Transcendent Aspects of Proton Channels

Abstract

A handful of biological proton-selective ion channels exist. Some open at positive or negative membrane potentials, others open at low or high pH, and some are light activated. This review focuses on common features that result from the unique properties of protons. Proton conduction through water or proteins differs qualitatively from that of all other ions. Extraordinary proton selectivity is needed to ensure that protons permeate and other ions do not. Proton selectivity arises from a proton pathway comprising a hydrogen-bonded chain that typically includes at least one titratable amino acid side chain. The enormously diverse functions of proton channels in disparate regions of the phylogenetic tree can be summarized by considering the chemical and electrical consequences of proton flux across membranes. This review discusses examples of cells in which proton efflux serves to increase pH_i, decrease pH_o, control the membrane potential, generate action potentials, or compensate transmembrane movement of electrical charge.

DISTINCTIVE PROPERTIES OF PROTON CHANNELS

Proton channels, defined in a strict sense as passive, gated, proton-selective proteinaceous pathways across cell membranes, exist in several varieties (Table 1). A much larger group of biological molecules includes proton pathways that are essential to their functions as proton transporters, pumps, or enzymes involved in bioenergetics or photosynthesis. Space limitations allow only superficial mention of this larger group here. This review aims to focus on what is known about the general features of proton-transporting molecules, specifically how the properties of protons are exploited by molecules whose business involves proton translocation. With respect to the proton pathway itself, understanding a voltage-gated proton channel, H_V, is informed more by learning about cytochrome c oxidase than about other voltage-gated ion channels, despite the much stronger structural similarity to the latter. Proton conduction has important applications beyond biology, some of which promise potential solutions to our existential crisis of climate change: proton-conducting batteries (1), solid electrolytes in electrochemical devices of energy storage and production such as fuel cells (2), electrolysers, and carbon dioxide (CO₂) converters (3). An efficient fuel cell must conduct a specific ion, e.g., a proton, rapidly and highly selectively (4), which could equally be said of proton channels. An exhaustive list of known or proposed functions of proton channels is not attempted; instead, the general consequences of proton flux across membranes are discussed and illustrated by a few specific examples.

Table 1.

Types of proton-selective ion channels

Name	Location	Structure	PDB ID	Gated by	References
HV	Widespread	2 × 4TM	3WKV	+Voltage, ΔpH	34, 168, 169
M2	Influenza virus	4 × 1TM	3BKD, 2RLF	Low pH	27, 170, 171
OTOP1, OTOP2, OTOP3	Widespread	2 × 12TM	6NF4, 6NF6, 6O84	Low or high pH, Zn2+	7, 11, 16, 17, 172
Channelrhodopsin-1, channelrhodopsin-2, Chrimson	Green algae	7TM	3UG9	Light	20, 22, 129, 173
TMEM175	Lysosomes	4 × 6TM	5VRE, 6SWR	Low pH	5, 174, 175
HCNL1	Zebrafish sperm	6TM		−Voltage	23
STING	Golgi	2 × 4TM	6NT7	Unknown	24

Structure gives oligomer status (2× is a dimer, 4× is a tetramer) and the number of TM domains in each monomer. The PDB ID is the first or a representative structure; hundreds of structures exist for M2. Many other channels conduct protons measurably but mostly less selectively (cf. 176, table 1).

Abbreviations: HCNL1, HCN-like-1; H_V, voltage-gated proton channel; OTOP, otopetrin; PDB ID, Protein Data Bank ID; TM, transmembrane.

Several types of biological proton-selective channels are listed in Table 1. There are voltage-gated proton channels [H_V and HCN-like-1 (HCLN1)], non-voltage-gated channels activated by low pH in viral membranes (M2) or lysosomes (TMEM175), channels active at low or high pH in vertebrates and invertebrates (OTOP), light-gated channelrhodopsins in algae, and a proton leak pathway in Golgi (STING). Most of these channels are extremely selective for protons. The expectation is that a functioning proton-selective channel must have a relative permeability to H⁺ over other physiological cations of 10⁶ or greater, because the relative concentrations of H⁺ versus K⁺ or Na⁺ differ by roughly that amount. Selectivity lower than this would mean protons permeate less often than other ions. An exception is TMEM175, a risk factor for Parkinson’s disease that was recently shown to comprise a proton leak channel in lysosome membranes. The lysosome lumen is acidified by a proton pump, but pH homeostasis appears to require a constitutive H⁺ leak of 0.1–1 pA through this pathway that is activated when the lumen becomes excessively acidic (5). Despite its relatively weak H⁺ selectivity (P_H/P_K = 4.8 × 10⁴ and P_H/P_Na > 2 × 10⁵), this channel leaks H⁺ selectively at physiological pH because the lysosome pH is 4.5–5.0. In its milieu, the abundance of protons compensates for its moderate selectivity.

The otopetrin family of proton channels is relatively newly discovered, first described in taste cells that detect sour (6), and identified molecularly in 2018 (7). In most vertebrates, the family has three members: OTOP1, OTOP2, and OTOP3 (7, 8). In contrast with the vast majority of H_V channels, whose sensitivity to depolarization and to the transmembrane H⁺ gradient results in exclusively outward proton currents (9), OTOPs are not voltage gated (10) and conduct proton currents in both directions, depending on subtype and species (11–13). Recently it was shown that OTOP3 is proton activated and zinc ion (Zn²⁺) activated, while OTOP2 is active over a large pH range, from pH 5 to pH 10 (11). The situation with OTOP1 is more complicated, as it appears to be both proton activated below pH 5 and to exhibit a separate peak of activation above ~pH 8.5 (11). The inward currents through OTOP1 channels promote activation of taste cells in response to acid stimuli (6),whereas the significance of the outward currents is not known. OTOP1 channels are required for gustatory responses to acid, such as is present in sour foods (14,15).OTOP channels are dimers of protomers with twofold symmetry (C and N domains) that are essentially concatenated dimers, so the final channel is in effect a pseudotetramer (16,17). The central cavity is filled with lipids and thus unlikely to represent a permeation pathway, which may instead be in either the N or C domains or an interdomain interface. Recent evidence indicates that the channels are gated, as opposed to being constitutively open, with opening potentiated by Zn²⁺ for OTOP1 and OTOP3 channels (11,18). The gating apparatus includes the linker between transmembrane domains 11–12 but may be distributed across multiple extracellular linkers (11, 18, 19).

Channelrhodopsins were first identified in green algae in 2002 (20). Their name reflects the observation that the core of the protein is homologous with the light-activated proton pump bacteriorhodopsin (21). These light-activated proton-selective channels have diverse applications in optogenetics (22). Identified quite recently in zebrafish sperm, HCLN1 channels are related to HCN channels but are proton selective and activated by hyperpolarization (23). The authors proposed that proton influx may offset K⁺ efflux through an alkaline-activated K⁺ channel. The most recently identified proton channel is the inflammasome activator STING, which resides in Golgi and conducts protons with unknown selectivity (24).

The most extensively studied channels listed in Table 1 are M2 and H_V. M2 is a viral channel that mediates proton flux into the virion, which results in viral uncoating that enables the viral ribonucleoproteins to invade the host cell nucleus (25). The M2 channel molecule is small, a homotetramer of 96-amino-acid monomers that spans the membrane. A crucial ring of four His³⁷ is thought to mediate both gating and proton selectivity (26–28). The His³⁷ ring occludes the pore, and successive protonation of the four His³⁷ results in electrostatic repulsion that opens the proton pathway. Nuclear magnetic resonance spectroscopy reveals hydrogen bonds and the protonation state of His³⁷, establishing unequivocally that the proton pathway is not a water wire (29). After protonation by hydronium ions in the external solution, a histidine in the ring tautomerizes (flips) to shuttle a proton to the distal side of the pore, ensuring proton-selective conduction (26). This mechanism is reminiscent of the proton entry pathway into carbonic anhydrase II, where His⁶⁴ undergoes an analogous ring flip to propel the proton toward the reaction center (30, 31).

H_Vs are extremely widespread, both phylogenetically (Table 2) and among human tissues. In many of the >550 papers published to date on voltage-gated proton channels, the channel is named H_V1, because only one gene per species had been identified. This name may be preceded by a single letter designating common species (e.g., mH_V1 for mouse or hH_V1 for human) or two letters designating genus and species (e.g., CiH_V1 for Ciona intestinalis H_V1 or DrH_V1 for Danio rerio H_V1). Recently it was demonstrated that the favorite species of neurobiologists, Aplysia californica, has three distinct H_V genes coding for three functionally different channels; hence AcH_V1, AcH_V2, and AcH_V3 now exist (32). Very recently, a fourth variety, CgH_V4, was described in oyster Crassostrea gigas (33). Crucial to all H_V functions is the unique property called 1pH-dependent gating (9). Over a wide range, the voltage at which the channel opens is set by the pH gradient, 1pH [extracellular pH (pH_o) – intracellular pH (pH_i)] with the result that in most species, H_Vs open only positive to the Nernst potential for H⁺, E_H. Consequently, a major function of H_V in most cells is acid extrusion (34). Although there are multiple gating states (9, 35–37), and the mechanism is not completely understood, the closed-to-open transition is governed by both voltage and 1pH (9, 35, 36, 38–41). H_Vs mediate proton action potentials that trigger the bioluminescent flash in dinoflagellates (42–44). They play a critical role in the global carbon cycle by enabling calcification in coccolithophores (45) and perhaps also in reef-building corals (46), a role that is threatened by ocean acidification (47, 48). In humans, H_V is best known for its synergistic relationship with NADPH oxidase (NOX2) and other NOXs (see below); NOX2 produces reactive oxygen species (ROS) by translocating electrons across the membrane (49), which requires both charge and pH compensation that H_Vs are perfectly designed to accomplish (50–53). Other roles of H_Vs include lung epithelial pH regulation (54), human sperm and oocyte function (55–57), and murine insulin secretion (58).

Table 2.

Voltage-gated proton channel genes confirmed by heterologous expression and voltage-clamp

Species	Common name	Abbreviation	Year	Reference
Homo sapiens	Human	hHV1, HsHV1	2006	177
Mus musculus	House mouse	mHV1, MmHV1	2006	178
Ciona intestinalis	Vase tunicate	CiHV1	2006	178
Coccolithus pelagicus a	Coccolithophore	CpHV1	2011	45
Emiliania huxleyi	Coccolithophore	EhHV1	2011	45
Karlodinium veneficum	Dinoflagellate	kHV1, KvHV1	2011	44
Phaeodactylum tricornutum	Diatom	PtHV1	2012	179
Strongylocentrotus purpuratus	Sea urchin	SpHV1	2016	180
Nicoletia phytophila	Parthenogenetic insect	NpHV1	2016	119
Clonorchis sinensis	Liver fluke	CsHV1	2016	181
Lingulodinium polyedrum	Dinoflagellate	LpHV1	2017	43
Biomphalaria glabrata	Snail	BgHV1	2017	182
Danio rerio	Zebrafish	DrHV1	2017	183
Helisoma trivolvis	Snail	HtHV1	2018	184
Suillus luteus	Boletus mushroom	SlHV1	2021	158
Aspergillus oryzae	Kōji mold	AoHV1	2021	158
Acropora millepora	Coral	AmHV1	2021	46
Acropora palmata	Coral	ApHV1	2021	46
Extatosoma tiaratum	Stick insect	EtHV1	2022	120
Aplysia californica	Sea hare	AcHV1	2023	32
Aplysia californica	Sea hare	AcHV2	2023	32
Aplysia californica	Sea hare	AcHV3	2023	32
Crassostrea gigas	Pacific oyster	CgHV4	2023	33

^aCoccolithus pelagicus subsp. braarudii.

UNIQUE PROPERTIES OF PROTON CONDUCTION

Protons exist in aqueous solutions as hydronium ions (H₃O⁺) or in other combinations with water, but they are conducted exclusively as protons (H⁺).Unlike other cations that diffuse around water molecules, protons move through water or proteins by some version of the classical Grotthuss mechanism (59, 60) in which a proton jumps from one water to the next, but crucially the identity oftheprotonthathopsnextmaybeanyofthethreeintheH₃O⁺ molecule. This ambiguity enables the proton to diffuse through water ~5 times faster than any other cation (61). A prerequisite for prototropic conduction is the formation of hydrogen bonds that comprise the conduction pathway (62). Proton conduction through water occurs by a process called structural diffusion in which a proton moves through a network of water molecules (63). Both proton transfer and hydrogen bond rearrangement are necessary for continuous proton transfer to occur (64–66). The rate-determining step in proton conduction in water is thought to be the breaking of a second shell hydrogen bond; first shell hydrogen bonds are too strong (67, 68). In stark contrast, other types of ion channels are simply water-filled pores. All other ions must wait for the water molecules in front of them to diffuse through any single-file region of the channel before they can permeate.

The Problem of the Supply of Protons

Despite the extraordinarily rapid conduction of protons in aqueous solutions, their scarcity limits the rate at which they can be transported across membranes. Figure 1 illustrates the reported unitary conductance for a number of proton-conducting molecules, plotted as a function of the H⁺ concentration. The highly studied gramicidin channel is particularly informative. Gramicidin is a cylindrical pore that is filled with a single-file row of up to a dozen water molecules (69–71). Although it conducts cations nonselectively, it conducts protons much more efficiently than other ions. The H⁺ permeability is two orders of magnitude greater than that of Na⁺, reflecting the unique ability of protons to hop in Grotthuss fashion across this water wire spanning the pore without displacing the water molecules (72). Normal cations must wait for the waters in front to permeate, but protons can hop through an intact water chain.

Figure 1.

Reported single-channel H⁺ currents or turnover rates for various proton-conducting channels or pathways, mostly obtained at symmetrical pH. Lines connect data from each source. Values have been scaled linearly with voltage to a driving force of 100 mV when possible. The dashed red line shows the maximum H⁺ current if diffusion of H⁺ to the channel were rate limiting, assuming a capture radius r_o of 0.87 Å as found for gramicidin (75), several forms of which are illustrated by black symbols. Note that the H⁺ conductance is proportional to [H⁺] over a wide range for several channels that are thought to be water-filled pores. The product of the conductivity of concentrated hydrochloric acid (HCl) at 25°C and its concentration is plotted for comparison after arbitrary scaling. M2 is the proton channel of the influenza A virus; 5-HT is the H⁺-selective leakage current through serotonin receptors; MotA is the flagellar motor torque generator; CF_o is the proton channel component of H⁺-ATPase; BRC is electron-coupled proton transfer rates for both proton transfer steps in the Rhodobacter sphaeroides bacterial reaction center; F_o is H⁺ flux through the F_o proton channel of H⁺-ATPase in vesicles; H⁺ flux through the R371H mutation of the Shaker K⁺ channel voltage sensor is shown (166); cytochrome C (cytC) is proton uptake by cytochrome c oxidase; and (LSLLLSL)₃ and (LSLBLSL)₃ are 21-amino-acid synthetic channels that conduct protons. CA2, CA3, and K64H are turnover rates of carbonic anhydrase variants; the hydration/dehydration of carbon dioxide (CO₂) requires proton translocation along a short water wire and is accelerated greatly by a histidine shuttle at the entrance to the reaction center. The voltage-gated proton channel (H_V1) values are from noise measurements in inside-out patches of human eosinophils. Sources of the data found in Reference 34 (figure 13). Figure adapted with permission from reference 34; copyright 2003 Am. Physiol. Soc.

The single-channel H⁺ current through gramicidin is the largest for any normal ion channel, up to 2 × 10⁹ H⁺/s (73). Gramicidin can function at pH <0, so its proton conductance has been studied over a vast concentration range of >10⁵ (pH 4.5 to −1); over this entire range the unitary H⁺ current is almost directly proportional to concentration. This conductance is close to that predicted by an equation that gives the access resistance for diffusion of protons to a pore in a planar membrane surface (34, 74). As shown in Figure 1, unitary currents through H_V channels and a number of other membrane proteins that transport protons are an order of magnitude or more above the diffusion limit, although these values are measured within a more physiological range of pH. This larger-than-expected flux raises the question of how the protons are supplied at a sufficient rate. Several mechanisms likely contribute. The presence of negatively charged groups located in or near the mouth of the pore can attract protons to the pore entrance electrostatically (75–78). It was proposed recently that Asp¹⁷⁴ comprises a fixed negative charge in the inner vestibule of hH_V1 and increases the capture radius 2.4-fold (79) compared to the 0.87-Å value assumed in Figure 1. The presence of buffer in the solution increases the availability of protons, but not by as much as one might expect (34); the main benefit of buffer is to restore protons depleted by proton flux across membranes. Another mechanism that increases the supply of protons is the antenna effect. By providing an efficient pathway for protons, titratable anionic groups, including the anionic headgroups of membrane phospholipids (80,81),can accelerate proton diffusion drastically at the membrane surface or the surface of a protein (82) compared to diffusion through bulk water. The combination of these and other factors is evidently sufficient to account for the data.

The ability of H_Vs to conduct protons at a high rate is vitally important in many cells that need to translocate protons in vast quantities, including coccolithophores (48) and human phagocytes (52). When large proton currents cross membranes, there is a strong tendency for the proton concentration (pH) to change, especially when there is diffusion limitation in one or both compartments. The latter occurs in cells with relatively small volumes. A universal problem that occurs when voltage-clamping small cells is that large proton currents tend to decay or droop (35,83,84). For example, the maximal H⁺ current in a rat alveolar epithelial cell dialyzed with 5 mM buffer depletes cytoplasmic protonated buffer at a rate of 1.5 mM/s (85). Beyond any aesthetic objection to the appearance of the currents, this phenomenon is a direct indication that proton efflux during outward currents removes intracellular protons and profoundly increases pH_i. Cytoplasmic proton depletion attenuates H⁺ currents by four mechanisms: (a) decreasing the unitary conductance (86), (b) shifting E_H positively, thus lowering the driving force (V-E_H), (c) slowing channel opening, and (d) shifting the g_H-V relationship positively so fewer channels are open. That H_Vs can change pH_i within seconds reflects their capacity to translocate protons at almost any rate cellular metabolism produces them. Because the voltage dependence and gating kinetics of H_Vs depend strongly on both pH_o and pH_i (9, 41), pH_i changes during the measurement are undesirable. To minimize such changes, experimentalists often use buffers at high concentration,100 mM or more, which ameliorates but does not prevent the phenomenon. A systematic study showed that extracellular buffer concentration has little effect on proton currents in whole-cell voltage-clamp studies, but intracellular buffer is critical. Lowering the intracellular buffer concentration from 100 to 10 to 1 mM progressively decreased the current measured during a given depolarizing pulse, and also distinctly changed the waveform, giving the appearance that the proton supply perceptively decreased during the pulse (87). This phenomenon can be minimized but not eliminated, and this limitation must be kept in mind when evaluating subtleties of proton current kinetics.

How Is Proton-Selective Conduction Achieved?

Because protons exist in biological solutions at ~10⁶ lower concentration than other biological cations, a proton channel must be extremely selective (by a factor >10⁶) in order to conduct protons rather than other, more mundane ions. Proton conduction in water or proteins occurs along a pathway comprised of hydrogen bonds (34, 62, 88, 89). Nagle & Morowitz (65) proposed that protons could cross a membrane selectively through membrane proteins via a hydrogen-bonded chain, a continuous string of hydrogen-bonded groups composed of any combination of protonatable side chains and water molecules (65, 66). Comparison of a number of proton-permeable channels or pathways through membrane proteins reveals that high proton selectivity results when at least one titratable side chain of an amino acid forms an obligatory part of the pathway (34); channels consisting entirely of water wires, such as gramicidin, are not highly selective. Water appears to hop readily through the aquaporin water channel, which excludes protons, likely due to an electrostatic barrier (90). A recent study concluded that water permeation through H_Vs is negligible (79). Intriguingly, aquaporin can be converted into a moderately proton-selective channel by mutation (91).

The idea of achieving proton selectivity when at least one titratable side chain of an amino acid lining the pore is an obligatory part of the pathway has numerous precedents. Protonatable amino acids that are critical for the functionality of proteins containing proton pathways include His⁶⁴ of carbonic anhydrase II (92); Asp⁶¹ of the F_o component of H-ATPase (93); His³⁷ of the M2 viral proton channel (26); Asp³² of MotB in the Escherichia coli flagellar motor (94, 95); Glu³²⁵ of lactose permease of E. coli (96, 97); Glu⁹⁰ of channelrhodopsin (98); Asp⁸⁵ and Asp⁹⁶ of bacteriorhodopsin (99, 100); His^H126, His^H128, Asp^L210, Asp^M17, Asp^L213, Ser^L223, and Glu^L212 of the Rhodobacter sphaeroides reaction center (101–103); Asp¹³² and Glu²⁸⁶ of the D channel (104, 105); and Lys³⁶² and Glu¹⁰¹ of the K channel of cytochrome c oxidase of R. sphaeroides (106). Wraight (89) discusses several other examples.

Nevertheless, several mechanisms have been proposed by which an aqueous pathway might be proton selective. In the frozen water hypothesis a water wire could potentially be proton selective if at some point the waters were prevented from permeating (107, 108) or if charged groups in the pore could somehow selectively exclude all ions except H⁺ (109). The frozen waters still need to be sufficiently mobile to reorient during proton conduction (65). One example of a proton channel that lacks titratable groups is a synthetic channel constructed with leucine and serine (110). This channel conducts protons, but high selectivity has not been demonstrated. Another suggestion is that proton selectivity might result though a water wire present only very transiently; protons might hop across rapidly while bulkier and more sluggish ions might fail to permeate. The introduction of a proton appears to induce water entry into hydrophobic pores, so H⁺ may be able to create its own conduction pathway (111). A final possibility is a water wire in which the delocalized protonic charge is distributed among multiple waters, just as Voldemort distributed his soul among multiple horcruxes in the Harry Potter book series. In a pore that is electrostatically hostile to all cations, delocalization of the protonic charge might enable a proton to squeeze through (112). A large and rapidly expanding literature exists on synthetic nanopores that conduct protons rapidly and selectively (e.g., 113, 114), but whether the extreme selectivity required for biological conditions exists in these structures is difficult to discern. Searching a structure for presumptive proton pathways often begins with identifying potential water wires (e.g., 17). In this context, it is instructive that the hydration profiles determined by molecular dynamics simulations (i.e., the averaged water occupancy along the pore) were indistinguishable among mutant hH_V1 constructs that were proton selective, anion selective, or nonconducting (115).

What Mechanism Produces Perfect Proton Selectivity in H_Vs?

The selectivity of H_Vs is essentially perfect; no evidence exists for the permeation of any other physiological ion. Among more than 200 different H_V1 mutants generated and studied by 2016 (116), H⁺ selectivity was compromised only when Asp¹¹² (in the human H_V1, hH_V1) was mutated (117). Identical effects are seen with mutations of the correspondingly located Asp⁵¹ in Karlodinium veneficum kH_V1 (44), Asp¹⁶⁰ in CiH_V1 from C. intestinalis (118), or Asp⁶⁶ in NpH_V1 from the parthenogenetic insect Nicoletia phytophila (119). The conservative Asp→Glu mutation preserves the carboxyl group and also H⁺ selectivity in hH_V1 (117), kH_V1 (44), and NpH_V1 (119); intriguingly, a recently reported stick insect channel EtH_V1 has a naturally occurring Glu at this position (120). Replacing Asp with almost any other amino acid not only eliminates the H⁺ selectivity, but changes the charge selectivity from cation to anion (117). Finally, replacing Asp¹¹² in hH_V1 with the extremely hydrophobic amino acids Val or Ile eliminates current altogether (117, 121). It is clear that this Asp in the middle of the S1 helix is crucial to proton selectivity. However, mutating Asp¹⁸⁵ that resides nearby within the hH_V1 pore does not compromise H⁺ selectivity (117). Therefore, and not surprisingly, something about the specific location of Asp¹¹² is also involved. This question was explored by neutralizing Asp¹¹² (replacing it with Ala or Val) and then introducing Asp successively at each location along the S1 helix from position 108 through 118 (122). Proton-selective current was seen only when Asp was moved to position 116. Gating kinetics, voltage dependence, and 1pH dependence were similar to wild-type (WT) channels. Molecular dynamics simulations of a homology model for the open hH_V1 channel indicated that Asp¹¹² interacts almost continuously with Arg²⁰⁸ in S4 of the WT channel (115). In mutants with Asp shifted from 112 to 116, the Asp still invariably preferred to engage in interactions with Arg in the S4 helix (Arg²⁰⁵ or Arg²⁰⁸ or both) (122). Other groups using somewhat different homology models also observe frequent interaction of the key Asp with Arg (123, 124). The importance of the interaction is suggested by electrostatics calculations showing that the energetic cost of moving a positive point charge through the pore is more than twice greater when the salt bridge is broken than when it is intact (122). A quantum mechanics/molecular mechanics study concluded that the proton pathway through hH_V1 includes transient protonation of four successive acidic groups, connected by intervening waters (125). Stated differently, Asp¹¹² allows the proton to overcome the electrostatic barrier posed by cationic Arg in the S4 helix (126).

This combined evidence points toward the Asp-Arg interaction as determining proton selectivity. To confirm that this mechanism is chemically plausible, quantum mechanical simulations were performed on a reduced model of the H_V selectivity filter (127). The Asp-Arg interaction results in two hydrogen bonds that occlude the channel (Figure 2). Introducing a proton in the form of a hydronium ion, H₃O⁺, results in protonation of the Asp, with the now-neutral water molecule positioned between and interacting with both side chains. Essentially, the proton parks its charge on Asp, then permeates by hopping onto a neutral H₂O, which might or might not be the same one it arrived on, completing the permeation cycle. When Na⁺ or Cl⁻ were positioned at the selectivity filter, they did not break the hydrogen bonds, but instead associated with their opposing charge on Asp or Arg, respectively. Although the model is simplified, it demonstrates that a correctly positioned charge pair may act as a selectivity filter, permitting permeation of protons while excluding other ions (127). Other molecules likely use this arrangement for a similar purpose. In the crystal structures of several other proteins, Asp-Arg pairs that are essential to the function of the molecule were located within 0.1 Å of being 4 Å apart, as they are in hH_V1. Similarly, in lactose permease, Arg³⁰² and Glu³²⁵ interaction is critical (97, 128). The authors proposed that charge pairs may comprise a structural motif that enforces selective proton conduction in many molecules (127).

Figure 2.

Proton-selective conduction in the voltage-gated proton channel (H_V1) occurs because the hydronium ion (H₃O⁺) is uniquely able to open its own conduction pathway by breaking the aspartic acid–arginine (Asp–Arg) connection in the selectivity filter. (a) A reduced model of the human H_V1 (hH_V1) selectivity filter comprises the side chains of Asp and Arg mounted on a ring scaffold (not shown) separated by 4 Å, the distance predicted by a homology model (115). All electron quantum mechanical calculations reveal that Asp and Arg interact via two hydrogen bonds (dashed lines), in two stable optimized configurations with Asp (upper) or Arg (lower) protonated; these configurations differed by <1 kcal/mol. Introducing H₃O⁺ into either starting configuration results in protonation of Asp, yielding a neutral water molecule that mediates interactions between side chains. Computed ΔG values are negative, indicating a favorable forward reaction. (b) Schematic of the proposed proton-selective conduction mechanism of H_V1. Negatively charged Asp is orange, neutral AspH⁰ and H₂O⁰ are green, and positively charged H₃O⁺ and Arg are light or dark blue, respectively. The dashed lines denote hydrogen bonds that occlude the pore. When H₃O⁺ approaches the selectivity filter (left), it breaks the hydrogen bonds and protonates the selectivity filter, resulting in neutral H₂O bridging AspH⁰ and Arg⁺ (middle). Transfer of a proton from the selectivity filter to H₂O completes the conduction cycle (right). Panel adapted with permission from Reference 127.

What Mechanisms Produce Proton Selectivity in Other Proton Channels?

The mechanism of proton conduction in OTOP is not resolved. There appear to be three conduction pathway candidates, identified as hydrated pores in the cryo-electron microscopy structures, one each in the C or N domains and a third located at the intersubunit interface (15, 16). These pathways might function independently or cooperatively. A salt bridge that might support proton conduction has been identified in the C domain. Specifically, mutating either member of the Glu⁴³³–Arg⁵⁸⁶ salt bridge (E433R or R586E) in mouse mOTOP1 impaired function, and the charge-swapped double mutation (E433R/R586E) rescued function (17), consistent with the selectivity mechanism proposed for H_V1 (127). The putative pathway through the intersubunit interface contains two conserved charged residues, the mutation of either of which abolished conduction without impairing trafficking: Glu²⁶⁷ and His⁵⁷⁴ in zebrafish DrOtop1 (17) and Glu³²¹ and His⁶⁶⁹ in Xenopus tropicalis XtOtop3 (16). Whether these residues are required for gating or permeation is not clear at this point. A structure of the channel in the open state might help resolve this question.

The proton selectivity of channelrhodopsin-2 is due to Glu⁹⁰ (22). Intriguingly, mutating this Glu to Lys changes the channel from proton to anion conducting (98), reminiscent of the Asp¹¹² mutants of hH_V1 (117). In channelrhodopsin Chrimson, proton selectivity is attributable to two Glu that may interact with Arg¹⁶² (129).

Proton selectivity in M2 viral proton channels is the result transient protonation of the ring of four His³⁷ residues (130) and thus differs qualitatively from the selectivity mechanisms of H_V1, OTOP, or channelrhodopsins.

FUNCTIONS OF PROTON CHANNELS

A detailed list of known or proposed functions of the various proton channels discussed here is far beyond the scope of this review and would be tedious in any case. However, elaborating on a previous effort (131), several general categories of functions can be described, and several examples are given for each. Because H_Vs have been studied extensively and are abundantly expressed, both in numbers of cell types and the phylogenetic range, and given the extremely varied protein varieties found in diverse species, these illustrations derive mainly from this channel. Analogous functions likely apply to other types of proton channels. These biological functions reflect the direct physical consequences of protons crossing cell membranes, namely chemical effects and electrical effects. Chemical effects consist of changes in proton concentration: when protons leave a cell, pH_i increases and pH_o decreases. The electrical consequences are of three varieties, namely regulating the membrane potential, generating action potentials, and charge compensation. Examples of cells in which each is important are discussed.

Increasing pH_i

A widespread and general function of H_V channels is acid extrusion, which occurs because the pathognomonic dependence of channel opening on the pH gradient, ΔpH = pH_o – pH_i, results in the channel opening almost exclusively when ΔpH is directed outwardly (9). Thus, under physiological conditions, whenever the H_V channel opens, it extrudes protons, increasing pH_i. The channel may be driven to open when metabolic activity generates large quantities of intracellular protons. A prime example of this is the phagocyte respiratory burst in which ROS generation by NOX2, the phagocyte NADPH oxidase, rapidly lowers pH_i (Figure 3). Unless both H_Vs and Na⁺/H⁺ antiport are activated, cytoplasmic acidification reaches extreme levels sufficient to inhibit NOX activity directly (53, 132). Another situation in which maintaining normal pH_i is important is the secretion of histamine by basophils (133). In an example that has implications for our survival of climate change, H_V channels maintain pH_i at levels conducive to the calcification process in coccolithophores (45), single-celled marine organisms that play a crucial role in atmospheric CO₂ handling (47, 48). OTOP channels play a similar role in enabling biomineralization in sea urchins (12).

Figure 3.

Stoichiometrically accurate graphic of the participation of protons and especially H_V1 in the respiratory burst of phagocytes. The tan double lines represent the plasma membrane of a leukocyte in the process of engulfing a bacterium. The NADPH oxidase complex assembles in the phagosome membrane and begins to produce superoxide anion, from which other reactive oxygen species are derived. Red solid arrows indicate proton pathways, while red dashed arrows indicate proton pathways that are either questionable or not stoichiometrically accurate. Blue arrows indicate reactive oxygen species pathways. Abbreviations: CIC-3, chloride channel 3; HMS, hexose monophosphate shunt; H_V1, voltage-gated proton channel; MPO, myeloperoxidase. Figure adapted with permission from Reference 167; copyright 2010 Am. Physiol. Soc.

There are also situations in which the cell needs pH_i to increase above its normal value for the purpose of signaling. One example is human sperm in which capacitation is triggered by an increase in pH_i mediated by H_V activation (55).

Decreasing pH_o

In many situations, acid excretion by a cell will have a negligible effect on pH_o because the extracellular volume is effectively infinite. However, when the proton channel is located in a membrane facing a confined space (134),such as an organelle membrane like a phagosome, acid extrusion will dramatically alter pH_o (52, 135). Airway epithelial cells regulate surface pH_o by excreting H⁺ into the airway lining fluid (54, 136). Another function of H_Vs in some cells may be to extrude acid indirectly in the form of CO₂, as may occur by facilitated diffusion of H⁺ and HCO₃⁻ (137–139).

There are several hints that H_V channels may play a role in cancer that may be related to the general propensity of tumor cells to acidify their microenvironment by using glycolysis preferentially, called the Warburg effect (140, 141). A short form of hH_V1 lacking the first 20 amino acids is highly expressed in a number of B cell malignancies and exhibits functional differences from the full-length protein (142). Several studies suggest an inverse relationship between H_V expression and cancer severity, metastasis, and progression, as well as lower tumor growth with H_V knockout (143–146). Cancer is extremely complex, however, and we are far from any definitive statement.

Regulating Membrane Potential

Several groups have proposed that H_Vs may serve to regulate membrane potential. In order to modulate membrane potential significantly, H_Vs must be the dominant conductance. This situation has been shown to occur in phagocytes during the respiratory burst (52,147).OTOP channels in taste receptor cells open during sour taste transduction, depolarizing the membrane potential (6, 14).

Generating Action Potentials

This function was proposed for H_V channels in bioluminescent dinoflagellates (148) when their existence was first imagined. In 1972, Hastings synthesized an array of experimental data to postulate the existence of a depolarization-activated proton-selective ion channel in the membranes of scintillons (42), the light-emitting organelles formed by evagination of the tonoplast (vacuole) membrane (149, 150). Hastings and colleagues had shown that the scintillons contain three components needed for bioluminescence: a light-emitting luciferin compound, a luciferin-binding protein that releases luciferin at low pH, and a luciferase that is activated by a decrease in pH (149, 151–153). A putative voltage-gated proton channel would open during the action potentials that had been recorded in the tonoplast membrane (150, 154, 155). The tonoplast encloses a large vacuole with pH ~3.5, and the scintillon contains normal cytoplasm with high pH closer to that of seawater, 8.0–8.2 (156). Given this enormous concentration gradient, opening a proton-selective channel would allow rapid H⁺ influx, activating the luciferase. A dinoflagellate proton channel gene product kH_V1 was identified in K. veneficum in 2011 (44), and an H_V in a bioluminescent species, Lingulodinium polyedrum, was identified in 2017 (43), unequivocally confirming Hastings’ prediction. At first, the proton channel was proposed to be activated (opened) by the action potential, triggering the flash. That the H_V itself might conduct the action potential was considered later (157). The identification of an H_V in dinoflagellates with electrophysiological properties consistent with conducting an action potential (43, 44) strongly supports this idea.

To mediate an action potential, a voltage-gated channel must carry inward current; hence, a proton channel must be activated at voltages negative to the Nernst potential for H⁺, E_H. This property is contrary to the defining feature of the vast majority of H_Vs that activate just positive to E_H and carry only outward current (9). A few H_V channels do activate in a sufficiently more negative range: kH_V1 (44), AoH_V1 (158), AcH_V2 (32), and EtH_V1 (120). Although LpH_V1 in a bioluminescent dinoflagellate does not activate negative to E_H at symmetrical pH, it nevertheless likely mediates proton action potentials that trigger the flash because it resides in the tonoplast membrane with a gargantuan 4.5-unit inward pH gradient and would readily conduct inward proton current under these extraordinary physiological conditions (43).

The OTOP channels are not voltage gated, but they do carry inward current in response to acid stimuli. Their activation is thought to depolarize the plasma membrane of taste cells, triggering an action potential that is mediated by voltage-gated Na⁺ channels (6, 14).

Charge Compensation

The phagocyte respiratory burst illustrates four of the main functions of proton channels, most particularly charge compensation (Figure 3). NADPH oxidase or NOX2, the enzyme that produces ROS, is electrogenic (49), extracting electrons from intracellular NADPH and shuttling them across the membrane to reduce molecular oxygen, O₂, to superoxide anion, O_{2^._}. The efflux of electrons can be measured directly as an electron current (51) that turns off completely at ⁺190 mV (50). During phagocytosis, the NOX2-generated electron efflux rapidly depolarizes the phagosome, whose membrane potential would reach +190 mV within ~20 ms and turn NOX2 off (50) if it did not simultaneously activate proton channels. The electrophysiological phenomena that occur during the respiratory burst (159–161) are fully and quantitatively explained by the symbiotic biophysical interaction between NOX2 and H_V channels (52, 162, 163). NOX2 activity depolarizes the membrane and acidifies the cytoplasm, both of which activate H_V channels. H_V activity counteracts both consequences of NOX2 activity by compensating charge, thus limiting depolarization, and extruding acid, thus minimizing pH changes. In addition, H⁺ efflux also provides substrate protons needed for the generation of ROS (e.g., H₂O₂ and HOCl) inside the phagosome and prevents the explosive swelling of the phagosome that would occur if charge were compensated by other cations (52).

Besides supporting ROS production by NOX2 in phagocytes for the purpose of killing phagocytosed microbes, H_V also facilitates ROS production for signaling purposes in other cells. For example, ROS produced by NOX5 in human sperm enhances their motility (56), and ROS produced by NOX2 in human B lymphocytes amplifies B cell receptor signaling pathways signaling (164).

CONCLUSIONS AND FUTURE STUDIES

Although some proton-selective ion channels have been studied extensively in recent decades, new varieties are still being discovered. Recent work on the viral M2 (and similar BM2) proton channels has focused on developing inhibitors that can control influenza. Channelrhodopsins, which are studied to better understand their mechanisms, have been considered for a number of optogenetic applications as photosensors (98, 165). Several major properties of H_V channels remain incompletely understood, some of which are controversial. Hundreds of mutations have been produced and studied, illuminating many of the mechanisms that define and are critical to the functions of H_V channels (116).Voltage-gated H_V channels of remarkably variable amino acid sequences are being identified in remote corners of the phylogenetic tree. By careful characterization of these channels already pretested by evolution, we can approach structure-function studies in a manner strategically different from the traditional method of generating mutations based on theory, models, and intuition. The newer members of the proton channel family—OTOPs, HCNL1, and STING—provide a vast array of questions. All proton-conducting channels share features that result from the unique chemical properties of protons.