Voltage-gated Proton Channels

Abstract

Voltage-gated proton channels, H_V1, have vaulted from the realm of the esoteric into the forefront of a central question facing ion channel biophysicists, namely the mechanism by which voltage-dependent gating occurs. This transformation is the result of several factors. Identification of the gene in 2006 revealed that proton channels are homologues of the voltage-sensing domain of most other voltage-gated ion channels. Unique, or at least eccentric, properties of proton channels include dimeric architecture with dual conduction pathways, perfect proton selectivity, a single-channel conductance ~10³ smaller than most ion channels, voltage-dependent gating that is strongly modulated by the pH gradient, ΔpH, and potent inhibition by Zn²⁺ (in many species) but an absence of other potent inhibitors. The recent identification of H_V1 in three unicellular marine plankton species has dramatically expanded the phylogenetic family tree. Interest in proton channels in their own right has increased as important physiological roles have been identified in many cells. Proton channels trigger the bioluminescent flash of dinoflagellates, facilitate calcification by coccolithophores, regulate pH-dependent processes in eggs and sperm during fertilization, secrete acid to control the pH of airway fluids, facilitate histamine secretion by basophils, and play a signaling role in facilitating B-cell receptor mediated responses in B lymphocytes. The most elaborate and best-established functions occur in phagocytes, where proton channels optimize the activity of NADPH oxidase, an important producer of reactive oxygen species. Proton efflux mediated by H_V1 balances the charge translocated across the membrane by electrons through NADPH oxidase, minimizes changes in cytoplasmic and phagosomal pH, limits osmotic swelling of the phagosome, and provides substrate H⁺ for the production of H₂O₂ and HOCl, reactive oxygen species crucial to killing pathogens.

Introduction

Proton pathways through proteins are a stalwart element in numerous bioenergetic molecules, in which proton and electron movements are elaborately choreographed (27, 177, 269, 272). In these molecules, protons generally cross partway through the membrane, perform some function or await a catalytic step, and then return or continue across the membrane. In contrast, the voltage-gated proton channel behaves like a simple, passive conduction pathway that spans the entire membrane. Like all biological channels, proton channels are gated; that is, the channel opens to conduct protons when required, and then closes. They are called “voltage-gated” because membrane potential regulates the probability that the channel will be open, with depolarization promoting opening.

Voltage-gated proton channels are attractive subjects of research for three broad reasons. First, there is a rapidly expanding list of important physiological roles played by voltage-gated proton channels in a variety of cells and species. Second, by virtue of their homology to the voltage-sensing domains of other voltage-gated ion channels, they provide a unique device for evaluating gating mechanisms. Third, they violate so many rules adhered to by other ion channels that they require a new vision and a whole new set of questions. Their characterization as ion channels has occasionally been questioned on several grounds, but by most criteria, they are unquestionably ion channels. This review will summarize current knowledge in these three areas, with an emphasis on their physiological roles.

History

The first explicit proposal of a voltage-gated proton channel by Woody Hastings and colleagues in 1972 (95), postulated a role for the proton channel in triggering a bioluminescent light flash in a dinoflagellate, Gonyaulax polyedra, when the cell is mechanically disturbed (Fig. 1). Further information on proton channels in dinoflagellates is presented below (Bioluminescence and other functions in dinoflagellates).

Figure 1.

Voltage-gated proton channels trigger the bioluminescent light flash in dinoflagellates. The proton concentration is high in the flotation vacuole (below the membrane in the diagram). An action potential depolarizes the membrane, opening proton channels, allowing protons to flow down their electrochemical gradient into the interior of the scintillon (upper compartment). The sudden drop in pH in the scintillon triggers the flash by two concerted mechanisms. Upon protonation, the light-emitting pigment, luciferin (LH₂) is released from luciferin binding protein (LBP) making it available as a substrate for luciferase, which is itself activated by acidification. [From (109). Originally published in Bioluminescence in Action, Ed. by P. Herring, Academic Press, London. pp. 129-170, 1978. Copyright Elsevier.]

Proton channels were identified as bona fide ion channels in 1982 by Roger Thomas and Bob Meech (260) who performed voltage-clamp studies of Helix aspersa snail neurons that had been impaled by a battery of microelectrodes. Protons in the form of HCl were injected into the cell bodies and outward membrane current carried by protons could be detected when the membrane potential was clamped to positive voltages. During sustained depolarization, the outward current declined as the intracellular pH (pH_i) increased. Despite the small single-channel conductance [See: Minuscule Single Channel Conductance], proton channels were sufficiently abundant in the membrane that their activity restored pH_i on a time scale of tens of minutes. These neurons were 100–200 μm in diameter (56); in small cells like leukocytes (6–10 μm diameter) proton current can restore pH_i after an acid load in tens of seconds (65, 76, 134), because of the more favorable surface/volume ratio. A thorough voltage-clamp characterization carried out by Byerly, Meech and Moody demonstrated a critical shift of the position of the g_H-V relationship along the voltage axis when either pH_o or pH_i was changed (29). Also at UCLA, Mike Barish and Christiane Baud identified and characterized similar proton currents in Ambystoma salamander oocytes (19). Substantial effort was expended in these early studies to establish convincingly that proton channels were genuine ion channels and that protons did not simply permeate other channels adventitiously (29, 170). Polyvalent cations were found to inhibit proton currents (260), and Martyn Mahaut-Smith (166) showed that Zn²⁺ inhibited proton currents 80 times more potently than Ca²⁺ currents in Helix neurons. Byerly and Suen (30) studied current fluctuations in membrane patches from Lymnaea stagnalis, and concluded that the single-channel proton conductance was extremely small (<50 fS). Based on measurements of pH changes and membrane potential, Henderson, Chappell, and Jones deduced the probable existence of proton channels in human neutrophils, where they were proposed to compensate for the electrogenic activity of NADPH oxidase (114–116).

Nearly a decade after the first voltage-clamp study, proton currents were discovered in mammalian cells, rat alveolar epithelial cells (54). These cells exhibit high levels of proton current, enabling a series of studies in which the main biophysical properties of proton channels were explored. The H⁺ current was not much larger with 100 mM buffer than with 10 mM or even 1 mM buffer, suggesting that the rate-determining step in permeation was not proton transfer from buffer to the channel, but occurred within the conduction pathway (61). Buffer concentration on the intracellular side of the membrane was, reasonably enough, shown to be important, with 100 mM buffer providing markedly better control of pH_i than 10 mM (61). Several studies explored the phenomenon of proton depletion as an inevitable consequence of the passage of large proton currents across cell membranes. It became clear that the proton channel does not inactivate, but that when proton depletion is pronounced, the currents decay with time as pH_i increases (54, 76, 105, 134, 184, 186). In the classical sense, “inactivation” means that a channel enters an altered conformational state from which it cannot open until recovery to a resting state has occurred (121). That proton current decay reflects changes in pH, and consequently in both the driving force and the pH-dependent gating of the channel is clear from studies showing that– whether assessed by pH electrode, pH sensing fluorescent dyes, or by measurement of the proton current reversal potential – pH_i changes roughly in proportion to the integral of the H⁺ flux during pulses (54, 65, 76, 105, 134, 144, 170, 186, 260). Separating the effects of proton depletion from “genuine” underlying proton current kinetics continues to challenge researchers in this field. However, because no proton channel has yet been shown to inactivate, it remains true that visible decay of proton currents constitutes prima facie evidence that significant pH changes occurred during the pulse.

In 1993, proton currents were recorded in three human cell types (24, 76), including human neutrophils (63), confirming the hypothesis of Henderson and colleagues. The number of cell types reported to express proton channels continued to expand. A breakthrough in physiological function was the discovery that in phagocytes with active NADPH oxidase, the properties of the proton conductance differed markedly from those in resting cells (16). The change was so profound [See: Enhanced gating mode] that it was at first thought to reflect the appearance of a novel type of proton channel, which was hypothesized to comprise the gp91^phox component of NADPH oxidase. First proposed explicitly in 1995 (112, 117, 118), this hypothesis gained momentum as additional members of the Nox (NADPH Oxidase) family of proteins were discovered, and added to the list of putative dual function molecules that could alternatively transport electrons and protons (14, 15, 169). However, evidence against this hypothesis accumulated as well, as neutrophils from chronic granulomatous disease patients who lacked functional gp91^phox protein were shown to have normal proton currents in both resting (204) and stimulated states (69). The phorbol ester activator of PKC, PMA, was shown to increase proton currents identically in PLB-985 cells with or without gp91^phox (69). COS-7 cells that lacked endogenous proton currents still lacked detectable proton current when gp91^phox was transfected into them together with other components to produce a functional NADPH oxidase complex (181). The coup de grâce for the hypothesis was the demonstration that neutrophils from proton channel (HVCN1) knockout mice exhibited no detectable proton currents, yet had normal electron currents when NADPH oxidase was activated, unambiguously demonstrating the presence of functioning gp91^phox in their membranes (88, 178). Ironically, in their first landmark paper that proposed the existence of voltage-gated proton channels in human neutrophils (115), Henderson and colleagues made six extremely accurate predictions that remain true today, “H⁺ ions … movement is via a channel in the membrane, the opening of which lags behind activation of the [NADPH] oxidase. The mechanism for initiating the opening of this channel is unknown at present, but it could be a voltage decrease [i.e., depolarization], an increase in the internal concentration of protons or a phosphorylation/conformational change.” If this had been the last word on the subject, a decade of controversy could have been avoided!

The discovery of proton channel genes in 2006 (224, 234) inevitably fomented research into several areas previously not possible, namely structure-function studies, as well as knock-out and knock-down studies, and the development of antibodies. Completely unexpected was that the proton channel molecule would so closely resemble the voltage sensing domain of most voltage-gated ion channels. Discovery of this feature immediately transformed the proton channel molecule into a new playground for the study of voltage-gating mechanisms. In the resulting flurry of activity, discoveries have been made at an astonishing pace.

Properties

Gene and molecular features

The Proton Channel Protein, H_V1

Identification of proton channel genes in human (224), mouse, and Ciona intestinalis (sea squirt) (234) in 2006 revealed surprising properties, the most remarkable of which is the similarity of the H_V1 protein to the voltage sensing domain (VSD) of most voltage-gated ion channels (Fig. 2). The first VSD to be identified that was not part of a voltage-gated ion channel was a voltage-sensing phosphatase (VSP) (191). The VSP senses membrane potential and exhibits “gating currents” like those associated with voltage-gated ion channels, but the response to depolarization is modulation (an increase) of phosphatase activity. The subsequent identification of the voltage-gated proton channel gene added impetus to the idea that the VSD might be viewed as a modular unit that could be attached to another molecule and provide voltage sensitivity to its function (143, 173, 176, 207). The Swartz group have further shown that the “paddle motif” (the S3b and S4 TM helices) is indeed modular, and continues to function when swapped among K⁺ channels, H_V1, and VSP (7).

Figure 2.

Membrane topology of voltage-gated K⁺ channels (left), voltage-gated proton channels (center), and voltage-sensitive phosphatases (right). The K⁺ channel (lower panel) is a tetramer of the 6 transmembrane domain monomer shown in the top panel; the proton channel is a homodimer of the 4 transmembrane domain monomer in the top panel, and the phosphatase is a monomer. The assembled K⁺ channel has a single conduction pathway, but the proton channel has two, one in each protomer. The VSP senses voltage, but does not conduct. The proton channel cartoon illustrates schematically that the dimer is held together by C-terminal coiled-coil interactions, and that the channel can be phosphorylated at Thr²⁹ in the intracellular N terminus to produce the enhanced gating mode (cf. Fig. 20). [Figure from (57). Originally published in Physiology 25:27-40, 2010; doi: 10.1152/physiol.00039.2009.]

Perhaps the most distinctive feature of the proton channel is the absence of S5–S6 domains that form the conduction pathway in ordinary ion channels. Because conduction can be demonstrated when purified H_V1 protein is reconstituted in synthetic liposomes, no accessory protein is required (151). The conduction pathway must exist within the H_V1 molecule itself. Homology models have been proposed, based on the similarity of the sequence to that of the VSD of K⁺ channels (199, 223, 270). The channel resembles an hourglass filled with water, with a constriction near the center. Proton selectivity is established by an essential aspartate residue in the S1 domain, Asp¹¹² in hH_V1 (200) and Asp⁵¹ in kH_V1 (251), which is thought to be located at or near the constriction when the channel is in the open state.

Three new proton channel genes were discovered in 2010, all in unicellular marine organisms; two coccolithophores, Emiliania huxleyi and Coccolithus pelagicus (259) and a dinoflagellate, Karlodinium veneficum (251). These channels greatly expand the phylogenetic representation of H_V1 (Fig. 3). One putative plant H_V1 is shown. Only seven of the 37 putative H_V1 genes have been confirmed by heterologous expression and electrophysiological recording; in addition to the six just discussed, Strongylocentrotus purpuratus was confirmed by Y. Okamura (personal communication). The confirmation that H_V1s in several unicellular species are genuine proton channels is an important validation of the criteria used to identify putative proton channels from sequence analysis alone. We proposed recently that a protein with four TM domains that includes (a) an appropriately positioned Asp residue in S1 to act as the selectivity filter, and (b) 2 or preferably 3 appropriately spaced Arg residues in S4, that include the WRxxR motif, can be considered highly likely to be a voltage gated proton channel (251).

Figure 3.

A phylogenetic analysis showing evolutionary relationships among 37 putative H_V1 VSD sequences. This maximum likelihood phylogenetic tree with 100 boostraps was constructed from a multiple sequence alignment of the VSD portion of 37 H_V1s; N- and C-termini as well as S5-S6 domains were truncated. Branch lengths are displayed to scale, and are proportional to the evolutionary distance between sequences. Bootstrap values >60 are shown. [Figure from (251). Originally published in Proc. Natl. Acad. Sci. USA. 000:0000-0000, 2011; doi ??????]

The phylogenetic tree in Fig. 3 is drawn to scale, so that the radial length of each branch reflects the extent of sequence differences. It is tempting to overinterpret this phylogram. Mammalian H_V1s are very similar to each other, but as one ventures toward more primordial species, variation increases. Invertebrate H_V1s exhibit a high degree of individuality, and H_V1s in unicellular organisms are riotously diverse. The most deviant among this group is kH_V1, a dinoflagellate proton channel. One might predict that in general, sequence diversity may presage functional diversity. As we will discuss below [See: Bioluminescence and other functions in dinoflagellates], the Karlodinium proton channel violates the hitherto sacred rule of conducting exclusively outward current (251), and thus its unique behavior matches its idiosyncratic sequence.

Nomenclature

The discoverers of the first proton channel genes adopted different systems for nomenclature. Ramsey and colleagues (224) christened the human channel H_V1: H for H⁺ selective, V for voltage-gated, and 1 for its being the first example identified in humans. Sasaki and colleagues adopted more evocative abbreviations, mVSOP and CiVSOP: m for mouse, Ci for Ciona intestinalis, VSOP for voltage-sensor-only protein (234), thus incorporating the most surprising feature of the gene product into its designation. The official gene name for all species is HVCN1. Here I will adopt the more austere convention and refer to the protein as H_V1, which will be qualified by a prefix that indicates species (hH_V1 for human, mH_V1 for mouse, etc.). Although this appears to limit the spectrum to 26 species (and then only if they have different initials!), the list can be expanded by including genus and species, e.g., CiH_V1, EhH_V1, or CpH_V1. To date, only one proton channel gene has been identified in any species, so for now, the suffix is 1 for all species.

The Proton Channel is a Dimer, Exhibiting Cooperative Gating

Several types of evidence led to the conclusion that the proton channel molecule exists mainly as a dimer, both in heterologous expression systems (139, 150, 261) and in native cells (217), with a conduction pathway in each protomer (139, 261). It is puzzling why the channel goes to the trouble of assembling as a dimer, in light of evidence that the monomer exhibits the main features of native proton current: activation upon depolarization, extreme proton selectivity, and exquisite sensitivity of gating to both pH_o and pH_i (103, 139, 199, 261). A precedent exists for a dimeric channel with conduction pathways in each monomer, in the ClC family of Cl⁻ channels and Cl⁻/H⁺ antiporters (1, 108, 159, 163, 171, 172, 174). Like ClC-0 (108), the two protomers comprising the proton channel dimer appear to interact during gating (103, 198, 199, 262). This interaction has been interpreted either as positive cooperativity in which the opening of a single protomer greatly increases the probability of the second protomer opening (262), or a strict requirement that both protomers undergo a conformational change before either can conduct (103, 198, 199). Coiled coil interactions at the C terminus appear to provide the main glue that holds the dimer together (103, 139, 151, 156, 198, 199, 261). Truncating (or replacing) the C terminus results in expression of monomers (103, 139, 198, 199, 261) that open 5–6 times more rapidly than the wild-type (WT) dimer (139, 198, 261). As illustrated in Fig. 4, the proton conductance of the dimer activates with a sigmoid time course, consistent with classical Hodgkin-Huxley n² kinetics (122), whereas the monomer activates exponentially (103, 199). Fig 4C shows that during depolarizing pulses a fluorescent probe attached to the S4 domain moves with an exponential time course, whereas the current turns on after a small delay and with a sigmoid time course. Reflecting the cooperativity of gating, the limiting slope of the g_H-V relationship of the dimer is twice as steep as that of the monomer (103). These properties of dimeric gating suggest an explanation for the utility of proton channels existing as dimers, at least in phagocytes (199). As will be seen [See: Charge compensation], proton channels open in phagocytes to prevent excessive depolarization caused by the electrogenic activity of NADPH oxidase. Consequently, it is critically important that proton channels activate with a steep voltage dependence, to prevent self-inhibition of NADPH oxidase (73). If steep voltage dependence occurs at the expense of slower opening, little is lost. Depending on the type of stimulus, the time course of the respiratory burst is relatively slow, usually preceded by a delay of seconds to minutes, and persisting for minutes to hours (72).

Figure 4.

Comparison of gating kinetics of the human proton channel dimer (A) and monomer (B), expressed in HEK-293 cells. Note the slower, sigmoid activation of the WT channel (A), and the faster, exponential turn-on of the monomer (C terminus truncated, H_V1ΔC) both during pulses to +50 mV at 23°C at pH_o 7.5, pH_i 7.5. [From (199). Originally published in The Journal of Physiology. doi: 10.1113/jphysiol.2010.188318.] C. The time course of movement of a fluorescent probe attached to the S4 domain of the Ciona proton channel is exponential (red). This time course raised to the second power (green) matches the proton current recorded at the same time (black). [From (103). Adapted by permission from Macmillan Publishers Ltd: Nature Structural & Molecular Biology 17: 51-56, copyright 2010; doi: 10.1038/nsmb.1739.]

The dimeric arrangement may not dictate an invariant interface. Two possible interfaces have been proposed, shown in Fig. 5. On the basis of cross-linking studies for mutants in which Cys was introduced at various locations in the hH_V1 molecule, several interactions at the external end of S1 (shown in red in Fig. 5A) were detected (150). However, based on Zn²⁺ effects on hH_V1 mutants in which the Zn²⁺-binding residues His¹⁴⁰ and His¹⁹³ were mutated, the interface shown in Fig. 5B was proposed to occur at least occasionally (199). In the Cys scanning study, weak cross-links were observed at position 194, right next to His¹⁹³, suggesting that this orientation can occur at least occasionally (150). If both kinds of interfaces are possible, then multimers could also occur (Fig. 5C). Given substantial evidence that the protomers interact during gating, one imagines that channel opening might occur only from a particular orientation. Because Zn²⁺ appears to prevent channel opening (36), it is reasonable to suppose that Zn²⁺ - which is thought to bind at the interface between protomers (199) - might act by preventing the dimer from adopting its preferred orientation.

Figure 5.

Possible dimer interfaces, with the transmembrane domains color coded as: S1 red, S2 yellow, S3 green, S4 blue. The dimer in A was proposed on the basis of cysteine cross-linking studies (150); the dimer in B was proposed to explain Zn²⁺ binding studies (199). External His residues that bind Zn²⁺ (224) are shown in aqua (cf. Fig. 14). [From (198). Originally published in Channels (Austin) 4: 260-265, 2010; doi: 10.4161/chan.4.4.12789.]

Although the dimer is normally stabilized by C terminal coiled-coil interactions (150, 156, 261), a tandem dimer in which the C terminus of one protomer was connected by a short linker to the N terminus of the second protomer functioned almost identically to the WT dimer (198, 199). The tandem dimer activated with a sigmoid time course (198), consistent with similar cooperative interactions between protomers observed in WT dimers (103). Its g_H-V relationship was also indistinguishable from WT, and the only difference detected was slower activation kinetics (198). Evidently, C terminal coiled-coil interactions tether the two protomers together, but appear not to contribute critically to the interactions between protomers that occur during channel gating (198). However, a role of the C terminus in modulating gating by direct mechanical linkage has been suggested (99).

In contrast to the structural role played by the C terminus, the N terminus appears to interact closely with the gating mechanism of the channel. The enhanced gating mode [See: Enhanced gating mode] is mainly the result of phosphorylation of Thr²⁹ in the N terminus of hH_V1 (194). In addition, a mutation that results in M91T substitution (Met replaces Thr at position 91) in the N terminus has been identified in the human population (128) [See: Acid Secretion in Epithelia of the Respiratory Tract]. This mutation shifts the g_H-V relationship positively by 20 mV, reducing the likelihood of channel opening. It remains unclear how these two residues scattered among the 100-residue N terminus of hH_V1 are able to modulate gating, which as discussed above presumably involves movement of the S4 domain relative to the other trans-membrane domains.

Selectivity and permeation by protons

Perfect selectivity

The voltage-gated proton channel appears to be perfectly selective, or specific, for protons. At present, no evidence even suggests that any other ion can permeate. Numerous studies in many cell types have demonstrated that when the dominant cation or anion in the bathing solution is changed, the measured reversal potential (V_rev) does not change, after liquid junction potentials are corrected (56, 58, 200). Any small deviation of V_rev from the Nernst potential for protons (E_H), is therefore attributable to imperfect control of pH. Even if the entire discrepancy between V_rev and E_H were due to permeation of other ions through the channel, the relative permeability for protons over other ions is >10⁶ when calculated with the Goldman-Hodgkin-Katz equation (102, 119, 123). How can such a high degree of proton selectivity be achieved?

Proton conduction through proteins is widespread, appears to occur fairly readily, and is presumed to be possible whenever water molecules are identified inside proteins within hydrogen bonding distance of other waters or titratable groups. Many membrane proteins have been shown to be capable of conducting protons, several of which were discussed previously in this context (56). However, despite the apparent facility of proton conduction, proton selective conduction is more demanding, and is discussed below. It is arguably more difficult to prevent protons from carrying current through an aqueous pathway, than to allow proton permeation. This challenge is faced by aquaporins, whose job is to facilitate rapid water permeation (4), but to prevent permeation by ions, especially protons. Precisely how protons are excluded remains debatable (28, 33, 51, 127, 175, 190, 256). Intriguingly, mutation of just three amino acids turns AQP1 into a cation conducting channel with a strong preference for protons (273).

Protons are chemically quite different from other cations. In aqueous solutions, protons exist as H₃O⁺ at least 99% of the time (47). By hopping from one water molecule to the next by the Grotthuss mechanism (3, 49, 52, 53), proton conduction through water occurs 5 times more rapidly than other cations the same size as H₃O⁺ (23, 86). Proton conduction through proteins occurs by a hydrogen-bonded chain (HBC) mechanism (202) in which protons hop from one molecule or group to the next (Fig. 6). The HBC may comprise titratable amino acid side chains or water molecules (202). Because protons are ubiquitous (the total concentration of hydrogen in water is 110 M) and interchangeable, multiple protons contribute to the net flux of a single proton across the membrane. Other ions diffuse around water and other molecules; protons use water and other molecules as stepping stones. The HBC mechanism potentially enables proton selective conduction, by virtue of the ability of protons to translocate along the chain without displacing the elements of the chain. Thus, protons are conducted through the gramicidin channel, which contains a linear row of a dozen water molecules (154), two orders of magnitude faster than any other ion (201), because other ions must wait for the waters to diffuse through the pore. The protons hop across, leaving the water wire intact.

Figure 6.

Hydrogen-bonded chain (HBC) conduction. In this schematic example, hydroxyl groups (e.g., from Ser residues) form a HBC that spans the membrane. (A) Proton conduction occurs when a proton enters the chain from the left to form a positive ion, which then moves to the right by hopping of successive protons to effect a reversal of the direction of the hydrogen bonds between each pair of oxygens. Proton conduction would also occur by loss of a proton on the right followed by movement of a negative ion (or fault) to the left. (B) To complete the process, reorientation of each hydrogen bond in the chain must occur, so that another proton can enter from the same side. [Redrawn with permission, from (202).]

Compared with proton conduction through the gramicidin channel, which is viewed as an archetypal water-filled pore, conduction through voltage-gated proton channels seems distinctly more challenging. The mobility of protons within gramicidin is roughly comparable to that in bulk water scaled to channel-like dimensions (48). In contrast, proton permeation through voltage-gated proton channels has a much higher deuterium isotope effect: g_H/g_D = 1.9 for proton channels (60) vs. 1.35 for gramicidin (6, 42). In addition, the temperature dependence for proton permeation through voltage-gated proton channels is 12–27 kcal/mol (64, 144) vs. 3–8 kcal/mol for proton permeation through gramicidin (6, 41). The strong isotope effect and temperature sensitivity can be explained if protons permeate via a HBC comprising titratable amino acid side chains in addition to water. Given these properties in addition to the extreme selectivity of the voltage-gated proton channel, and the indication from buffer effects that the rate-limiting step in H⁺ permeation occurred in the pore (61), the general opinion was that the permeation pathway through voltage-gated proton channels likely contained at least one titratable group (56, 60–62, 64, 65, 77, 117, 169). Recently, this hypothesis was tested by mutating every likely HBC element in the human H_V1 proton channel (223). No single residue was found to be indispensable for conduction. By default, the mechanism of proton selectivity was suggested instead to involve non-diffusible waters “frozen” at a narrow constriction in the channel. These waters could transfer H⁺ but block permeation of ordinary ions. Even more recently, Asp¹¹² located in the S1 TM domain was identified as the proton selectivity filter of the human voltage-gated proton channel (200). Neutralizing mutations not only eliminated proton specific conduction, they resulted in permeability to anions. The conservative replacement of Asp with Glu did not alter proton selectivity, but the D112H mutant was anion permeable. The corresponding residue in a dinoflagellate proton channel, Asp⁵¹, also in the S1 domain, was found to exhibit identical phenomenology; neutral substituents produced a strongly Cl⁻ permeable channel, Glu acted like Asp, and the His mutant conducted Cl⁻ (251). Evidently, when the Asp at the selectivity filter is neutralized, the rest of the channel molecule excludes cations. How proton specific conduction is enforced is not yet understood in atomic detail, but an acid at this key location in the first transmembrane domain (S1) is essential.

A similar problem has been confronted by the M₂ proton channel of influenza A virus. This channel shares many biophysical properties with the voltage-gated proton channel, including high selectivity (43, 158, 189), small conductance (158, 189), and a large deuterium isotope effect (189). However, surprising new studies indicate that the M₂ channel is not exclusively proton selective, but that Na⁺ and K⁺ can occasionally permeate in the opposite direction, compensating electrically for H⁺ influx (152, 215). If M₂ is less selective than the voltage-gated proton channel, then its direct relevance as a model for the selectivity mechanism of the voltage-gated proton channel may be questioned, but it remains heuristically useful. The M₂ channel is a homotetramer with a ring of four His³⁷ residues that comprise an acid-activated gate. One conceptual model postulates that protonation of these His³⁷, through electrostatic repulsion, expands the channel allowing waters to transfer protons down the center of the pore while still preventing other cations from permeating by a frozen water mechanism (136, 142, 233, 252). Alternatively, His³⁷ might be protonated and deprotonated with each conduction event, shuttling protons through the pore (124, 137, 148, 218, 241, 245, 248). Recently, a hybrid mechanism was suggested, in which the permeating proton is delocalized among the four His³⁷ and associated water molecules (2). Whether a proton delocalization mechanism might also be applicable to the voltage-gated proton channel remains to be determined. The idea seems seductively vague, but it may actually reflect reality. Surprisingly, mutation of Asp¹¹² in hH_V1 to His (D112H) failed to produce proton selective conductance; instead, the D112H mutant was anion permeable (200). Precisely the same result was seen in the D51H mutant of the dinoflagellate proton channel kH_V1 (251). Evidently, His can shuttle protons, but even at a constriction, does not guarantee proton specificity.

Minuscule Single Channel Conductance

One of the most exciting possibilities enabled by the discovery of the patch-clamp technique was the direct measurement of current flowing through a single molecule, an open ion channel (106). Most ion channels have a conductance of 2–400 pS (119), and conduct currents of a few picoamperes, that are resolvable by direct recording from a patch of membrane, or even from whole cells that are electrically tight (such as human lymphocytes (59) or eosinophils (90)). However, the voltage-gated proton channel has a conductance ~3 orders of magnitude smaller than most other channels, and for many years single channel currents were thought to be too small to resolve by direct measurement. Due to the difficulty in resolving single-channel currents, a preferred approach is to evaluate current fluctuations, which provides similar information, as illustrated in Fig. 7. At subthreshold voltages, the current exhibits very little noise. Above threshold, the random gating of proton channels introduces noise into the current, from which the single-channel current amplitude, i_H, can be derived: i_H = σ²/[I_H × (1−P_open)], where σ² is the variance and P_open is the open probability (119). These parameters are all directly measurable, excepting P_open, which can be accounted for by selecting conditions that minimize its effect; by recording just above V_threshold where P_open ≈ 0 (63), or by analyzing data from multiple voltages simultaneously (39, 244). Determined in this way, the unitary conductance increased as pH_i decreased, from 38 fS at pH_i 6.5 to 140 fS at pH_i 5.5 at room temperature (39). The unitary conductance was independent of pH_o, as might be expected for exclusively outward currents. Incidentally, this behavior is consistent with H⁺ being the conducted species, rather than OH⁻, which would be expected to have the opposite dependence on pH changes. In some patches, Vladimir Cherny and colleagues observed step-like events that resembled single-channel currents, with amplitudes 7–16 fA fluctuations (39). These were just on the boundary of being resolvable, and required seal resistances in the TΩ (10¹⁵ Ω) range; because at the low frequencies of proton channel gating, the resolution is limited by extraneous noise contributed by the seal resistance (153). The amplitude of these apparent unitary currents was about double that calculated in the same patches from fluctuations (39). This result is intriguing in light of the proton channel existing as a dimer (139, 150, 261), whose two subunits each have separate conduction pathways and function with strong cooperativity (103, 198, 199, 262).

Figure 7.

Determination of the single channel conductance from proton current fluctuations that reflect the stochastic opening and closing of proton channels. A shows families of proton currents in an excised, inside-out patch from a human eosinophil at three pH_i values. At subthreshold voltages, the current is quiet, but just above V_threshold, proton channels begin to open, and the current becomes distinctly noisy. It is noteworthy that (at low pH_i) the noise first increases with depolarization, but then decreases for large depolarizations. B shows g_H-V relationships from this patch. The variance of the current fluctuation, measured at quasi-steady-state, is plotted in C. The variance increases more than 100-fold at voltages where the proton conductance is active, and is maximal near the midpoint of the g_H-V relationship at each pH_i (indicated as V_1/2). D shows that the expected variance given the simplest possible two-state model of gating (closed ↔ open) coincides with the observed voltage dependence. The non-monotonic increase in variance with depolarization is consistent with the maximum P_open limiting to ~0.95 at pH_i 5.5. [From (39) ©Cherny et al., 2003. Originally published in The Journal of General Physiology. doi: 10.1085/jgp.200308813.]

Once the unitary conductance is known, it is possible to calculate the open probability of the channel, P_open. The limiting P_open for large depolarizations, P_max, was found to be 0.75 at pH_i 6.5, increasing to 0.95 at pH_i 5.5 or lower (39). Thus, lowering pH_i increases proton currents both by increasing the unitary conductance and by increasing P_max.

Regulation of Gating by the pH Gradient, ΔpH

A feature displayed universally by voltage-gated proton channels is that the voltage dependence of their gating shifts drastically when pH is changed on either side of the membrane. Lowering pH_o shifts the g_H-V relationship positively and tends to slow channel opening (larger activation time constant, τ_act) (19, 29). Byerly and colleagues (29) observed that the slowing of activation at lower pH_o was greater than could be accounted for by the size of the g_H-V relationship shift (i.e., on the assumption that all gating parameters shifted equally), and concluded that external protons directly inhibited channel opening in snail neurons. Conversely, lowering pH_i shifts the g_H-V relationship negatively (29, 167). Fig. 8 illustrates the effects of changes in pH_i on proton currents in inside-out patches of membrane from rat alveolar epithelial cells (67). Qualitatively, the effects are like those described in snail neurons, but there are subtle differences. Effects of pH_o on τ_act appear to be accounted for by the 40 mV/unit pH shift of the g_H-V relationship, in contrast with snail. However, lowering pH_i in rat increased the current amplitude and greatly accelerated activation. In rat epithelium, τ_act decreased 5-fold per unit decrease in pH_i, which cannot be ascribed to a voltage shift. These minor differences probably reflect species differences, but some subtle discrepancies seem to exist between properties observed in excised membrane patches compared with those in whole-cell configuration. Nevertheless, the qualitative effects of pH_o and pH_i are quite consistent among all species studied to date. The physiological implication of this intricate regulation by pH is that the proton channel normally conducts only outward current, and consequently, its primary function must be acid extrusion from cells.

Figure 8.

A–D Effects of pH_i on voltage-gated proton currents in an inside-out membrane patch from a rat alveolar epithelial cell. The pipette pH (i.e., pH_o) was 7.5. A shows proton currents in a cell-attached patch that increase gradually with time during each pulse because the single channel proton currents are too small to resolve. Superimposed are large single channel currents most likely due to Kv1.3 delayed rectifier channels that dominate macroscopic currents in these cells (71). After this patch was excised into K⁺-free solutions, the same population of proton channels generated the currents shown in B–D at the indicated pH_i. As pH_i was decreased, the currents became larger and activated much more rapidly (note the changing calibrations). The graph in E shows average activation time constants (τ_act) obtained by single exponential fits in many patches. Changing pH_o shifts the kinetics along the voltage axis with little change in the range of τ_act values. In contrast, changes in pH_i profoundly affect τ_act, with a ~5-fold slowing per unit increase in pH_i. [From (67). Originally published in The Journal of Physiology 489:299-307, 1995.]

In stark contrast to the effects of pH_o on channel activation, deactivation was completely insensitive to pH_o in human THP-1 cells (66). A weak sensitivity to pH_o was seen in rat alveolar epithelial cells if the tail currents were force-fit by a single exponential, but when a double exponential fit was used, the rapid τ_tail component was pH_o independent. Near V_threshold a slower component of tail current decay appeared that was kinetically closer to τ_act and was more steeply voltage dependent, as well as being shifted ~40 mV/unit change in pH_o (38).

In a systematic study of the regulation of the voltage-activation curve by pH, it was discovered that increasing pH_o or decreasing pH_i by one unit shifted the g_H-V relationship by −40 mV (38), as illustrated in Figs. 9A & 9B. This relationship was quantified as the Rule of Forty:

$$V_{\text{threshold}}=-40\mathrm{mV}\times \left(\Delta \mathrm{pH}\right)+20\mathrm{mV}$$ Eq. (1)

where V_threshold is the “threshold” voltage at which proton currents are first evident and ΔpH = pH_o - pH_i. Although V_threshold is an artificial construct, when used rationally and consistently it is an extremely useful tool for evaluating changes in the position of the g_H-V relationship. A metric more in vogue for other voltage-gated channels is to fit their g-V relationship to a Boltzmann function, to obtain a slope factor and a midpoint voltage. However, for proton channels this approach provides only the illusion of quantification. Under most experimental conditions, proton depletion combined with slow activation kinetics precludes straightforward measurement of the g_H-V relationship, so that Boltzmann parameters become meaningless (196). An alternative form of Eq. 1 that does not require precise knowledge of pH_i (which often deviates from its nominal value, even when attempts are made to control it) is:

$$V_{\text{threshold}}=0.76\times V_{\text{rev}}+18\mathrm{mV}$$ Eq. (2)

in which V_rev is the measured reversal potential (60). Both eqs. (1) and (2) were determined in rat alveolar epithelial cells. When data from all studies published by 2003 were included, encompassing 15 cell types, the result was quite similar, as illustrated in Fig. 10:

$$V_{\text{threshold}}=0.79\times V_{\text{rev}}+23\mathrm{mV}$$ Eq. (3)

The physiological meaning of this relationship, as alluded to above, is that the proton channel evidently evolved into an ideal acid extrusion mechanism. In fact, the regulation of its gating by pH suggests that the voltage-gated proton channel was engineered to solve “the central problem of pH_i regulation: the neutralization of intracellular acid derived from a variety of sources” (228). Over a very wide range of pH_o and pH_i encompassing most or all physiological conditions, proton channels open only when there is an outward electrochemical gradient. They open only when doing so will result in outward H⁺ current, i.e., acid extrusion (56).

Figure 9.

A. Families of proton currents at different pH_o//pH_i in three rat alveolar epithelial cells (one in each row). The most obvious effect of pH is to vary the voltage range in which proton channels open. B. Average current-voltage relationships at the indicated ΔpH, where ΔpH = pH_o - pH_i, illustrate that the position of the g_H-V relationship is completely determined by ΔpH. The “Rule of Forty” (Eqs. 1–3) expresses the 40-mV shift in the position of the g_H-V relationship that occurs when ΔpH changes by one unit, regardless of whether pH_o or pH_i is changed. [From (38) ©Cherny et al., 1995. Originally published in The Journal of General Physiology 105:861-896.]

Figure 10.

Regulation of the position of the g_H-V relationship by ΔpH occurs in all known voltage gated proton channels. In most species, this regulation results in V_threshold always being positive to the reversal potential, V_rev, thus ensuring that when proton channels open, acid will be extruded. The blue line shows the result of linear regression on data reported in 15 cell types that are listed in (56); the dashed red line shows equality between V_threshold and V_rev for comparison. In the recently identified kHv1 channel in the dinoflagellate, Karlodiniun veneficum, V_threshold is shifted by −60 mV relative to all other species (green data points and line), with the result that depolarization above V_threshold will produce inward H⁺ current in kH_V1 at all ΔpH (251).

Now that we have enshrined an Absolute Rule we will immediately identify an exception. A spectacular deviation from the universality of Eq. 3 was identifed recently in dinoflagellates (251). The proton channel from Karlodinium veneficum, kH_V1, is regulated by ΔpH with an offest of −37 mV, so that its V_threshold is always 60 mV more negative than in other proton channels:

$$V_{\text{threshold}}=0.79\times V_{\text{rev}}-37\mathrm{mV}$$ Eq. (4)

Fig. 10 illustrates that V_threshold data from kH_V1 (green dots) always fall below the red dashed line of equality between V_rev and V_threshold . The consequence is that kH_V1 conducts inward rather than outward H⁺ current over a wide voltage range. The function of the dinoflagellate proton channel is thus clearly very different from that in all other species, and will be discussed below [Bioluminescence in Dinoflagellates]. However, despite its anomalous voltage dependence, not to mention its evolutionary distance from humans, it is noteworthy that kH_V1 still obeys the Rule of Forty (251). It is difficult to escape the conclusion that the gating of all voltage-gated proton channels is regulated by ΔpH by a similar mechanism. The details of this mechanism remain obscure. The possibility that a single titratable residue might govern the ΔpH dependence was eliminated by a comprehensive study in which most such groups were individually mutated to Ala or another neutral residue (223). Every mutant studied obeyed the Rule of Forty, retaining a roughly 40 mV/unit pH shift when pH was varied, although the absolute positions of the g_H-V relationships varied widely. The mechanism responsible for the pH dependence of gating is evidently subtler and substantially more robust than a simple modulatory site.

Gating kinetics

Voltage-gated proton currents appear to electrophysiologists very similar to many other voltage-gated channel currents. In fact, to someone familiar with delayed rectifier K⁺ currents, proton currents evoke a sense of déjà vu, with the exception of the time scale being orders of magnitude slower. Both currents activate sigmoidally; this appears to reflect n² kinetics for dimeric proton channels (103), but n⁴ kinetics for tetrameric K⁺ channels (122). Both H⁺ and K⁺ channels exhibit the Cole-Moore effect, in which the delay before the current turns on during depolarization increases with preceding hyperpolarization (46, 65). Both H⁺ and K⁺ currents deactivate exponentially upon repolarization, and τ_tail is generally faster than τ_act, presumably because a closing conformational change in just one subunit suffices to abolish conduction. Both H⁺ and K⁺ currents are inhibited by extracellular divalent cations like Zn²⁺ in most species by a mechanism in which τ_act is slowed drastically, the g-V relationship is shifted positively, and τ_tail is affected only weakly (36, 98, 101).

Proton currents in different species differ most obviously in gating kinetics. Mammalian proton currents activate with τ_act in the range of seconds at room temperature, whereas in snail neurons, τ_act is a few milliseconds. These kinetic differences are consistent with the proposed functions of proton channels in these cells. In most mammalian cells proton currents extrude acid, which is not a job that needs to be rushed. Even when proton currents compensate charge in phagocytes [See: Charge compensation], the kinetics of NADPH oxidase activation is sufficiently slow that a rapid compensatory response is not necessary. However, in snail neurons, the proposed function is to open during action potentials (29, 167, 170, 242, 260), and consequently, speedy activation is essential.

Gating kinetics provides clues to and constraints on possible gating mechanisms. For example, as was discussed previously [See: Gene and molecular features] activation of proton current during depolarizing pulses is sigmoid in most species. As shown in Fig. 4, this property reflects the dimeric nature of the channel, together with the apparent cooperativity in gating (103, 139, 198, 199, 261, 262). The greater complexity of channel opening in the dimer is reflected in the twice-stronger temperature dependence of both opening and closing kinetics in the dimer compared with the monomer (199). One intriguing feature of the temperature dependence of gating of native proton currents (Fig. 11) is that the Q₁₀ (the factor by which a kinetic parameter changes with a 10°C change in temperature) was the same for all three parameters determined: delay, τ_act, and τ_tail; activation was fitted by a delay followed by a single rising exponential (64). All three parameters had Q₁₀ 6–9, which is extraordinarily high for ion channel gating kinetics, which most frequently are near 3 (64, 119). The similarly high Q₁₀ for all three gating parameters was interpreted to reflect a channel with multiple independent subunits, each of which must undergo a complex gating transition during opening, with a reverse conformational change in only one subunit being sufficient to close the channel (64). This class of mechanism again resembles the classical Hodgkin-Huxley model for Na⁺ and K⁺ channel gating. The revelation that in many species, the proton channel functions as a dimer in which both protomers must undergo an opening conformational change before either can conduct (103, 199, 262) is in excellent agreement with the predictions that were based on temperature effects on gating kinetics.

Figure 11.

Proton channel gating is strongly temperature dependent. A, B. In a human neutrophil at pH_o 7.0, pH_i 5.5 at the indicated temperatures, a depolarizing prepulse activated the g_H, then the voltage was stepped to test voltages, illustrated in 20-mV increments. Note the change in time-base. The “tail current” (deactivation or closing) time constant, τ_tail, was obtained by fitting a single decaying exponential to each current. The Q₁₀ was identical at all voltages (C) and was 8.5 in this cell, corresponding to an activation energy of ~38 kcal/mol. [From (64) ©DeCoursey & Cherny, 1998. Originally published in The Journal of General Physiology. doi: 10.1085/jgp.112.4.503.]

Another clue to the gating mechanism comes from deuterium isotope effects. In most respects, proton channels functioned similarly in normal and heavy water. However, channel opening was 3-fold slower in deuterium, whereas channel closing was at most weakly affected (60). Because deuterons bind with higher affinity to most titratable groups, the slowing of activation by deuterium is consistent with deprotonation being a rate-determining step in channel opening. A three-fold slowing is consistent with a 0.5 unit increase in the pK_a of the group in D₂O (60), which in turn is more consistent with a carboxylic or ammonium group being involved than a sulfhydryl acid (60, 238). Intriguingly, and perhaps coincidentally, both pH_o and D₂O have large effects on τ_act and small effects on τ_tail. Deprotonation of hypothetical groups accessible to the internal solutions was proposed to be a prerequisite step in channel opening in a model proposed to explain the dependence of the g_H-V relationship on both voltage and the pH gradient, ΔpH (38). In this model, the protonation state of titratable groups accessible to one side of the membrane or the other regulates proton channel gating (38).

Given their structural parallels (Fig. 2), it is tempting to assume that gating of H_V1 will entail the same outward movement of the S4 domain that is thought to occur in the VSD of K⁺ and Na⁺ channels (131, 147, 161, 253, 275, 276). Indeed, Cys scanning supports the general similarity of S4 movement in CiH_V1 to that of other VSDs (103), as do paddle motif swaps (7). However, a remarkable study of mH_V1 revealed that mutants in which the entire C terminus was truncated along with S4 as far as between the second and third Arg residues in the S4 domain still retained proton channel function (230). Although each of the three S4 Arg residues in H_V1 contributes to voltage sensing (104), their movement must not be as extensive as in K⁺ channels, unless the movement in the C truncated mutant was reduced compared to the WT channel. The charged residues in S4 of the Shaker K⁺ channel VSD are proposed to move through a “gating charge transfer center” (258); the first (R1) is in this center in closed channels, but the fifth (K5) resides there in the open state (157).

At present, most fundamental questions regarding gating remain unanswered or incompletely answered. What is the structural difference between open and closed proton channels? How does the molecule move during gating? How do the two protomers interact during gating in the dimer? How does pH regulate the voltage-dependence of gating? Although recent discoveries resulting from structure-function studies seem to have complicated rather than clarified these questions, the voltage-gated proton channel continues to be an exceptional system in which to explore fundamental questions about voltage gating mechanisms.

Inhibition of Proton Currents by Zn²⁺ is Surprisingly Revealing

Essentially every study that has reported the existence of proton currents in a new cell type has used inhibition by Zn²⁺ as a defining feature. Occasionally, Cd²⁺ or other polyvalent metals are included; a large number have generally similar effects, including Cu²⁺, Ni²⁺, Co²⁺, Mn²⁺, Be²⁺, Pb²⁺, Hg²⁺, La³⁺, Gd³⁺, and Al³⁺, with Zn²⁺, Cu²⁺, and La³⁺ considered the most potent (56). In fact, in the absence of other high affinity inhibitors, Zn²⁺ is frequently used to antagonize physiological effects of proton channel function in cells. The advent of the HVCN1 knock-out mouse has provided an alternative to this classical pharmacological lesion experiment. As more homologs of proton channels are identified and characterized, some have been discovered to be relatively insensitive to inhibition by Zn²⁺. The proton channel of the sea squirt, Ciona intestinalis, can be inhibited by Zn²⁺, but only at concentrations 27-fold higher than the mouse channel (234). A proton channel in Emiliania huxleyi, a marine phytoplankton, is also only weakly inhibited by Zn²⁺ (259), as is kH_V1, a dinoflagellate channel (unpublished data with S.M.E. Smith, D. Morgan, B. Musset, V.V. Cherny, A.R. Place, and J.W. Hastings). There is thus a hint of a pattern of weaker metal sensitivity for proton channels in marine than terrestrial life. However, proton currents in the nudibranch mollusc, Tritonia diomedea, appear to exhibit high Zn²⁺ sensitivity (271), as do proton currents in Coccolithus pelagicus, a marine coccolithophore (259). In any case, the idea that lower Zn²⁺ sensitivity in marine species might be a protective adaptation to the heavy metal content of seawater seems unlikely, because the metal concentrations are too low to produce much inhibition of even the more sensitive human proton channel: 76 nM Zn²⁺, 112 nM Ni²⁺, 14 nM Cu²⁺, 1 nM Cd²⁺, 21 pM La³⁺ (263). Low sensitivity to metal will become more important as the seas become polluted, but at present there seems to be a reasonable safety factor.

The effects of polyvalent cations on voltage-gated proton channels resemble the effects of these metals on most voltage-gated ion channels. Metal ions slow channel opening, shift the g-V relationship positively, and may reduce the maximum conductance (36, 89, 119). As illustrated in Fig. 12, although Zn²⁺ appears to make the current smaller, most of this is the result of slower activation kinetics (larger τ_act) and a positive shift of the g_H-V relationship. Both of these effects were quantified over a range of pH_o and the pattern was evaluated using several possible models of competition. Ironically, despite the widespread use of Zn²⁺ to characterize proton currents, one of the most remarkable features of the inhibition of proton currents by Zn²⁺, namely the strong modulation of Zn²⁺ effects by pH, was overlooked until 17 years after the first report of voltage-gated proton channels (36). To a first approximation, the inhibition produced by 1 μM Zn²⁺ at pH_o 7.0 requires 10 μM Zn²⁺ at pH_o 6.0, and ~1000 μM Zn²⁺ at pH_o 5.0 (Fig. 12). The antagonism between low pH and Zn²⁺ suggests that H⁺ and Zn²⁺ complete for a binding site on the proton channel molecule.

Figure 12.

Modulation of Zn²⁺ effects on proton currents by pH_o reveals strong competition between Zn²⁺ and H⁺ for external binding sites. Identical families of pulses were applied in each row at the indicated pH_o and Zn²⁺ concentrations. Zn²⁺ slows activation of the proton current at a given voltage and shifts of the g_H-V relationship positively. [From (36) © Cherny & DeCoursey, 1999. Originally published in The Journal of General Physiology. doi: 10.1085/jgp.114.6.819.]

Fig. 13 illustrates the predictions of the preferred model in which Zn²⁺ binds competitively with H⁺ at three sites with pK_a 6.3, a metal binding constant pK_M 6.5, and a cooperativity factor of 0.03, meaning that Zn²⁺ can still bind if one of the three groups is protonated, but with ~30-fold lower affinity. There were no other adjustable parameters in the model, which simply assumed that the proton channel cannot open when Zn²⁺ is bound (36). Two other, even simpler models also provided good fits to the data; these had 2–3 titratable groups that coordinated Zn²⁺ strictly competitively with protons. In contrast, simple 1:1 competition could not explain the data, because the effectiveness of Zn²⁺ decreases by a maximum of 10-fold/unit pH in such a model, and the data clearly indicated a 100-fold shift between pH_o 6 and pH_o 5 (Figs. 12 & 13). The outcome of this exercise was a prediction that Zn²⁺ inhibits proton currents by binding simultaneously to at least 2 groups with pK_a 6–7, in the range of His (36). This prediction appeared to be borne out nicely when the human proton channel was identified seven years later. Ramsey and colleagues (224) showed that two His residues, His¹⁴⁰ and His¹⁹³, that were expected to be accessible to the external solution, both contributed to Zn²⁺ inhibition.

Figure 13.

The pH_o dependence of the slowing of proton current activation can be explained if Zn²⁺ binds to a site consisting of three groups with pK_a 6.3, with affinity 316 nM, and cooperativity factor a 0.03 (see text). All curves were drawn with these values and no other adjustable parameters. Adequate fits were also obtained by assuming two titratable groups, but not with only one. At high pH_o competition with protons disappears and limits to simple metal binding. See text for further details. [From (36) © Cherny & DeCoursey, 1999. Originally published in The Journal of General Physiology. doi: 10.1085/jgp.114.6.819.]

To our consternation, a homology model of the human proton channel structure, based on the crystal structure of the VSD from K⁺ channels, indicated that the two Zn²⁺-binding His residues, His¹⁴⁰ and His¹⁹³, were ~14 Å apart, too far to plausibly coordinate a single Zn²⁺ atom (199). Another homology model predicted a comparable separation between the two His residues (223). However, the discovery that the proton channel was a dimer (139, 150, 261), presented a possible explanation. As shown in Fig. 5B, if the dimer adopts a position with the complementary His residues facing each other, Zn²⁺ could bind at the interface between the protomers. This hypothesis was explored by creation of a series of channels with mutations to His¹⁴⁰ and His¹⁹³ as well as tandem dimers. Figs. 14A and 14B illustrate the generally similar pattern observed for Zn²⁺ shifting the g_H-V relationship and slowing activation, respectively. Mutation of either His¹⁴⁰ (H140A) or His¹⁹³ (H193A) individually attenuated Zn²⁺ effects, indicating that both residues contribute to the effects observed in WT channels. H140A, H193A and the tandem dimer H193A-H140A all exhibited distinct Zn²⁺ effects that were weaker than in WT. Remarkably, the tandem dimer WT-H140A/H193A was as insensitive to Zn²⁺ as the double mutant that lacked both His (H140A/H193A). In particular, the slowing effect of Zn²⁺ occurs only when there is at least one His in each protomer, and is absent when both His are present in the same protomer. This pattern suggests that Zn²⁺ slows proton channel opening when it binds at the interface between protomers. Presumably, the Zn²⁺-bound channel dimer cannot undergo a conformational change required for opening.

Figure 14.

Effects of Zn²⁺ on V_threshold (A) and on the slowing of proton current activation (B) in constructs with several His mutations. Dimeric proton channels are illustrated with diagrams in which solid symbols indicate His and open symbols indicate Ala substituted for His. Three tandem dimers are shown with their C-and N-termini linked to constrain the His positions in the dimer. V_threshold shifts were estimated from g_H-V relationships plotted semilogarithmically. Statistical comparisons are to WT channel parameters (^⌘p<0.05, *p<0.01). [From (199). Originally published in The Journal of Physiology 588:1435-1449, 2010; doi: 10.1113/jphysiol.2010.188318.]

Functions

As discussed above [See: Regulation of Gating by the pH Gradient, ΔpH] the properties of the voltage-gated proton channel suggest that in most species, this molecule is ideally suited to solve “the central problem of pH_i regulation: the neutralization of intracellular acid derived from a variety of sources” (228). Proton channels have been shown to contribute to pH_i recovery following an acid load in a variety of cells: snail neurons (260), rabbit osteoclasts (211), human neutrophils (75), mouse mast cells (145), rat microglia (187), a human bronchial epithelial cell line (16HBE14o-) (264), rat alveolar epithelial cells (192), rat hippocampal neurons (35, 246), murine osteoclasts (183), in TsA201 cells transfected with the mouse proton channel gene mVSOP (234), and in a rat microglial cell line GMI-R1 (185). In many other cells, proton channels extrude acid under specific conditions that occur physiologically. Hypotonic shock leads to depolarization and alkalinization in bovine articular chondrocytes, that may be mediated by proton channels (231, 232). Roles in regulating both pH_i and membrane potential have been proposed in human cardiac fibroblasts (87). Although acid extrusion remains the main general function of proton channels, several quite different functions have been identified in a number of cells, which are described below.

Bioluminescence and Other Functions in Dinoflagellates

As mentioned above (Fig. 1), the first explicit proposal of a voltage-gated proton channel was in 1972 by Fogel & Hastings in the bioluminescent dinoflagellate, Gonyaulax polyedra (95). It had been shown previously that current injection into the large central vacuole of another bioluminescent dinoflagellate, Noctiluca scintillans (N. miliaris), elicited an all-or-nothing action potential that appeared to be inverted (34, 120). Elegant studies by Roger Eckert and colleagues showed that the Noctiluca action potential occurs across the tonoplast, the membrane surrounding the large central flotation vacuole, and is inverted when recorded by a pipette inserted into the vacuole, because this is the topological equivalent of extracellular recording (81) (83). The action potential triggers luminescence that originates in numerous discrete sources (80–83, 220). These microsources have been named “scintillons” (78, 111), and are filled with luciferin, luciferase, and luciferin binding protein (95, 96, 237). Fig. 15 illustrates the mechanism by which proton channels are thought to participate in dinoflagellate bioluminescence. Upon mechanical stimulation (e.g., by a breaking wave, the wake of a passing boat, or the splashing of a nocturnal swimmer), an action potential travels along the tonoplast (81) and invades the scintillon. Scintillons are connected by a neck of membrane that is continuous with the tonoplast (209). The action potential depolarizes the scintillon membrane, opening proton channels, which mediate proton flux from vacuole into the scintillon. The vacuole has very low pH, estimated to be 3.5 in Noctiluca (206), whereas the scintillon is near neutral, so there is a large chemical driving force. The protons trigger a light flash by changing the activities of two proteins. First, luciferin binding protein releases luciferin, the light emitting substrate of luciferase, at low pH (25, 96, 141). Second, luciferase has a pH optimum at pH 6.3, and is inactive at pH 7.5 (141, 237). It seems very likely that voltage-gated proton channels play a crucial role in triggering bioluminescence in dinoflagellates, by orchestrating the concerted activation of both mechanisms.

Figure 15.

Proposed mechanism by which proton flux through voltage-gated proton channels triggers light flashes in dinoflagellates. Mechanical stimulation elicits an action potential conducted along the tonoplast, the membrane separating the central vacuole and a thin layer of cytoplasm (gray). When the action potential invades the scintillons (grape-like structures formed by evagination of tonoplast), proton flux from the vacuole at low pH into the scintillon activates luciferase, triggering the flash. [From (110). Originally published in Hastings JW. Bioluminescence. In: Cell Physiology Sourcebook: A Molecular Approach (3rd ed.), edited by Sperelakis N. San Diego: Academic Press, 2001, p. 1115-1131.]

A further, more spectacular role has been speculated to exist, namely mediating the action potential itself (209). The flash-triggering action potential coincides with decreased impedance, indicating that a conductance increase is responsible (83). The peak of the action potential is not affected by various ion substitutions in the vacuole (206), but when pH in the vacuole was increased, the peak decreased (205) as expected if proton flux mediates the action potential. A voltage-gated proton channel gene from a non-bioluminescent dinoflagellate, Karlodinium veneficum (22) has been identified and expressed heterologously (250). K. veneficum is a predatory dinoflagellate notorious for producing a variety of toxins (247); its blooms coincide with fish kills in the coastal waters worldwide (13, 74). Remarkably, although the expressed kH_V1 protein exhibits all of the main features of voltage gated proton channels in other species, including obeying the Rule of Forty (Eq. 4), it differs in the absolute position of the g_H-V relationship. As illustrated in Fig. 10, at any given ΔpH, the kH_V1 channel opens at voltages 60 mV more negative than in all other species (251). Consequently, there is a substantial voltage range within which inward H⁺ current occurs. These properties are reminiscent of voltage gated Na⁺ channels and strongly suggest the ability of these channels to mediate an action potential. In any case, the negative voltage range of activation of kH_V1 indicates a stark difference in function, compared with all other known proton channels. Whereas other proton channels are designed to allow only proton extrusion; kH_V1 activation will produce H⁺ influx. H⁺ influx could initiate or propagate action potentials, or produce a drop in pH_i that might have signaling functions. A role of H⁺ flux into the cytoplasm was proposed to trigger tentacle flexion in N. miliaris (212). As originally proposed (95), H_V1-mediated H⁺ flux from the vacuole into the scintillon triggers the flash (251) (Fig. 15).

A variety of functions might exist for proton channels in non-bioluminescent dinoflagellate species. For example, in mixotrophic species like K. veneficum, H⁺ flux might be involved in capturing prey by extrusion of trichocysts or prey digestion. A phylogenetic analysis of VSD regions from several dozen H_V1 sequences revealed high sequence diversity among the single-celled species and among invertebrates (Fig. 3). Although this observation may partially reflect the genetic lability of dinoflagellates and other unicellular species (11), diverse sequences could well produce other novel functions of H_V1. As in multicellular organisms, ion channels in dinoflagellates undoubtedly play a variety of roles in regulating basic life functions, which makes them excellent targets for controlling dinoflagellate populations and behavior.

Calcification by Coccolithophores

New proton channel genes were recently identified in two marine unicellular algae, the coccolithophores Emiliania huxleyi and Coccolithus pelagicus ssp braarudii (259). The properties of these channels in heterologous expression systems were quite similar to those of other species, with the exception that the E. huxleyi channel (EhH_V1) was only weakly inhibited by Zn²⁺. Coccolithophores are abundant phytoplankton that play a major role in calcification; i.e., the conversion of HCO₃^Ȣ (derived from CO₂ dissolved in ocean water) into calcium carbonate, CaCO₃ via the intermediate reaction HCO₃^Ȣ → CO₃²⁻ + H⁺. Each Ca²⁺ that is converted into CaCO₃ generates a proton, which must be eliminated from the cytosol of the coccolithophore. Inhibiting the Zn²⁺-sensitive CpH_V1 with 30 μM Zn²⁺ produced an immediate drop in pH_i and a sustained decrease in calcite production (259). Decreasing pH_o lowers pH_i, suggesting that ocean acidification due to dissolution of atmospheric CO₂ derived from human activity (or, for those who harbor doubts as to its source, by magic) could inhibit calcification directly. However, a stronger correlation was found between coccolith mass and [CO₃²⁻] than with pH or pCO₂, indicating complex control (21). Nevertheless, since the Last Glacial Maximum (“Ice Age”), ocean CO₂ levels have increased by ~50% (from ~190 to ~280 parts per million), and coccolith mass has decreased by almost 50% (from 13.6 to 7.5 pg) (21).

Acid Extrusion during Action Potentials in Snail Neurons

The first voltage-clamp studies of proton channels were in garden snail neurons, Helix aspersa (260). Similar proton currents exist in the pond snail, Lymnaea stagnalis (29). The most unusual feature of proton currents in snail neurons is their rapid gating kinetics (Fig. 16), with both activation and deactivation occurring in a few tens of milliseconds or less (29, 167). The rapid activation kinetics is consistent with the proposal that proton channels open during action potentials, and extrude acid to counteract the rapid intracellular acidification that occurs in these neurons during trains of action potentials (5, 29, 167, 260). Activation of proton channels in Helix aspersa axons by depolarization generates alkalinization that is non-uniform due to geometrical factors, resulting in pH_i microdomains (214, 242).

Figure 16.

Rapidly activating proton currents in a Lymnaea snail neuron, 120 μm in diameter. Currents during pulses to the voltages shown at pH_o 7.4, pH_i 5.9 at room temperature. [From (29). Originally published in The Journal of Physiology 351:199-216, 1984.]

A similar role of opening to extrude acid during action potentials was suggested for proton channels in human skeletal muscle myotubes (24).

Amphibian Oocytes and Fertilization

Voltage-gated proton currents in Ambystoma oocytes have slower kinetics than those in snail neurons, but faster than in most mammalian cells, with τ_act of a few hundreds of milliseconds (19). The idea that proton channels might contribute to the alkalinization of oocytes during fertilization (19) is supported by a gradual decrease in proton current density within hours after fertilization (20). Xenopus laevis oocytes alkalinize by an average of 0.18 pH unit during progesterone-induced maturation; and exogenously induced alkalinization triggers maturation; however, the correlation is not absolute (149).

Frog (Rana esculenta) oocytes have slowly activating proton currents that were suggested to play a role during fertilization similar to that in Ambystoma (126). A more specific role of promoting inositol triphosphate (InsP₃) receptor mediated intracellular Ca²⁺ oscillations was later elaborated (125).

The Zinc Theory of Fertilization by Human Sperm

Human sperm are blessed with abundant proton channels, with a proposed role somewhat like that postulated in oocytes, namely the production of an intracellular alkalinization during events associated with fertilization. Sperm from humans and other animals undergo several changes in their motility as they undergo “capacitation,” the maturation process that occurs after sperm enter the female reproductive tract and that is prerequisite to fertilization. Some of those changes are mediated by increased pH_i, which stimulates metabolic activity and motility of sperm (9). Studied in vitro, the alkalinization is sensitive to membrane potential and in some species appeared to reflect an increase in membrane proton permeability (8). Indirect monitoring of the activity of voltage-gated proton channels by pH measurements, however, did not provide definitive demonstration of their presence in sperm (8, 9). The discovery of a way to patch-clamp sperm finally confirmed that voltage-gated proton channels are present in human sperm (160). Proton currents were larger in capacitated sperm, and capacitation in vitro was inhibited by Zn²⁺ (160), leading to a “zinc hypothesis” that proposes that Zn²⁺ is necessarily removed from sperm during the course of capacitation. Seminal fluid has a high concentration of zinc at 2.2 mM (229), but this value reflects total concentration, not free Zn²⁺. (For instance, equine plasma contains 8 μM total zinc, but only 210 pM free Zn²⁺ (165)). Assuming that sufficient Zn²⁺ is present to inhibit proton channels, pH_i would remain low until the sperm approach their destination inside the female reproductive tract where they encounter a more hospitable environment replete with proteins to buffer Zn²⁺. Thus relieved of their inhibitions, proton channels would conduct H⁺ out of the cytoplasm, pH_i would increase, and the sperm would capacitate and forge ahead to fertilize the ovum. This remarkable hypothesis may apply to humans, but cannot explain sperm alkalinization in mouse, whose sperm apparently lack proton channels (160). No fertility defects have been reported in HVCN1 knockout mice. In the Japanese eel, Anguilla japonica, spermatogenesis as well as sperm motility require Zn²⁺ (274).

Acid Secretion in Epithelia of the Respiratory Tract

Voltage-gated proton currents are present throughout the epithelium of the respiratory system, from alveolar epithelium (54) to tracheal epithelium (94) to nasal and sinus mucosa (44). Both the alveolar lining fluid and airway surface liquid are maintained at relatively low pH (93, 210). The primary mechanism of H⁺ secretion by airway epithelial cells - both human airway cell lines and primary cultured airway tissues - is the voltage-gated proton channel, with smaller contributions from other transporters (44, 94, 128, 240). Proton secretion measured in chronic rhinosinusitis patients increases in asthma, and the fraction contributed by proton channels decreases (44).

Acid secretion in airway epithelia was inhibited equally by Zn²⁺ or by NADPH oxidase inhibitors, suggesting that oxidase activity generates a proton gradient that drives H⁺ secretion via proton channels (240). The main NADPH oxidase system in human fetal lung epithelial cells is DUOX1 (92); Duox is also present in adult human airway epithelium (97). DUOX activity may serve a host defense mechanism in airway cells akin to that in phagocytes (97, 208, 240). In airway epithelial cells, ATP-stimulated H₂O₂ production corresponding to ~3–340 fA (97, 188) and in alveolar cells hormone-stimulated H₂O₂ production of ~21 fA of electron current per cell were found (92). The disparate values for H₂O₂ production in cultured airway cells may reflect a strong dependence of Duox expression on time in culture (221). Because these cells have abundant other conductances, it is not clear whether proton channels are necessary to compensate such a small electrogenic flux (91).

It was hypothesized that proton channel-mediated H⁺ efflux in alveolar epithelial cells might contribute to CO₂ elimination by the lung (55), although this would occur slowly due to the absence of carbonic anhydrase in apical subphase liquid (84). Several attempts to evaluate this mechanism have failed to support the idea (85, 133, 255), but conclusive studies have not been reported.

The first proton channel mutation to be identified in a human subject was Met⁹¹ to Thr (M91T) (128). The source of the M91T mutation was deceased and had an unremarkable clinical history. Only one heterozygous M91T genotype was detected in 95 human genomic DNA samples, placing an upper limit on its frequency of occurrence. Channels with the M91T mutation in heterologous expression systems were activated in a more positive voltage range than normal (WT) proton channels. Because airway epithelial cells are thought to maintain a relatively constant membrane potential, under physiological conditions proton channels are most likely activated by an increased outward pH gradient (cytoplasmic acidification or mucosal alkalinity or both) (128). The M91T mutant would require a pH gradient ~0.5 unit larger to activate the g_H to the same extent (128).

B-Cell Receptor Signaling in Human B Lymphocytes

Voltage-gated proton channels are expressed in every leukocyte or related cell line that has been studied, but expression levels vary greatly. Human T lymphocytes have barely detectable proton currents, averaging 1.5 pA at +60 mV in whole-cell recordings at pH_o 7.5 and pH_i 6.0, but proton currents in human B lymphocytes are >100 times larger (236). Staining with an hH_V1 antibody reveals high levels of protein expression in peripheral B cells, but undetectable levels in T cells (32). The expression of hH_V1 protein in human tonsillar tissue was high in resting cells, but down-regulated in proliferating cells in the germinal centers, suggesting a role in early stages of B cell activation (32). The proton channel protein co-localized with the B cell receptor (BCR) during capping and receptor internalization and was co-immunoprecipitated with the BCR-associated protein Igβ, indicating physical proximity of the two molecules (32).

Unlike the bactericidal role of ROS produced by phagocytes [See: The Phagocyte Respiratory Burst], B cells produce ROS mainly for signaling. The somewhat daunting pathways involved are summarized in Fig. 17. ROS production by splenic B cells stimulated by either PMA or anti-IgM was greatly reduced in the HVCN1 knockout mouse (32). ROS are required to inhibit phosphatases temporarily and thus allow correct B cell activation. Most of the deficits in B cell function in the HVCN1 knockout mouse could be explained as consequences of perturbed downstream signaling due to impaired oxidation of the protein tyrosine phosphatase SHP-1 (32, 227).

Figure 17.

Schematic representation of the role of the proton channel (HVCN1) in the context of B cell receptor (BCR) stimulation. Antigen binding to the BCR results in phosphorylation of ITAMs (immunoreceptor tyrosine activation motif) in the Ig-α/β heterodimer by LYN, creating docking sites for Syk. This serves to amplify BCR signaling by further recruitment and activation of Syk, which leads to PI3K activation, activation of Akt and increased glucose uptake and metabolism. Amplification of signaling is negatively regulated by CD22, which is also phosphorylated by LYN, providing a docking site for protein tyrosine phosphatase SHP-1. In resting cells, SHP-1 inhibits B cell activation by dephosphorylating Syk, thus counterbalancing ITAM-Syk mediated signal amplification. BCR stimulation results in ROS generated by NADPH oxidase. The Ȯ₂⁻ that is produced combines with protons to form H₂O₂ and O₂, which freely diffuse through the membrane. ROS generate a localized oxidizing environment causing inhibition of SHP-1, which results in amplification of BCR signal. HVCN1 sustains NADPH oxidase activity. In the absence of HVCN1, the oxidizing environment cannot be maintained and consequently SHP-1 remains active, reducing BCR-signal strength. [From (32). Originally published in Nature Immunology 11: 265-272, 2010; doi: 10.1038/ni.1843.]

The Phagocyte Respiratory Burst

Charge compensation

The best-known specific function of voltage-gated proton channels is charge compensation in phagocytes. This role is quite distinct from acid extrusion, although the latter inevitably occurs at the same time. Henderson, Chappell, and Jones pioneered this area of study in the late 1980s, publishing a series of groundbreaking papers (114–116) that drew three crucial conclusions. First, they showed that NADPH oxidase, the enzyme activated in neutrophils, macrophages, and eosinophils during phagocytosis (Fig. 18), is electrogenic. The electron transfer pathway is located in the gp91^phox component of NADPH oxidase, which has been renamed Nox2 (sometimes the entire NADPH oxidase complex is imprecisely called Nox2). Electrons are transferred from intracellular NADPH across a redox chain comprising FAD and two heme groups to extracellular (or intraphagosomal) O₂ to form superoxide anion (O_2̇⁻) (10). The flux of electrons across the membrane has two immediate consequences: depolarization and pH changes. The membrane potential will depolarize rapidly; without compensation, the electron current in a human eosinophil at 37°C would depolarize the membrane at a rate greater than 18 Volts/s (56). It is obvious that charge compensation is necessary to prevent membrane dielectric breakdown. Somewhat less obviously, depolarization itself inhibits NADPH oxidase activity, especially positive to +50 mV (73), which happens to be roughly the level of depolarization attained by intact human neutrophils at the peak of the respiratory burst (100, 130, 222). Depolarization to +200 mV abolishes NADPH oxidase activity altogether (73, 216). The depolarization of the plasma membrane during NADPH oxidase activity is exacerbated by Zn²⁺ (or Cd²⁺) at concentrations that inhibit proton currents, supporting the idea that proton channels contribute significantly by limiting the extent of depolarization (16, 17, 77, 88, 115, 146, 193, 222). Similarly, the phagosome membrane depolarizes rapidly after phagosome formation (182). Limiting the extent of depolarization is also important for Ca²⁺ signaling, because it influences the driving force for Ca²⁺ influx through store-operated Ca²⁺ channels. The rise in intracellular Ca²⁺ elicited by chemoattractants was greatly attenuated in HVCN1 knockout mice, but could be restored by collapsing the membrane potential with gramicidin (88).

Figure 18.

Proton husbandry during the phagocyte respiratory burst. Upon stimulation, the components of the NADPH oxidase complex assemble at the membrane and begin to relay electrons from NADPH across the membrane to reduce O₂ to superoxide anion, O_2̇⁻. Because of the massive increase in O₂ consumption, this process is called the “respiratory burst” (12) despite the fact that most of the O₂ is consumed to make ROS (reactive oxygen species) and is not affected by mitochondrial inhibitors (235). Protons left behind in the cytoplasm exit mainly through voltage-gated proton channels (193), although other transporters also play important roles (178). Inside the phagosome, protons are consumed during the spontaneous dismutation of O_2̇⁻ to H₂O₂ as well as during HOCl generation by myeloperoxidase (MPO) (138). [From (57). Originally published in Physiology 25: 27-40, 2010; doi: 10.1152/physiol.00039.2009.]

Regulation of pH

The second major consequence of NADPH oxidase activity is to change pH on both sides of the membrane, decreasing pH_i and increasing pH_o. For each electron that leaves the cytoplasm, one proton is left behind, either during the reduction of FAD or during the reconstitution of NADPH by the hexose monophosphate shunt (HMS in Fig. 18) (129). The HMS must operate continuously during the respiratory burst. To sustain an electron current of −10 pA, it was estimated that the entire cytoplasmic NADPH supply (50–100 μM) must turn over 4–8 times per second (180); the respiratory burst lasts minutes to hours depending on the cell type and stimulus (72). The complex events depicted in Fig. 18 can be best understood by realizing that NADPH oxidase is the driving force for nearly everything else. Although at the peak of the respiratory burst, proton and electron fluxes are nearly identical (26, 257, 265), the NADPH oxidase gets off to a faster start than proton channels do. The most rapid consequence of oxidase activity is membrane depolarization. Because relatively few electrons must cross the membrane to change the membrane potential, compared with the amount of ROS that must be produced for biochemical detection, many early studies concluded incorrectly that membrane depolarization preceded and perhaps triggered the respiratory burst (31, 45, 132, 140, 243, 249, 268). The main charge compensators, voltage-gated proton channels, do not open until there has been sufficient depolarization, and thus will inevitably lag behind. Fig. 19A illustrates that a rapid spike of acidification occurs simultaneously with phagocytosis in human neutrophils (178). In most cells, this was followed by rapid recovery. Pharmacological inhibition of several candidates indicated that the Na⁺/H⁺ antiporter and voltage-gated proton channels were both required for recovery. When H-ATPase was inhibited, only a small additional acidification of 0.14 units occurred, which became evident ~10 minutes after the initial phagocytotic event, indicating that the proton pump plays at best a minor role. Inhibiting Na⁺/H⁺ antiport or proton channels resulted in profound monotonic acidification to pH_i 6.14 or 5.87, respectively, after 14 min, levels that would strongly inhibit NADPH oxidase (180). As may be seen in the example in Fig. 19D, among the inhibitors tested, only Zn²⁺ affected the initial rate of acidification, doubling it (178). This indicates that the voltage-gated proton channel is the first H⁺ transporter to respond during phagocytosis.

Figure 19.

The pH_i in four individual human neutrophils during phagocytosis of opsonized zymosan (OPZ) is plotted, with pseudocolor confocal images of each cell at corresponding times. The cells were loaded with the pH sensing dye, SNARF-1. In control cells, pH_i drops rapidly when OPZ is ingested and often recovers rapidly (A), but in some cells does not (B). Recovery is prevented by 20 μM dimethylamiloride, which inhibits the Na⁺/H⁺ antiporter (C) or 100 μM Zn²⁺, which inhibits proton channels (D). Among the inhibitors tested, only Zn²⁺ increased the rate of acidification, indicating that proton channels contribute significantly at early times. [From (178). Originally published in Proceedings of the National Academy of Sciences, USA 106: 18022-18027, 2009; doi: 10.1073/pnas.0905565106.]

Other roles in phagocytes

In summary, there is clear evidence that proton channels contribute to both charge compensation and pH_i regulation during the phagocyte respiratory burst. Proton currents play at least two additional roles. Protons conducted into phagosomes act as substrate for the production of H₂O₂ and HOCl, the predominant early products of NADPH oxidase activity (Fig. 18) (107, 193). The choice of protons, rather than other ions, as the main charge compensating ion is sensible when one consider the magnitude of the task. Within a few minutes, the electrons entering through NADPH oxidase would reach a virtual concentration of up to 4 M inside the phagosome (107, 226), if they were not immediately consumed in ROS production. If these electrons were compensated by any cation besides H⁺, the osmotic effect would be a ~20-fold swelling of the phagosome, which does not occur. However, when H⁺ compensates the electron flux, the protons and electrons end up mainly in the form of H₂O, HOCl, or H₂O₂, all of which are membrane permeant, thus minimizing osmotic swelling of the phagosome. It is unclear, and perhaps mainly of semantic importance, which of these four functions of proton channels is most crucial during the respiratory burst, because they occur simultaneously.

The importance of proton channel function to NADPH oxidase activity has been demonstrated by the inhibition by Zn²⁺ of the production of superoxide or its product H₂O₂ in a large number of studies (90). The development of HVCN1 knockout mice has enabled independent tests of this proposed role, which have revealed substantial, but incomplete loss of ROS production by neutrophils (88, 213), bone marrow cells (225), or B lymphocytes (32). In support of the idea that charge compensation can be rate-limiting during NADPH oxidase activity, the reduced ROS production in hH_V1 knockdown PLB-985 cells can be overcome by addition of the ionophore amphotericin B (217). Similarly, the nonselective gramicidin channel restores ROS production in mH_V1 knockout mouse neutrophils (88). Normal ROS production in a cell-free system from mH_V1 knockout mice is consistent with either charge compensation or pH regulation functions of H_V1 (225). Similarly, either mechanism could explain the restoration of ROS production by CCCP (a protonophore) in human neutrophils or eosinophils inhibited by Zn²⁺ (73, 115, 116).

Enhanced gating mode

Proton channels in phagocytes are modulated profoundly during the respiratory burst. Agonists that activate the oxidase also produce what has been called the “enhanced gating mode” of proton channels. Fig. 20A illustrates the dramatic increase of H⁺ currents recorded during a series of identical depolarizing pulses applied at 30-s intervals in a human eosinophil stimulated with PMA. Analyzed quantitatively (Fig. 20C), the enhanced gating response consists of a −40 mV shift of the g_H-V relationship, a 2-4-fold increase in maximum proton conductance (gH,max) at large positive potentials, much faster activation kinetics (2-4-fold smaller τ_act), and much slower deactivation kinetics (2-6-fold larger τ_tail). A roughly identical constellation of changes occurs in human neutrophils (70), human eosinophils (16, 18, 37, 68), PLB-985 cells (69), murine osteoclasts (183), murine granulocytes (179), and human monocytes (195) after stimulation with PMA or a variety of more physiological agonists including arachidonic acid, fMLF, IL-5 (interleukin-5), LTB₄ (leukotriene B₄), and opsonized zymosan (which triggers phagocytosis), or during spontaneous activation by adherence to glass. The enhanced gating response occurs over several minutes with a characteristic time course (Fig. 20A & 20B). Although the impression is that this response is all-or-none, intermediate responses are often seen with physiological agonists like arachidonic acid (37), fMLF (179), or in basophils, anti-IgE (197) whereas PMA elicits a maximal response.

Figure 20.

The enhanced gating mode of proton channels in human eosinophils. A, the onset of enhanced gating in a human eosinophil stimulated with 30 nM PMA in perforated-patch configuration at symmetrical pH 7.0. Test pulses to +60 mV applied at 30-s intervals are superimposed, before and up to 6 min after addition of PMA, illustrating the increasing current, faster activation, and slower tail current decay. The downward shift of the baseline current at −60 mV reflects electron current generated by NADPH oxidase activity. B, the time course of proton current enhancement (top) after stimulation with PMA (arrow) and subsequent inhibition of PKC by 3 μM GF109203X (GFX) (arrow) in a different human eosinophil. Repeated test pulses to +60 mV were applied, and only the peak current is evident at this time scale. The inward electron current at −60 mV in this cell can be seen as a downward deflection in the holding current, which is amplified in the lower panel (the pulses eliciting proton current are blanked). C shows individual currents during test pulses applied at times indicated in B by lower-case letters. The inward electron current at the holding potential is clearly evident. [From (68). (A) Originally published in The Journal of Physiology 535: 767-781, 2001; doi: 10.1111/j.1469-7793.2001.00767.x and from (179) (B, C) Originally published in The Journal of Physiology 579: 327-344, 2007; doi: 10.1113/jphysiol.2006.124248.]

The enhanced gating mode phenomenon is closely associated with the activation of NADPH oxidase (16), and the interactions between these two molecules have not yet been completely elucidated (195). Interactions between NADPH oxidase and proton channels are evident when one compares the responses of neutrophils from normal human subjects and from chronic granulomatous disease (CGD) patients. Any of several hundred specific mutations to the NADPH oxidase complex (Fig. 18) that result in loss of function produce CGD. Without medical supervision, CGD patients often die in childhood of chronic, recurrent infections (79). The first suggestion of a link between proton channels and CGD was a study showing that stimulation of CGD neutrophils with PMA failed to activate Zn²⁺ sensitive H⁺ efflux through proton channels (203). Although one interpretation was that proton channels were defective, which in turn suggested that proton current was mediated by the NADPH oxidase itself [See: History], this idea was soon disproved by a later study showing normal proton currents in these cells (204). The reason H⁺ efflux did not occur in CGD cells was that without NADPH oxidase activity to produce intracellular protons and membrane depolarization, there is little reason for proton channels to open. However, in later studies, the enhanced gating mode was found to differ subtly but distinctly in normal and CGD neutrophils (69). Although most of the changes characterizing enhanced gating occurred in CGD phagocytes, the negative shift of the g_H-V relationship was only half as large as in normal cells, and there was almost no slowing of deactivation (channel closing). One explanation was that a product of NADPH oxidase activity modulates proton channel properties. The effects of lowering pH_i on gating are consistent with the culprit being intracellular H⁺ production. Consistent with this hypothesis, enhanced gating in cells like basophils, which lack NADPH oxidase activity (50), closely resembles the gestalt seen in CGD phagocytes (197).

Until recently, the final activator of proton channels in phagocytes was thought to be arachidonic acid (AA) (113). Stimulation of phagocytes with PMA activates cPLA₂α, which liberates AA (219). Although the PKC activator, PMA, produces enhanced gating of proton channels (68–70), AA also enhances the gating of phagocyte proton channels to a limited extent in whole-cell studies (63, 105, 135, 239) (254), but to an extent comparable to the effects of PMA in perforated-patch studies (37). Finally, cPLA₂α knockout mouse neutrophils or PLB-985 cells failed to exhibit Zn²⁺-sensitive pH changes reflecting H⁺ efflux (155, 162, 168). However, Deri Morgan and colleagues (179) found that the enhanced gating mode elicited in cPLA₂α knockout mouse neutrophils stimulated with PMA or with the more physiological agonist, fMetLeuPhe, a chemotactic peptide, was indistinguishable from that seen in normal cells. Furthermore, three specific cPLA₂α inhibitors failed to affect the onset of enhanced gating or NADPH oxidase activity (179). The prevention or reversal of some of the enhanced gating features by PKC inhibitors GF109203X (GFX, Fig. 20B, (18)) or staurosporine (179) strongly supports the idea that phosphorylation by PKC is responsible for most of the response. Finally, it was shown that when human eosinophils were stimulated by AA, the resulting enhanced gating of proton channels was partially reversed by the PKC inhibitors GF109203X or staurosporine (179). Thus, strong evidence supports the conclusion that phosphorylation of the proton channel by protein kinase C (PKC) produces enhanced gating of proton channels.

That PKC produces enhanced gating by phosphorylating the proton channel itself, not an accessory molecule, was demonstrated by a study in a B cell line, LK35.2, that does not normally exhibit proton current (194). Two candidate phosphorylation sites in the intracellular N terminus of the human proton channel (Thr²⁹ and Ser⁹⁷) were mutated individually and together. Although both sites were detectably phosphorylated when various mutant HVCN1 genes were expressed in LK35.2 cells, the PMA enhancement of proton currents was lost in cells expressing channels with the T29A but not the S97A mutation (194). Thus, Thr²⁹ appears to be responsible for producing enhanced gating. Furthermore, at least part of the enhanced gating response is due to direct phosphorylation of the proton channel molecule.

One might wonder whether it is necessary for the enhanced gating mode to exist, given that proton channels would be activated eventually by the effects of NADPH oxidase activity (depolarization and pH changes). This question was examined in a mathematical model of the electrical events occurring during the respiratory burst in human neutrophils and eosinophils, that was based entirely on measured parameter values (193). In the model, if proton channels were assigned their resting properties so that enhanced gating did not occur, the peak depolarization during the respiratory burst was nearly 30 mV greater. This larger depolarization reduced the NADPH oxidase turnover rate calculated in eosinophils by 18% (193), due to the inhibition of electron translocation by depolarization (73). Thus, enhanced gating limits the extent of membrane depolarization, and thereby maximizes ROS production.

Histamine Secretion by Human Basophils

Like most other leukocytes, human basophils have large voltage-gated proton currents (40). However, the best-known role of proton channels in phagocytes, namely, compensating for the electrogenic activity of NADPH oxidase [See: The Phagocyte Respiratory Burst] cannot apply to basophils, because these leukocytes are not phagocytic and they lack NADPH oxidase (50). Nevertheless, upon stimulation with agents that trigger histamine release, including anti-IgE, fMLF, and PMA, distinct enhanced gating of proton channels was observed (197). The responses were generally smaller and more variable with the physiological agonists fMLF and anti-IgE; although only about half the cells responded, non-responding cells were still capable of responding to PMA. Quantitative comparison of the PMA response in basophils, compared with phagocytes, revealed similarly increased g_H,max and faster τ_act, but the g_H-V relationship shift was only half as large (−20 mV vs. −40 mV in phagocytes), and at most a very weak slowing of τ_tail was detected (197). Intriguingly, this attenuated PMA response is almost identical to that observed in human neutrophils from chronic granulomatous disease patients (69) or in other leukocytes that lack significant NADPH oxidase activity, such as LK35.2 cells (32). It appears that there is a mysterious functional link between proton channels and NADPH oxidase that results in more profound enhancement of proton channel gating in cells that also have high levels of NADPH oxidase activity (195).

That proton channels in basophils respond to agonists that trigger histamine release suggested possible involvement in this process. Zn²⁺ inhibited histamine release at concentrations that inhibit proton currents, but in a somewhat different range for PMA than for anti-IgE, consistent with these agonists acting by different signaling pathways. One might expect that protons channels might facilitate histamine release by compensating charge if Ca²⁺ influx were involved in the response, or by limiting pH_i changes. Because histamine release triggered by anti-IgE requires Ca²⁺ influx, but the responses to fMLF or PMA do not (164, 267), a general role of charge compensation seems unlikely. However, confocal imaging of pH_i using SNARF-1 revealed that basophils undergo significant acidification upon anti-IgE simulation (Fig. 21), which had not been reported previously. The acidification was exacerbated by Zn²⁺ (197), implicating proton channel activity during the response. Evidently, proton channels limit acidification during histamine release by human basophils. The step(s) in this process that are sensitive to pH_i have not yet been identified.

Figure 21.

Confocal images of four human basophils taken at 30 s intervals (left to right, then top to bottom) during their response to anti-IgE stimulation. The pseudo-colors indicate pH_i detected with SNARF-1 in confocal slices ~0.2 μm thick, with alkaline to acidic shown as blue to green to red to yellow. Voltage-gated proton channels are active during the response and limit the extent of acidification. [Image provided by Deri Morgan with permission.]

Do Proton Channels Cause Cancer?

A recent study reported a correlation between hH_V1 expression and the propensity for metastasis in breast cancer tissue and cell lines (266). Knockdown of hH_V1 using siRNA decreased basal pH_i and the migration of cells, suggesting that hH_V1 is constitutively active in these cells, and that hH_V1 suppression might inhibit metastasis of cancer cells. These novel results raise many questions that need to be pursued.

Conclusion

Voltage-gated proton channels are widely expressed in species ranging from unicellular dinoflagellates and coccolithophores to humans. The properties of these channels are unique, although they vary among species. They exhibit perfect proton selectivity. The protein is similar to the voltage sensing domain of other voltage-gated ion channels, but it lacks an explicit pore domain. In most species, proton channels are dimers with separate conduction pathways in each protomer, but if forced, they can function as monomers. Their voltage gating mechanism in all species is exquisitely sensitive to pH_o and pH_i with the result that in multicellular species they open almost exclusively when there is an outward electrochemical gradient. The list of intriguing and diverse functions in various cells and species is growing apace. Voltage gated proton channels trigger bioluminescence and may mediate action potentials in dinoflagellates. They promote calcification in coccolithophores. They open to extrude acid during action potentials in snail neurons. They regulate the pH in human airways and inside phagosomes. They compensate for the electrogenic activity of NADPH oxidase in mammalian and human neutrophils and eosinophils. They appear to trigger maturation of human sperm. They facilitate histamine secretion by basophils. They participate in signaling cascades in human B lymphocytes. Despite the exciting progress in recent years, many questions remain unanswered. In dozens of species, putative proton channel genes have yet to be studied. Important fundamental functions of proton channels (particularly voltage gating and pH dependence) lack mechanistic explanations. The future is positively charged!