The ability to discriminate between different ionic species, termed ion selectivity, is a key feature of ion channels and forms the basis for their physiological function. Members of the degenerin/epithelial sodium channel (DEG/ENaC) superfamily of trimeric ion channels are typically sodium selective, but to a surprisingly variable degree. While acid-sensing ion channels (ASICs) are weakly sodium selective (sodium:potassium ratio ∼10:1), ENaCs show a remarkably high preference for sodium over potassium (>500:1). This discrepancy may be expected to originate from differences in the pore-lining second transmembrane segment (M2). However, these show a relatively high degree of sequence conservation between ASICs and ENaCs, and previous functional and structural studies could not unequivocally establish that differences in M2 alone can account for the disparate degrees of ion selectivity. By contrast, surprisingly little is known about the contributions of the first transmembrane segment (M1) and the preceding pre-M1 region. In this study, we used conventional and noncanonical amino acid–based mutagenesis in combination with a variety of electrophysiological approaches to show that the pre-M1 and M1 regions of mASIC1a channels are major determinants of ion selectivity. Mutational investigations of the corresponding regions in hENaC show that these regions contribute less to ion selectivity, despite affecting ion conductance. In conclusion, our work suggests that the remarkably different degrees of sodium selectivity in ASICs and ENaCs are achieved through different mechanisms. These results further highlight how M1 and pre-M1 are likely to differentially affect pore structure in these related channels.
Acid-sensing ion channels (ASICs) and epithelial sodium channels (ENaCs) are members of the degenerin (DEG)/ENaC superfamily of trimeric ion channels and play important roles in neurotransmission and salt homeostasis, respectively. The six human ASIC isoforms known to date (1a, 1b, 2a, 2b, 3, and 4) can form homo- or heterotrimeric proton-gated ion channels and are found primarily in the nervous system, where they contribute to the depolarization of post-synaptic neurons. They are implicated in numerous physiological and pathophysiological processes, including in nociception (Chen et al., 2002; Duan et al., 2007; Mazzuca et al., 2007), learning and memory (Wemmie et al., 2002), fear (Ziemann et al., 2009), and neuronal death in ischemic stroke (Wang et al., 2015; Xiong and Xu, 2012). ENaCs comprise a family of related channels formed by three homologous subunits, α/δ, β, and γ, but the canonical channel formed by the α, β, and γ subunits is the most physiologically understood isoform (Haerteis et al., 2009). Canonical ENaCs are located in the apical membrane of epithelial cells, particularly in the kidney, colon, and lung, where they contribute to the control of body electrolyte and water homeostasis as well as the volume and composition of lung epithelial lining liquid, respectively (Hanukoglu and Hanukoglu, 2016; Rossier et al., 2015).
Both ASICs and ENaCs are voltage-insensitive channels and share a similar overall subunit topology with intracellular N- and C-termini, two membrane-spanning helices (M1 and M2), and a large extracellular domain (Cheng et al., 2018; Hanukoglu and Hanukoglu, 2016). They are inhibited by amiloride and contain sodium-selective pore modules, although the degree of Na+/K+ permeability greatly varies from ∼10:1 in ASICs (Lingueglia et al., 1997) to >500:1 in ENaCs (Kashlan and Kleyman, 2011; Palmer, 1982). Functional and structural evidence suggests that in both channel types, the ion-conducting pore is lined by M2 (Jasti et al., 2007; Lynagh et al., 2017; Noreng et al., 2018; Schild et al., 1997). Interestingly, this is also the region that displays the highest level of sequence conservation between ASICs and ENaCs. As such, the molecular basis for the pronounced discrepancy in sodium selectivity between the two types of channels remains unclear. In fact, both channel types contain a highly conserved G/SXS motif situated roughly in the middle of the M2, which had been proposed to form a size-exclusion filter (Jasti et al., 2007; Kellenberger et al., 1999a; Palmer, 1985). Previous work indeed suggested that the selectivity filter of ENaCs is formed by the S of the G/SXS motif (Kellenberger et al., 1999a), and most ASIC structures available to date display a marked constriction at the level of this highly conserved M2 sequence feature (Baconguis et al., 2014; Yoder and Gouaux, 2020). However, we have recently shown that the ASIC GAS sequence is unlikely to make a major contribution to ion selectivity in ASIC1a and ASIC2a, while acidic side chains in the lower M2 appear to be more important (Lynagh et al., 2020; Lynagh et al., 2017). Additionally, the latest ASIC structures show that the lower pore is lined by a reentrant loop formed by side chains from the N-terminus (Yoder and Gouaux, 2020), and this pre-M1 region has previously been implicated in ion selectivity in ASICs (Coscoy et al., 1999; Pfister et al., 2006), ENaCs (Gründer et al., 1999; Kellenberger et al., 2005) and the related MEC-4/MEC-10 channel (Arnadóttir et al., 2011). Unlike for ASICs, we lack reliable structural information on the pore module of ENaCs, and mutations to conserved acidic side chains in the lower M2 did not have a major impact on ion selectivity (Yang and Palmer, 2018). Overall, most previous studies have focused primarily on M2 with regard to contributions to ion selectivity; yet, little is known about the contribution of other parts of the pore module, such as M1 and the preceding pre-M1 region. Here, we therefore set out to conduct a comparative analysis of the potential contribution of M1 and pre-M1 to ion selectivity in mouse ASIC1a (mASIC1a) and human ENaCs (hENaCs).
We employed a combination of site-directed mutagenesis and noncanonical amino acid (ncAA) incorporation in whole-cell, single-channel, and high-throughput electrophysiological recordings. Our data suggest that two aromatic residues in ASIC1a M1 play an important role in maintaining a ∼10-fold Na+/K+ permeability ratio. Removal of the bulk of these side chains resulted in a marked decrease in relative Na+/K+ and Na+/Cs+ permeability ratios, likely by enlarging the pore diameter. Additionally, single amino acid substitutions of multiple residues in ASIC1a pre-M1 abolish ion selectivity. In contrast, neither the M1 nor the pre-M1 side chains appear to be important for ion selectivity in hENaC, but single-channel recordings suggest that some side chains may play a role in ion conduction. Together, our results reveal important yet different roles of the M1 and pre-M1 segments in ASICs and ENaCs, possibly due to differences in their lower pore structures. This consolidates the notion of fundamentally different mechanisms by which these related channel types achieve sodium selectivity.
Materials and methods
NaCl, MgCl2, KCl, BaCl2, CsCl, HEPES, EGTA, D-glucose, amiloride hydrochloride hydrate >98%, methylammonium chloride, dimethylamine hydrochloride, ethylammonium chloride, diethylamine hydrochloride, triethylammonium chloride, tetraethylammonium, propylamine hydrochloride, and tetrapropylammonium chloride were purchased from Sigma-Aldrich. Dipropylamine hydrochloride was purchased from TCI Chemicals.
mASIC1a cDNA, cloned between the BamHI and SacI restriction sites of the pSP64 vector, was a gift from Dr. Marcelo Carattino (University of Pittsburgh, Pittsburgh, PA). hENaC α, β, and γ subunits in the pTNT vector were a gift from Dr. Diego Alvarez de la Rosa (University of La Laguna, San Cristóbal de La Laguna, Spain). The α subunit contained the T334 and A663 polymorphisms. The construct referred to as WT hENaC contains C-terminal truncations in the β and γ subunits (β_R566STOP and γ_K576STOP) to increase expression. Site-directed mutagenesis was performed using custom-designed primers (Eurofins Genomics) and regular PCR with PfuUltra II Fusion HS DNA Polymerase (Agilent Technologies). Plasmids were linearized with EcoRI (for mASIC1a) and BamHI (for hENaC) and used as a template for synthesis of mRNA with the Ambion mMESSAGE mMACHINE SP6 Transcription Kit (Thermo Fisher Scientific). ENaCα-mASIC1aM1 chimera DNA cloned between the NheI and Xhol restriction sites in the pUNIV vector was purchased from Twist Bioscience.
Incorporation of the ncAA naphthalene
The nitroveratryloxycarbonyl-protected ncAA naphthalene esterified with 5′-O-phosphoryl-2′-deoxycytidylyl-(3′→5′)adenosine was incorporated via the nonsense suppression method (Dougherty and Van Arnam, 2014). Modified Tetrahymena thermophila tRNA was prepared by annealing full-length 5′ and 3′ DNA strands (Integrated DNA Technologies); RNA was synthetized using the T7-Scribe transcription kit (CELLSCRIPT) and purified with Chroma Spin DEPC-H2O columns (Clontech). The ncAA naphthalene was ligated to the tRNA with T4 DNA ligase (New England Biolabs). The aminoacylated tRNA was purified with phenol-chloroform extraction and ethanol purification, air dried, and stored at −80°C until use. The pellet was resuspended in 1 μl water and deprotected with 400-W UV light (Newport UV lamp , including Newport power supply ) immediately before injection into oocytes. The deprotected aminoacylated tRNA was mixed 1:1 with mASIC1a mRNA containing a TAG codon at position 50 and injected into oocytes.
Oocyte preparation and mRNA injection
Stage IV oocytes were extracted from female Xenopus laevis frogs (anesthetized in 0.3% tricaine under license 2014-15-0201-00031, approved by the Danish Veterinary and Food Administration) by surgery and washed thoroughly in OR-2 medium (in mM: 2.5 NaCl, 2 KCl, 1 MgCl2, and 5 HEPES, pH 7.4). The oocytes were digested with type I collagenase (1.5 mg/ml; Roche) in OR-2. Stage V and VI oocytes were manually isolated and stored in OR-2 medium. For the injection, oocytes were lined up in a 35-mm dish containing OR-2. For the microinjection, glass capillaries (1.14 mm, 3.5 in; World Precision Instruments) were pulled using a horizontal puller (P-1000; Sutter Instruments) and filled with mineral oil (Sigma-Aldrich). The tips of the capillaries were broken with fine tweezers to allow RNA uptake. RNA was injected into the oocytes at the oocyte equator (Nanoliter 2010 Injector; World Precision Instruments). RNA was injected in amounts as mentioned in the relevant figure legends, and oocytes were incubated in OR-3 medium (Leibovitz’s L-15 Medium [Gibco] supplemented with 3 mM L-glutamine [Gibco], 0.25 mg/ml gentamicin [Gibco], 15 mM HEPES, pH 7.6) for 1–5 d at 18°C. For ENaC single-channel experiments, ovary lobes were purchased from the European Xenopus Resource Centre, and procedures were approved by the Animal Welfare and Ethical Review Body at Newcastle University (project identifier 630). RNA-injected oocytes were incubated at 16°C in a low-sodium solution as previously described (Wichmann et al., 2019).
Electrophysiological experiments and data analysis
For ion selectivity measurements in hENaC constructs, WT or mutant α, β, and γ subunits were injected in a 1:1:1 ratio with a final amount of 5–15 ng mRNA. For constructs that showed small macroscopic current sizes, the total RNA amount was increased to 50 ng. Oocytes injected with ENaC constructs were incubated in the presence of 100 µM amiloride to avoid cell death due to excess Na+ uptake. For ion selectivity measurements, cells were continuously perfused with bath solution (in mM: 96 XCl, 2 MgCl2, 1.8 CaCl2, and 5 HEPES; “X” represents Na, K, Li, or Cs) containing 100 µM amiloride. Currents were elicited upon switching the solution to one without amiloride. After 1 min, when the currents had reached a steady-state level, a 1-s voltage step protocol was applied, ranging from −140 mV to 40 mV. To determine the amiloride-sensitive current, the protocol was repeated in the presence of 100 µM amiloride. Relative ion permeability ratios were calculated as relative amiloride-sensitive current amplitudes at −100 mV. Data were analyzed in Clampfit 10.7.
Statistical analysis was performed using GraphPad Prism version 7.0 software. An unpaired t test was used to compare two groups. Multiple groups were compared by one-way ANOVA with Dunnett’s multiple comparisons test for comparison with a control group (WT). A P value <0.05 was considered significant. All bars and data points are presented as mean ± SD.
For cell-attached patch-clamp experiments, Xenopus oocytes were injected with 10–50 ng mRNA. After 24–48 h, the oocytes were mechanically devitellinized and placed in a recording chamber containing bath solution (in mM: 145 KCl, 1.8 CaCl2, 10 HEPES, 2 MgCl2, and 5.5 glucose at pH 7.4 adjusted with KOH). Patch pipettes were pulled from borosilicate glass capillaries (6–11 MΩ), heat polished, and filled with pipette solution (in mM: 145 NaCl, 1.8 CaCl2, 10 HEPES, 2 MgCl2, and 5.5 glucose). The pH of the pipette solution was adjusted with HCl and NaOH. Current signals were amplified using an LM-PC patch-clamp amplifier (List-Medical), low-pass filtered at 100 Hz (Frequency Devices), and recorded at 2 kHz with Axon Clampex software (Axon Instruments) using an Axon 1200 interface amplifier. Single-channel analysis was performed with Clampfit version 10.7 (Axon Instruments). Recordings were performed at room temperature. Single-channel amplitudes were plotted as a function of voltage and fitted to a linear curve, and values for slope conductances were determined.
Statistical analysis was performed using GraphPad Prism version 7.0 software. To compare two groups, an unpaired t test was used. Multiple groups were compared by one-way ANOVA with Dunnett’s multiple comparisons test for comparing a control group (WT). A P value <0.05 was considered significant. All bars and data points are presented as mean ± SD.
Cell culturing and transfection of HEK293-T for automated patch-clamp recordings
ASIC1a knockout HEK293-T cells (provided by Dr. Nina Braun; Borg et al., 2020) were grown in monolayers in T75 and T175 flasks (Orange Scientific) in cDMEM (Gibco DMEM supplemented with 10% FBS [Thermo Fisher Scientific] and 1% penicillin-streptomycin [Thermo Fisher Scientific]) and incubated at 37°C in a humidified 5% CO2 atmosphere. Cells were split at near confluency by treatment with trypsin-EDTA (Thermo Fisher Scientific) after the cells were washed with PBS. Cells were counted with an EVE automatic cell counter (NanoEntek) and seeded into 10-cm dishes (Orange Scientific) at a density of 2.0 × 106 cells/dish.
For transient transfections in 10-cm dishes, 7 µg of DNA was mixed with 21 µl of Trans-IT LT1 transfection reagent (Mirus Bio) or LipoD transfection reagent and 1,750 µl DMEM and incubated at room temperature for 20 min to allow formation of DNA–transfection reagent complexes before addition to the cells. Transfected cells were incubated for 24 h.
Transfected cells were harvested by trypsin-EDTA treatment (2 ml/10-cm dish). Detached cells were resuspended in DMEM and centrifuged at 2,000 g for 2 min. The cells were resuspended in a 1:1 mixture of DMEM and Nanion Technologies physiological solution (in mM: 140 NaCl, 4 KCl, 1 MgCl2, 2 CaCl2, 10 HEPES, and 5 glucose, pH 7.4). Automated patch-clamp experiments were performed 24 h after transfection.
Automated patch-clamp recordings
The permeability of ASIC1a to a range of monovalent cations (MA+, methylammonium+; DMA+, dimethylamine+; EA+, ethylamine+; DEA, diethylamine+; TEA, triethylamine+; TetraEA+, tetraethylamine+; PA+, propylamine+; DPA+, dipropylamine+; and TetraPA+, tetrapropylamine) different in size and shape was assessed with automated whole-cell patch clamping on the SynchroPatch 384PE. The experiment was designed such that the intracellular solution (ICS) contained only cations impermeable (or with very low permeability) for ASIC1a, whereas the ECSs contained one of a range of test ions. Solutions used for the automated patch-clamp experiment were as follows: ECS: 140 mM XCl, 4 mM CsCl, 1 mM MgCl2, 2 mM BaCl2, 10 mM HEPES pH 7.4 and 6, 290 mosmol/kg; ICS: 80 mM NMDG2-SO4, 20 mM CsCl, 10 mM HEPES pH 7.2, 270 mosmol/kg.
The SynchroPatch 394PE consists of a Biomek FXP liquid-handling station (Beckman Coulter) and a planar patch-clamp module (Nanion Technologies), which can be controlled simultaneously via a computer terminal using the PatchControl384 and Biomek software (version 4.1). 25 ml of cells expressing hASIC1a WT or hASIC1a F50A (>4.0 × 106 cells) were loaded into a Teflon reservoir. Cells were incubated at 20°C and shaken at 200 rpm in the integrated cell compartment of the SynchroPatch 384PE.
For patch-clamp recordings, 15 µl of cells were loaded into each well of an NPC-384 medium-resistance single-hole plate (Nanion Technologies). The plate contains 384 individual wells with a single patch-clamp orifice where the cells and ECSs are delivered. The design of the plate and the patch-clamp module allows perfusion of the ICS during an experiment. Cells were caught on the patch-clamp holes by applying a brief pressure of −100 mbar, and another pulse of pressure (−150 mbar) was applied to reach whole-cell configuration. Cells were clamped at 0 mV under atmospheric pressure. Seals of 50 MΩ and above were included in the analysis.
ECSs were provided in 12-well Teflon reservoirs, with each column of wells containing an ECS with 140 mM of one of the following cations provided as chloride salts shown in Fig. 1 C. One 12-well dish contained ECSs with resting pH (7.4), and another dish contained ECSs with activating pH (pH 6). ECSs were applied using a liquid stacking approach as described elsewhere (Braun et al., 2021 Preprint). The pipette tips were loaded with 25 µl pH 7.4 solution followed by 15 μl pH 6.0 solution. For determination of I-V relationships, four sweeps were recorded. For each sweep, the baseline current was recorded for 1 s (clamped at 0 mV) before the application of the activating solution, followed by a delayed dispersion with the pH 7.4 solution. This was followed by two washing steps with pH 7.4 solution, after which the protocol was repeated at a different voltage.
Online supplemental material
Fig. S1 contains sequence logos for M1 and M2, along with an overview of the nonsense suppression method. Fig. S4 shows current traces and a plot of ion selectivity versus pH50 of mASIC1a M1 mutants. Fig. S5 illustrates the underlying principle and data averages for automated patch-clamp recordings. Figs. S2, S3, and S6 show sequence alignments. Fig. S7 shows chicken ASIC1 (cASIC1) structures and current traces for selected mASIC1a pre-M1 mutants.
M1-swapped chimeras show altered function
Unlike the highly conserved protein sequences of the pore-lining M2 segments of ASICs and ENaCs, there is noticeable sequence divergence in the M1 segments (Fig. S1, A and B; see Figs. S2 and S3 for further information). We hypothesized that differences in the M1 segments between the two channel types may partially account for the observed difference in Na+/K+ selectivity. To directly test this notion, we swapped the M1 segment of mASIC1a with that of the hENaC α subunit and vice versa (residues A44–F69 in mASIC1a and the equivalent residues in hENaC α) and expressed the resulting chimeras in Xenopus oocytes. The αENaC–ASIC1a (M1) chimeric construct, when coexpressed with WT hENaC β and γ, yielded amiloride-sensitive currents (Fig. 1, A and B) but required injection of 100 ng RNA and 3 d of incubation for detection of currents. Notably, the αENaC-ASIC1a (M1) construct appeared to be inwardly rectifying. Although the small current magnitudes carried by K+ and Cs+ prevented us from accurately determining the effects of the M1 swap, the data tentatively suggest an increased permeability for K+ and Cs+ (IK+/INa+ = 0.09 ± 0.06 and ICs+/INa+ = 0.10 ± 0.10 at −100 mV). By contrast, the Li+/Na+ amiloride-sensitive current ratio was markedly increased (ILi/INa = 4.2 ± 0.06 compared with 1.2 ± 0.19 in WT hENaC at −100 mV).
The ASIC1a–αENaC (M1) chimera formed constitutively active, pH-sensitive channels that underwent severe tachyphylaxis, rendering an accurate analysis of relative ion permeability ratios of this construct infeasible (Fig. 1, C and D). However, the measured reversal potential with Na+ as the primary cation in the ECS was markedly reduced compared with WT mASIC1a, suggesting that the M1 swap reduces Na+ selectivity (Fig. 1 E).
The altered function of these M1-swapped chimeras suggests that the M1 region contributes to ion selectivity in both channel types. This prompted us to investigate the role of M1 side chains in further detail.
Residue swapping in the variable region of M1 does not affect ion selectivity of mASIC1a
Because pronounced tachyphylaxis precluded obtaining detailed information on relative ion permeabilities of the mASIC1a-hαENaC(M1) chimera, we resorted to swapping nonconserved amino acids in the M1 segment of mASIC1a, one at a time, with those found at corresponding positions in hENaC-α (highlighted in red in Fig. 2 A). We expressed WT or mutant channels in Xenopus oocytes and determined ion selectivity by measuring relative ion permeabilities of Na+, Li+, K+, and Cs+. Mutations in the more variable region of M1 (residues C3′L and G6′C-Q20′G) had minimal effects on relative ion permeability ratios (Fig. 2 C, Fig. S4, and Table 1), suggesting that these residues do not play critical roles in ion selection in mASIC1a.
Substitution of aromatic residues in the lower M1 region of mASIC1a affects ion selectivity
Next, we tested the possibility that bulky aromatic side chains in the M1 segments are important in maintaining the structural integrity and/or size of the pore of mASIC1a (highlighted in blue, Fig. 2 A). We found that substitutions of W0′ and F4′ altered ion selectivity markedly and decreased functional expression, particularly for the W0′ position (Fig. 2 C and Tables 1 and 2). Decreasing the side chain volume of W0′ to F or L reduced the Na+/K+ permeability ratio to unity, caused an alkaline shift in the pH50, decreased the rate of desensitization, and induced a constitutive current component (Fig. S4), whereas incorporation of the slightly bulkier ncAA naphthalene with the in vivo nonsense suppression method (Fig. S1 C) did not increase ion selectivity but resulted in an alkaline shift in the pH50 and increased the rate of desensitization (Fig. S4). Thus, the bulk of W0′ appears important for both ion selectivity and gating (Fig. 2, B and C; Table 2; and Fig. S4). Substitution of F4′ for smaller side chains (L and A) reduced ion selectivity, but increasing the bulk with a W mutation or incorporation of naphthalene did not increase ion selectivity (Table 2). Contrary to expectations, the relative Na+/Cs+ permeability ratio was significantly reduced in F4′Nap compared with WT. This was not investigated further. None of the F4′ mutants affected the pH50 (Fig. 2 C and Table 2). Hence, the F4′ mutants do not affect gating, but decreasing the bulk of the side chain appears to be detrimental to ion selectivity. Substitution of the aromatic residues in the upper M1 (Y21′E-F23′A) did not affect ion selectivity or proton sensitivity, although the role of the Y22′A mutant could not be deciphered due to lack of current. In light of the available toxin-stabilized open structure of cASIC1, the lower pore is too wide for the W0′ and F4′ side chains to contribute to ion permeation (Fig. S7 A). However, according to the latest cASIC1 structure in the desensitized state, both W0′ and F4′ are interacting with lower pore-forming residues (Fig. S7 B). In this structure, the absolutely conserved W0′ is making contacts both with a reentrant loop that forms the lower pore of the channel according to this structure and with residues in the neighboring M2 helix (Fig. S7 B). This complements our data showing that F4′ is important for ion selectivity, whereas the absolutely conserved W0′ is critical for both ion selectivity and gating (Fig. 2 C and Table 2).
Permeability properties of WT ASIC1a and the nonselective F4′A mutant
To test the hypothesis that the removal of side chain bulk of aromatic residues in the lower ASIC1a M1 affects the size of the pore, we set out to measure currents carried by a range of monovalent cations with different sizes through WT and the nonselective F4′A mutant. We chose the F4′A mutant because it produced larger currents in Xenopus oocytes and because it does not obviously affect channel gating (unlike W0′F and W0′L). HEK293T cells without endogenous hASIC1a channels (Borg et al., 2020) were transfected with WT or F4′A hASIC1a DNA, and currents were recorded using an automated patch-clamp system (SyncroPatch 384PE). Similar to conventional patch clamping, the SyncroPatch 384PE system allows control over both the ICS and the ECS, but with greater overall throughput. We used N-methyl-D-glucamine (NMDG+) as the only cation in the ICS to ensure minimal outward ionic flow from the ICS and thus to facilitate isolation of small inward currents carried by the extracellular test ions. pH 6–induced currents were recorded at four test potentials (−20, 10, 40, and 70 mV) to estimate the reversal potentials with different extracellular monovalent test ions (Fig. S5). Fig. 2 E shows the pH 6–induced inward currents at −20 mV carried by the different monovalent test ions normalized to the current carried by Na+. Based on these data, the F4′A mutant conducts larger monovalent cations more readily than WT hASIC1a (Fig. 1 E and Table 3). For example, the current carried by MA+ in F4′A is almost equal in magnitude to the current carried by Na+ (IMA/INa = 0.81 ± 0.45; P = 0.29), whereas in WT, the magnitude of the MA+ current is only one-third of the Na+ current (IMA+/INa+ = 0.31 ± 0.14; P < 0.001; Fig. 2, D and E). Also, the permeability of DMA+ is significantly increased in F4′A (IDMA+/INa+ = 0.40 ± 0.31) compared with WT (IDMA+/INa+ = 0.14 ± 0.07; P < 0.05). A similar pattern was observed when we examined the reversal potentials with the respective extracellular test ions. Fig. S5 C shows reversal potentials estimated with different monovalent cations. The corresponding relative ion permeability ratios are reduced in the F4′A mutant for all ions except for EA+ (Fig. S5 D). This suggests that the pore of the F4′A mutant is larger than that of the WT channel.
Our mASIC1a data show that substitution of the F4′ side chain significantly affects the size of the pore and increases the permeability toward larger monovalent cations. This finding supports the notion that F4′, and possibly W0′, can indirectly affect ion selectivity. Because aromatic residues at the 0′ positions are conserved in all three hENaC subunits and the 4′ residue in hENaC α is also an aromatic, we hypothesized that these residues in hENaC might play an equally important role in ion selectivity.
The 0′ and 4′ residues of hENaC M1 are unlikely to be critical for ion selectivity
Based on the mutational screen performed on the M1 segment of mASIC1a, we concluded that side chains at the 0′ and 4′ positions are important for ion selectivity. We therefore decided to investigate the role of the corresponding residues in hENaC. hENaC α, β, and γ subunits were expressed in Xenopus oocytes, and we measured currents using two-electrode voltage clamp. The voltage protocol and I-V relationships for WT hENaC and a designated mutant in either 96 mM NaCl, KCl, LiCl, or CsCl in the bath solution are shown in Fig. 3, A–C. For WT hENaC, we found that Na+ and Li+ currents reverse at ∼25 mV, whereas Cs+ and K+ are virtually nonpermeable, because the Cs+ and K+ currents reverse at a membrane potential of less than −100 mV. This is consistent with previously observed values for Na+/K+ and Na+/Cs+ current ratios (Palmer, 1982, 1990). Because reversal potentials could not be accurately determined, relative ion conductance ratios were estimated as amiloride-sensitive K+, Li+, and Cs+ currents measured at −100 mV normalized to the amiloride-sensitive Na+ current for all hENaC constructs (Kellenberger et al., 2001; Kellenberger et al., 1999a).
Single amino acid substitutions were introduced at the 0′ and 4′ positions in one, two, or all three hENaC subunits. W0′L or W0′A substitutions in hENaC α coexpressed with WT β and WT γ rendered the channel nonfunctional, but the W0′F mutation in hENaC α expressed with WT β and WT γ yielded amiloride-sensitive currents and showed relative amiloride-sensitive currents similar to WT hENaC (Fig. 3 E; see Table 4). W0′F substitutions in both α and β ENaC (expressed with WT γ) increased the amiloride-sensitive Li+/Na+ but had no significant effect on the K+/Na+ or Cs+/Na+ currents (Fig. 3 E and Table 4). With W0′F substitutions in all three hENaC subunits, amiloride-sensitive currents could not be determined with any of the four ions due to lack of current. Although we cannot comment on the triple-mutant scenario, W0′ does not appear to contribute to ion selectivity in hENaC, at least upon W0′ substitutions in one or two of the three subunits. Substitution of the amino acids at the 4′ position in α, β, and γ ENaC yielded measurable amiloride-sensitive currents for all three mutants, but the relative amiloride-sensitive current profiles of all tested 4′ position mutations are similar to that of WT hENaC (Fig. 3 E and Table 4).
Although mutations at the 0′ and 4′ positions of hENaC had no significant effect on ion selectivity, there is a possibility that other amino acids in M1 of hENaC contribute to ion selectivity. The M1 of hENaC contains a higher number of aromatic residues than mASIC1a (Fig. 1 A). We set out to test the hypothesis that the larger number of aromatic residues explains why hENaC is more Na+ selective, possibly by indirectly decreasing the pore diameter. To this end, single amino acid substitutions were introduced individually into the M1 of hENaC. For positions that contained an aromatic amino acid in two or three subunits, substitutions were made in two or all three subunits. Fig. 3 E depicts amiloride-sensitive currents with Li+, K+, or Cs+ relative to the current carried by Na+. None of the mutants exhibited significantly altered ion selectivity profiles, and all remained nonpermeable to K+ and Cs+, except for α(F19′A)βγ and α(Y23′A)βγ, which showed a slight permeability for K+ and Cs+ (Fig. 3 E and Table 4). Some mutants exhibited modestly altered Li+/Na+ current. These include α(F19′A)βγ, α(Y22′A)βγ, and αβ(F22′A)γ (Fig. 3 E; see Table 4).
Given that 0′ and 4′ mutations in mASIC1a alter channel gating, apparent pore diameter, and ion selectivity and that M1 aromatic residues are highly conserved (especially in ENaCs), we were surprised that 0′ and 4′ mutations in hENaC had no effect on ion selectivity (note that only the 4′ position in hENaC α is aromatic in nature, whereas 4′ is occupied by threonines in β and γ subunits). To further test the potential contributions of conserved 0′ and 4′ aromatic side chains to currents through hENaC, we measured single-channel conductance in WT and mutant channels. We expressed hENaC WT and 0′ and 4′ mutants in Xenopus oocytes and measured the single-channel conductances of WT and 0′/4′ mutants containing amino acid substitutions in one subunit (α) or two subunits (α and β). The single-channel conductances of the 0′ mutants were significantly increased from 4.7 ± 1.0 pS in WT hENaC to 7.4 ± 3.2 pS in α(W0′F)βγ and 8.6 pS ± 1.0 pS in α(W0′F)β(W0′F)γ (Fig. 2, F and G; and Table 5), whereas the single-channel conductances of the 4′ single and double mutants were not significantly different from WT hENaC (Fig. 3, F and G; and Table 5). These results implicate the highly conserved W0′ but not the 4′ side chain in conductance.
Our data show that although hENaC M1 mutations can affect conductance, their impact on ion selectivity is generally small. We thus conclude that M1 contributes to ion selectivity to a lesser degree in hENaC than in mASIC1a.
Determinants of ion selectivity in the pre-M1 segment of mASIC1a
We next wanted to expand our investigation into other regions possibly contributing to ion selectivity in ASICs and ENaCs. Because the most important determinants of ion selectivity in M1 were found to be located toward the intracellular end of M1, we wanted to probe side chains further upstream of these positions. Work by others had already highlighted the functional importance of this pre-M1 region (Coscoy et al., 1999; Pfister et al., 2006), and the latest cASIC structures reveal that residues in the N-terminus preceding M1 form part of the lower pore structure in ASICs (Yoder and Gouaux, 2020). We used comparative sequence analysis of 33 ASIC and ENaC sequences, as well as other members of the DEG/ENaC superfamily (Fig. 4 A and Fig. S6), to identify potentially important determinants of ion selectivity in the pre-M1 segment. The resulting sequence logo indicates the degree of amino acid conservation in this region (Fig. 4 A).
We started by investigating the N-terminal end of this sequence stretch, which in mASIC1a comprises a FASSST sequence motif: substitution of F21 (mASIC1a numbering) for the smaller L or A reduced relative ion permeability to unity and caused an alkaline shift in the proton sensitivity (pH50 6.8 ± 0.0 in WT, 7.0 ± 0.0 in F21A, and 7.0 ± 0.1 in F21L), whereas substitution with the larger W decreased ion selectivity to a lesser degree but caused a similar alkaline shift in pH50 (7.1 ± 0.1; Fig. 4 C and Table 6). Substitution of the neighboring A22 for G and L also abolished ion selectivity and shifted the pH50 values to 7.1 ± 0.0 in A22G and 7.1 ± 0.0 in A22L. The S23A, S23E, and S24A mutants were not significantly different from WT in terms of ion selectivity or the pH50 values, but the S24E mutant exhibited a modest increase in the Na+/K+ permeability ratio (10.7 ± 3.9). Previous work had suggested hASIC1a S25 to be phosphorylated (Duan et al., 2012), so we mutated S25 to A and E in order to either remove a potential phosphorylation (A) or mimic the negative charge of a phosphate group (E). However, Na+ selectivity in S25A was only slightly decreased compared with WT (PNa+/K+ permeability = 5.6 ± 1.6, with WT-like proton sensitivity: pH50 6.8 ± 0.1), while the S25E mutant abolished Na+ selectivity and proton sensitivity was markedly increased (pH50 7.1 ± 0.1). This argues against a direct role for phosphorylation of S25 in channel function. By contrast, substitution of the neighboring T26 for C, V, or E reduced ion selectivity to unity and drastically shifted the pH50 to more alkaline pH (Fig. 4, B and C; and Table 6), while the T26S ion selectivity profile resembled that of WT. For this mutant, the pH50 value was not different from WT, and it displayed only modest differences in the Na+/K+ permeability ratio (5.7 ± 1.8 compared with 6.8 ± 0.3 in WT) and in the Na+/Cs+ permeability ratio (9.7 ± 1.7 compared with 6.8 ± 0.3 in WT; Fig. 3 C and Table 6). Together, this indicates that the T26 hydroxyl (or its T26S equivalent) is crucial to both Na+ selectivity and proton-dependent gating. However, we found that the isosteric T26V mutant did not alter the amplitude of Na+-mediated single-channel currents, suggesting that the loss of selectivity in this mutant does not reduce relative permeability of Na+, but is rather more likely to increase relative permeability of K+ (Fig. 4, D and E).
The highly conserved H28 and G29 positions were recently shown to be located near the apex of a reentrant loop formed by part of the cASIC1 N-terminus (Yoder and Gouaux, 2020). Consistent with the notion that both residues are crucial to function (and lining the permeation pathway), mutations at both sites disrupted ion selectivity and proton sensitivity: substitution of either of the two amino acids with A resulted in channels that were equally permeable to Na+, K+, Li+, and Cs+. Additionally, the pH50 values of both mutants were drastically left-shifted to 7.3 ± 0.1 for H28A and 7.2 ± 0.0 for G29A, respectively. This is in agreement with previous work on the equivalent positions in ENaC (Kellenberger et al., 2002; Kucher et al., 2011). While alanine substitution at the conserved I33 resulted in a similar phenotype (loss of selectivity, left-shifted pH50), mutation of K42 and R43 at the bottom of M1 showed a more modest reduction in Na+ selectivity and a less pronounced effect on proton dependence: the K42Q mutant showed WT-like gating with slightly reduced ion selectivity (PNa+/PK+ = 3.7 ± 0.3 and PNa+/PCs+ = 3.6 ± 1.7), while the R43Q mutant was slightly less sensitive to protons (pH50 6.6 ± 0.1) and also displayed modestly reduced ion selectivity (PNa+/PK+ = 6.8 ± 2.3 and PNa+/PCs+ = 15.0 ± 6.3). Interestingly, the mutants that exhibit large alkaline shifts in pH50 tend to also show distinct current phenotypes, where the peak current was typically followed by a sustained current (Fig. S7). Together, this complements the latest structural findings, which suggest that these residues contribute to the ion permeation pathway (Fig. S7), and mutational disruptions appear to cause severe consequences, manifested in abolished ion selectivity and altered gating.
In summary, our data on mASIC1a clearly show that the pre-M1 region is important for both ion selectivity and channel gating.
Determinants of ion selectivity in the pre-M1 of hENaC
Intrigued by the pronounced effects of mutations in the stretch of amino acids comprising the FASSST sequence motif in the pre-M1 region of mASIC1a, we sought to probe the potential role of the equivalent amino acids in the hENaC α, β, and γ subunits. To this end, alanine substitutions were introduced in one, two, or all three subunits of the FASSST sequence equivalent in hENaC. Additionally, two threonines immediately following this stretch of amino acids in the β and γ subunits of hENaC were mutated to valine (Fig. 5 A). None of the mutants were substantially different from WT hENaC in terms of the relative K+/Na+ and Cs+/Na+ current ratios, and Li+/Na+ current ratio was only modestly altered (Fig. 5 D). The only exception was the α(F62A)β(Y29A)γ double mutant, for which the Li+/Na+ current ratio was significantly increased from 1.2 ± 0.2 in WT to 3.7 ± 2.1 in α(F62A)β(Y29A)γ (P < 0.0001; Fig. 5 D and Table 7). The α(F62A)β(Y29A)γ(Y32A) and α(T66V)β(T33V)γ(T36V) triple mutants did not generate detectable currents.
Although these data suggest that the pre-M1 segment does not make a major contribution to ion selectivity in hENaC, we next tested if the amino acids corresponding to the mASIC1a FASSST sequence contribute to ion conduction. We thus determined single-channel conductances for (1) α(F62A)βγ and α(F62A)β(Y29A)γ (Table 8), because the equivalent aromatic amino acids at this position in mASIC1a play a role in ion selectivity and proton sensitivity (Fig. 4 C); and (2) mutants containing single T-to-V substitutions at all positions in this region containing the residue T (α(T66V)βγ, α(T66V)β(T33V)γ, α(T67V)βγ, αβ(T35V)γ, and αβγ(T38V); Fig. 5 A). Single-channel currents were not detectable for α(T67V)βγ, αβ(T35V)γ, and αβγ(T38V), but there was a marked increase in the single-channel conductance of the α(T66V)β(T33V)γ mutant compared with WT hENaC (Fig. 5, E and F; and Table 8). By contrast, the α(F62A)β(Y29A)γ double mutant showed only a small but insignificant increase in single-channel amplitude (Fig. 5, E and F; and Table 8).
Together, our data suggest that the pre-M1 region in hENaC is not a major determinant of ion selectivity but contributes to ion conduction.
Recent advances in structural biology have resulted in the publication of multiple structures of cASIC1 and hENaC (Baconguis et al., 2014; Jasti et al., 2007; Noreng et al., 2018; Noreng et al., 2020; Yoder and Gouaux, 2020; Yoder et al., 2018). However, structures of open, full-length ASICs have not been obtained, and especially the conformation of the lower pore varies considerably among ASIC structures (Baconguis et al., 2014; Jasti et al., 2007; Yoder and Gouaux, 2020; Yoder et al., 2018). Similarly, the resolution of the pore-forming M1 and M2 segments in the available ENaC structures is comparatively low, giving rise to uncertainty regarding their absolute and relative positions, also with respect to the (unresolved) N- and C-termini (Noreng et al., 2018; Noreng et al., 2020). It is therefore difficult to reconcile the stark differences in ion selectivity among these channel types solely on the basis of available structural data. Here we use a combination of electrophysiological approaches to assess the contribution of M1 and pre-M1 to ion selectivity in mASIC1a and hENaC. We demonstrate that these regions contribute differently in both channel types: although mutations have severe consequences for ion selectivity in mASIC1a, they mostly affect conductance in hENaC. These findings support the notion that despite their similarities in sequence and structure, ASICs and ENaCs achieve ion selectivity via different mechanisms.
Both M1 and pre-M1 segments of mASIC1a contain key determinants of ion selectivity
In agreement with previous work, we found that mutations (Pfister et al., 2006; Poët et al., 2001) at the three native cysteines in ASIC1a M1 (C3′L, C13′W, and C15′F) do not have a major impact on ion selectivity or channel gating (Fig. 2 C). Similarly, other mutations along M1 between F23′ and G6′ do not change ion selectivity. By contrast, we found that mutations of aromatic side chains toward the intracellular end of M1 (F4′ and W0′) abolish sodium selectivity. We also observed significant changes in apparent proton sensitivity with W0′ mutants, which is consistent with earlier reports and supportive of the notion that this highly conserved side chain is important for channel gating (Kasimova et al., 2020; Li et al., 2011). F4′ mutants, however, had WT-like gating and only affected ion selectivity. We therefore considered the F4′A mutant a reasonable model to test the notion that reduced bulk in the lower M1 may lower ion selectivity by increasing pore diameter. The finding that larger cations, such as MA+, EMA+, and EA+, permeate F4′A mutant channels more readily than WT channels lends direct support to this hypothesis. We conclude that the two aromatic residues in lower M1, W0′ and F4′ indirectly contribute to the pore diameter and likely the overall structure of the mASIC1a lower pore.
Additionally, these results pointed toward the lower pore as a major determinant for ion selectivity in ASICs. Although the exact location of the selectivity filter in ASICs has remained a matter of debate, these findings support the notion that the selectivity filter might be located in the lower pore (Lynagh et al., 2020; Lynagh et al., 2017). This led us to postulate that M2-E18′ and M2-D21′ are likely to form the selectivity filter in ASICs. However, the recent cASIC1 structures clearly show that, at least in closed and desensitized conformations, it is the pre-M1 region that lines the lower ion permeation pathway, thus calling into question whether M2-E18′ and M2-D21′ make direct contributions to ion selectivity in the open state of ASICs (Yoder and Gouaux, 2020). The structures raise the possibility that the ASIC selectivity filter might be formed, at least in part, by the pre-M1 segment, specifically the narrow constriction formed at the histidine from the conserved HG motif that is involved in a network of both intra- and intersubunit interactions (Yoder and Gouaux, 2020). This is in agreement with our finding that the pre-M1 segment of mASIC1a is extremely sensitive to mutations, such that even minor mutational disruptions abolish ion selectivity, including those to the HG motif (Fig. 4 C). In fact, this protein segment contains numerous side chains that, when mutated, have drastic consequences for both ion selectivity and channel gating, therefore underlining its functional importance. By comparison, we found relatively fewer side chains in the pore-lining M2 segment to affect ion selectivity (Lynagh et al., 2020; Lynagh et al., 2017).
Together, these data are in agreement with the possibility that the selectivity filter is located at the lower end of the pore, likely formed by the pre-M1 segment and at least indirectly supported by E18′ and D21′ in M2 and W0′ and F4′ in M1. However, in the absence of an open channel structure with a fully resolved pre-M1 region, we cannot rule out the possibility that other side chains, including M2-E18′ and M2-D21′, may directly line the permeation pathway in the channel open state. Interestingly, none of the mutations tested in this study increases the ion selectivity of ASIC1a. Because ENaC is more Na+ selective than ASIC, but none of the numerous ASIC mutations, even to ENaC-like identity, markedly increase Na+ selectively, ASICs seem to have developed, independently of ENaCs, a fragility in their molecular blueprint for ion selectivity.
ENaC ion selectivity is comparatively impervious to mutations in M1 and pre-M1
Although structural information on the ENaC pore is of relatively low resolution (Noreng et al., 2018; Noreng et al., 2020), there is strong functional evidence that the G/SXS motif in the middle of the pore is critical for ion selectivity in ENaCs (Kellenberger et al., 2001; Kellenberger et al., 2002; Kellenberger et al., 1999b; Yang and Palmer, 2018). Additionally, and in contrast to the findings for ASICs, the negatively charged side chains in the M2-18′ and M2-21′ positions do not appear to play a crucial role in ion selectivity (Yang and Palmer, 2018). Yet, it is interesting to note that mutations to W0′ in the ENaC α and β subunits affected both selectivity and single-channel conductance. Although the triple mutant was nonfunctional, this points toward W0′ being—directly or indirectly—involved in ion conduction in ENaCs (Fig. 2), although likely to a lesser degree than in ASICs. Furthermore, important functional contributions by side chains in the (putative) lower pore of ENaCs are highlighted by the findings that mutations to the conserved HG motif affect gating by decreasing ENaC open probability (Kellenberger et al., 2002) and that mutations of the G in this motif cause pseudohypoaldosteronism type 1 in humans (Kucher et al., 2011). Our results expand on these results by showing that selected pre-M1 mutations can affect ion selectivity or conductance (Fig. 5). This shows that, in contrast to ASICs, mutations at a few positions in both the middle and lower parts of the pore have relatively minor effects on ion selectivity but contribute significantly to conduction in ENaCs. Additionally, reducing the bulk of aromatic side chains along the entire length of at least the α subunit M1 in hENaC, such as W0′, F8′, F19′, Y22′, and Y23′, can lead to (albeit small) changes in ion selectivity.
This indicates that although the overall ion selectivity profile is maintained in most ENaC mutants, subtle changes can be elicited by mutations throughout the M1 and pre-M1 segments. By contrast, mutations in the upper and middle sections of both M1 and M2 of ASICs do not result in changed ion selectivity (the only exception being C15′F; see Fig. 2 C). We thus conclude that the open pore structures of ASICs and ENaCs likely display important structural differences.
Limitations and outlook
In the majority of our work on ASICs, we rely on relative ion permeability ratios as a proxy for ion selectivity. More subtle changes in conductance, as revealed for selected ENaC mutants, are therefore harder to assess in measurements of relative ion permeability ratios. But previous studies demonstrate that relative ion permeability measurements can result in robust estimates of ion selectivity (Kellenberger et al., 2001; Kellenberger et al., 1999a; Lynagh et al., 2020; Lynagh et al., 2017). Also, and in line with work by others (Yang and Palmer, 2018), some of the ENaC triple mutants did not generate measurable currents. We therefore cannot exclude the possibility that we may have underestimated the contributions of some side chains to selectivity and/or conductance. Additionally, it has been reported that the ENaC pore helices might be arranged asymmetrically and that mutations at equivalent sites may not cause equivalent effects (Li et al., 2003). Finally, the lack of definitive structural information on the open pore structures of both ASICs and ENaCs means that it is difficult to interpret our functional data in a detailed structural context.
Overall, however, the data at hand clearly support the notion of ASICs and ENaCs achieving sodium selectivity by different mechanisms. Our work therefore highlights how related channels can achieve different functional properties despite obvious sequence conservation in, for example, the (at least partially) pore-lining M2 segments. Instead, the data indicate that the non–pore-lining M1 segment indirectly contributes to ion selectivity in both ASICs and ENaCs.
Christopher J. Lingle served as editor.
This work was supported by Lundbeckfonden (R171-2014-558, to T. Lynagh; R139-2012-12390, to S.A. Pless), the Danmarks Frie Forskningsfond (4092-00348B, to T. Lynagh), and Carlsbergfondet (2013_01_0439, to S.A. Pless).
S. Friis is a full-time employee of Nanion Technologies. The other authors declare no competing financial interests.
Author contributions: Z.P. Sheikh, T. Lynagh, M. Wulf, and S. Friis conducted functional experiments; Z.P. Sheikh, T. Lynagh, M. Wulf, S. Friis, and M. Althaus analyzed data; Z.P. Sheikh, T. Lynagh, M. Althaus, and S.A. Pless conceptualized the work; Z.P. Sheikh and S.A. Pless drafted the manuscript and finalized it with input from all authors.
M. Althaus’s present address is Department of Natural Sciences, Institute for Functional Gene Analytics, Bonn-Rhein-Sieg University of Applied Sciences, Rheinbach, Germany.
T. Lynagh’s present address is Sars International Centre for Marine Molecular Biology, University of Bergen, Bergen, Norway.