A conserved peptide-binding pocket in HyNaC/ASIC ion channels

We used the HyNaC2/3/5 heteromer (31) as a model to study the ligand-binding site of HyNaCs. Because subunit interfaces are different in a trimeric complex wherein subunits are arranged clockwise in a 2-3-5 or a 2-5-3 order when viewed from the top, we first determined the order of the subunits in the heteromeric complex. Inspection of homology models of the 2-3-5 and 2-5-3 complexes, which were based on the resting conformation of chicken ASIC1 (cASIC1; PDB ID 6AVE) (24), identified two residues in HyNaC3 (N213 and F330) that are located at subunit interfaces on the two opposite sides of the protomer in both models (Fig. 1A). We substituted these two amino acids individually with the unnatural amino acid azido-phenylalanine (azF), which allows the cross-linking of these residues by UV irradiation to other amino acids around them. Both mutant channels were functional (SI Appendix, Fig. S1) and were tagged with a C-terminal HA epitope. When we coexpressed HA-tagged HyNaC3 carrying the azF substitution at position 213 or 330 together with HyNaC2 and HyNaC5 in HEK293 cells and subjected the proteins to SDS-PAGE, an anti-HA antibody revealed a strong signal with an apparent molecular weight of ~65 kDa, corresponding to a single HyNaC3 subunit. However, in cells treated with UV light, another band with a molecular weight that corresponds to a dimer was detected. This dimer was absent when cells were not treated with UV light and in cells expressing wild-type HyNaC3, showing that the dimer was only formed when UV light was applied and when azF was present, suggesting that azF at positions 213 and 330 cross-linked HyNaC3 to its neighboring subunit (Fig. 1B). To identify the subunit that was cross-linked, we probed the same western blots with an anti-Flag antibody; either HyNaC2 or HyNaC5 carried a Flag epitope. This revealed that HyNaC3 carrying an azF at position 213 was cross-linked to HyNaC2. In contrast, HyNaC3 carrying an azF at position 330 was cross-linked to HyNaC5 (Fig. 1B). Both results are consistent with a clockwise 2-5-3 arrangement of subunits when viewed from above (Fig. 1A). We found no evidence in favor of the 2-3-5 arrangement, suggesting that a channel with the 2-5-3 arrangement is the predominant (and perhaps the only) form of HyNaC2/3/5.

We used refined homology modeling based on the crystallographic structure of cASIC1 in the open conformation (PDB ID 4NTW) to generate 200 models of HyNaC2/3/5 having a clockwise 2-5-3 order when viewed from the top (Fig. 1A). The 10 models with the highest score (see Methods for details) were chosen for further analyses. The models predicted an overall structure for HyNaC2/3/5 that was similar to that of cASIC1 with conserved α-helices and β-sheets. Analyses of possible binding sites with the SiteMap tool (32, 33) predicted a druggable cavity at subunit interfaces at a position corresponding to the acidic pocket of ASICs (SI Appendix, Table S1 and Fig. S2). Therefore, we substituted at least six different amino acids at each of the three different interfaces (2-5, 5-3, and 3-2, respectively). We introduced initially amino acids with side chains that might interfere with putative molecular interactions with RFamides or cause steric constraints. When these substitutions altered apparent affinity for RFamide I, we introduced additional amino acids with milder chemical differences of their side chains. We coexpressed mutant subunits individually with the two other wild-type subunits in Xenopus oocytes. Then, we determined the apparent affinity to RFamide I using a two-electrode voltage clamp (TEVC), revealing that substitutions at the 2-5 and 3-2 interfaces did not affect apparent peptide affinity (Fig. 2), except for the K214E substitution in HyNaC2, which increased apparent affinity ~10-fold (from 5 µM to 0.7 µM; P = 0.84). In contrast, many mutations at the 5-3 interface reduced apparent peptide affinity, some more than 2,000-fold (Fig. 2). Most mutations that reduced EC₅₀ were on HyNaC3, suggesting that HyNaC3 contributes the principal side of the ligand-binding pocket and HyNaC5 the complementary side. In particular, substitutions of 3-D122, 3-Q188, 3-R229, 3-D314, 3-W334, and 3-E338 on the principal side and of 5-E213 on the complementary side reduced sensitivity to RFamide I, highlighting that the lower finger, β-ball, and thumb domains make major contributions to peptide sensitivity. Importantly, substitutions of equivalent residues on the 3-2 and 2-5 interfaces did not affect peptide sensitivity. For example, the substitution of 3-D314 by Arg reduced EC₅₀ ~200-fold (P = 0.058), whereas substitutions of the equivalent residues 2-S311 and 5-S310 by Arg did not affect EC₅₀. Likewise, substitution of 3-W334 by Arg reduced EC₅₀ ~1,000-fold (P < 0.001), whereas substitutions of the equivalent residues 2-M332 and 5-F331 by Arg did not affect EC₅₀. These results indicate that subunit interfaces do not equally contribute to the activation of HyNaC2/3/5 by RFamide I. The strong reduction in apparent affinity suggested that RFamide I may bind to the lower finger, β-ball, and thumb domains at the 5-3 interface to activate HyNaC2/3/5.

To assess which residues of Hydra RFamides are important for HyNaC activation, we conducted an alanine scan of RFamide II (pQWFNGRFamide). We selected this peptide because of its 10-fold higher apparent affinity compared with RFamide I (31). We substituted each amino acid of the peptide individually with alanine and assessed EC₅₀ values by TEVC in oocytes. Strikingly, all substitutions led to a loss of function (up to 600 µM peptide), except for the G5A substitution, but which also reduced the potency and efficacy of the peptide (EC₅₀ = 1.6 ± 0.2 μM compared with 0.3 ± 0.06 μM, P = 0.004; Fig. 3). These results indicate that nearly all residues of RFamide II are essential for its function. However, RFamide I (pQWLGGRFamide) differs from RFamide II at positions 3 and 4, yet has an EC₅₀ of ~5 µM (31), indicating that at least at these two positions there is some structural flexibility and that the exchange of F3 by Leu and of N4 by Gly is better supported than the exchange by Ala.

To corroborate the mutagenesis results, we used the identified cavity at the 5-3 interface in the ECD of HyNaC2/3/5 as the docking site for RFamide I and RFamide II. Each neuropeptide had 100 binding poses generated across the 10 homology models of the channel with the highest score, resulting in 1,000 poses per neuropeptide. The resulting set of poses was filtered based on their Glide score (34) and on a Tanimoto similarity score, which represents the overlap between the interacting residues in the pose and a subset of seven HyNaC3 residues found in mutagenesis experiments (Methods). None of the poses could interact simultaneously with all the HyNaC residues in the reference group. Nonetheless, the selected poses showed a common binding configuration and recurrent interactions with a conserved set of residues (Fig. 4A; for a comparison with all docking results, see SI Appendix, Fig. S3). The best poses for both neuropeptides were found deep in the analyzed pocket, completely surrounded by HyNaC3 and HyNaC5 residues (Fig. 4 B–D and SI Appendix, Fig. S4). This is in contrast to the shallower poses observed for FMRFamides on FaNaCs (17, 18). In our models, the ligand-binding pocket is shifted closer to the membrane and core of the protein. The ligand backbone partially aligns with the channel’s main axis: The C-terminal Phe residue (F7) is anchored in the 5-3 interface via π–π interactions, while the N-terminal pyroglutamate (pQ1) extends extracellularly (Fig. 4D and SI Appendix, Fig. S4). This set of selected poses also corroborates the experimental results, given that almost all the side chains of the neuropeptides play a relevant role in proper binding and recognition. The most consistent interactions between RFamides and HyNaC2/3/5 include both hydrophobic and electrostatic interactions (Fig. 4E and SI Appendix, Fig. S5) and were as follows: i) a π–π interaction between W2 and 5-Y154, especially maintained in the selected RFamide II poses (SI Appendix, Fig. S5); ii) H-bonds between R6 and 5-E213 and 3-E338; iii) an H-bond between the backbone carbonyl of R6 and 3-Q188; iv) a π–π interaction between F7 and 3-W186; and v) an H-bond between the backbone carbonyl of F7 and 3-K249. While these interactions are conserved for both RFamide I and RFamide II, the side chains of residues 3 and 4, which differ between the two peptides (LG in RFamide I and FN in RFamide II), did not make consistent interactions in our models, consistent with the fact that both peptides are agonists at HyNaC2/3/5, although with different potencies (~5 µM for RFamide I and ~0.4 µM for RFamide II) (31). These results identify a binding pocket that largely overlaps with the pocket identified by site-directed mutagenesis of HyNaCs. Moreover, some specific residues, for example 3-Q188, 5-E213, and 3-E338, were identified by both methods to make important contributions to peptide binding or apparent affinity.

To obtain more direct evidence for a binding site at the 5-3 interface, we turned to photo-cross-linking employing noncanonical amino acids in living cells. This approach has previously been successfully used to confirm the Big Dyn-binding site in ASIC1a (28). We substituted 25 amino acids not only close to the suspected binding pocket at the 5-3 interface but also at the corresponding location in the other two interfaces by azF. Functional analyses in HEK293 cells revealed that 16 out of the 25 mutants were functional with current amplitudes >200 pA and were almost fully activated by 1 µM RFamide II (SI Appendix, Fig. S6). We chose positions 3-F330 and 3-E338 for further analyses because they localize to opposite sides of the putative binding pocket at the 5-3 interface and because RFamide II elicited relatively large currents with these mutants (SI Appendix, Fig. S6). HEK cells expressing 3-F330azF or 3-E338azF together with wt HyNaC2 and HyNaC5 were treated with UV light for 20 min, either in the absence or presence of 5 µM RFamide II. Then, we conducted patch-clamp recordings of these cells. We reasoned that the covalent attachment of RFamide II to its binding site might constitutively activate the channels (Fig. 5A); therefore, to avoid damage to the cells by the constant influx of Na⁺, we kept the cells in a bath solution in which we replaced Na⁺ with the impermeant cation NMDG⁺. For cells that had been treated with UV light in the absence of the peptide, we recorded only small inward currents when we exchanged the NMDG solution with the standard bath containing Na⁺. Furthermore, the application of RFamide II to these cells elicited large currents with amplitudes of 445 ± 40 pA (3-F330azF) and 403 ± 61 pA (3-E338azF), respectively (n = 5; Fig. 5). In striking contrast, for cells that had been treated with UV light in the presence of the peptide, we recorded large inward currents (438 ± 86 pA and 387 ± 58 pA, respectively; n = 5) already upon exchanging the NMDG solution with the standard bath containing Na⁺, and the application of RFamide II did not elicit larger currents (Fig. 5). These results show that UV light in the presence of RFamide II–induced constitutive Na⁺ currents in cells expressing 3-F330azF or 3-E338azF together with HyNaC2/5. Moreover, they demonstrate that tethering a single peptide ligand to the channel is sufficient for full activation. Importantly, in control cells expressing HyNaC2/3/5 wt (carrying no azF substitution), UV light did not induce constitutive currents, even when the peptide was present during UV treatment (SI Appendix, Fig. S7A), illustrating that the constitutive currents relied on the presence of azF at position 3-330 of 3-338, close to the entrance of the putative peptide-binding pocket. Moreover, the amplitude of the constitutive currents was reduced by the HyNaC pore blockers amiloride and benzamil (Fig. 5), confirming that they were indeed mediated by HyNaCs. It is noteworthy that the potency of the two blockers was lower than anticipated (31). We speculate that the conformation of the ion pore was slightly changed by the covalent attachment of the peptide ligand. To further prove the specificity of these results and for the presence of a single ligand-binding pocket at the 5-3 interface, we substituted positions 5-F327 (corresponding to 3-F330) and 5-E335 (corresponding to 3-E338), which are located at the 2-5 interface, by azF. Strikingly, UV light in the presence of RFamide II did not induce constitutive currents when these subunits were coexpressed with the other two wt subunits (SI Appendix, Fig. S7 B and C), providing independent proof that HyNaC2/3/5 has a single peptide-binding pocket at the 5-3 interface.

Peptides were purchased from ProteoGenix (Schiltigheim, France) with a purity of >83 %. Depending on their properties, they were reconstituted in a 3:1 or 2:1 mixture of deionized water:DMSO at concentrations of 1.6 to 50 mM. For electrophysiological and biochemical analyses, the peptides were diluted to their final concentrations, either in standard bath solution, Hank’s Balanced Salt Solution (HBSS), or phosphate-buffered saline (PBS). Experimental solutions were freshly prepared before their use.

Stage V-VI oocytes were collected from anesthetized Xenopus laevis adult females (2.5 g/L tricainemethanesulfonate for 20 to 30 min). Anesthetized frogs were killed after the final oocyte harvest by decapitation. Animal care, surgery, and experiments were performed in accordance with the German Law on the protection of animals and were approved by the ministry for environment, nature and consumer protection (LANUV) of the State North Rhine-Westphalia (NRW) under number 81-02.04.2019.A356.

cRNA for HyNaC2, HyNaC3, and HyNac5 was synthesized from linearized cDNA using the mMessage mMachine SP6 kit (Thermo Fisher Scientific, Grand Island, NY) and purified with RNA Clean & Concentrator-TM-25 (Zymo Research Europe GmbH, Freiburg, Germany). RNA concentrations were determined at 260 nm, and quality was confirmed on a 1% agarose gel. The cRNA concentration was adjusted to ~200 ng/µL, and aliquots were stored at −80 °C. Approximately 1.6 ng cRNA of a 1:1:1 mixture of the three subunits was injected into defolliculated oocytes, and oocytes were incubated at 19 °C for 2 d in OR-2 solution containing (in mM): 85 NaCl, 2.5 KCl, 1 Na₂HPO₄, 1 MgCl₂, 1 CaCl₂, 5 HEPES, 0.5 g/L PVP, 1,000 U/L penicillin, and 10 mg/L streptomycin; pH was adjusted to 7.4 using NaOH.

TEVC experiments were performed using a Turbo Tec-03X amplifier (npi electronic GmbH, Tamm, Germany), and data were recorded with CellWorks 6.2.1 (npi electronic GmbH) at 1 kHz. Glass electrodes filled with KCl (0.3 to 1.5 MΩ resistance) were employed with a holding potential of −70 mV throughout the recordings. Fast solution exchange was facilitated by the programmable pipetting robot ScreeningTool (npi electronic GmbH).

The standard bath solution contained (in mM): 140 NaCl, 10 HEPES, 1.8 CaCl₂, and 1 MgCl₂; pH was adjusted to 7.4 using NaOH. Peptide stocks were added to the bath solution just before each experiment to achieve the desired working concentration. Due to the Ca²⁺ permeability of HyNaCs (43), oocytes were injected with 50 nL of 20 mM EGTA before measurements to avoid activation of endogenous Ca²⁺-activated Cl^- channels (CaCCs). Oocytes from at least two frogs were used in the experiments.

The sequences of HyNaC2, HyNaC3, and HyNaC5 were retrieved from Uniprot (IDs: A8DZR6, D3UD58, and A8DZR7). A blast search in the Protein Data Bank database revealed that the best template to model the trimeric protein in the open state was ASIC1 of the species Gallus gallus, with PDB ID 4NTW. To enhance the alignment quality between the sequence to be modeled and the template, we incorporated the biological information from the two taxonomic groups. Namely, the alignment with clustalW was repeated by including the 1,750 most similar sequences belonging to the Hydroidolina (taxid: 37516) and Aves groups (taxis: 8782). These sequences were obtained from the database of nonredundant protein sequences via blastp. The process was repeated for every monomer and the alignment to the ASIC structure sequence was used as input in Modeller 9.25 to generate 200 models scored with the DOPE-HR metric. The models were additionally evaluated with Procheck 3.5.4 and ranked by the product of their procheck and DOPE-HR score.

The ten best models were imported in Maestro version 12.6.144 (Schrödinger Release 2020-4: Maestro, Schrödinger, LLC) and prepared using the default settings in the Protein Preparation Wizard menu. The structures were analyzed using SiteMap and, in all the models, a pocket was found at the interface of HyNaC5 and HyNaC3 (SI Appendix, Table S1 and Fig. S2), which was used to build a receptor grid for peptide docking. The three-dimensional structures of RFamide I and II were built in UCSF Chimera 1.14 and prepared for docking using LigPrep (Schrödinger Release 2020-4: LigPrep, Schrödinger, LLC). The Bioluminate Peptide Docking module was used to return 100 docking poses per each ligand, per each receptor. The interactions between each pose and the receptor were analyzed using the Interaction Fingerprint tool in Maestro. For each pose and each residue, the presence and type of the interaction (backbone, side chain, polar, hydrophobic, hydrogen bond acceptor or donor, aromatic, and charged) were saved in a tabular format.

The interaction fingerprint of each ligand was compared to a binary vector representing the only theoretical case where the ligand forms, at the same time, all possible interactions with a set of selected residues, based on the experimental results and consisting of HyNaC3 residues 122, 188, 229, 314, 316, 334, and 338. The agreement between this theoretical fingerprint and one of the poses was quantified with a Tanimoto similarity score. Poses with a Glide docking score <−10.0 and Tanimoto similarity >10 % were selected for further analysis. In particular, within the selection, the same configuration was present in the three best scoring poses for both RFamide I and II, and thus, it was selected as main hypothesis and discussed in depth in the Results section.

The protein and docking results were visualized with Maestro and rendered with UCSF Chimera 1.14. The interaction fingerprints were analyzed with Python 3.9.7 and pandas 1.3.2 and visualized with matplotlib 3.4.3.

HEK293 T cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10 % fetal bovine serum and 2 mM L-glutamine, and maintained at 37 °C in a humidified atmosphere with 5 % CO₂. The cells were grown as an adherent monolayer and passaged every 2 to 3 d; they were replaced after ~30 passages.

For expression in HEK293 cells, codon-optimized HyNaC constructs were used (50). HyNaC subunits carried a TAG stop codon at the position, where an unnatural amino acid should be placed. To introduce p-azido-L-phenylalanine (azF) into HyNaC subunits in HEK293 cells, the pIRE4-Azi plasmid from Iris Biotech GmbH was utilized. This plasmid contains four copies of a tRNA essential for bypassing stop codons, along with an azF-specific aminoacyl-tRNA synthetase. A 0.5 M stock solution of azF in 0.5 M NaOH was diluted to a working concentration of 0.5 mM in DMEM and added to the cell culture 1 h before transfection. Photo-cross-linking experiments were conducted 36 to 48 h after transfection using a homemade UV lamp (equipped with six Philips Actinic Bl TL 8 W bulbs) for 20 min. For optimal cross-linking efficiency, cells were at a distance < 3 cm to the UV source.

For biochemical determination of the subunit stoichiometry, HEK cells were maintained in 10 cm dishes with up to 50 % confluency and transfected with a total of 10 μg DNA (2 μg of a 1:1:1 mixture of the three HyNaC subunits and 8 μg pIRE4-Azi) mixed with 30 μg polyethylenimine (PEI). For assessing azF incorporation efficiency, the GFP stop codon mutant Y40* was cotransfected with HyNaC constructs. Twenty-four hours after transfection, the medium in 10 cm dishes was replaced with fresh DMEM containing 0.5 mM azF. For the experiment, cells were washed with cold PBS, resuspended in PBS containing 5 μM RFamide-II, and exposed to UV light for 20 min. Cells were then resuspended in 500 μL of ice-cold lysis buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 10 % glycerol, 1% Triton X-100, 1.5 mM MgCl₂, 1 mM EGTA, 1 mM DTT, and 1X Protease Inhibitor Roche, Germany) for further analysis.

For patch clamp analysis, HEK cells were maintained in 35 mm dishes and transfected with 0.5 μg each of HyNaC2, HyNaC3, and HyNaC5, 1.2 μg pIRE4-Azi, 0.3 μg GFP mutant, and 6 μL PEI. After transfection, cells were transferred to poly-D-lysine-coated coverslips in six-well plates and provided with DMEM containing 0.5 mM azF. Coverslips were washed with PBS, followed by the addition of 2 mL of a cold, Na⁺-free solution containing (in mM): 128 N-Methyl-D-Glucamine (NMDG), 5.4 KCl, 1 MgCl₂, 2 CaCl₂, 10 HEPES, and 5.5 glucose; depending on the experimental condition, 5 μM RFamide-II was added. After UV exposure for 20 min, cells were washed with PBS and cultured in fresh DMEM for recovery and subsequent patch-clamp recording.

Approximately 36 to 48 h after transfection, HEK 293 cells from 35 mm dishes transfected as above were washed twice with ice-cold PBS (pH 7.4), exposed for 20 min to UV light while on ice in PBS, scraped from the dishes and centrifuged (4 °C, 5 min, 1,000 rpm). After discarding the supernatant, 500 µL HEPES lysis buffer was used to lyse the cells on ice for 1 to 1.5 h with intermittent vortexing. Samples were centrifuged (4 °C, 10 min, 13,000 rpm) and the supernatant diluted in 5X sample buffer (250 mM Tris-HCl pH 6.8, 10 % SDS, 0.05 % bromphenol blue, 50 % glycerol, and 500 mM DTT). After boiling (95 °C, 7 min), 40 µL of each sample was loaded onto a 10% polyacrylamide SDS-PAGE gel and run for 1.5 to 2 h at 130 V in running buffer (25 mM Trisma Base, 200 mM glycine, and 0.1 % SDS). Next, proteins were blotted onto a PVDF membrane (ThermoFisher, Massachusetts) for 1 h at 85 V in transfer buffer (25 mM Trisma Base, 200 mM glycine, 20 % methanol), followed by 30 min incubation in blocking buffer (137 mM NaCl, 2.7 mM KCl, 25 mM Tris, pH 7.4, 0.1 % Tween®-20, and 5 % milk powder) and 30 to 120 min incubation in blocking buffer with anti-FLAG (M2, mouse monoclonal, Sigma-Aldrich, Missouri). Blots were then washed 3 × 10 min in TBS-T (137 mM NaCl, 2.7 mM KCl, 25 mM Tris, pH 7.4, and 0.1 % Tween®-20), after which a secondary antibody (HRP AP16072 goat anti-mouse polyclonal, ThermoFisher, Massachusetts) in TBS-T was added. Finally, the blot was washed 3 × 10 min in TBS-T and results were analyzed by adding Clarity Western ECL substrate (BioRad, California). Blots were imaged in a Chemi-Smart 5000 Image station (Vilber Lourmat, Eberhardzell, Germany). Afterward, the blot was exposed to a sequence of i) stripping buffer (200 mM glycine, 0.1 % SDS, and 0.1 % Tween®-20, pH 2.2) for 2 × 10 min, ii) PBS for 2 × 10 min, and iii) TBS-T for 2 × 5 min. This was followed by the same procedure as above starting with blocking buffer to analyze signal from an anti-HA antibody (3F10, rat monoclonal, Sigma-Aldrich, Missouri) and a secondary antibody (HRP A136P goat anti-rat polyclonal, Merck, Darmstadt, Germany).

Approximately 48 h after transfection, and after the UV treatment, the coverslip with attached cells was mounted on the stage of an inverted phase-contrast microscope (IX71, Olympus). The recording chamber was perfused with the following bath solution (in mM): NaCl 128, KCl 5.4, HEPES 10, glucose 5.5, MgCl₂ 1, and CaCl₂ 2; pH was adjusted with NaOH to 7.4 at RT (22 to 25 °C). Patch-clamp experiments were performed in the whole-cell configuration, using an Axon-200B amplifier (Molecular Devices; San Jose, CA) and an Axon Digidata 1440 A acquisition system controlled by the Clampex 10.0 software (Molecular Devices). Signals were low pass filtered at 1 kHz and digitized at 4 kHz. Micropipettes (4 to 6 MΩ) were prepared from borosilicate glass capillaries with a micropipette puller (DMZ-Universal Electrode Puller; Zeitz Instruments, Martinsried, Germany) and filled with an intracellular solution containing (in mM): NaCl 10, KCl 121, HEPES 10, EGTA 5, and MgCl₂ 2. Holding potential was −70 mV. Capacitance and series resistances were compensated electronically at 80 %, and digital data were stored in a compatible PC for off-line analysis using Clampfit 10.0 software (Molecular Devices).

HEK cells and oocytes were randomly assigned to experimental groups without blinding of the experimenter. For electrophysiological experiments, we used cells from a minimum of two independent transfections or at least two batches of oocytes isolated on different days from distinct animals. Photo-cross-linking experiments were conducted with cells from two independent transfections. Each cell and oocyte served as a biological replicate for electrophysiology, while 10 cm dishes of cells were considered biological replicates for cross-linking studies.

Electrophysiological data were analyzed offline, and current amplitudes were assessed using CellWorks Reader 6.2.2 (npi electronic GmbH, Tamm, Germany) for TEVC data and ClampFit 10.0 (pClamp 10.0, Molecular Devices, Sunnyvale, CA) for patch clamp data, calculating current density by dividing maximal current amplitude (pA) by cell capacitance (pF). Data analysis was conducted using Excel (Microsoft) or Prism (GraphPad).

Dose–response curves from TEVC and patch clamp experiments were plotted as normalized current (I/I_max) against peptide concentration. The half-maximal effective concentration (EC₅₀) was estimated by fitting the dose–response curves in Prism using the following equation:

where I is the normalized current (I/I_max), [peptide] the peptide concentration, EC₅₀ the peptide concentration required for the half-maximal effect, and n_Hill the Hill coefficient. Data are presented as mean ± SD, and statistical significance was determined using two-tailed t tests or one-way ANOVA in Prism 10.0 (GraphPad, San Diego, CA) with a significance threshold set at P = 0.05.

A.M.O.-R., M.B., A.S., G.R., and S.G. designed research; A.M.O.-R., S.A., M.B., A.S., M.P.-S., M.A., and K.A. performed research; A.M.O.-R., S.A., M.B., A.S., G.R., and S.G. analyzed data; and S.G. wrote the paper.

The authors declare no competing interest.