The S4 segment of voltage-sensing domains (VSDs) directly responds to voltage changes by reorienting within the electric field as a permion. A narrow hydrophobic “gasket” or charge transfer center at the core of most VSDs focuses the electric field into a narrow region and catalyzes the sequential and reversible translocation of S4 positive gating charge residues across the electric field while preventing the permeation of physiological ions. Mutating specific S4 gating charges can cause ionic leak currents through the VSDs. These gating pores or omega currents play important pathophysiological roles in many diseases of excitability. Here, we show that mutating D129, a key countercharge residue in the Ciona intestinalis voltage-sensing phosphatase (Ci-VSP), leads to the generation of unique anionic omega currents. Neutralizing D129 causes a dramatic positive shift of activation, facilitates the formation of a continuous water path through the VSD, and creates a positive electrostatic potential landscape inside the VSD that contributes to its unique anionic selectivity. Increasing the population or dwell time of the conducting state by a high external pH or an engineered Cd2+ bridge markedly increases the current magnitude. Our findings uncover a new role of countercharge residues in the impermeable VSD of Ci-VSP and offer insights into mechanisms of the conduction of anionic omega currents linked to countercharge residue mutations.
Introduction
Voltage-sensing domains (VSDs) undergo defined conformational transitions in response to membrane potential changes to orchestrate the gating of the pore domain of ion channels (Tombola et al., 2006; Bezanilla, 2008; Catterall, 2010; Ahern et al., 2016), the catalytic activity of phosphatases (Murata et al., 2005; Kohout et al., 2010; Grimm and Isacoff, 2016; Okamura et al., 2018; Mizutani et al., 2022), the activation of ion exchange of solute carriers (Wang et al., 2003; Windler et al., 2018), and the proton-specific conduction of proton channels (Ramsey et al., 2006; Sasaki et al., 2006; Tombola et al., 2008; DeCoursey, 2013; Li et al., 2015; Mony et al., 2020; Zhao et al., 2021). VSDs share high sequence and structural similarity, in particular, the conserved positive gating charge residues on S4, the presence of negative countercharge residues on S1–S3 (Yu and Catterall, 2004; Palovcak et al., 2014; Bezanilla, 2018), and a layer of in-plane hydrophobic residues on S1–S3 (Li et al., 2014; Campos et al., 2007; Catterall, 2014; Tao et al., 2010) thought to electrically separate the extracellular and intracellular aqueous crevices of the VSDs. These characteristics prevent passage of water and ions through VSDs of ion channels, phosphatases, and solute carriers, termed impermeable VSDs, and set-up a focused electric field (Starace and Bezanilla, 2004; Ahern and Horn, 2005; Tao et al., 2010; Freites et al., 2012). The number and location of these gating charges and countercharges vary in different VSDs, resulting in diverse voltage gating set points and sensitivity (Sakata and Okamura, 2014; Sun and MacKinnon, 2017; Clark et al., 2020; Li et al., 2021c) and defining the overall physiological response of excitable cells.
The omega current was first identified as an alternative voltage-elicited current in ion channels, distinct from the pore currents evoked by the opening of the central pore domain or the transient gating currents carried by the movement of gating charges across the focused electric field (Starace and Bezanilla, 2004). Rather than originating from a canonical pore aqueous path, omega currents permeate through the VSDs themselves, typically displaying proton and/or cation selectivity when the VSDs have a missense mutation on gating charge residue(s). In Shaker potassium channels, mutating the first S4 gating arginine to histidine can create steady proton currents at hyperpolarization potentials (Starace and Bezanilla, 2004). Subsequent studies showed that neutralization of a single, two or three adjacent gating charge residues in S4 of potassium channels (Tombola et al., 2005, 2007; Gamal El-Din et al., 2010; Delemotte et al., 2010; Khalili-Araghi et al., 2012; Jensen et al., 2012), as well as sodium and calcium channels (Sokolov et al., 2005, 2007; Gosselin-Badaroudine et al., 2012; Capes et al., 2012; Monteleone et al., 2017; Jiang et al., 2018, 2020), can generate a monovalent cation selective current through the VSDs when those mutated gating charge residue(s) approach the hydrophobic gasket at depolarizing or hyperpolarizing potentials, reaching as much as 1% of the central pore current magnitude (Tombola et al., 2005; Jiang et al., 2018). Large organic monovalent cations, such as tetraethylammonium (TEA+) and N-methyl-D-glucamine (NMDG+), are much less permeable, and the monovalent cation currents can be blocked by divalent cations at millimolar concentrations (Sokolov et al., 2007). Cationic omega currents have also been observed at hyperpolarized potentials in a truncated wild type (WT) VSD of the Shaker potassium channel (Zhao and Blunck, 2016), and the WT VSD in the α3 subunit of the ascidian CatSper calcium channel (Ci-CatSper3; Arima et al., 2018) in both truncated and full-length forms. The WT VSD of Ci-CatSper3 can even conduct divalent cations Ca2+, Ba2+, and Sr2+ (Arima et al., 2018). Although voltage-sensitive phosphatases can neither conduct ions nor permeate protons, mutations of S4 gating charge residues in Ci-VSP have been shown to conduct protons as well (Villalba-Galea et al., 2013; Shen et al., 2022). The important role of these cationic omega currents caused by missense mutations of gating charge residues has been highlighted in certain inherited channelopathies, e.g., hypokalemic periodic paralysis, normokalemic periodic paralysis, and cardiac arrhythmias, and may also be linked to other pathologies (Sokolov et al., 2007; Struyk and Cannon, 2007; Delemotte et al., 2010; Gosselin-Badaroudine et al., 2012; Jurkat-Rott et al., 2012; Groome et al., 2014; Moreau et al., 2015; Jiang et al., 2018, 2020).
Countercharge mutations of VSDs have also been identified in association with a number of disease phenotypes; however, the biophysical characteristics of missense mutations of countercharge residues continue to be understudied (Groome and Bayless-Edwards, 2020). Neutralization of the countercharge residue (D112) of the human proton channel hHv1 has been reported to change the selectivity of hHv1 by conducting anion currents instead (Musset et al., 2011). However, in the absence of a consensus atomic structure of hHv1, the underlying molecular mechanism and determinant regulating its ion selectivity and permeability remains elusive. In the study of a fluorescent voltage sensor, a chimera protein derived from the VSD of Ci-VSP (Ciona intestinalis voltage-sensing phosphatase; Ci-VSD), Piao et al. found that a countercharge residue mutant (D129N) can display omega currents, but the nature of the charge carrier has not been clearly characterized because current traces carried by anions were visually indistinguishable from those carried by proton or cations (Piao et al., 2015). Musset et al. initially expected the currents evoked in the D112 mutant of hHv1 to be non-selective cation currents (Musset et al., 2011). Previously, we determined two crystal structures of Ci-VSD in the “down” and “up” states (Li et al., 2014). Our electrophysiological and computational studies demonstrated the existence of two extra states of Ci-VSD: the “down-minus” state (the final resting state) and “up-plus” state (the fully activated state), and Ci-VSD translates sequentially from the down-minus state, the down state, and the up state to the up-plus state in a stepwise way during activation (Shen et al., 2022). It allows us to study the biophysical functionals of the countercharge mutants of Ci-VSP at the molecular level. In the present study, we have explored the role of countercharge residues in voltage sensitivity and the generation of omega currents by focusing on residue D129 at the extracellular side of Ci-VSD. We demonstrate here that single neutralizing mutations at D129 in S1 can lead to anionic omega currents, mostly through the down state VSD. These mutations cause large positive shifts in the voltage dependence of activation, enabling the VSD to populate the down state and elicit outward anionic omega currents at high voltages, as well as amplify the influence of external pH on activation kinetics. Transient inward anionic omega currents have also been observed upon deactivation, which can be enhanced by engineered Cd2+ bridges between S4 and lipids (immobilizing S4 slows down the transition of VSD from the conducting down state to the impermeable down-minus state during deactivation). We show through MD simulations and free-energy calculations that the neutralization of D129 or R226 generates significant changes in the electrostatic potential landscape inside the gating pore. In turn, these catalyze the anionic currents observed in the D129 mutants and the cationic currents in the R226 mutants.
Materials and methods
Molecular biology and electrophysiological recordings
The cDNA of Ci-VSP with catalytic center mutation C363S was subcloned into the pSP64T vector. This construct has no enzymatic activity but displays similar outward and inward sensing currents as the original one and has been used in the study of voltage sensitivity of Ci-VSP (Murata et al., 2005; Villalba-Galea et al., 2013; Shen et al., 2022) and referred to as the WT hereafter. The point mutations were generated using site-directed mutagenesis based on this construct. For multiple mutants containing a cysteine substitution, the background cysteine on the VSD (C159) was mutated to a serine to prevent the formation of an internal disulfide bond. The VSD alone construct (residues 1–239) was generated by deleting the cytosolic enzyme region of Ci-VSP. Plasmids purified from miniprep were linearized and transcribed in vitro using the mMessage mMachine SP6 transcription kit (Ambion, Invitrogen). The mRNA was then diluted in RNase-free water to a final concentration of ∼1 μg/μl. 50 nl of mRNA was injected into each freshly isolated Xenopus laevis oocyte, which was incubated in the standard oocyte saline (SOS) solution for 16–24 h at 18°C before recording.
Electrophysiological recordings were performed using a cut-open oocyte voltage-clamp setup (Taglialatela et al., 1992; Stefani and Bezanilla, 1998). Currents were filtered at 10 kHz using a low-pass four-pole Bessel filter within the CA-1 amplifier (Dagan Corporation). A holding potential of −60 mV was applied together with a 200 ms prepulse to −100 mV to maximally deactivate the WT and mutants of Ci-VSP. Test pulses from 120 or 160 mV to −100 mV in −10-mV decrements were used to evoke outward currents, which stepped back to −100 mV to generate inward currents. In-house software GPatch and Analysis were used for data acquisition and analysis, respectively.
The basic external solution contained 120 mM N-methyl-D-glucamine (NMDG), 2 mM Ca(OH)2, 0.5 mM EDTA, and 10 mM buffering agent: HEPES for pH 7.4, MES for pH 6.5, and CHES for pH 9.0. Methanesulfonate acid (CH3SO3−) was used to adjust the pH to the corresponding value. The internal solution contained 120 mM NMDG, 2 mM EGTA, and 10 mM HEPES, and the pH was adjusted to 7.4 using CH3SO3−. To change the external anion, HCl, HBr, HI, or glutamate was used for pH adjustment. To modify the external cation, a fraction of NMDG (12 mM) was replaced with KOH (12 mM) while keeping other components the same. Cd2+ was diluted to the external solution without 0.5 mM EDTA at the designed concentration from 100 mM CdCl2 stock solution. Freshly prepared DTT solution of 1–2 mM was added to the external solution to chelate the Cd2+ ions.
The time constant of activation of the WT Ci-VSP was calculated using the equation τON = (A1τ1 + A2τ2)/(A1 + A2) (Villalba-Galea et al., 2013), where A1, A2 and τ1, τ2 are the amplitudes and time constants for the first and second exponentials, respectively, to fit the decay phase of outward sensing currents with the in-house software Analysis. Each individual Q-V curve was fitted and normalized using the non-linear least-squares minimization and curve-fitting (lmfit) package in Python (https://lmfit.github.io/lmfit-py/model.html) based on the Boltzmann distribution Q(V) = 1/{1 + exp[ze0(V − V1/2)/kBT]}, where z is the apparent gating charge, V1/2 is the half-maximum activation voltage, e0 is the elementary charge, kB is the Boltzmann constant, and T is the absolute temperature in Kelvin (Clark et al., 2020). The normalized Q-V curves were then averaged and fitted to get the best-fit values and standard deviations for the parameters z and V1/2 using the maximum likelihood by the Monte Carlo–Markov Chain method in the lmfit package (Clark et al., 2020).
Molecular dynamics (MD) simulations
Ci-VSD D129A and R226A mutants at four different states were generated with the program VMD (Humphrey et al., 1996) using the corresponding centroid structures of the WT from the previous MD simulations (Shen et al., 2022). The ions in the bulk solution were readjusted to make the final systems electrically neutral. Following 5,000 steps of energy minimization, each system was initially equilibrated for 50 ns. The coordinates of the last frame were then used to launch four independent 50-ns simulations with a different initial set of velocities in the NPT ensemble to study the water occupancy and water wire propensity, and an additional 50-ns simulation in the NVT ensemble to study the electrostatic potential landscape. Positional restraints were applied to the α carbon atoms of residues 149 and 170 to prevent the drift of the protein, and “harmonicWalls” restraints of the collective variables interface (Colvars) module (Fiorin et al., 2013) were used to prevent the ions from entering the water crevices. Projection of the distance between an ion in the solution and a dummy atom in the center of the VSD along the z-axis (“distanceZ”) was used as the variable of the harmonicWalls restraint, with the “lowerWalls” and “lowerWallConstant” parameters being set to 22 Å and 10 kcal/mol/Å2, respectively, to ensure that the restraint has only been applied to an ion when the distanceZ value is <22 Å. The MD simulations were performed using the program NAMD (Phillips et al., 2005). The CHARMM36 force field was used for protein, lipids, and ions (MacKerell et al., 1998; MacKerell et al., 2004), and the TIP3P model was used for water (Jorgensen et al., 1983). The Langevin dynamics and the Nose–Hoover Langevin piston method were employed to keep the temperature and pressure at 300 K and 1 atm, respectively (Martyna et al., 1994; Feller et al., 1995). The van der Waals interactions were smoothing switched between 10 and 12 Å. The long-range electrostatic interactions were calculated using the particle mesh Ewald (PME) method with a grid density of at least 1/Å3 (Essmann et al., 1995). The timestep was 2 fs.
Water molecules near the principal axis of the VSD (−12 Å < x < 12 Å and −12 Å < y < 12 Å) were used to calculate the water occupancy with an interval of 2.0 Å in all three dimensions. A total of 500 snapshots from one 50-ns simulation trajectory were used to calculate the average water occupancy. The continuous water wire propensity was calculated using the breadth-first algorithm (Han et al., 2014; Shen et al., 2022). Results from all four trajectories of each state of each mutant were used to calculate the mean value and the standard deviation. We further performed longer simulations to verify the result of the continuous water wire propensity at the four major states of the WT and D129A and R226A mutants of Ci-VSD. The initial equilibration of each system was extended to 700 ns to relax the lipids surrounding the protein and the water molecules in the vestibules of the VSD. Positional restraints were first applied to the backbone atoms of the protein (first 500 ns) and later only to the α carbon atoms of residues 149 and 170 (last 200 ns) to prevent the drift of the protein with a weak force constant of 0.1 kcal/mol/Å2. Six snapshots from the last 100 ns simulation were selected to perform the production simulations, and each production simulation lasted 100 ns. No restraint was applied to the ions to speed up the simulations as the penetration event of ions into the deeper center of VSD is rare and the occasional occupancy of ions at the entrance of the vestibules has little effect on the calculation of the continuous water wire propensity. Extending the duration of the simulation didn’t lead to a significant change in the result.
Electrostatic potential calculations
Electrostatic potentials were calculated using the PMEPot plugin of VMD (Aksimentiev and Schulten, 2005). 800 snapshots from the last 40-ns trajectory were used to calculate the three-dimensional (3-D) time-average electrostatic potentials. The 2-D and 1-D electrostatic maps were extracted from the 3-D map along a plane or a vertical line passing through the center of the VSD (Khalili-Araghi et al., 2010). For the 1-D map, 800 snapshots were divided into four blocks and electrostatic potential calculation was performed individually. The 1-D results from the four blocks were then averaged to get the mean value and the standard deviation.
Free energy calculations using umbrella sampling simulations
The umbrella sampling simulations were performed to calculate the 1-D PMF for ion permeation through the D129A and R226A mutants of Ci-VSD at the down state (Shen et al., 2010; Shen and Guo, 2012; Li et al., 2021a; 2021b). The reaction coordinate was defined as the projection of the distance between the center-of-mass of the α carbon atoms of the hydrophobic gasket residues (I126, F161, and I190) and the permeating ion K+ or Cl− or the sulfur atom of CH3SO3− along the z-axis (distanceZ) using the Colvars module (Fiorin et al., 2013).
The last frame from the 50 ns equilibration simulation trajectory was used to generate starting configurations for the umbrella sampling simulations. Initially, a conducting K+, Cl−, or CH3SO3− ion was manually relocated into the center of the VSD, which was later slowly moved to the target positions or sampling window centers using the distanceZ restraint of Colvars module in a 1 ns equilibration simulation with a force constant of 1.0 kcal/mol/Å2. Then, 10-ns production simulation was performed for each sampling window. A total of 141 windows (−30 Å < z < 40 Å, with an increment of 0.5 Å) were used for each PMF with a force constant of 10.0 kcal/mol/Å2 for the harmonic restraint. The movement of the conduction ion in the X and Y directions was also regulated with a harmonicWalls restraint using “distanceXY” as the variable. The upperWalls and upperWallConstant parameters were set to 10 Å and 2 kcal/mol/Å2, respectively, to ensure that a restraint will be applied to the ion when its distanceXY value or its distance from the z axis in the XY plane is >10 Å. In addition, an extra force (5 kcal/mol/Å) will be applied to other ions in the bulk solution when they get into a cylinder region along the z-axis (r = 15 Å, −20 Å < z < 30 Å) using the tcl plugin of NAMD to prevent them from getting into the VSD water crevices. The sampling data were unbiased and combined using the weighted histogram analysis method (WHAM) to calculate the PMF (Kumar et al., 1992). The last 8-ns sampling data were evenly divided into four blocks to get four 1-D PMFs. The calculated PMFs were corrected first to remove the free energy offset between the two ends of bulk water regions (Hub and Groot, 2008) and then averaged to get the mean and standard deviation values. The free energy offset ΔGoff was removed by adding a correction term δGi,i+1 to the PMF calculated using WHAM. δGi,i+1 is proportional to 1/(nini+1)2, where ni and ni+1 are the number of data points in the adjacent bins i and i + 1 from histograms of the total sampling, and fulfills the requirement of ΔGoff = −ΣiδGi,i+1 (Hub and Groot, 2008).
Online supplemental material
Fig. S1 displays currents from uninjected Xenopus oocytes and oocytes injected with cRNA encoding the WT and D129A mutant of Ci-VSP. Fig. S2 shows the effect of buffer components and Cd2+ ions on the outward and inward currents of Ci-VSP mutants. Fig. S3 shows anion conduction through single D129 mutants of Ci-VSP. Fig. S4 shows the electrophysiological characterization of uninjected Xenopus oocytes and oocytes injected with cRNA encoding the D129V mutant of Ci-VSP. Fig. S5 demonstrates anion conduction through the D129V mutant of Ci-VSD. Fig. S6 shows the effect of external pH on outward and inward currents of single D136 and D151 mutants of Ci-VSP. Fig. S7 characterizes proton conduction through the R226A mutant of Ci-VSP. Fig. S8 displays the continuous water pathway through the down state VSD in D129A and R226A mutants of Ci-VSP from MD simulations. Fig. S9 shows that D129A and R226A mutations change the electrostatic potential landscape across the Ci-VSD from MD simulations.
Results
Anionic omega currents in the D129A mutant of Ci-VSP and modulation by external pH
Voltage-sensing phosphatases contain an N-terminal transmembrane VSD and a C-terminal cytosolic membrane-targeted enzyme domain. The ascidian VSP, Ci-VSP, was the first phosphatase that exhibited voltage-dependent catalytic activity (Murata et al., 2005). Phylogenetic analysis shows that the VSDs of VSPs are closer to the VSDs of voltage-gated proton channels (Hv1s) than those of voltage-gated ion channels (Musset et al., 2011; Li et al., 2015). A sequence alignment shows that both VSPs and Hv1s have a conserved countercharge aspartate in S1 (Fig. 1, A and B), which is D112 in hHv1 and D129 in Ci-VSP, a helix-turn above the hydrophobic gasket residues (Li et al., 2014). It should be noted that WT Ci-VSP doesn’t conduct proton currents as hHv1 despite their high sequence similarity (Musset et al., 2011; Piao et al., 2015).
We started to investigate the functional role of D129 in Ci-VSP by mutating it to alanine and characterizing its functional behavior using the cut-open oocyte voltage-clamp technique (Taglialatela et al., 1992; Stefani and Bezanilla, 1998). The D129A mutant displays “abnormal” currents in external 120 mM N-methyl-D-glucamine (NMDG+) methanesulfonate (CH3SO3−) at various pH conditions (Fig. 1, C–F): (1) it activates slowly without typical sharp peak at the very beginning of the outward currents of the WT Ci-VSP (Fig. 1 G), a sign of ON sensing currents (Murata et al., 2005); (2) it conducts an outward current at depolarization potentials; and (3) the magnitude of the maximum outward current (Imax) and inward net charge (Qmax) are tightly regulated by external pH (pHo). As a clarification, the term I is the outward current at the end of the depolarization pulse after linear leak subtraction, when the ON sensing current has decayed, and the term Q is the net translated charge calculated by integrating each inward current upon repolarization. Since the external buffer contains only putatively impermeant large organic ions (NMDG+ and CH3SO3−), we first suspected that the currents were carried out by protons or hydroxides. If D129A is proton selective, inward currents should increase at low pHo, as is the case for omega currents in Ci-VSP sensing charge mutants (Villalba-Galea et al., 2013). However, we see clear reductions in inward currents at low pHo. If, on the other hand, D129A is permeable to hydroxides, inward currents should decrease at high pHo (inhibition of the efflux of cytosolic hydroxide). Yet, currents increased at high pHo (Fig. 1, C and D).
To define D129A omega current charge carrier, we replaced the external cation from 120 mM NMDG+ to a combination of 108 mM NMDG+ and 12 mM K+ in one experiment and the external anion from CH3SO3− completely to Cl− in the other experiment (Fig. 1 E) at pHo 7.4. The inclusion of K+ did not change the Imax or Qmax. However, the Imax at 160 mV increased by two to three times after replacing CH3SO3− with Cl− without a substantial decrease in the Qmax. We note that Xenopus oocytes have endogenous cation and chloride channels that generate outward currents at large positive potentials (Vasilyev et al., 2002; Machaca et al., 2002), but these endogenous channels do not generate inward currents, are less influenced by external pH, and the maximum magnitude and increment of outward currents in the external Cl− solution is smaller than what we see with the D129A mutant (Fig. S1). Rather, we suggest that the D129A mutant generates anion-selective omega currents regulated by pHo and that these anionic omega currents turn off once the VSD deactivates. Furthermore, rapid closure of the gating pore might prevent conduction of any inward anionic omega currents upon deactivation (the inward currents include only inward sensing currents with no omega currents; Arima et al., 2018) or only conduct very small inward anionic currents (changes in inward omega currents are negligible compared with the inward sensing currents; Tombola et al., 2005; Jiang et al., 2018) in the D129A mutant of Ci-VSP.
We also report large positive shifts in the midpoint of D129A voltage dependence, with the half-maximum activation potential V1/2 shifts of >100 mV relative to WT (Fig. 1 F). We noticed that the activation kinetics of the WT and mutants of Ci-VSP can be tuned by external pH (Fig. 1, G and H; Shen et al., 2022), where higher pHo leads to faster gating kinetics with the largest effect taking place at the V1/2 region. In other words, the transition rate between adjacent conformational states increases at high pHo. However, there was no significant change in the Qmax of the WT Ci-VSP, and all the Q-V curves reached a plateau at pHo between 6.5 and 9.0 (Fig. 1 F; Shen et al., 2022). We also argue that under these conditions the D129A mutant VSD is hard to reach the fully activated up-plus state (Shen et al., 2022), as the maximum depolarization voltage (+160 mV) is lower than the V1/2 of D129A (Fig. 1 F). Therefore, at high pHo, the D129A mutant VSD has a faster transition rate out of the non-conducting resting state (down-minus state) upon activation, shifting the equilibrium toward the conducting intermediate state(s) (down state and/or up state) and thus increasing the magnitude of outward omega currents.
Augmenting of inward anionic omega currents by an engineered Cd2+ bridge
We pursued an alternative way to characterize the inward anionic omega current (extending the dwell time of the conducting state[s] during deactivation) while decoupling the pHo modulation effect. We took advantage of the finding that a cysteine mutation at the extracellular side of S4 can form Cd2+ bridges with lipid headgroups, immobilizing S4 in its activated conformations (Shen et al., 2022). To that end, we introduced a cysteine at position V219 in the background of the D129A mutant (Fig. 2 A). Control experiments showed that V219C cannot catalyze omega currents and external Cd2+ has little or no effect on D129A gating (Fig. S2). The V219C/D129A double mutant displayed a similar electrophysiological behavior as the single D129A mutant and could conduct anionic currents upon activation (Fig. 2 B and Fig. S2 D).
The V219C mutant showed a Cd2+ concentration-dependent reduction of its net charge Qmax, attributing it to the Cd2+ bridge-induced gating charge immobilization (Fig. S2 B and Fig. 2 C). In contrast, the addition of external Cd2+ leads to a significant increase in inward currents or net charge Qmax for the V219C/D129A double mutant with a positive correlation (Fig. S2 D and Fig. 2 C), while the outward currents only showed a slight increase even at 100 μM Cd2+ (Fig. 2 C). Potentiation of inward currents in the V219C/D129A double mutant can be attributed to the efflux of cytosolic anions since the engineered Cd2+ bridge slows down the closure of the gating pore, allowing more internal anions to pass through the VSD before it reaches the impermeable resting state (down-minus state) upon deactivation. As expected, Cd2+-driven increase in the net charge Qmax of the V219C/D129A double mutant readily reverses by washout with the chelating buffer (Fig. 2 D). In comparison, Cd2+ has little effect on the net charge Qmax of the L225C/D129A double mutant (Fig. 2 C) as L225C cannot form Cd2+ bridges with the lipid head groups (Shen et al., 2022).
Anionic omega currents vary in single D129 mutants
Additional substitutions at D129 also led to distinct outward anionic currents under positive potentials in the external NMDG+/CH3SO3− buffer at neutral pH (Fig. 3 A and Fig. S3). For asparagine, leucine, and valine substitutions, their Imax and Qmax are larger than those of D129A in a positive correlation way (Fig. 3 B), in contrast to the WT which yields a larger Qmax but a smaller Imax because of its impermeability to omega currents. Replacing the external buffer NMDG+/CH3SO3− with NMDG+/Cl− results in an increase in the outward currents and a decrease in the inward currents for all of the three D129 mutants, demonstrating that these inward currents are a mixture of transient inward sensing currents and transient inward anionic currents (Fig. 3, C and D; and Fig. S3, A–C). Notably, D129N, a naturally occurring mutation in the Callithrix jacchus (Cj) VSP (Fig. 1 A), has about a fivefold increase of the Imax at +160 mV in the Cl− solution than in the CH3SO3− solution. Since the D129V mutant led to the largest omega currents, we performed additional experiments under a variety of conditions to verify its anion selectivity and permeability (Fig. S3, D–F). Replacing a fraction of external NMDG+ with K+ has little or no effect on both the outward and inward currents. Decreasing the external ionic strength (Musset et al., 2011) by dilution of the external solution NMDG+/CH3SO3− with 50% isotonic sucrose decreases the outward currents (about 30% at +160 mV) but has little effect on the inward currents. Changing the external anion CH3SO3− with glutamate (Glu−) significantly decreases the outward currents (about 80% at +160 mV), while it has little effect on the inward currents either (Fig. 3, C and D). These results confirm that the anion selectivity of the D129 mutants (with higher permeability for Cl− than CH3SO3−), the presence of outward and inward anionic omega currents during activation and deactivation, respectively, and the decrease of the inward anionic omega currents with the presence of high concentration Cl− in the extracellular buffer are attributed to the suppression of cytosolic Cl− efflux.
We further tested the permeability of other common anions Br−, I−, and SCN− using the D129V mutant of Ci-VSP as an example and non-injected oocytes as the control (Fig. S4, A and B). Due to the existence of endogenous anion channels in Xenopus oocytes, non-injected oocytes display noticeable outward currents for all the test anions with a permeability series of SCN− > I− > Br− > Cl− (Fig. S4 C), which is in accord with previous reports (Young et al., 1984; Ferrera et al., 2011). It should also be noted that SCN− behaves as a classical hydrophobic ion and can directly permeate “naked” lipid vesicles (Perozo and Hubbell, 1993), so, it is also expected to permeate the oocyte membrane even in the absence of functional anionic channels. In oocytes expressing the D129V mutant, substitution of external CH3SO3− with Br−, I−, and SCN− increased outward currents while decreasing inward currents, as when substituting with Cl− (Fig. S4, D and E). By considering the currents conducted by endogenous anion channels, Cl−, Br−, and I− have similar permeability through the D129V mutant, which is much smaller than SCN− (Fig. S4, F and G). A D129V mutant based on the construct of Ci-VSD (residues 1–239) by removing the cytosolic enzyme region of Ci-VSP shows similar electrophysiological phenotypes as D129 mutants of Ci-VSP (Fig. S5), confirming the critical role of the VSD domain in the conduction of anionic omega currents.
Neutralized D136 and D151 mutants are impermeable to ions and protons
We evaluated other potential ion-conducting phenotypes by focusing on countercharges D136 and D151 located at the extracellular entrance of the gating pore water crevices (Fig. 1 B). In contrast to D129 mutants, neutralization of D136 and D151 (with either alanine or asparagine) did not result in omega currents in NMDG+/CH3SO3− under acidic, neutral, or basic conditions, as the external pH value has little or no effect on their Imax and Qmax (Fig. 4, A and B; and Fig. S6). However, their activation kinetics and voltage sensitivity were modulated by the external pH as with WT and the D129 mutants. Particularly, at low pHo, we observed a positive shift of the Q-V curve for the D136A, D136N, and D151N mutants. We have previously shown that 100 μM Cd2+ in the external buffer does not change the net charge Qmax in the V219C/D136A and V219C/D151A mutants of Ci-VSP (Shen et al., 2022), further confirming that neutralization of countercharge residues D136 and D151 cannot conduct omega currents.
Although countercharge residues D136 and D151 are not directly involved in the conduction of omega currents, they still regulate the anionic omega current in D129 neutralization mutants, likely by further changing the voltage sensitivity of gating (Fig. 4 C). Asparagine substitution of D136 and/or D151 in a D129N background amplifies the pHo effect on the magnitude of the anionic omega current. For example, the net charge Qmax of the triple mutant D129N/D136N/D151N increased about 2.5-fold at +160 mV when the pHo changed from 7.4 to 9.0, while it is only about a 1.2-fold change for the single mutant D129N in the same condition.
A molecular mechanism for state dependency and ion selectivity of anionic omega currents
For voltage-gated ion channels and phosphatases, there exists a narrow and hydrophobic region at the center of their VSDs resembling an hourglass-like internal morphology. Hydrophobic residues from transmembrane segments S1–S3 form a layer of “hydrophobic gasket,” with the help of bulky sidechains of gating charge residues on S4, preventing the connection of water molecules from the extracellular and intracellular water crevices during the movement of the S4 segment, and hence the conduction of omega currents through WT VSDs. The formation of a continuous water pathway connecting the extracellular and intracellular solutions has been considered to be a prerequisite for omega current conduction through VSDs having gating charge mutation(s). Generally, this takes place when the mutated gating charge residue (by a neutral amino acid with a small sidechain) reaches close to the hydrophobic gasket region during activation or deactivation, leading to the state dependency of omega currents (Tombola et al., 2005; Delemotte et al., 2010; Gamal El-Din et al., 2010; Khalili-Araghi et al., 2012; Jiang et al., 2018; Shen et al., 2022). However, in contrast to the gating charges, countercharge residues are located in the S1–S3 segments of VSDs, which have been associated with limited conformational rearrangement during voltage gating (Li et al., 2014; Wisedchaisri et al., 2019; Xu et al., 2019; Clairfeuille et al., 2019; Lee and MacKinnon, 2019). To understand the molecular mechanism for the state-dependent and selective conduction of anionic omega currents, we performed MD simulations of the D129A and R226A mutants of Ci-VSD. As a control, the R226A mutant of Ci-VSP can also conduct transient omega currents, but with a cation selectivity (Fig. S7; Shen et al., 2022).
First, we examined water occupancy of the D129A and R226A mutants of Ci-VSD at the four major functional states: down-minus, down, up, and up-plus using the WT Ci-VSD as a control (Shen et al., 2022). The water path was disconnected in the hydrophobic gasket region of the WT Ci-VSD in all four states (Fig. S8 A), which is consistent with the electrophysiological result that WT Ci-VSP doesn’t conduct any omega currents. In addition, there is an extra restriction site near the position of 129 (Fig. S8 A), corresponding to the reorientation of a nearby gating charge residue to form a salt bridge interaction with D129 (Shen et al., 2022). We found that both the D129A and R226A mutants can form a continuous water pathway at the down state, but not the other three states (Fig. S8, B and C). We further quantitatively evaluated the propensity of the formation of a continuous water wire connecting the extracellular and intracellular sides of VSD (Fig. 5 A). Similar to the water occupancy result, the D129A and R226A mutants have a high propensity to form continuous water wires in the down state. In addition, the up state R226A mutant, and the up state and up-plus state D129A mutant can also form continuous water wires with a much lower propensity. Occasionally formed water wires in the down-minus and down states WT Ci-VSD have also been observed. These simulation results suggest that the down state VSD is mostly responsible for the conduction of omega currents in the D129A and R226 mutants of Ci-VSP.
Analysis of the MD trajectories showed that the occupancy of the R226 side-chain in the hydrophobic gasket region prevents the formation of continuous water wires through the down state WT Ci-VSD (Fig. 5, B and C; and Fig. S8 A). Neutralizing D129 and R226 enables the hydration of the entire down state VSD by creating an open portal at the hydrophobic gasket region (Fig. 5 B). As shown in Fig. 5 C, the position of the CZ atom of R226 shifts toward D186 in the D129A mutant compared with the WT due to the absence of the electrostatic interaction between R226 and D129. Although the CB atom of A226 in the R226A mutant is still located at a similar level of the hydrophobic gasket, the small side chain of A226 cannot fully block the water pathway.
Electrostatic potential calculations show that neutralizing D129 makes the gating pore surface more positive than in WT, while neutralization of the gating charge R226 creates a negative electrostatic surface (Fig. 5 C and Fig. S9). We suggest that these changes in the VSD electrostatic landscape help define the ion selectivity of the gating pore (Bavi et al., 2021). One-dimensional potential of mean force (PMF) calculations were carried out to study the free energy for ion conduction through the down state D129A and R226A mutants of Ci-VSD. Consistent with our experimental observations, the energy barrier is about 4 kcal/mol for Cl− and K+ ions to permeate the D129A and R226A mutants. But in the D129A mutant, K+ encounters an energy barrier of over 16 kcal/mol to pass through the gating pore, while the cost of Cl− permeation is about 20 kcal/mol in the R226A mutant (Fig. 5 D). These results provide a satisfying explanation for the different ion selectivity in these two mutants. The large anion CH3SO3− experiences a deep energy basin of about −8 kcal/mol at the extracellular side of the hydrophobic gasket region, while the highest energy barrier is close to that for Cl−, which helps explain the lower permeability of CH3SO3− relative to Cl−.
Discussion
VSDs are conserved structural and functional modules that transduce transmembrane potential changes into defined conformational rearrangements of its upstream or downstream effectors. A narrow hydrophobic “gasket” or charge transfer center at the core of most VSDs focuses the electric field into a narrow region and catalyzes the sequential and reversible translocation of positive gating charge residues across the electric field while preventing the permeation of physiological ions in voltage-gated ion channels, phosphatases, and solute carriers (Starace and Bezanilla, 2004; Ahern and Horn, 2005; Tao et al., 2010; Freites et al., 2012; Tombola et al., 2005). However, under certain conditions, omega currents have been observed in the VSDs of voltage-gated ion channels containing site-directed as well as naturally occurring mutations of S4-gating charge residues. These cation-selective omega currents in sodium and calcium channels have been linked to certain channelopathies (Sokolov et al., 2007; Jurkat-Rott et al., 2012; Wu et al., 2021). In the present study, we report the discovery of a state-dependent, anion-selective omega current in single countercharge D129 mutants of Ci-VSP.
Just as with gating charge mutants, neutralization of the D129 countercharge in Ci-VSP leads to a continuous water permeation path (mostly linked to its down state) by changing the side-chain orientation of the gating charge residue occupying the hydrophobic gasket region. Gating charge residues in VSDs make extensive electrostatic interactions with countercharge residues, polar residues, as well as lipid headgroups during voltage gating (Palovcak et al., 2014; Delemotte et al., 2015; Henrion et al., 2012; Yarov-Yarovoy et al., 2012; Groome and Bayless-Edwards, 2020). Neutralization of countercharge residues modifies this complicated network of interactions and can flip the side-chain orientation of the nearby gating charge. This, in turn, opens a transient aqueous portal at the hydrophobic gasket region allowing ion conduction. In contrast, S4 translocation brings mutated gating charge residues across the electric field (i.e., in histidine substitutions; Starace and Bezanilla, 2004) or to the hydrophobic gasket region, making a continuous water pathway directly and leading to proton and/or cation currents. Neutralization of gating charges and countercharges can also modify the electrostatic landscape of the gating pore of VSDs in a way that defines gating pore selectivity. Importantly, the hourglass-like morphology of the gating pore poses additional steric constraints that drive selectivity toward smaller ions, e.g., small physiological ions like K+, Na+, and Cl− have a higher permeability than large organic ions like guanidinium, NMDG+, and CH3SO3−.
Neutralizing countercharge residues also change the voltage sensitivity of the gating of VSDs. The half-maximum activation potential of D129A shifts by over 100 mV in the positive direction compared with the WT. This increases the population of the anion conducting conformation (the down state) at high voltage ranges of the current study (up to +160 mV), as much as larger voltages are required to trigger the subsequent movement of the VSD from the down state to more activated but less permeable conformations (e.g., the up and up-plus states). In addition, high external pH increases the conformational transition rate of the VSD from the resting down-minus state to the down state, increasing the down state population at the same depolarization potentials. As shown in Fig. 1 H, external pH has the largest effect on the activation kinetics near the half-maximum activation potential region. Positive shifts in the Q-V curve of the D129 mutants, with V1/2 reaching the highest record potential range in the current study (Fig. 1 F), further enhance the effects of pHo on gating voltage sensitivity and outward anionic omega current magnitude. As expected, immobilizing the S4 segment by an engineered Cd2+ bridge leads to an increase in the dwell time of the down state, which in turn increases the inward anionic omega currents (outward permeation of cytosolic anions) during deactivation.
Neutralization of the countercharge D112 of hHv1 (D129 in Ci-VSP) has been shown to switch hHv1 from a proton-selective channel to an anion-selective channel. Neutralization of the D185 countercharge (T107 in Ci-VSP) has little effect on the proton selectivity, consistent with the observation that neutralization of D136 and D151 in Ci-VSP (E119 and H140 in hHv1, respectively) is unable to permeate anionic omega currents as in the D129 mutants. However, it should be noted that the D185 mutant and WT hHv1 can conduct proton currents, while the D136 and D151 mutants and WT Ci-VSP don’t conduct any type of omega currents at all. Interestingly, currents were abolished in the D112V mutant of hHv1, while the D129V mutant of Ci-VSP conducts the largest anionic omega current in all the tested D129 mutants. These results suggest that although countercharge mutants of hHv1 (D112) and Ci-VSP (D129) can both conduct anionic currents, their underlying mechanisms and molecular determinants are unique.
The present findings define the determinant and energy for state-dependent anionic omega currents via countercharge mutants at position D129 of Ci-VSP. The underlying molecular mechanism here proposed extends our understanding of the biophysical roles of countercharge residues in the impermeable VSDs. We expect that the present framework will help us uncover some of the unique mechanisms fundamental to the understanding of channelopathies caused by countercharge mutations (Groome and Bayless-Edwards, 2020), develop new drugs targeting these diseases, and design novel engineered proteins for scientific, industrial, and therapeutic applications.
Data availability
All data are available in the main text or the supplementary materials.
Acknowledgments
Christopher J. Lingle served as editor.
We thank the Francisco Bezanilla laboratory at the University of Chicago (Chicago, IL) for providing oocytes and access to in-house software. We thank Drs. Tian Li, Michael David Clark, and the members of the Perozo lab and the Roux lab for helpful advice and discussions.
This work was supported by the National Institutes of Health grants GM057846 (to E. Perozo) and GM062342 (to B. Roux).
Author contributions: R. Shen, B. Roux, and E. Perozo designed the whole study and analyzed the data. R. Shen performed the electrophysiological experiments and MD simulations. R. Shen wrote the initial manuscript. R. Shen, B. Roux, and E. Perozo revised the manuscript.
References
Author notes
Disclosures: The authors declare no competing interests exist.