBS, and homogenized in a high‐pressure homogenizer (Constant System Ltd, Northants, UK). The precursor thioredoxin‐6×His‐toxin fusion protein was immobilized in a Ni column (Sangon Biotech, Shanghai, China), washed, and eluted with imidazole. The eluent was subjected to ultrafiltration to remove salts and imidazole, and the purified fusion protein was digested with enterokinase (Sangon Biotech, Shanghai, China) for 16 hr at 25°C to release the toxin. Finally, the toxins were purified to homogeneity using RP‐HPLC.

2.3. Constructs

In this study, voltage‐gated sodium channel (Na_V) cDNAs were kind gift from Professor Theodore R. Cummins (Stark Neurosciences Research Institute, Indiana University School of Medicine, Indianapolis, IN, USA). The H_V1 channel cDNA was a kind gift from Professor David E. Clapham (Janelia Research Campus, Howard Hughes Medical Institute, Ashburn, VA, USA). Channel and toxin mutants were constructed by site‐directed mutation using KOD FX PCR kit (Toyobo Co., Ltd., Osaka, Japan) and FastDigest DpnI (Fermentas, Burlington, Canada). Briefly, the parental channel or toxin plasmid was amplified by a pair of oppositely directed mutation primers with a 15‐bp overlap, the PCR product was digested by DpnI to remove the template, and 10‐μl digestion mix was used to transform 100‐μl DH5α chemical competent cells. Plasmids from several transformants were extracted and sequenced to confirm that appropriate mutants were made.

2.4. Cell culture and transfection

HEK293T (ATCC Cat# CRL‐3216, RRID:CVCL_0063) and ND7/23 (ECACC Cat# 92090903, RRID:CVCL_4259) cells were maintained in standard culture condition (5% CO₂, saturated humidity) in DMEM supplemented with 10% FBS. Transfection was performed when the cell confluence reached 70–90%. Wild‐type or mutant H_V1 channel (Homo sapiens), https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=579 (Rattus norvegicus), https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=580 (R. norvegicus), https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=581 (R. norvegicus), https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=582 (H. sapiens), https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=583 (Mus musculus), or https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=584 (H. sapiens) plasmid was transiently co‐transfected with pEGFP‐N1 plasmid into HEK293T cells using Lipofectamine ₂₀₀₀ following the manufacturer's instructions (Invitrogen Corporation, Carlsbad, CA, USA). https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=585 (R. norvegicus) or https://www.guidetopharmacology.org/GRAC/ObjectDisplayForward?objectId=586 (H. sapiens) channel plasmid was transiently co‐transfected with pEGFP‐N1 plasmid into ND7/23 cells using X‐tremeGENE HP DNA Transfection Reagent following the manufacturer's instructions (Roche, Basel, Switzerland). Cells were seeded onto PLL‐coated coverslips 4–6 hr after transfection, and 24–36 hr after seeding, cells were ready for patch clamp analysis.

2.5. Primary culture of dorsal root ganglion (DRG) neurons

All animal care and experimental procedures complied with the guidelines of the National Institutes of Health for care and use of laboratory animals and were approved by the Animal Care and Use Committee of the College of Medicine, Hunan Normal University. Animal studies are reported in compliance with the ARRIVE guidelines (Kilkenny et al., 2010) and with the recommendations made by the British Journal of Pharmacology. Sprague–Dawley (SD) rats of either sex (4‐weeks old; Hunan SJA Laboratory Animal Co., Ltd., Changsha, Hunan, China; RRID:RGD_70508) were used to obtain DRG neurons. Acutely dissociated DRG neurons were prepared from SD rats and maintained in short‐term primary cultures as previously described (Hu & Li, 1996). The combined pool of DRG neurons from four animals was used for TTX‐R Na_V current recording.

2.6. Electrophysiology

Whole‐cell current recordings were performed in an EPC10 patch clamp platform (HEKA Elektronik, Lambrecht, Germany). The recording pipettes were pulled from glass capillaries (wall thickness = 0.225 mm) in a PC‐10 puller (NARISHGE, Tokoyo, Japan). To minimize the capacitance effect, only the tip of the pipette was filled with pipette solution, and artificial capacitance was cancelled by sequential fast and slow capacitance compensation using the computer‐controlled circuit of the amplifier. To minimize voltage error, the series resistance in the whole‐cell recording circuit was kept to be less than 10 MΩ, and 80% series resistance compensation was used, with a compensation speed value of 100 μs. Positively transfected cells, as indicated by EGFP fluorescence, were randomly selected for whole‐cell current recording, and equal group sizes were employed for experiments in this study (the number of separate experimental cells for each data set is five to 10). For recording H_V1 currents, the bath solution contained (in mM) 150 NaCl, 5 KCl, 1 MgCl₂, 2 CaCl₂, 10 d‐glucose, and 100 HEPES (pH = 7.2) or MES (pH = 6.0), and the pipette solution contained (in mM) 120 CsCl and 100 MES (pH = 6.2) or Homopipes (pH = 5.0). As the extrusion of protons through H_V1 channels can produce intracellular proton depletion and extracellular proton accumulation on the outside membrane of the recording cell, even in well‐buffered solutions, in H_V1 kinetic analysis in Figure 3, ΔpH was set to 1 (pH_o7.2/pH_i6.2 or pH_o6.0/pH_i5.0; pH_o and pH_i refer to the pH of bath solution and pipette solution, respectively) to minimize such effects. The H_V1 G–V relationship was determined as described by Musset el al (Musset et al., 2008). Briefly, the slope conductance gH was calculated using the equation gH = (I_end − I_tail)/(V_test − V_hold) and was plotted as a function of depolarizing voltage. The curve was fitted with the Boltzmann equation: y = y_steady + (y₍₀₎ − y_steady)/(1 + exp[(V − V_1/2)/K]), where V_1/2, V, and K represents midpoint voltage of activation, test potential, and slope factor, respectively. For Na_V channel current recording, the bath solution contained (in mM) 124 NaCl, 20 TEA‐Cl, 10 d‐glucose, 0.3 CdCl₂, 2 BaCl₂, and 5 HEPES (pH = 7.3), and the pipette solution contained (in mM) 140 CsF, 10 NaCl, 1 EGTA, and 10 HEPES (pH = 7.3). For recording of TTX‐R Na_V current in DRGs, 1‐μM https://www.guidetopharmacology.org/GRAC/LigandDisplayForward?ligandId=2616 was added to the bath solution. Data were acquired using the Patchmaster software (HEKA Elektronik, Lambrecht, Germany; Patchmaster, RRID:SCR_000034) and analysed by Igor Pro 6.10A (WaveMetrics, Inc., Lake Oswego, OR, USA; IGOR Pro, RRID:SCR_000325), Origin 8 (OriginLab Corp., Northampton, MA, USA), Sigmaplot 10.0 (Systat Software, Inc., San Jose, CA, USA; SigmaPlot, RRID:SCR_003210), and GraphPad Prism 5.01 (GraphPad Software, Inc., La Jolla, CA, USA; GraphPad Prism, RRID:SCR_002798). Dose–response curves were fitted using the Hill logistic equation to estimate the potency of the toxin (IC₅₀).

Figure 3.

AGAP/W38F is a gating modifier of H_V1 channel, stabilizing the channel in resting state. (a) IC₅₀ values of AGAP/W38F and Zn²⁺ inhibiting H_V1 current are 1.0 ± 0.1 μM and 102.9 ± 7.9 nM at pH_o7.2/pH_i6.2 and 1.5 ± 0.2 μM and 154.6 ± 22.4 nM at pH_o7.2/pH_i5.0, respectively (n = 5–8). (b) Sample H_V1 current traces elicited by a cluster of depolarizations from −60 mV to +60 mV (in 10 mV increment) before and after AGAP/W38F treatment at pH_o7.2/pH_i6.2 and pH_o6.0/pH_i5.0. The tail currents were recorded at −80 mV, and the sweep interval was 10 s (in each pH condition, n = 5–8, and experimental cells were from two separate transfections). (c) The G–V relationships of H_V1 channel before and after rAGAP/W38F treatment. The V_1/2 values are 38.9 ± 2.6 mV and 34.5 ± 2.0 mV for control and 60.4 ± 2.9 mV and 51.2 ± 1.4 mV for toxin‐treated channels, at pH_o7.2/pH_i6.2 and pH_o6.0/pH_i5.0, respectively (n = 5–8). (d) The inhibitory effects of AGAP/W38F on H_V1 in Figure 3c is reduced at large depolarizations in both pH conditions. *P < .05, significantly different as indicated; one‐way ANOVA). (e) Dose–response curves of AGAP/W38F inhibiting H_V1 current at +30 mV and +60 mV at pH_o7.2/pH_i6.2, IC₅₀ values are 1.0 ± 0.1 μM and 2.2 ± 0.5 μM, respectively. *P < .05, significantly different comparing the inhibition ratio of 1, 5, and 10 μM toxin at +30 mV with that at +60 mV; unpaired t test; n = 5–8. (f) H_V1 outward currents at +30 mV (upper panel) and tail currents (lower panel) at −80 mV before and after 10‐μM AGAP/W38F treatment were normalized to 1, demonstrating that the toxin slows channel activation but accelerates channel deactivation (n = 10, experimental cells were from three separate transfections). (g) The deactivation time constant of I_tail in Figure 3f was calculated by fitting the trace with the equation y = y₀ + A₁*exp(−x/t₁). The τ_deactivation was markedly decreased by the toxin. *P < .05; significantly different as indicated; paired t test. (h) IC₅₀ values of Zn²⁺ and AGAP/W38F inhibiting H_V1 channels are 102.9 ± 7.9 nM and 1.0 ± 0.1 μM at pH_o7.2/pH_i6.2, 1.5 ± 0.3 μM and 3.9 ± 0.5 μM at pH_o6.0/pH_i5.0, respectively (n = 5–8). Currents were recorded at +30 mV. (i)The inhibition ratio of 1, 5, and 10‐μM AGAP/W38F was decreased by lowering the pH from pH_o7.2/pH_i6.2 to pH_o6.0/pH_i5.0 (see Figure 3h). *P < .05, significantly different as indicated; unpaired t test. In (e), (f), and (g), pH_o = 7.2 and pH_i = 6.2. In (a), (c), (e), and (h), each value on the curve represents mean ± SEM of data obtained from n experimental cells (number in parentheses) from two separate transfections

2.7. Homology modelling

For toxins, the crystal structure of scorpion toxin (PDB entry: 1KV0) was used as a template (Guan et al., 2004) to generate a model of the AGAP/W38F (the sequence of the wild‐type AGAP comes from GenBank No. AF464898). The sequence identity between the template and the target is 73.4%. For the resting‐state human H_V1 channel used here, the crystal structure of H_V1 channel from M. musculus (PDB entry: 3WKV) was employed as a template (Takeshita et al., 2014). A similar procedure was used by Gianti, Delemotte, Klein, and Carnevale (2016). The SWISS‐MODEL (Waterhouse et al., 2018), a fully automated protein structure homology modelling server, was used for building the atomic structure of AGAP/W38F and H_V1 channel.

2.8. Docking

For building the H_V1‐AGAP/W38F complex, initially, the two components were fully relaxed by using molecular dynamics (MD) simulations (please see below for more details). Three representative structures of H_V1 channel and two representative structures of AGAP/W38F were selected, leading to a combination of six groups. For each group, the HADDOCK webserver (van Zundert et al., 2016) was used for molecular docking, where R58 on AGAP/W38F and H193, H140, on H_V1 channels were selected as key residues involved in binding. In each of the six groups, the top three ranked configurations were selected. Then the 18 produced complex models were carefully selected for subsequent relaxation by MD simulations, with the criterion that specific patterns of interactions could be maintained.

2.9. System set‐up

For AGAP/W38F, the atomic structure obtained from homology modelling was embedded in a water box. In addition of a few ions added to neutralize the system, 150‐mM NaCl was added to mimic the physiological condition of salt concentration. The system contained a total of 41,425 atoms, including 13,463 water molecules, 36 Na⁺ ions, and 37 Cl⁻ ions.

For the H_V1 channel, the protein was embedded in a fully hydrated 1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phosphocholine (POPC) bilayer. The membrane bilayer system was solvated by 8,397 water molecules, 22 Na⁺ ions, and 23 Cl⁻ ions, where the extra one anion was used to neutralize the whole system. The total number of atoms in the periodically replicated simulation box was 45,599, and the initial size of the box was 75 × 75 × 95 ų.

For the H_V1‐AGAP/W38F complex, the system was initialized from the snapshot obtained from MD simulations for Hv1, and the AGAP/W38F molecule was added to the system by aligning to the docking position obtained. Then more water molecules were added along the membrane normal to fully solvate the system, with the size on x–y plane remaining intact. Additional Na⁺ and Cl⁻ ions were added to neutralize the system and to mimic the salt concentration of 150 mM.

2.10. Molecular dynamics (MD) simulations

For both systems, energy minimization was used to remove bad contact, where harmonic restraints were applied to the protein Cα atoms. Next, the simulation system was equilibrated, with gradually decreased harmonic position restraints applied to heavy atoms of the protein backbone, in constant temperature and pressure (NPT) ensemble. After equilibration, two independent MD simulations were performed for each system, and data from each were accumulated for 400 ns for production. All the analyses used in this work were based on the well‐equilibrated system. To be specific, the snapshot used for Figure 5 is the structure at 280 ns. MD simulations were carried out using the CUDA‐accelerated NAMD program version 2.12 (Phillips et al., 2005; NAMD, RRID:SCR_014894). The Charmm36 force field parameters (Best et al., 2012) and the TIP3P water model were used. Periodic boundary conditions were applied, and the particle mesh Ewald method was used to treat long‐range electrostatic interactions.

Figure 5.

Molecular dynamics simulations of AGAP/W38F‐H_V1 interaction. (a) The structure model of H_V1‐toxin complex. H_V1 channel and toxin are shown in cartoon mode and coloured in yellow and purple, respectively. The key interactions at the binding interface are highlighted, where R2, K54, and R58 come from AGAP (labelled in purple), and the rest residues are from H_V1 channel (labelled in black). (b) The interactions at the ICN before (left) and after (right) binding of toxin. The S1 helix on H_V1 protein was hidden for clarity. (c) The protein backbone RMSD in 500‐ns MD simulations after equilibration. The H_V1 channel and the H_V1‐AGAP/W38F complex are in blue and yellow, respectively. (d) The distribution of key interactions between AGAP and H_V1 channel. The dashed line shows the average distance. Seemingly, the strength of the interaction is consistent with the intensity of experimental determined mutation effect. (e) E153 and R208 in H_V1 channel form salt bridges upon toxin binding

2.11. Data analysis

The data and statistical analysis comply with the recommendations of the British Journal of Pharmacology on experimental design and analysis in pharmacology. Data are expressed as mean ± SEM of n separate experimental cells. Statistical analysis was undertaken only for studies where the group size was at least n = 5. Statistical significance was assessed using two‐tailed unpaired/paired t test or one‐way ANOVA, and multiple comparisons between groups were performed using Tukey's method if F was significant and if no variance inhomogeneity occurred. A P value of less than .05 was considered statistically significant.

2.12. Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2019/20 (Alexander et al., 2019).

3. RESULTS

3.1. The scorpion toxin AGAP is a novel H_V1 channel antagonist

Venoms of toxic animals are rich sources for mining toxins that act on ion channels. We screened RP‐HPLC fractions of several spider and scorpion venoms to identify novel H_V1 channel antagonists. Finally, we characterized an active fraction from the venom of B. martensii (Figure 1a; upper panel, left). It has a retention time of 30.8 min in the RP‐HPLC purification spectrum (Figure 1a; upper panel, right). A second round of activity‐guided analytical RP‐HPLC purified the active component to homogeneity (Figure S1a). We performed MS/MS analysis of the toxin (Figure S1b), and searching the MS/MS data in the Swiss–Prot database matched several candidate toxins with high homology, including the antitumour analgesic peptide AGAP (other hit toxins are bukatoxin, BmKa3 and alpha‐toxin 4). We then unequivocally identified the toxin by its precise molecular weight. As shown in the ESI‐MS spectrum, the m/z 809.1373, 910.1516, 1,040.0311, 1,213.1988, and 1,455.6433 were multiple‐charged ion peaks corresponding to [M + 9H]⁹⁺, [M + 8H]⁸⁺, [M + 7H]⁷⁺, [M + 6H]⁶⁺, and [M + 5H]⁵⁺, respectively (Figure S1c). The calculated toxin molecular weight (7,273.2317 Da) matched well with AGAP (theoretical MW = 7,273.18 Da). Its cDNA and full protein sequences are shown in Figure 1a (GenBank No. AF464898). This toxin was reported as a Na_V channel antagonist which potently inhibits the peak currents of heterologously expressed Na_V1.4–1.5 and Na_V1.8–1.9 channels, as well as TTX‐R Na_V channels in DRG neurons (Li et al., 2016; Xu et al., 2017). In animal studies, AGAP reduces pain behaviour in murine models of pain and exhibits anti‐tumour activity in S‐180 fibrosarcoma and Ehrlich ascites tumour models in mice (Liu et al., 2003). We demonstrated that AGAP also potently inhibited H_V1 currents with an IC₅₀ of 2.5 ± 0.4 μM at +30 mV (Figure 1b,c, n = 5–6). However, the Na_V channel activity of AGAP in this study differed from previous reports. Our data showed that 1‐μM AGAP apparently inhibited the fast inactivation of Na_V1.2–1.7 heterologously expressed in HEK293T cells, but did not affect Na_V1.8–1.9 currents expressed in ND7/23 cells and TTX‐R Na_V currents in rat DRG neurons (Figure 1d,e; Figure S1d; n = 5 for each Na_V subtype). The underlying mechanism is currently unknown. However, another study reported that BmK NT1, a scorpion toxin whose protein sequence is highly homologous to AGAP, except for the glycine residue deletion at its extreme C‐terminus, potently activates Na_V channels in cerebellar granule cells possibly by acting on Na_V site 3 (Zou, He, Qiao, Zhang, & Cao, 2016).

Figure 1.

Purification and characterization of AGAP. (a) Upper panel: left, the scorpion Buthus martensii; right, RP‐HPLC profile of Buthus martensii venom, the red circle‐labelled fraction contains AGAP; lower panel: cDNA and protein sequence of AGAP, partial sequence of AGAP (DGYIADDKNCAYFCGR) was determined by MS/MS spectrometry analysis and shown in cyan. (b) Representative traces showing 10‐μM AGAP fully inhibits H_V1 current (n = 5; experimental cells were from two separate transfections). Currents were elicited by depolarizations to +30 mV from the holding potential of −60 mV; pH_i and pH_o are 5.0 and 7.2, respectively. (c) The dose–response curve of AGAP inhibiting H_V1 currents, the IC₅₀ value are 2.5 ± 0.4 μM at +30 mV. Each value on the curve represents mean ± SEM of data obtained from five to six experimental cells from two separate transfections. (d) Sample current traces showing AGAP inhibits the fast inactivation of NaV1.2 channel; currents were elicited by voltage protocol as shown (n = 5; experimental cells were from two separate transfections). (e) Sample current traces showing 1‐μM AGAP do not affect TTX‐R Na_V currents in acutely dissociated DRG neurons. Currents were elicited by voltage protocol as shown (n = 5) [Colour figure can be viewed at http://wileyonlinelibrary.com]

3.2. W38F mutation in AGAP makes it a selective H_v1 channel antagonist

We produced AGAP using recombinant expression (rAGAP) and tested its inhibitory effect on H_V1 peak currents and the fast inactivation of Na_V channels. The AGAP cDNA sequence was optimized for expression in E. coil BL21(DE3) and cloned into the pET‐32a(+) vector (Figure S2a). After IPTG induction, the thioredoxin‐6×His‐AGAP fusion protein was expressed in soluble form, as shown by SDS‐PAGE analysis of Ni bead immobilized proteins from the supernatant of the bacteria homogenate (Figure S2b, lane 2, the black arrow indicated band). Enterokinase digestion of the fusion protein released the toxin monomer (Figure S2b, lane 3, red arrow indicated band). After two rounds of RP‐HPLC purification, rAGAP was purified to homogeneity (Figure 2a). As the native AGAP purified from the venom (Figures 1 and S1), rAGAP dose‐dependently inhibited H_V1 currents with an IC₅₀ of 2.1 ± 0.6 μM at +30 mV (Figure 2b,c, n = 5). It also inhibited the fast inactivation of Na_V1.2–1.7 but not Na_V1.8–1.9 channels (Figures 2d and S2c; n = 5 for each NaV subtype). These data suggested that rAGAP was successfully produced and correctly refolded in E. coli. Previously, it has been indicated that the W38G mutation in AGAP reduced its activity on Na_V1.4–1.5 channels (Xu et al., 2017). In our experiment, W38, as well as two other residues (K10 and R58) in AGAP, were mutated, and the effect of the mutations on toxin activity was evaluated. All mutants were successfully produced in E. coli and purified to homogeneity (Figures S2b and 2a). As shown in Figure 2b,c, 5‐μM rAGAP/K10A and 10‐μM rAGAP/W38F fully inhibited H_V1 currents, with IC₅₀ values at +30 mV determined as 1.0 ± 0.1 μM and 1.8 ± 0.3 μM, respectively (n = 5–6). For rAGAP/R58D, however, significantly reduced activity on H_V1 currents was observed, with high doses of toxin only partly inhibiting the currents (Figure 2b,c, n = 5). Unlike rAGAP, rAGAP/R58D and rAGAP/W38F did not affect Na_V1.2–1.7 currents even at 10‐μM concentration (Figures 2d and S2e,f; n = 5 for each Na_V channel subtype). For AGAP/K10A, its activity on Na_V1.2–1.4 and Na_V1.6–1.9 resembled rAGAP, but it did not affect Na_V1.5 currents (Figures 2d and S2d; n = 5 for each Na_V channel subtype). Furthermore, AGAP/W38F reversibly inhibited H_V1 currents, as did Zn²⁺ (Figure 2e). Overall, the mutated toxin rAGAP/W38F demonstrated H_V1 channel antagonism with optimized selectivity.

Figure 2.

Recombinant expression and activity assay of AGAP and AGAP mutants. (a) RP‐HPLC profiles of recombinant AGAP and AGAP mutants. Toxin monomers were released from the precursor fusion proteins by enterokinase digestion and purified by RP‐HPLC; the star‐labelled peaks contain desired toxins. (b) Sample current traces showing 10‐μM rAGAP, 5‐μM rAGAP/K10A, and 10‐μM rAGAP/W38F fully inhibit H_V1 currents, while 10‐μM rAGAP/R58D only partly inhibits H_V1 currents. Currents were elicited by depolarizations to +30 mV from −60 mV holding (for each toxin, n = 5–6, and experimental cells were from two separate transfections). (c) The dose–response curves of AGAP and AGAP mutants inhibiting H_V1 currents at +30 mV. The IC₅₀ values are 2.1 ± 0.6 μM, 1.0 ± 0.1 μM, and 1.8 ± 0.3 μM for rAGAP, rAGAP/K10A, rAGAP/W38F, respectively. The IC₅₀ value for rAGAP/R58D could not be precisely determined with 30‐μM toxin inhibiting H_V1 currents by 66.5 ± 3.7%. Each value on the curve represents mean ± SEM of data obtained from five to six experimental cells from two separate transfections. (d) Sample current traces showing rAGAP (1 μM) and rAGAP/K10A (1 μM) but not rAGAP/W38F (10 μM) and rAGAP/R58D (10 μM) inhibit the fast inactivation of Na_V1.2 channel; currents were elicited by voltage protocol as shown (for each toxin, n = 5, and experimental cells were from two separate transfections). (e) Time course of block (red bar) and unblock of H_V1 channel on acute application and washout of 10‐μM AGAP/W38F or 1‐μM Zn²⁺. Currents were recorded at +30 mV at pH_o7.2/pH_i5.0 (τ_on = 5.7 ± 1.1 s, τ_off = 16.9 ± 2.2 s for AGAP/W38F; τ_on = 4.4 ± 0.5 s, τ_off = 19.2 ± 1.4 s for Zn²⁺; each value on the curve represents mean of data obtained from five experimental cells from two separate transfections)

3.3. AGAP/W38F is a gating modifier stabilizing the H_V1 voltage sensor in a deactivated state

The inhibition by AGAP/W38F of H_V1 channel might employ a voltage‐sensor trapping mechanism, as used by Zn²⁺, or by simply blocking the pore, as observed with guanidine derivatives (Hong et al., 2013; Qiu et al., 2016). We can discriminate between these possibilities by comparing the channel gating kinetics before and after toxin treatment. We first analysed the effect of AGAP/W38F on the voltage‐dependent activation of H_V1 channel at two pH values, pH_o7.2/pH_i6.2 and pH_o6.0/pH_i5.0; see Section 2). We demonstrated that intracellular pH changes (pH_i6.0 and pH_i5.0) did not affect the potency of AGAP/W38F as observed for Zn²⁺ (Figure 3a). However, 10‐μM AGAP/W38F fully inhibited H_V1 currents at pH_o7.2/pH_i6.2 but not pH_o6.0/pH_i5.0 at +30 mV, suggesting that changes in extracellular pH could markedly reduce AGAP/W38F potency. Therefore, we analysed the toxin effect on G–V relationships of H_V1 channel, using 10‐ and 20‐μM toxin at pH_o7.2/pH_i6.2 and pH_o6.0/pH_i5.0, respectively. As shown by the representative traces in Figure 3b, AGAP/W38F failed to fully inhibit H_V1 currents at large voltages in both pH conditions. The G–V curves showed that AGAP/W38F positively shifted the V_1/2 by 21.6 ± 2.3 mV at pH_o7.2/pH_i6.2 and 16.8 ± 0.8 mV at pH_o6.0/pH_i5.0, respectively (Figure 3c, n = 5–8). Moreover, AGAP/W38F potency was markedly attenuated by strengthening depolarization, as revealed by comparing the inhibitory effect at 40, 50, and 60 mV with that at 20 mV (Figure 3d). Collectively, these data suggested that AGAP/W38F inhibited the H_V1 channel by impairing its voltage‐dependent activation, with strong depolarizations counteracting the toxin effect. Consequently, the dose–response curve of AGAP/W38F inhibiting H_V1 was shifted rightward at +60 mV compared with that at +30 mV (Figure 3e). Analysing the effect of AGAP/W38F on the activation and deactivation rates of H_V1 channel showed that it slowed channel activation but accelerated channel deactivation (Figure 3f,g). Hence, these data supported the toxin trapping the H_V1 voltage sensor in a deactivated state. Among the limited number of H_V1 antagonists, AGAP/W38F resembled Zn²⁺ and Corza6 in its mode of action.

We therefore compared AGAP/W38F and Zn²⁺ inhibition of the H_V1 channel. At pH_o7.2/pH_i6.2 and pH_o6.0/pH_i5.0, Zn²⁺ inhibited H_V1 currents with an IC₅₀ of 102.9 ± 7.9 nM and 1.5 ± 0.3 μM, respectively (Figure 3h, n = 5–8), consistent with previous studies showing extracellular acidification markedly attenuates Zn²⁺ potency (Capasso, DeCoursey, & Dyer, 2011; Cherny & DeCoursey, 1999). Interestingly, reducing the pH_o from 7.2 to 6.0 only slightly increased AGAP/W38F IC₅₀ from 1.0 ± 0.1 μM to 3.9 ± 0.5 μM (Figure 3h, n = 5–8), but indeed significantly impaired AGAP/W38F potency (Figure 3i, n = 5–8). These data suggested that AGAP/W38F inhibited H_V1 channel with less pH‐dependence than Zn²⁺.

3.4. Molecular mechanism of AGAP/W38F‐H_v1 interaction

An alanine scan strategy was used to characterize the key residues in the extracellular loops of the H_V1 channel for toxin binding. Figure 4a shows the protein sequence of the H_V1 S1–2 and S3–4 loops, with the designed mutations indicated by arrows. Most mutations did not reduce AGAP/W38F potency (D123A, K125A, Q128A, D130R, K131A, N132A, F139A, Y141A, D185A, L189A, F190A, E192A, F195A, L201A and I202A; Figure 4b, n = 5–6). However, three mutations in H_V1 protein, H140A, H193A, and E196A, significantly reduced the inhibitory effect of the toxin (Figure 4b,c, n = 6). The dose–response curves of AGAP/W38F inhibiting H140A, H193A, and E196A mutants at +30 mV are shown in Figure 4e (n = 5–6). The IC₅₀ value was 4.7 ± 0.3 μM for E196A. For H140A and H193A, the IC₅₀ values could not be determined from the curve, with 10‐μM toxin inhibiting their currents by approximately 45% and 22%, respectively. We concluded that these three are key binding sites for AGAP/W38F. H140 and H193 in H_V1 protein (analogous to H136 and H189 in mH_V1) are key residues for binding with Zn²⁺ (Ramsey et al., 2006; Takeshita et al., 2014), suggesting that the binding pockets of AGAP/W38F and Zn²⁺ in H_V1 channel might overlap. Hence, we evaluated the involvement of the other two Zn²⁺ binding residues (E119 and D123) on toxin binding. As shown in Figure 4b,d, E119S/D123S mutation in H_V1 markedly reduced the inhibitory effects of Zn²⁺, but did not affect AGAP/W38F potency. Collectively, these data suggested that Zn²⁺ and AGAP/W38F binding pockets in H_V1 channel only partly overlapped. Hanatoxin is a tarantula toxin with indiscriminate K_V, Na_V, Ca_V, and H_V1 channel activity (Alabi et al., 2007; Bosmans & Swartz, 2010). The D185A mutation in H_V1 protein effectively eliminated the inhibitory effects of hanatoxin (Alabi et al., 2007), although it did not reduce AGAP/W38F potency (Figure 4b,c, n = 6). Hence, the molecular determinants for binding hanatoxin and AGAP/W38F to H_V1 channels were different.

Figure 4.

The molecular mechanism of rAGAP/W38F‐H_V1 interaction. (a) The location of designed mutations in H_V1 channel (red arrows indicated), only partial sequence of S1(TM1), S2(TM2), S3(TM3), and S4(TM4) were shown for clarity. The borders of TM1–4 were determined by protein sequence information in: https://www.ncbi.nlm.nih.gov/protein/NP_001035196.1. (b) Summary data of the normalized residual currents of WT and H_V1 channel mutants at +30 mV after 10‐μM rAGAP/W38F treatment. The H140A, E153A, H193A, and E196A mutations in H_V1 channel significantly reduced rAGAP/W38F inhibition of its currents. *P < .05; significantly different from WT channels (WTH_v1); one‐way ANOVA. For each mutant, data were obtained from five to six experimental cells from two to three separate transfections). (c) Sample current traces showing that the H193A, H140A, E196A ,and E153A, but not the D185A mutation, in H_V1 channel reduced AGAP/W38F potency (for each mutant, n = 6; experimental cells were from two to three separate transfections). (d) Sample current traces showing that the E119S/D123S mutations in H_V1 channel reduced Zn²⁺ but not AGAP/W38F potency, with 1‐μM Zn²⁺ inhibiting E119S/D123S and WT channel currents by 45.5 ± 4.0% and 82.8 ± 1.8% (P < .05, unpaired t test, n = 5–7), and 10‐μM AGAP/W38F inhibiting E119S/D123S and WT channel currents by 80.4 ± 6.7% and 88.0 ± 1.6%, respectively (P > .05, unpaired t test, n = 5). Experimental cells were from two separate transfections. (e) Dose–response curves of AGAP/W38F inhibiting the currents of H_V1 and H_V1 mutants, IC₅₀ values are 1.8 ± 0.3 μM and 4.7 ± 0.3 μM for WT and E196A channel, respectively. For H_V1/H140A, H_V1/E153A, and H_V1/H193A channels, 10‐μM toxin inhibits their currents by 44.7 ± 3.2%, 29.9 ± 5.2%, and 21.7 ± 3.2%, respectively (each value on the curve represents mean ± SEM of data obtained from five to six experimental cells from two transfections)

A recent study showed that E153A mutation at the highly conserved intracellular Coulombic network (ICN) of the H_V1 channel reduces Zn²⁺ potency as an inhibitor of the channel (De La Rosa et al., 2018). As the AGAP/W38F binding pocket partly overlapped with that of Zn²⁺, we evaluated the effects of AGAP/W38F on the currents of the E153A mutant H_V1 channel and found that the mutation reduced toxin potency, with 10‐μM toxin inhibiting the currents by 29.9 ± 5.2% at +30 mV (Figure 4b,c,e, n = 6). The apparent loss of toxin potency induced by the E153A mutation was comparable to that of H193A mutation (Figure 4e). These data suggested that AGAP/W38F used a mechanism like that of Zn²⁺ , to inhibit H_V1 channel, a suggestion supported by MD simulations (see below). Finally, H140A and H193A mutations in H_V1 channel also reduced the inhibitory effect of wild‐type AGAP, as observed for AGAP/W38F (Figure S3a), implying that AGAP and its mutant AGAP/W38F, inhibited H_V1 channel by the same mechanism.

3.5. AGAP/W38F binding to H_V1 channel rearranges residue interactions in the ICN region

By using molecular modelling at an atomic level, the AGAP/W38F‐H_V1 interaction (Figure 5a) and the reorganization of the highly conserved intracellular Coulombic network (ICN) in H_V1 channel after toxin binding (Figure 5b) could be well characterized. The AGAP/W38F‐H_v1 complex is stable within 500‐ns simulations (Figure 5c), with the interface between the two being at the extracellular surface of the membrane, where both the N‐ and C‐terminal segments of the toxin and the extracellular mouth of the channel pore are involved (Figure 5a). The toxin completely covered the entrance of the tunnel and was stabilized by specific interactions between the two components: The positively charged guanidino group on R2 of toxin formed a hydrogen bond with H193 on the loop between S3 and S4 of H_V1; in the peripheral region, K54 on the C‐terminal of toxin formed a salt bridge with E196 on H_V1, and R58 hydrogen‐bonded to H140 on H_V1 through a bridging water molecule (Figure 5d). All these interactions were well conserved in MD simulations, locking the channel in the resting state structure by suppressing the upward motion of the S4 helix during activation. Moreover, toxin insertion at the extracellular entrance of H_V1 channel triggered certain conformational changes in the pore (Figure 5b), especially the highly conserved ICN (De La Rosa et al., 2018). The interaction between R205 and D112 was conserved. However, before binding the toxin, R208 on S4 formed a salt bridge with D174. It switched to close contact with E153 and E171 upon the binding of toxin (Figure 5e), and now, D174 formed a bi‐dentate salt bridge with R211, the third gating residue on S4. Consequently, the S4 was less tilted upon toxin binding. Similar changes were observed on Zn²⁺ binding, suggesting the coupling between binding at the extracellular side of the pore and the reorganization of the helices at the intracellular region (De La Rosa et al., 2018). Hence, MD simulations also suggested that the inhibition mechanism of the AGAP/W38F and Zn²⁺ had certain similarities and can be used for reference.

4. DISCUSSION

The present study has identified the scorpion toxin AGAP as a novel H_V1 channel antagonist. This toxin selectively blocked H_V1 channel after a W38F mutation in its sequence (AGAP/W38F). The mutant AGAP/W38F inhibited the H_V1 channel by impeding its S4 activation, with its binding pocket on this channel partly overlapping that of Zn²⁺. However, AGAP/W38F inhibited H_V1 channel with less pH‐dependence than Zn²⁺. Notably, mutating E153 (E153A) in the H_V1 ICN region also attenuated the effects of AGAP/W38F and MD simulations revealed that E153 formed salt bridge with R208 on S4 upon toxin binding, as observed for Zn²⁺. These data emphasized the Zn²⁺ binding site in the H_V1 channel as a promising docking pocket for the design of ligands for this channel. Moreover, AGAP/W38F could be a useful probe for exploring the structure–function relationship of H_V1 channel and a valuable drug candidate.

We compared the mechanism of AGAP/W38F inhibiting H_V1 channel with that of Zn²⁺, as well as that of some other gating modifier toxins inhibiting Na_V and K_V channels. Our electrophysiological data suggest that AGAP/W38F inhibition of H_V1 channel shared a canonical voltage‐sensor trapping mechanism with gating modifier toxins inhibiting Na_V and K_V channels (Lee, Wang, & Swartz, 2003; Tang et al., 2017; Xiao et al., 2008). AGAP/W38F and Zn²⁺ act on H_V1 channel with certain similarities. On the one hand, it has been proposed that Zn²⁺ inhibits H_V1 channel by stabilizing the channel in its activated closed state or resting closed state by binding to Site 1 or Site 2, respectively (Qiu et al., 2016). The binding pocket of AGAP/W38F in H_V1 channel partly overlapped with that of Zn²⁺, with H140 and H193 being the common sites. However, the two other acidic residues (E119 and D123) in the Zn²⁺ binding pocket which place water molecules around Zn²⁺ (Iwaki et al., 2018; Takeshita et al., 2014), and the D185 residue which is part of Zn²⁺ binding site 2 and the key binding site for hanatoxin (Alabi et al., 2007; Qiu et al., 2016), were not involved in AGAP/W38F binding (Figure 4). On the other hand, the binding modes of AGAP/W38F and Zn²⁺ with H_V1 channel might be different. The interaction of Zn²⁺ with H140 and H193 was mediated by coordinate bonds, and acidification of the bath solution attenuated Zn²⁺ potency as the protonation of these two histidine residues at low pH largely attenuated Zn²⁺ binding. Although inhibition by AGAP/W38F of H_V1 currents was also pH dependent, the effect of reducing extracellular pH was minimal compared with its effects on inhibition by Zn²⁺, and the underlying reasons are still to be investigated. One interesting finding is that H140A and H193A mutation also profoundly reduced the inhibitory effect of AGAP/W38F at pH_o6.0/pH_i5.0 (Figure S3b), suggesting that the protonated H140 and H193 were also involved in toxin binding.

In the gating pathway of H_V1 channel, there are complicated electrostatic interactions between the S4 gating charges and negatively charged acidic residues (counter charges) in S1–3 segments. A recent study demonstrated that a mutation (E153A) which impairs such interaction in the ICN markedly reduced Zn²⁺ potency (De La Rosa et al., 2018). Interestingly, this mutation also significantly reduced AGAP/W38F potency in the present study. In H_V1 channel, E153 is inaccessible to Zn²⁺ and the bulky AGAP/W38F toxin, from the extracellular side. Hence, it is clear that the E153A mutation reduced AGAP/W38F potency by allosterically changing the conformation of the toxin pocket and reducing toxin binding. In our MD simulations, AGAP/W38F binding to H_V1 channel profoundly rearranged the electrostatic interaction network in the ICN (Figure 5b). These data strongly supported that AGAP/W38F and Zn²⁺ inhibited H_V1 currents using a similar long‐range conformational coupling mechanism.

We therefore proposed a model of AGAP/W38F inhibiting H_V1 channel, as follows: (a) AGAP/W38F binding to its extracellular binding pocket in the H_V1 protein induced local conformation change; (b) then the energy was transduced into the ICN region and induced its reorganization, in which interactions stabilizing the H_V1 S4 voltage sensor in the deactivated state (e.g., E153–R208 salt bridge) were strengthened and thus the channel is stabilized in a resting closed state. It is of particular interest that the small‐sized Zn²⁺ ion and the bulky AGAP/W38F toxin share their binding sites in H_V1 channel and even their mechanism of inhibiting the channel. It would seem that occupying the Zn²⁺ pocket in the H_V1 channel is bound to inhibit the channel, which emphasizes the importance of this binding pocket for designing H_V1 channel ligands.

Inhibiting H_V1 channel activity is considered to be a promising strategy for treating H_V1‐related diseases by regulating intracellular pH or ROS production. H_V1‐deficient (H_V1^−/−) mice are almost as normal as wild‐type mice, except with substantially reduced ROS production. However, animal infection models have shown that the H_V1‐deficient mice are able to clear some bacterial infections as effectively as wild‐type animals (Ramsey et al., 2009). Although another study reported an increased tendency towards autoimmunity in aged H_V1‐deficient mice (Sasaki et al., 2013), the existing data suggest that inhibiting H_V1 channel should not produce severe side effects, which adds the value of the H_V1 channel as a drug target. The identification of novel modulators of H_V1 channel would be valuable either as drug candidates or as pharmacological tools to investigate the structure and function of this channel. Recently, Zhao et al. identified a novel H_V1 channel toxin, Corza6, from the ICK toxins permutation library using a phage‐display strategy (Zhao et al., 2018). Both Corza6 and AGAP/W38F inhibit H_V1 currents by trapping the S4 voltage sensor in deactivated states, and these effects could be counteracted by strong depolarization. AGAP/W38F in this study is a new member of the family of H_V1 channel peptide antagonists and further studies evaluating the therapeutic potential of AGAP/W38F as a H_V1 channel blocker will be of value.

AUTHOR CONTRIBUTIONS

C.T., Z.L., and D.T. developed the concept and designed the study; D.T., C.T., Z.X., J.X., Q.Y., and H.Da. (Han Dai) performed the experiments and data acquisition. H.Do. (Hao Dong) and Y.Y. performed the molecular dynamics simulations. S.L. contributed to insightful discussions. C.T., Z.L., H.Do, and D.T. wrote the manuscript.

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

DECLARATION OF TRANSPARENCY AND SCIENTIFIC RIGOUR

This Declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research as stated in the BJP guidelines for https://bpspubs.onlinelibrary.wiley.com/doi/full/10.1111/bph.14207, and https://bpspubs.onlinelibrary.wiley.com/doi/full/10.1111/bph.14206, and as recommended by funding agencies, publishers and other organisations engaged with supporting research.

Supporting information

Figure S1 (a). The fraction containing AGAP in Figure 1a was purified to homogeneity by analytical RP‐HPLC. Inset is the RP‐HPLC profile of purified AGAP. (b) AGAP was reduced, alkylated and digested with trypsin, shown is a representative MS/MS spectrum of the parental ion with the m/z of 1924.8591. (c) The molecular weight of AGAP was determined by electrospray ionization (ESI) mass spectrometry. (d) The activity of native AGAP on NaV subtypes (n = 5 for each NaV subtype, experimental cells were from two transfections)

Figure S2 (a). The architecture of the AGAP recombinant expression vector, the background vector was pET‐32a(+) and the AGAP coding sequence was optimized by E.coli condons. (b) SDS‐PAGE analysis of recombinant AGAP and AGAP mutant fusion proteins (strongly stained bands at 27 KD, indicated by black arrow). Enterokinase digestion released toxin monomers between 6.5 k and 9.5 k marker (indicated by red arrow). (c) ‐ (f) The activity of rAGAP, rAGAP/K10A, rAGAP/W38F and rAGAP/R58D on NaV subtypes (n = 5 for each NaV subtype, experimental cells were from 2 to 3 transfections)

Figure S3 (a). Sample current traces showing that H140A and H193A mutations in H_V1 remarkably reduce the inhibitory effect of wild‐type (WT) AGAP. 10 μM toxin inhibits the currents of WTH_V1, H_V1/H140A and H_V1/H193A by 91.9 ± 2.4%, 70.4 ± 1.7% and 34.6 ± 2.1%, respectively(P < 0.001 when comparing H_V1/H140A and H_V1/H193A with WTH_V1; n = 5 for each channel, experimental cells were from two separate transfections; currents were recorded at +30 mV at pH_o7.2/pH_i6.2). (b) Sample current traces showing H140A and H193A mutations in H_V1 remarkably reduce the inhibitory effect of AGAP/W38F at pH_o6.0/pH_i5.0, with 20 μM toxin inhibiting the currents of WTH_V1, H_V1/H140A and H_V1/H193A by 87.4 ± 1.7%, 53.5 ± 2.4% and 38.5 ± 2.0%, respectively (P < 0.001 when comparing H_V1/H140A and H_V1/H193A with WTH_V1; n = 5 for each mutant, experimental cells were from two separate transfections; currents were recorded at +30 mV)

Tang D, Yang Y, Xiao Z, et al. Scorpion toxin inhibits the voltage‐gated proton channel using a Zn²⁺‐like long‐range conformational coupling mechanism. Br J Pharmacol. 2020;177:2351–2364. 10.1111/bph.14984