Exotic properties of a voltage-gated proton channel from the snail Helisoma trivolvis

Voltage-gated proton channels, H_V1, remain relative newcomers to the ion channel family. Although the idea of a depolarization-activated proton-selective ion channel was proposed in 1972 by J. Woodland Hastings and colleagues (Fogel and Hastings, 1972), the first voltage-clamp study that established the existence of this channel type occurred a decade later in the snail Helix aspersa (Thomas and Meech, 1982). An H_V1 gene was not identified until 2006 (Ramsey et al., 2006; Sasaki et al., 2006). Strong interest in this channel has arisen for two main reasons. First, its structure, with just four transmembrane helices, closely resembles the voltage-sensing domain of other voltage-gated ion channels, making it a unique model for voltage-gating mechanisms. By combining voltage sensing, gating, and conduction into a single module, H_V1 uniquely provides a direct readout of its gating state. Second, exceedingly diverse functions have been identified for H_V1 in many species and in many human tissues (DeCoursey, 2013).

The first systematic voltage-clamp characterization of voltage-gated proton currents was in Lymnaea stagnalis snail neurons (Byerly et al., 1984). When mammalian proton currents were identified a decade later (DeCoursey, 1991), the most obvious difference was that H_V1 in snails activated two to three orders of magnitude faster. Here, we investigate the properties of the Helisoma trivolvis snail H_V1 gene product. We searched a transcriptome of H. trivolvis and found a putative HtH_V1; we then cloned the gene from a cDNA pool constructed from H. trivolvis brain tissue. We find many similarities to native proton currents studied in situ in other snail species, including rapid gating kinetics and other significant differences from mammalian H_V1. HtH_V1 currents differ from mammalian H_V1 in having exponential (vs. sigmoid) activation, similarity of τ_act and τ_tail at overlapping voltages, and maximal time constants near the midpoint of the proton conductance–voltage (g_H-V) relationship, all features suggestive of simple first-order gating kinetics expected of a monomeric protein. However, the existence of an extensive coiled-coil motif in the C terminus together with steep voltage dependence suggests “cooperative” gating of a dimeric protein. Potent inhibition of HtH_V1 by Zn²⁺ and Cd²⁺ is explained by conservation of three of four members of the Zn²⁺-binding site (Takeshita et al., 2014). The most remarkable property of the HtH_V1 channel is that its sensitivity to pH_i is anomalously weak. The voltage-gating mechanism of all H_V1 identified to date is unique in being nearly equally responsive to pH_o and pH_i, such that a one-unit change in either shifts the g_H-V relationship by 40 mV. This “rule of forty” (DeCoursey, 2013) has the biologically crucial effect of ensuring that H_V1 channels open only when there is an outward electrochemical gradient for H⁺. In other words, H_V1 channels open only when doing so will result in acid extrusion from cells. Extensive mutation of hH_V1 has failed to produce any significant violation of the rule of forty (Ramsey et al., 2010; DeCoursey, 2016). In this issue, Cherny et al. identify a single amino acid difference between HtH_V1 and hH_V1 that appears to be responsible for the anomalous ΔpH dependence of the snail channel.

H. trivolvis, a pulmonate snail (order: Basommatophora; family: Planorbidae) from an albino stock maintained and continuously bred in aquaria at Georgia State University, was used for experiments. Snails were originally caught in the wild and introduced as an experimental model animal by S.B. Kater (Kater, 1974).

Basic Local Alignment Search Tool searches of a transcriptome from H. trivolvis (unpublished data) yielded a hit that matched the criteria for an H_V1 sequence (Smith et al., 2011). Brains were dissected from H. trivolvis (Cohan et al., 2003), RNA was extracted from brain tissue using the RNeasy kit (Qiagen), and a cDNA pool was constructed using the SuperScript III kit (Life Technologies) according to the manufacturer's instructions. Primers designed against the transcriptome hit were used to clone the putative HtH_V1 coding sequence; the sequence was confirmed by commercial sequencing (SourceBio Science). This coding sequence was subcloned into eukaryotic expression vector pCA-IRES-eGFP. Site-directed mutagenesis of HtH_V1 was performed and sequence verified commercially (Genewiz). Antibody was raised in rabbit to a synthetic peptide (RSPSDHGEGFEEPLC) based on the predicted HtH_V1 epitope and affinity purified (Genscript) with a final concentration of 0.904 mg/ml. Total lysate was prepared from H. trivolvis brains that had been stored whole in Qiagen RLT buffer at −80°C for 12 mo. Brains were thawed and triturated briefly on ice; the lysate was cleared by centrifugation at 10,000 × g for 5 min. Proteins from H. trivolvis brain lysate were separated by SDS-PAGE, Western blotted, and probed with anti-HtH_V1 antibody (diluted 1:10,000 in blocking buffer) either alone or preincubated with 1,000-fold molar excess of synthetic peptide corresponding to the epitope.

HEK-293 cells were grown to ∼80% confluence in 35-mm culture dishes. HEK-293 cells were transfected with 0.4–0.5 µg cDNA using Lipofectamine 2000 (Invitrogen) or polyethylenimine (Sigma). Plasmids that did not include GFP were cotransfected with GFP. After 24 h at 37°C in 5% CO₂, cells were trypsinized and replated onto glass coverslips at low density for patch-clamp recording. We selected green cells under fluorescence for recording. Because HEK-293 cells often have small endogenous H_V1 currents (Musset et al., 2011), cells that exhibited small currents suspected to be native were exposed to 1 µM Zn²⁺, which has generally weaker effects on HtH_V1 (20% slowing of τ_act, ∼5 mV shift of the g_H-V relationship, and a 24% decrease in g_H,max in three to four cells at pH_o 7) than on hH_V1 (more than a twofold slowing of τ_act, ∼20 mV shift of the g_H-V relationship; Musset et al., 2010b). Cells determined on this basis to exhibit native currents were excluded from the study.

Micropipettes were pulled using a Flaming Brown automatic pipette puller (Sutter Instruments) from Custom 8520 Patch Glass (equivalent to Corning 7052 glass; Harvard Apparatus), coated with Sylgard 184 (Dow Corning Corp.), and heat polished to a tip resistance range of typically 3–10 MΩ with highly buffered TMA⁺ pipette solutions. Electrical contact with the pipette solution was achieved by a thin sintered Ag-AgCl pellet (In Vivo Metric Systems) attached to a Teflon-encased silver wire, or simply a chlorided silver wire. A reference electrode made from a Ag-AgCl pellet was connected to the bath through an agar bridge made with Ringer’s solution. The current signal from the patch clamp (EPC-9 from HEKA Instruments or Axopatch 200B from Axon Instruments) was recorded and analyzed using Pulse and PulseFit software (HEKA Instruments), or P-CLAMP software supplemented by Sigmaplot (SPSS). Seals were formed with Ringer’s solution (in mM: 160 NaCl, 4.5 KCl, 2 CaCl₂, 1 MgCl₂, 5 HEPES, pH 7.4) in the bath, and the potential was zeroed after the pipette was in contact with the cell. Current records are displayed without correction for liquid junction potentials.

Whole-cell or excised inside-out patch configurations of the patch-clamp technique were performed. Bath and pipette solutions were used interchangeably. They contained (in mM) 2 MgCl₂, 1 EGTA, 80–100 buffer, and 75–120 TMA⁺ CH₃SO₃^–, adjusted to bring the osmolality to ∼300 mOsm, and were titrated using TMAOH. Buffers with pK_a near the desired pH were used: homo-PIPES for pH 4.5–5.0, Mes for pH 5.5–6.0, BisTris for pH 6.5, N,N-bis(2-hydroxyethyl)-2-aminoethanesulfonic acid for pH 7.0, HEPES for pH 7.5, Tricine for pH 8.0, and N-cyclohexyl-2-aminoethanesulfonic acid for pH 9.0. Experiments were done at room temperature (∼20–25°C). Current records are shown without leak correction.

Reversal potentials (V_rev) in most cases were determined from the direction and amplitude of tail current relaxation over a range of voltages after a prepulse that activated the proton conductance, g_H. When the g_H was activated negative to V_rev, the latter could be determined directly from families of currents. Currents were fitted with a single exponential to obtain the activation time constant (τ_act), and the fitted curve was extrapolated to infinite time to obtain the steady-state current amplitude (I_H), from which the g_H was calculated as g_H = I_H/(V − V_rev). Thus we assume that the time-dependent component reflects H⁺ current, and time-independent current represents leak. Because of the strong voltage dependence of activation kinetics, we frequently applied longer pulses near threshold voltages and shorter pulses for large depolarizations to resolve kinetics and avoid proton depletion associated with large H⁺ flux. The voltage at which g_H was 10% of g_H,max (V_gH,max/10) was determined after defining g_H,max as the largest g_H measured.

The gene coding for a putative voltage-gated proton channel was identified based on criteria established previously, namely the presence of four transmembrane helices homologous to S1–S4 of voltage sensor domains with an Asp in the middle of the S1 transmembrane helix and the RxWRxxR motif in S4 (Musset et al., 2011; Smith et al., 2011). We cloned the putative HtH_V1 gene from a cDNA pool of brain tissue, verifying that this gene is expressed at the RNA level. Protein level expression was verified by Western blotting of H. trivolvis brain lysate probed with a commercially raised antibody to a synthetic peptide based on a HtH_V1 epitope (Fig. 1, inset). The single protein detected ran at ∼50 kD, somewhat larger than the predicted size of 40 kD. Glycosylation at the five putative N-glycosylation sites in the S1–S2 linker could account for this discrepancy, given that N-linked oligosaccharides range from 1,884 to 2,851 D (Imperiali and O’Connor, 1999). Excess synthetic peptide abolished the binding of antibody to brain lysate, establishing the specificity of the antibody. The antibody did not significantly bind to two proteins that do not contain the epitope: human glutathione S-transferase and luciferin binding protein from Lingulodinium polyedrum.

The HtH_V1 channel protein (Fig. 1) is substantially larger than the human hH_V1, with 360 amino acids (hH_V1 has 273). Much of this excess resides in the S1–S2 extracellular linker with 73 residues (vs. eight in hH_V1), which contains five potential N-glycosylation sites (vs. 0 in hH_V1). Focusing on the transmembrane regions, HtH_V1 has charged amino acids nearly identical to those of hH_V1. One exception is at the outer end of the S1 helix, where hH_V1 has basic Lys¹²⁵ but snail HtH_V1 has acidic Glu¹²⁰. There is extensive predicted coiled-coil in the C terminus: 36 residues (positions 289–324) with 90% stringency, and 28 residues (294–321) with 99% stringency according to MARCOIL (Delorenzi and Speed, 2002). H_V1 in several species have been shown to exist as dimers, largely because of coiled-coil interactions in the C terminus (Koch et al., 2008; Lee et al., 2008; Tombola et al., 2008).

The HtH_V1 gene was transfected into HEK-293 cells. Under voltage clamp, transfected cells displayed depolarization-activated currents. The selectivity of these currents was established by measuring the reversal potential, V_rev, over a range of pH_o and pH_i values (Fig. 2). The measured values of V_rev are close to the Nernst potential for H⁺, E_H, shown as a dashed line. Clearly, the HtH_V1 channel is highly proton selective over the pH range studied.

A family of currents generated by HtH_V1 at symmetrical pH 6.0 is illustrated in Fig. 3 A. The currents activate rapidly with depolarization, and activation becomes much faster at higher voltages. Although H_V1 currents in all species activate more rapidly at more positive voltages, the τ_act of HtH_V1 currents (solid and open red squares in Fig. 3 C) exhibits noticeably steeper voltage dependence. The maximum slope of the τ_act-V relationship in seven cells at pH_o 6 was 13.0 ± 3.4 mV/e-fold change in τ_act (mean ± SD). In several mammalian H_V1, τ_act changes e-fold in 40–72 mV (DeCoursey, 2003). Channel closing in HtH_V1 was also steeply voltage dependent (Fig. 3 B and blue diamonds in Fig. 3 C), with τ_tail changing e-fold in 14.2 ± 1.9 mV in six cells. In mammalian cells, the slope is typically much flatter, 26–44 mV/e-fold change in τ_tail (DeCoursey, 2003).

A remarkable feature of HtH_V1 is that at intermediate voltages where the measurements overlap, the time constants of H⁺ current turn-on (τ_act) and deactivation (τ_tail) essentially superimpose (Fig. 3 C). This behavior is suggestive of simple first-order kinetics, such as a two-state system:

in which α is the rate of channel opening and β is the rate of channel closing, and the time constant τ is (α + β)⁻¹ (Hodgkin and Huxley, 1952). Another feature suggestive of first-order kinetics is evident in the g_H-V relationship from this cell (Fig. 3 D). The voltage at which the g_H is half-maximal is ∼40 mV, where the time constants are maximal (Fig. 3 C). However, the limiting slope of the g_H-V relationship in Fig. 3 D, i.e., the slope of the most negative values obtained, indicates a gating charge of ∼6 e₀. The mean gating charge in 18 limiting slope measurements was 5.5 ± 0.9 e₀ (mean ± SD). Because the range of g_H values resolved did not exceed three orders of magnitude, these gating charge estimates should be considered lower limits. In most species, cooperative gating of the dimeric H_V1 channel doubles the gating charge from 2–3 to 4–6 e₀ (Gonzalez et al., 2010, 2013; Fujiwara et al., 2012).

Mean gating kinetics determined at symmetrical pH 7.0 is shown in Fig. 4 A. As was also seen at pH 6.0 (Fig. 3 C), at voltages where τ_act and τ_tail overlap, they have similar values, suggestive of first-order gating kinetics. In the first description of proton currents in snail neurons, the activation time to half-peak current was 25 ms or less at pH_o 7.4 (Byerly et al., 1984). With this in mind, the activation kinetics of HtH_V1 is quite similar to that reported in neurons from L. stagnalis. When proton currents were first identified in mammalian species, they were found to be radically slower (DeCoursey, 1991; Bernheim et al., 1993; DeCoursey and Cherny, 1993; Demaurex et al., 1993; Kapus et al., 1993). The activation kinetics of HtH_V1 is two to three orders of magnitude faster than that of hH_V1 (Fig. 4 B), averaging 476 times faster between 50 and 90 mV at pH_o 7.

The polyvalent metal cations Zn²⁺ and Cd²⁺ were among the first H_V1 inhibitors identified (Thomas and Meech, 1982; Mahaut-Smith, 1989b). Zn²⁺ in particular has been used widely on H_V1 identified in new species and remains the most potent inhibitor (Cherny and DeCoursey, 1999). Fig. 5 illustrates the effects of 100 µM Zn²⁺ or Cd²⁺ on HtH_V1 currents. Three main effects are evident: the current amplitude is reduced, the current activates more slowly (scaled currents in Fig. 5 D), and the g_H-V relationship is shifted positively along the voltage axis. These three parameters are interrelated in that a positive shift of the g_H-V relationship will in itself decrease the current and slow τ_act at any given voltage. The mean changes in these three parameters produced by 10 or 100 µM of the two metals are summarized in Fig. 5 E.

These three effects of polyvalent metal cations have been observed for H_V1 from many species. As in rat H_V1 (Cherny and DeCoursey, 1999), Zn²⁺ is more potent than Cd²⁺ in HtH_V1. Focusing on the three main effects, HtH_V1 was more sensitive, similar to, or less sensitive than human H_V1 (hH_V1). The reduction of g_H,max is glaringly obvious for HtH_V1, whereas in mammalian H_V1 this effect is small and difficult to detect because of the interrelatedness of the three effects (Cherny and DeCoursey, 1999). Zn²⁺ at 10 µM slows τ_act by four- to fivefold in both hH_V1 (Musset et al., 2010b) and HtH_V1 (Fig. 5). In contrast, the shift of the g_H-V relationship by Zn²⁺ is far more profound in human hH_V1, with a 20-mV shift produced by 1 µM Zn²⁺ (Musset et al., 2010b) compared with a 12-mV shift by 10 µM Zn²⁺ in HtH_V1 (Fig. 5).

Families of proton currents generated by the H. trivolvis proton channel gene product, HtH_V1, in a cell studied at four pH_o values with pH_i 6 are illustrated in Fig. 6 (A–D). The currents activate with depolarization, and activation becomes much faster at higher voltages. Both voltage dependence and kinetics were exquisitely sensitive to pH_o. At higher pH_o, the proton conductance, g_H, turned on at more negative voltages and turned on much more rapidly (note the different time bases). Fig. 6 (E–H) shows deactivation kinetics at each pH_o during tail current measurements in this cell. Channel closing becomes much more rapid at more negative voltages.

In Fig. 6 I, time constants of H⁺ current turn-on (activation, τ_act) and turn-off (deactivation, τ_tail) from the same cell are plotted. Several intriguing features emerge. Unlike H_V1 in other species, at intermediate voltages where τ_act and τ_tail overlapped, they were of similar magnitude (as was seen in Figs. 3 and 4). Because this behavior suggests first-order kinetics (Scheme 1), the data in Fig. 6 I were analyzed in this way, and the expressions for each rate constant are given on the graph in the following form:

and

The voltage dependence of τ_act is steep (small k_α) and appears to become steeper at lower pH_o. The voltage dependence of channel closing, τ_tail, is also steep (small k_β) but appears to be independent of pH_o. To a first approximation, β is independent of pH_o, whereas α is markedly influenced by pH_o. Mean values for the rate constants are plotted in Fig. 7. Confirming the impression from Fig. 6 I (where k_α was 11.4, 18, 26, and 30 mV at pH_o 5, 6, 7, and 8), k_α increased with pH_o, and k_β was pH independent. But by far the strongest effect of pH is that increasing pH_o massively increases the opening rate constant α, which increased more than an order of magnitude per unit increase in pH_o. Stated differently, protonation at the external face of the HtH_V1 channel strongly inhibits channel opening. More subtly, it is evident that for any given pH_o, α is higher and β is lower at pH_i 6 than at pH_i 7; hence, lower pH_i both promotes opening and slows closing.

The g_H-V relationships from the cell in Fig. 6 (A–I) are plotted in Fig. 6 J. Like other H_V1, HtH_V1 exhibits robust pH_o-dependent shifts with increasing pH_o shifting the g_H-V relationship negatively. The shifts for pH_o 5 → 6 and 6 → 7 are closer to 50 than 40 mV, indicating that HtH_V1 exceeds the rule of forty for changes in pH_o. To reconstruct g_H-V relationships using the simple first-order assumption (Scheme 1 and Eqs. 1 and 2), which predicts that P_open = α/(α + β), the solid curves in Fig. 6 J were drawn from the rate constant equations in Fig. 6 I scaled by g_H,max. Their limiting slope at negative voltages is shallower than observed. Squaring the P_open-V relationships, as in the classic Hodgkin–Huxley n² approach (Hodgkin and Huxley, 1952), produces the steeper dashed curves, which better approximate the data. Without pushing the model too far, we conclude that it is probable that, like several other H_V1 (Gonzalez et al., 2010; Musset et al., 2010b; Tombola et al., 2010; Fujiwara et al., 2012), HtH_V1 functions as a dimer in which both protomers must activate before either one conducts.

The effects of changes in pH_i were explored in inside-out patches, as illustrated in Fig. 8. The resolution was limited somewhat by the typically small current amplitude combined with rapid activation kinetics at some pH. Activation kinetics could be resolved at low but not at high pH_i. For example, at pH_i 8 (Fig. 8 C), inward current is clearly activated, but the kinetics is ambiguous. Nevertheless, it is evident in Fig. 8 D that activation kinetics depends only weakly on pH_i, in stark contrast to the strong dependence seen for pH_o (Figs. 6 and 7), and in contrast to mammalian H_V1, in which lowering pH_i speeds activation fivefold/unit (DeCoursey and Cherny, 1995; Villalba-Galea, 2014). Deactivation kinetics was poorly resolved in most patches. The most surprising feature (Fig. 8 E) is that the heretofore universal rule of forty governing ΔpH-dependent gating is violated by HtH_V1. Changing pH_i shifts the g_H-V relationship of HtH_V1 by just 20 mV/unit or less. The aberrant behavior of HtH_V1 provides clues to the mechanism of ΔpH-dependent gating.

Fig. 9 summarizes the ΔpH dependence of HtH_V1. For a variety of reasons discussed elsewhere (Cherny et al., 2015), we have adopted V(g_H,max/10), the voltage at which the g_H is 10% of its maximal value, as a parameter to define the position of the g_H-V relationship. We find this preferable to other parameters that have been used for this purpose, such as the midpoint of a Boltzmann curve (which frequently does not fit the data well or is ill determined) or the threshold voltage at which current is first detectable (which is arbitrary, depends on the signal-to-noise ratio, and is particularly difficult to resolve when it occurs near V_rev, as frequently occurs in HtH_V1). It is evident in Fig. 9 that when pH_o < 7, changes in pH_o shift V(g_H,max/10) by more than 40 mV/unit (for reference, this slope is shown as a dashed green line in Fig. 9). H_V1 in two other species (coccolithophore EhH_V1 and insect NpH_V1) also exhibit shifts with pH_o greater than 40 mV/unit (Cherny et al., 2015; Chaves et al., 2016). At pH_o higher than 7, the shift decreases, which may reflect saturation of the response caused by the ambient pH approaching the pK_a of a critical titratable group. Saturation of ΔpH dependence has been observed previously in hH_V1 at pH > 8 (Cherny et al., 2015).

The most striking result in Fig. 9 is the data for changes in pH_i (dark red diamonds), which reveal that the position of the g_H-V relationship depends only weakly on pH_i. There is no clear indication of saturation, although the slope appears to increase at larger ΔpH (i.e., lower pH_i). This is qualitatively like the whole-cell pH_o response, which is steepest at low pH_o and saturates at high pH_o. Over the entire ΔpH range, the mean slope is only 15.3 mV/unit change in pH_i. HtH_V1 is the first H_V1 in which such weak ΔpH dependence has been identified.

The snail H_V1, HtH_V1, exhibits all of the major features of H_V1 in all species studied thus far. It is highly proton selective and it is voltage gated, opening with depolarization, and opening more rapidly at more positive voltages. Furthermore, its voltage dependence is strongly modulated by pH, such that increasing pH_o or decreasing pH_i shifts the g_H-V relationship negatively, in what has been called ΔpH-dependent gating (Cherny et al., 1995). Beyond these qualitative similarities, however, HtH_V1 differs markedly from H_V1 in humans and other mammalian species. The main differences include very rapid activation kinetics, steeply voltage-dependent activation kinetics, activation in a more negative voltage range, exponential rather than sigmoid activation, and distinctly aberrant ΔpH dependence. These properties are discussed below.

The first voltage-gated proton channels to be characterized by voltage clamp were in neurons from the snails L. stagnalis (Byerly et al., 1984), H. aspersa (Thomas and Meech, 1982; Mahaut-Smith, 1989b), and Helix pomatia (Doroshenko et al., 1986). All activated rapidly, with time constants, τ_act, of a few milliseconds. When mammalian proton currents were identified, the most obvious difference was much slower activation, with τ_act in the range of seconds (DeCoursey, 1991; Bernheim et al., 1993; Demaurex et al., 1993; Kapus et al., 1993) or even minutes (DeCoursey and Cherny, 1993). A more subtle difference was that mammalian H_V1 activate with a distinct delay, whereas snail H_V1 activate exponentially. We show here that the HtH_V1 channel shares both properties with other snail H_V1. Byerly et al. (1984) reported half-times for activation of less than 25 ms for proton currents in L. stagnalis neurons at pH_o 7.4, as observed here at pH_o 7 (Fig. 4 A).

Activation and deactivation time constants in HtH_V1 are of similar magnitude at voltages where they overlap. This property is typical of a simple first-order system (Scheme 1). In mammalian H_V1 (DeCoursey, 1991; Cherny et al., 1995, 2001; DeCoursey and Cherny, 1996, 1997; Cherny and DeCoursey, 1999; Schilling et al., 2002), activation tends to be slower than deactivation. This asymmetrical behavior is typical of cooperatively gated multimeric channels (Hodgkin and Huxley, 1952; Hille, 2001), because all subunits must activate before the channel conducts, whereas only one subunit needs to deactivate to close the pore. In rat H_V1, deactivation was rapid, pH_o independent, and weakly voltage dependent at large negative voltages (Cherny et al., 1995). However, near the threshold voltage for g_H activation, a second slower component of τ_tail appeared that was pH_o dependent and of comparable magnitude to τ_act. Also suggestive of a first-order system in HtH_V1 is that τ_act and τ_tail were slowest at the midpoint of the g_H-V relationship (Figs. 3 and 6). Finally, activation kinetics was well described by a single exponential and could not be fitted reasonably with a higher-order function.

There is general agreement that the H_V1 dimer in several species gates “cooperatively,” but it is less clear what this word means; in drug binding, cooperativity can be produced by quite different mechanisms (Colquhoun, 1973). One sense is that, like the Hodgkin–Huxley model, multiple subunits must move before the channel can conduct. Another sense is that, like oxygen binding to the four hemes of hemoglobin, the movement of one H_V1 protomer promotes the movement of the other. The sigmoid activation kinetics of H_V1 in several species (Gonzalez et al., 2010; Musset et al., 2010b; Tombola et al., 2010; Fujiwara et al., 2012) appears to reflect that both protomers must undergo a conformational change before either can conduct (first sense; Gonzalez et al., 2010) or highly cooperative gating (second sense; Tombola et al., 2010). When H_V1 is forced to exist as a monomer, by splicing it with the N terminus of Ciona intestinalis voltage-sensing phosphatase or by truncating the C terminus, the current turns on exponentially and five to seven times faster than with the WT dimeric protein (Koch et al., 2008; Musset et al., 2010a,b; Tombola et al., 2010; Fujiwara et al., 2012). The dimerization of H_V1 in several species appears strongly dependent on coiled-coil interactions in the C terminus (Koch et al., 2008; Lee et al., 2008; Tombola et al., 2008; Li et al., 2010). HtH_V1 has extensive predicted coiled-coil in its C-terminal region. This complicates the interpretation for HtH_V1, because the exponential activation and the apparently first-order kinetics suggest monomeric behavior. One explanation might be that HtH_V1 exists in the membrane as a dimer because of the coiled-coil region, but the coupling between C terminus and S4 segment is dysfunctional, as can be achieved experimentally by introducing a flexible linker between S4 and the C terminus (Fujiwara et al., 2012). However, this appears unlikely to be the case, because the apparent gating charge of HtH_V1 is nearly 6 e₀ (5.5 ± 0.9, mean ± SD in a sample of 18 g_H-V curves), based on the limiting slope of the g_H-V relationship. Monomeric H_V1 typically exhibit gating charge roughly half that of the dimer, 2–3 versus 4–6 e₀. When the coupling between the C terminus and the S4 helix was disrupted by a flexible linker, the gating charge was halved (Fujiwara et al., 2012). Given that HtH_V1 has charged amino acids in its transmembrane regions similar to those of other H_V1, we assume that its gating charge has analogous origins. The g_H-V relationships in Fig. 6 J also are compatible with Hodgkin–Huxley-type gating. One possibility is that a concerted rate-limiting step in opening occurs late, presumably after the conformational changes in each monomer (Gonzalez et al., 2010; Musset et al., 2010b; Villalba-Galea, 2014). The voltage-dependent movement of monomers may be so rapid in HtH_V1 that the concerted opening step becomes rate limiting. Another speculative explanation for its exponential activation is that HtH_V1 enjoys tighter coupling between protomers than H_V1 in other species; in essence, both S4 helices move together.

Perceptibly different from mammalian H_V1, the activation kinetics of snail H_V1, HtH_V1, is more steeply voltage dependent, giving a family of currents a distinctive gestalt (Fig. 3 A). In HtH_V1, τ_act decreased e-fold in 13.8 mV, in contrast to several mammalian H_V1, where τ_act changes e-fold in 40–72 mV (DeCoursey, 2003). In addition, τ_tail increased e-fold in 14.0 mV in HtH_V1, compared with a slope typically 26–44 mV/e-fold change in τ_tail in mammalian cells (DeCoursey, 2003). The steeply voltage-dependent gating kinetics of HtH_V1 is strikingly reminiscent of voltage-gated K⁺ channel behavior (Cahalan et al., 1985).

Byerly et al. (1984) noted that activation kinetics in snail LsH_V1 slowed at lower pH_o more than could be accounted for by the shift of the g_H-V relationship. This is clearly true of HtH_V1 as well. The τ_act-V relationship shifts positively with lower pH_o (Fig. 6 I), but its maximum increases by roughly an order of magnitude per unit decrease in pH_o. In stark contrast, in rat H_V1, the τ_act-V relationship mainly shifted along the voltage axis with changes in pH_o, with little change in kinetics. However, the τ_act-V relationship in rat was strongly affected by pH_i, slowing fivefold per unit increase in pH_i (DeCoursey and Cherny, 1995). In one study of human hH_V1, gating was described by three exponentials with activation generally faster at lower pH_i and deactivation faster at higher pH_i (Villalba-Galea, 2014). Qualitatively similar results were reported in mouse macrophages, but the largest change was a two- to threefold slowing of τ_act for a 1.5-unit increase in pH_i or a 2.1-unit decrease in pH_o (Kapus et al., 1993). An insect H_V1, NpH_V1, however, exhibited nearly as strong pH_o dependence of kinetics as found here in HtH_V1 (Chaves et al., 2016). In contrast to the strong dependence of activation kinetics on pH_i in rat (DeCoursey and Cherny, 1995), in HtH_V1, τ_act was in a similar range at all pH_i values from 5 to 8. These differences in gating kinetics among species may make it challenging to produce a single universal model that describes the voltage and pH dependence of gating in all species.

HtH_V1 was moderately sensitive to inhibition by Zn²⁺, the classic (Thomas and Meech, 1982; Mahaut-Smith, 1989a) and still most potent (Cherny and DeCoursey, 1999) H_V1 inhibitor. Somewhat weaker effects were observed for Cd²⁺ (Fig. 5). The principal effects of Zn²⁺ on mammalian H_V1 are a slowing of activation, a positive shift of the g_H-V relationship, and possibly a reduction of the maximum H⁺ conductance, g_H,max (Cherny and DeCoursey, 1999). These effects are also observed in HtH_V1, but the decrease in g_H,max is much more obvious, whereas the shift of the g_H-V relationship is substantially weaker in HtH_V1 than in hH_V1. Thus, the g_H-V relationship in human hH_V1 is shifted more by 1 µM Zn²⁺ (Musset et al., 2010b) than HtH_V1 is shifted by 10 µM Zn²⁺ (Fig. 5).

In mammalian H_V1, Zn²⁺ binds mainly to two His: His¹⁴⁰ and His¹⁹³ in hH_V1 (Ramsey et al., 2006; Musset et al., 2010b). Surprisingly, when mH_V1 was crystallized, it contained a Zn²⁺ atom, coordinated by the corresponding two His with contributions from two acids, Glu¹¹⁵ and Asp¹¹⁹, given in Table 1 (Takeshita et al., 2014). Mutation of both acids simultaneously decreases Zn²⁺ affinity of mH_V1, but neutralizing either alone does not (Takeshita et al., 2014). As indicated in Table 1, three of these four corresponding residues are conserved in HtH_V1: Glu¹¹⁴, Glu¹¹⁸ (conservatively replacing Asp), and His²⁰¹, with Val²⁵⁴ replacing the second His. Consistent with the partial conservation of the mammalian Zn²⁺ site, Zn²⁺ was generally less potent in HtH_V1 but still quite effective. The main difference in the presumed Zn²⁺ binding residues in HtH_V1 is the lack of His¹⁹³. One might therefore speculate that His¹⁹³ in human hH_V1 is important in Zn²⁺ shifting the g_H-V relationship positively. Evidently, when the binding site includes His¹⁹³ located in the external S2–S3 linker, Zn²⁺ binding biases the membrane potential more effectively. Coordination by four amino acids is more typical of a structural Zn²⁺ binding site, whereas catalytic Zn²⁺ binding sites usually have three amino acids and one water as a ligand (Auld, 2001). Of interest is a study showing that the metal transport site of ZnT transporters is selective for Zn²⁺ over Cd²⁺ when the four ligands are 2 His + 2 acids, but cannot discriminate the two metals with 1 His + 3 acids (Hoch et al., 2012). The HtH_V1 channel has 1 His + 2 acids and is moderately selective for Zn²⁺ over Cd²⁺. As shown in Table 1, the NpH_V1, CiH_V1, CpH_V1, SpH_V1, and DrH_V1 channels share a 1 His + 3 acids scheme and are much less sensitive to Zn²⁺ than mammalian H_V1 (Cd²⁺ was not tested) and generally less sensitive than HtH_V1. Intriguingly, the D145H mutation in NpH_V1 results in 2 His + 2 acids, which markedly increases its Zn²⁺ sensitivity (Chaves et al., 2018). Empirically, Table 1 indicates that the configurations of H_V1 for Zn²⁺ binding to H_V1 in order of decreasing efficacy are: 2 His + 2 acids > 1 His + 2 acids > 1 His + 3 acids. It appears that the 1 His + 3 acids motif is somewhat less favorable for Zn²⁺ binding than 1 His + 2 acids as found in HtH_V1, which seems paradoxical, because the 1 His + 3 acids motif has four ligands instead of three, possibly plus water. Perhaps geometrical factors can be more important than the number of ligands.

A unique property of H_V1 is that its voltage-dependent gating is strongly modulated by pH in a manner called ΔpH dependence (Cherny et al., 1995). The g_H-V relationship is shifted equally by increasing pH_o or decreasing pH_i, by −40 mV/unit change, thus responding to the pH gradient (ΔpH) rather than to the absolute pH (Cherny et al., 1995). The practical consequence is that H_V1 opens only when the electrochemical gradient for H⁺ is outward, such that when the channel opens it will always extrude acid from the cell (Doroshenko et al., 1986; DeCoursey and Cherny, 1994). To a rough approximation, all H_V1 appear to shift by 40 mV/unit at all pH values (DeCoursey, 2003). Until recently, the rare exceptions to this rule of forty were ignored as anomalies, perhaps reflecting difficulties of the measurements, in particular with control over pH (DeCoursey and Cherny, 1997). However, measurements explicitly addressing this point revealed that the ΔpH-dependent gating of hH_V1, kH_V1, and EhH_V1 does indeed deviate by saturating at high pH, namely above pH_o 8 or pH_i 8 (Cherny et al., 2015). Byerly et al. (1984) reported little shift between pH_o 7.4 and 8.4 in L. stagnalis, and this observation is consistent with the saturation at high pH_o observed here for HtH_V1 (Fig. 9). The slope in HtH_V1 begins to decrease above pH_o 7 (Fig. 9), suggesting that saturation begins at lower pH_o than in hH_V1. Saturation of ΔpH-dependent gating suggests that pH is approaching the effective pK_a of one or more titratable groups that sense pH_o. Given this interpretation, the effective pK_a is roughly 1 unit lower in HtH_V1 than in hH_V1.

Another deviation from the rule of forty is that the g_H-V relationship in HtH_V1 shifted ∼60 mV/unit change in pH_o between pH_o 5 and 7 (Fig. 9), well above the classic value of 40 mV/unit change in pH_o (Cherny et al., 1995). This unusual property is shared by several disparate species, including other snails. In H. pomatia, the shift was 63 mV from pH_o 7.5 to 6.6 (Doroshenko et al., 1986). In L. stagnalis, the shift was 46 mV from pH_o 7.4 to 6.4 (Byerly et al., 1984). Changes in pH_o in a coccolithophore EhH_V1 produced shifts of ∼50 mV/unit (Cherny et al., 2015). An insect H_V1 (NpH_V1) shifts 54 mV/unit change in pH_o (Chaves et al., 2016).

More dramatically, changes in pH_i produced much smaller shifts of the g_H-V relationship in HtH_V1 than the 40 mV in mammalian species (Cherny et al., 1995). The mean shift in HtH_V1 between pH_i 5 and 9 was only 15.3 mV/unit (Fig. 9). This is in remarkable agreement with the 15 mV/unit reported in the snail LsH_V1 between pH_i 5.9 and 7.3 (Byerly et al., 1984). Meech (2012) recently emphasized the stronger effects of pH_o over pH_i after reanalyzing old data. However, in another snail, H. pomatia, HpH_V1 apparently shifted normally, roughly 30–50 mV/unit change in pH_i (Doroshenko et al., 1986), so on this point it is not possible to generalize about molluscan H_V1.

The ΔpH dependence of mammalian H_V1 results in only outward H⁺ currents under most circumstances, which is crucial to many if not all of its functions (DeCoursey, 2003). One striking consequence of the anomalous ΔpH dependence of HtH_V1 and perhaps of other snail H_V1 is that inward currents are readily observed at certain ΔpH. Even at symmetrical pH, there are often inward currents. More conspicuously, because of the weak dependence on pH_i, an inward pH gradient (ΔpH < 0) produces inward currents over an extensive voltage range (e.g., Fig. 8 C). Inward currents would affect neuronal excitability by providing a depolarizing current. They at first appear incompatible with an early proposal that proton currents in snail neurons function to extrude protons that enter via Ca²⁺/H⁺ exchange after each Ca²⁺-mediated action potential (Ahmed and Connor, 1980; Thomas and Meech, 1982; Byerly et al., 1984), but under normal conditions of an outward H⁺ gradient, inward currents would likely not be activated. Nevertheless, the possibility arises that H_V1 might mediate action potentials in molluscan neurons under certain conditions, although Ca²⁺ channels are thought to be primarily responsible (Hagiwara and Byerly, 1981). H_V1 appears to mediate action potentials in bioluminescent dinoflagellates (Fogel and Hastings, 1972; Smith et al., 2011; Rodriguez et al., 2017).

As L. stagnalis and H. trivolvis snails live in similar habitats, we expect that their proton channels should function similarly. This view is supported by the similarity of the LsH_V1 sequence to that of HtH_V1, especially in the S2/S3 region that we have identified for its importance in pH_i sensing in the accompanying paper (Cherny et al., 2018). In that paper, we identify a single amino acid difference between hH_V1 and HtH_V1 that appears largely responsible for the difference in pH_i sensing.

We thank Kristie Bishop (Albany State University) for technical assistance with Western blotting experiments. We gratefully acknowledge heroic efforts by David Colquhoun to disambiguate our fuzzy thoughts on cooperativity.

This work was supported by the National Institutes of Health (grants GM121462 to T.E. DeCoursey and GM102336 to T.E. DeCoursey and S.M.E. Smith) and the National Science Foundation (grant MCB-1242985 to T.E. DeCoursey and S.M.E. Smith and Neuroscience Cluster award 0843173 to V. Rehder). Ms. Bishop was a participant in the Kennesaw State Research Experience for Undergraduates, Chemistry and Biochemistry Summer Undergraduate Research Experience (National Science Foundation CHE-1560329).

The authors declare no competing financial interests.

Author contributions: Conceptualization: S.M.E. Smith and T.E. DeCoursey; data curation: S.M.E. Smith and T.E. DeCoursey; formal analysis: V.V. Cherny, D. Morgan, L.R. Artinian, V. Rehder, and T.E. DeCoursey; funding acquisition: V. Rehder, S.M.E. Smith, and T.E. DeCoursey; investigation: S. Thomas, V.V. Cherny, and D. Morgan; project administration: S.M.E. Smith and T.E. DeCoursey; resources: S. Thomas, L.R. Artinian, V. Rehder, and S.M.E. Smith; visualization: S.M.E. Smith and T.E. DeCoursey; writing (original draft): T.E. DeCoursey; writing (review and editing): V.V. Cherny and S.M.E. Smith.

Richard W. Aldrich served as editor.