Computational analysis of nucleotide sequences encoding transcript variants of the Mus musculus Hvcn1 gene, located on Chromosome 5–[NC_000071.6 (122206805-122242297)], predicts the presence of up to 13 exons (Table 1). These exons have the potential for alternative splicing, resulting in the coding of at least six Hv proton channel mRNA transcripts (Figure 1A, left). Previous in vitro studies have functionally confirmed three transcripts coding for the same 269-amino acid mHv1.1 protein: variants 1 (NM_001042489.2), 2 (NM_028752.3), and 3 (NM_001359454.1). Notably, there is still a predicted isoform (variant X3: XM_006530472.3) encoding the same amino acid sequence, suggesting that the expression of the mHv1.1 isoform plays a significant role, possibly related to its translational redundancy among different mRNA transcripts.

The selective splicing of mHv1 channel isoforms can be explained by considering alternative transcription mechanisms. Within the six mRNA transcripts of the Hvcn1 gene, four (transcript variants 1, 2, 3, and X3) begin processing at exons 2, 4, 3, and 1, respectively. The untranslated regions (UTRs) are located in these exons. Although the initiation codon for protein translation is found in exon 6, all of these transcripts encode the same 810-base pair (bp) product up to exon 13 (Figure 1A, left; Supplementary Figures S1A–C). This product corresponds to the canonical mHv1.1 isoform, consisting of 273 amino acids (Figure 1A, right).

For Vx1 (XM_011248243.2), transcription processing begins at exon 8, where the start codon for translation is also located. This results in an elongated ORF that codes for a 936-bp product (Figure 1A, left; Supplementary Figure S1B). This product corresponds to a predicted mHv1.2 isoform of 311 amino acids (Figure 1A, right).

Similarly, for Vx2 (XM_006530471.4), the transcription processing occurs at exon 7, where the transcription initiation site is located. This produces an mRNA with an ORF of 822 pb (Figure 1A, left; Supplementary Figure S1C). This allows the translation of the putative mHv1.3 channel isoform, which is only four amino acids longer than the canonical mHv1.1 (Figure 1A, right).

After the N-terminal variable region, all of the ORFs include the same coding sequence from exons 9 to 13. This results in all of the amino acid differences among mHv1 isoforms being concentrated in the initial N-terminal domain (Figure 1A, right; Supplementary Figure S1).

A comprehensive analysis of the primary sequences of the predicted longer isoforms revealed a high degree of conservation in the typical structural features for functional Hv1 proton channels (Supplementary Figure S1D), which include the selectivity filter at the residue D108 (Musset et al., 2011; DeCoursey et al., 2016); the voltage sensor domain (VSD), encompassing the S4 helix and its three arginines (referred to as R1 to R3, from outside to inside of the cell), which is the sensor of both voltage and ΔpH (Gonzalez et al., 2010; Gonzalez et al., 2013; Schladt and Berger, 2020; Carmona et al., 2021); and the coiled-coil region that promotes dimerization between Hv1 monomers (Koch et al., 2008; Gonzalez et al., 2010; Fujiwara et al., 2012; Fujiwara et al., 2012).

In the earlier prediction, it was mentioned that the Hvcn1 gene has the potential for alternative splicing, which could result in the production of up to six mRNA transcript variants. These variants, in turn, translate into three isoforms of the mHv1 channel protein with different N-terminal region lengths: mHv1.1, mHv1.2, and mHv1.3.

To further investigate this, we designed specific primers targeting the full length of the six predicted transcript variants, including the 5′ and 3′UTRs. We then screened a messenger library derived from mouse MDSCs using RT-PCR with these primers (Table 2).

The RT-PCR analysis results revealed the presence of four bands matching the expected sizes for variants 1–3 and variant X3 mRNAs, which encode the mHv1.1 isoform (2,790 pb for V_1, 2,652 pb for V_2, 2,711 pb for V_3, and 2,949 pb for Vx3, Figure 1B, indicated by black arrows). Another band of the expected size was detected for the transcript variant X1, which expresses mHv1.2 (2,807 pb for Vx1, Figure 1B, indicated by a blue arrow). Finally, a band corresponding to the transcript variant X2 was observed, which expresses the mHv1.3 isoform (2,489 pb for Vx3, Figure 1B, indicated by a red arrow).

To confirm the identity of these amplified DNAs, they were sequenced and determined to be mRNA transcripts derived from the Hvcn1 gene. Based on this confirmation, the ORFs that encode hypothetical mHv1.1, mHv1.2, and mHv1.3 proton channels were cloned.

To assess the functionality of these hypothetical ion channels, we conducted electrophysiological recordings in Xenopus laevis oocytes, which does not express endogenous proton currents (Supplementary Figures 1E,F), and injected them with mHv1.1, mHv1.2, and mHv1.3 cRNA. As predicted, these sequences should contain all the structural components of functional mHv1 channels (Supplementary Figure S1D). By using the voltage patch-clamp technique, we tested their hypothetical function as voltage-sensitive channels using excised macro-patches in the inside-out configuration. We recorded voltage-activated outward currents for mHv1.1, mHv1.2, and mHv1.3 (Figure 1C, left to right, respectively. Methods; Eq. 1).

While the products expressed by the Hvcn1 gene in Xenopus oocytes exhibited ion channel behavior, the question remains: Do these channels function as mammalian voltage-gated proton channels? To address this question, we investigated whether the cloned ORFs exhibit the hallmarks of mHv1 channels. In addition to voltage-dependent gating (Sasaki, M., Takagi, M., & Okamura, Y., 2006; Ramsey, I. S., et al., 2006), these include ΔpH-dependent gating (Musset, B., et al., 2008; Schladt, T. M., & Berger, T. K., 2020; Carmona, E. M., et al., 2021), unitary currents in the order of femtoamperes (fA) (Cherny, A., et al., 2003), proton selectivity (Musset, B., et al., 2011), and a dimeric oligomer state (Lee, S. Y., Letts, J. A., & MacKinnon, R., 2008; Koch, H. P., et al., 2008; Gonzalez, C., et al., 2010; Fujiwara, Y., et al., 2012; Gonzalez et al., 2013) in the cloned ORFs.

All ORF currents exhibited a clear ΔpH dependence (Methods, Eq. 2), where the IV curves in all isoforms shifted to hyperpolarized potentials with increase in the ΔpH and toward depolarized potentials when the ΔpH was lower, similar to mammalian Hv1 channels (Figure 2A; Supplementary Figure S2A; Table 3; Musset, B., et al., 2008). These results suggest that the gating of the cloned ORFs is ΔpH-dependent.

The unitary conductance of mammalian Hv1 channels is estimated to be in the order of fS (Cherny, A., et al., 2003), with single-channel current values below the limit of detection of traditional electrophysiology techniques, including inside-out macro-patches. Consequently, direct single-channel current measurements are not feasible.

Using nonstationary noise analysis (Alvarez, O., Gonzalez, C., & Latorre, R., 2002), we determined the unitary conductance from the macroscopic currents of the three cloned ORFs. In Figure 2B (top), the time course for the mean current and variance for mHv1.2 and mHv1.3 is shown, both obtained at ΔpH = 1 (pHin = 6 and pHout = 7) at 100 mV and 80 mV, respectively. Both variances exhibited a biphasic time course, reflecting the open probability of the channels (Gonzalez, et al., 2000). From these data, variance versus mean current plots were generated (Figure 2B, bottom), which were fitted to Eq. 3 (see Methods). The unitary conductance values obtained for mHv1.2 and mHv1.3 were 115.1 ± 34.8 fS (n = 3) and 164.4 ± 24.8 fS (n = 4), respectively. Additionally, the maximum open probabilities for mHv1.2 and mHv1.3 were 0.86 ± 0.03 (n = 3) and 0.80 ± 0.03 (n = 4), respectively.

In Supplementary Figures S2B, nonstationary noise analysis for mHv1.1, mHv1.2, and mHv1.3 at ΔpH = 2 (pHin = 6 and pHout = 8) and at 0 mV is displayed. The fits to the variance versus mean current data exhibited unitary conductance values of 93.6 ± 16.9 fS (n = 4) for mHv1.1, 63.1 fS for mHv1.2 (n = 1), and 165.4 ± 79 fS for mHv1.3 (n = 3), with a maximum open probability of 0.85 ± 0.04 (n = 4) for mHv1.1, 0.99 for mHv1.2, and 0.89 ± 0.08 (n = 3) for mHv1.3. All these data are consistent with previously reported unitary properties for Hv1 channels (Cherny et al., 2003).

To confirm the proton selectivity of all cloned isoforms, we employed variable duration voltage protocols coupled to a fast ramp ranging from depolarizing to hyperpolarizing potentials (Figure 2C), as previously utilized (Gonzalez, C., et al., 2010; Carmona et al., 2021; Alvear-Arias, J., Carrillo, C., et al., 2022). Using this protocol, we measured mHv1.2 and mHv1.3 currents at ΔpH = 1 (Figure 2C, upper panel). The red boxed region highlights the current during the fast ramp (Figure 2C). The lower panels in Figure 2C provide a closer examination of the currents elicited by the fast repolarization ramp, revealing an overlap of the currents for all the cloned channels. To determine the exact voltage at which the current reverses, we calculated the variance of the current over time and focused on the point where this variance was minimal (Figure 2C, lower panel, red straight line, Methods, Eqs 4–10). This allowed us to estimate reversal potential values near the Nernst equilibrium for protons at ΔpH = 1 (Figure 2C, lower panel). We repeated the measurements at ΔpH = 2 and ΔpH = 0 (Supplementary Figure S2C), and the estimated reversal potentials, along with those previously obtained at ΔpH = 1, were plotted against ΔpH (Figure 2D). Every cloned channel exhibited a straight-line slope between −55.8 mV and 60.2 mV/ΔpH (Figure 2D). As the Nernst equilibrium for protons predicts a slope of −58 mV/ΔpH, our data strongly suggest that currents from mHv1 isoforms are selective for protons.

The previous information we have collected is consistent with the idea that all cloned ORFs are functional voltage-gated proton channel isoforms. In mammals, Hv1 is naturally found as a dimer (Fujiwara, Y., et al., 2012). The novel cloned mHv1 isoforms contain a predicted C-terminal domain that is expected to fold into a coiled-coil region (CCR) (Figure 1A, right), driving the dimerization process. For all isoforms, we observed a sigmoidal pattern on activation currents (Figure 1C), a characteristic that is consistent with the strong cooperativity previously reported for dimeric Hv1 formation (Gonzalez et al., 2010).

Next, to determine the oligomeric state of the cloned Hv1 isoforms, we applied the limiting slope technique method at very low open probabilities to estimate the effective gating charge (zδ_eff) coupled to the channel opening (Gonzalez, C., et al., 2013). The three arginines in the S4 helix contribute directly to the value of the zδ_eff in Hv1 channels (Gonzalez, C., et al., 2013), with a monomer yielding a value of ∼3 charges and a Hv1 dimer reaching a value of ∼6 charges (Gonzalez, C., et al., 2010; Gonzalez, C., et al., 2013). In mice, limiting slope measurements of Hv1 indicated charge values of 4 e₀ for dimers and 2 e₀ for monomers (Fujiwara, Y., et al., 2012).

We used slow voltage ramps (ranging between 0.5 mV/s and 1 mV/s) to ensure gating equilibrium during measurements of mHv1.1, mHv1.2, and mHv1.3 currents, which were then transformed into macroscopic conductance and graphically represented in a log-linear plot against voltage (Supplementary Figure S3A). This analysis followed the limiting slope theory for Hv1, as discussed elsewhere (Gonzalez et al., 2010; Gonzalez et al., 2013; Methods, Eqs 11, 12). We used a Boltzmann curve to fit the previous data and estimated a zδ_eff of 4.19 ± 0.13 e₀, 4.30 ± 0.18 e₀, and 4.44 ± 0.17 e₀ for mHv1.1, mHv1.2, and mHv1.3, respectively (Supplementary Figure S3B). These values are consistent with all these novel isoforms of mHv1 having a dimeric oligomerization, as observed in mammalian mHv1 channels.

All the previous gathered evidence suggests that the three cloned ORFs from Hvcn1 gene expression in MDSCs are indeed three mammalian isoforms of mHv1 channels: mHv1.1 (the shortest), mHv1.2 (the longest), and mHv1.3 (of mid-length). Although these isoforms exhibit clear hallmarks of mammalian mHv1 channels, do they exhibit functional differences among them?

The Hvcn1 gene in mice expresses up to six mRNA transcripts, from which three ORFs are identified, expressing three different mHv1 isoforms. Upon inspecting their primary sequences, one can observe that nearly all sequences are conserved. However, at the very beginning of the N-terminal domain, there are 42 additional amino acids in mHv1.2 compared to mHv1.1 (Figure 3A) and four additional amino acids in mHv1.3 compared to mHv1.1 (Figure 3A). Could these differences in amino acids result in functional differences among mHv1 isoforms?

The sole reported isoform of Hv1, HVCN1_S, has 20 fewer amino acids than the canonical Hv1, HVCN1_L, particularly at the beginning of the N terminus (Hondares, et al., 2014). HVCN1_S displays slightly slower activation kinetics when compared to HVCN1_L. Given the evident and localized differences in the beginning of the N-terminal domain of mHv1 isoforms, we hypothesize that each mHv1 isoform might possess distinctions in the gating properties. To address this hypothesis, we conducted a comprehensive study of the activation of mHv1 isoforms, centering on the functioning at the thermodynamic level (conductance/voltage relationship) and at the kinetic level (activation kinetics).

First, we calculated and fitted conductance vs. voltage curves for all three isoforms (Methods; Eqs 13, 14) and compared them at ΔpH = 2 (Figure 3B). We did not observe any difference between the V_0.5 of mHv1.1 and the V_0.5 of mHv1.3 in the GV plots (Student’s t-test; p = 0.68; Figure 3B; Table 4), with a clear rightward shift in the GV curve for mHv1.2 (Figure 3B; Table 4). The rightward shift in the GV curve of mHv1.2 is consistent with nearly a 1-kT displacement of the V_0.5 to the right when compared to mHv1.1, and this difference is statistically significant (Student’s t-test; ***p < 0.005; Figure 3B; Table 4). This suggests that the 42 extra amino acids in mHv1.2 could stabilize the channel’s closure and/or destabilize the channel’s opening. On the other hand, the presence of four extra amino acids in mHv1.3 does not contribute to channel closure or destabilize the opening.

Next, we investigated the activation kinetics of these isoforms. We determined voltage-dependent time constants, which were then plotted against voltage (τ-V curves, Figure 3C; Methods; Eqs 15–17). It can be observed that mHv1.1 exhibits the fastest τ-V, followed by mHv1.3, and finally, the slowest, mHv1.2 (Figure 3C). At 0 mV, the time constants are statistically different (data were analyzed using Student’s t-test; **p < 0.01 for mHv1.1 vs. mHv1.2, **p = 0.01 for mHv1.1 vs. mHv1.3, and p = 0.06 for mHv1.2 vs. mHv1.3). These kinetics findings suggest that a longer N-terminal end leads to slower activation of mHv1.

Are these cloned mHv1 channels responsible for the proton currents seen in native murine tissues? The validation of cloned Hvcn1 transcripts expressing functional mHv1 isoforms could support that mHv1 currents in MDSCs are produced by the combination of mHv1 isoforms. Previous studies have reported that Hv1 heterologous expression in HEK-293 or COS-7 cells shifts their V_0.5 toward hyperpolarized potentials, reaching values near −30 mV when compared to mHv1 recorded in a system where it is expressed natively (Musset B., et al., 2008). To confirm that X. laevis oocytes are a suitable model for mimicking native mHv1 expression in MDSCs, we measured mHv1 currents from MDSCs by whole-cell patch-clamp and computed their normalized conductance as a function of voltage (Supplementary Figures S4A, B). We observed that the GV curve of mHv1 in MDSCs closely resembles the GV curves found for mHv1 isoforms expressed in X. laevis oocytes (Supplementary Figure S4B; Table 5). This suggests that mHv1 isoforms expressed in oocytes from X. laevis can effectively replicate the function observed in native mHv1 in MDSCs.

Subsequently, we investigated whether the mHv1 function in MDSCs is a result of the combination of mHv1 isoforms. We noted that all three isoforms exhibited different τ_activation at 0 mV (Figure 3C) and assumed that if all three mHv1 isoforms are expressed, fitting the mHv1 MDSC currents at 0 mV to the sum of three simple exponential functions (Methods, Eq. 18) should allow the estimation of three time constants, each related to a single isoform expressed by the Hvcn1 gene. By performing this fit to mHv1 currents from MDSCs at 0 mV (Supplementary Figure S4C, inset), we obtained three ? and compared them to the τ_activation at 0 mV for all mHv1 isoforms (Supplementary Figure S4C). We observed that the three estimated time constants are quite similar to those of each individual mHv1 isoform expressed by the Hvcn1 gene in the mHv1 MDSC current.

All these data are consistent with the three mHv1 isoforms exhibiting distinct voltage-gating activation properties, with the most efficient activation observed in mHv1.1, followed by mHv1.3, and finally, the least efficient activation, mHv1.2. The mHv1 isoforms are expressed and function in combination to produce mHv1 currents in MDSC.

Thus far, mHv1.1, mHv1.2, and mHv1.3 have displayed distinct activation patterns, which may be influenced by their N-terminal regions. It is noteworthy that all three mHv1 isoforms have varying lengths in their N-terminal segments. Since alternative splicing occurs at the initiation of exon processing, could this imply that the initial segment of the N-terminal region regulates the gating of Hv1 isoforms? We hypothesized that shortening the N-terminal region of mHv1.2 channels would result in a more favorable gating compared to the full-length mHv1.2, while still maintaining less favorable gating compared to mHv1.1. To test this hypothesis, we removed 20 amino acids from the N-terminal region of the mHv1.2 isoform, resulting in a truncated isoform referred to as mHv1.2 Δ20N (a 291-amino acid protein, as depicted in Figures 4A, B).

To confirm the aforementioned findings, we measured voltage-dependent proton currents from inside-out excised patches containing mHv1.2 ΔN20 channels (Figure 4C). The biophysical characterization of this construct was performed, confirming proton selectivity (Supplementary Figure S5A) and showing proton conduction coupled to the movement of six effective charges by a dimeric mHv1 channel (Supplementary Figure S5B).

We found that the GV curve of mHv1.2 ΔN20 was slightly left-shifted in comparison to mHv1.2 GV (*p < 0.05; Figure 4D), supporting the notion that mHv1.2 ΔN20 macroscopic conductance was higher than that of mHv1.2. Additionally, mHv1.2 ΔN20 V_0.5 was slightly right-shifted compared to mHv1.1 (*p < 0.05; Figure 4D), suggesting that mHv1.2 ΔN20 exhibited less favorable proton conduction compared to mHv1.1.

At the microscopic level, by nonstationary noise analysis of mHv1.2 ΔN20 current fluctuations at ΔpH = 2 and pHi = 6 (Supplementary Figure S5C), we estimated a unitary conductance of 109.3 ± 27.2 fS (n = 4), with a maximum open probability of 0.92 ± 0.04 (n = 4).

Finally, with regard to mHv1.2 ΔN20 activation kinetics, its τ-V curve (Figure 4E) closely resembled the opening kinetics of mHv1.3, with nonstatistical differences at 0 mV. However, it was faster than mHv1.2 (*p < 0.05) and slower than mHv1.1 (*p < 0.05), suggesting that mHv1.2 ΔN20 activation kinetics was more accelerated than that of mHv1.2 and less accelerated than that of mHv1.1 (Figure 4E).

These findings align with the initial prediction that mHv1.2 ΔN20 gating falls between the mHv1.1 and mHv1.2 phenotypes. Together, these results suggest that the first 20 amino acids of mHv1.2 are involved in the less efficient behavior observed in mHv1.2 gating. These amino acids likely influence either the stabilization of channel closure or the destabilization of the channel opening (Figure 3).

On the other hand, the subsequent 22 amino acids of mHv1.2 ΔN20 seem to contribute to further stabilizing the channel’s closure and/or destabilizing the open state without affecting the activation kinetics (Figure 4).

The canonical isoform, mHv1.1, exhibited the most efficient activation, with the first seven amino acids displaying a positive net charge (Figure 3A). Notably, the beginning of the N-terminal region in mHv1 is closely associated with gating. This is evident in the case of mHv1.3, where the alternative ORF adds four residues and modifies the subsequent seven residues. In total, 11 amino acids with a predominantly negative net charge slow the activation while not affecting the conductance (Figure 3). Therefore, the N-terminal extension and content could be implicated in the differential activation of mHv1 isoforms. Hence, the longer the N terminal end, one should expect a reduced conductance and slowed opening kinetics, as demonstrated by the behavior found in mHv1.2.

By the ClustalW^® program (https://www.genome.jp/tools-bin/clustalw, Thompson et al., 1994), multiple alignments for sequences reported at the NCBI database (https://www.ncbi.nlm.nih.gov/gene/74096) corresponding to the Hv1 transcripts or their open-reading frame were performed to determine similitudes among sequences and define structural motifs. A putative alternative edition among transcripts was predicted using the Splign program (https://www.ncbi.nlm.nih.gov/sutils/splign/splign.cgi). Query blast and identity percentage were calculated using BLAST-p (https://blast.ncbi.nlm.nih.gov/Blast.cgi). Primers used to amplify the full transcript or the codifying of the sequence (Supplementary Table S1) were designed using AmplifX 2^® (version 2.0.6) software, and their selectivity was evaluated by the Primer-BLAST program (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) using the RefSeq mRNA database for M. musculus as the target organism. Prediction of secondary structure was done by using PSIPRED v4.0 (http://bioinf.cs.ucl.ac.uk/psipred, Jones DT, 1999). The mHv1 isoform N terminal structure was obtained by performing de novo structure prediction with the Rosetta AbinitioRelax application (Bradley et al., 2005). The models with the lowest pseudo-energy score, selected from a range of 5,000 to 10,000 models, were chosen as representatives for N-terminal beginning isoforms and drawn using VMD (Humphrey et al., 1996), with licorice representing the beginning of the protein (methionine, white) and the last residue (Arg, Lys, and Arg for mHv1.1, mHv1.2, and mHv1.3, respectively, blue).

Mice C57BL/six mice (8–12 weeks old) were maintained in the University of Valparaiso bioterium (Valparaiso, Chile). All experiments followed the institutional guidelines of the University of Valparaiso’s Animal Care and Use Committee. Bone marrow-derived cells (BM cells) were isolated from the femurs and tibias of C57BL/6 mice and cultured to obtain MDSCs following the protocol previously described by Alvear-Arias et al. (2022). Cellular differentiation was obtained 4 days after incubating 2.5 × 10⁵ BM cells with 40 ng/mL of GM-CSF at 37°C in a humidified atmosphere of 5% CO₂.

Once MDSC differentiation was observed, cells were detached from the Petri dish using 2 mM PBS/EDTA and then centrifuged at 200 g for 5 min to re-suspend them in PBS on ice. Total RNA was extracted using the RNA-Solv kit (OMEGA Biotek) and lysed using a 20G needle and syringe. To degrade genomic DNA, samples were treated with dsDNAseI (ThermoFisher Scientific). cDNA was synthesized using the Maxima H Minus First-Strand cDNA Synthesis Kit (ThermoFisher Scientific) with 0.7–1 µg of total RNA. PCR and overhang PCR (oPCR) reactions were performed according to the instructions of the Platinum SuperFi DNA Polymerase Kit (ThermoFisher Scientific) in a final volume of 20 µL. The PCR products were run on a 1% agarose gel, stained with GelStar (Lonza) 2X, and visualized using a 3UV transilluminator (Epi Chemi Darkroom). The O’Gene Ruler 1 Kb DNA Ladder (ThermoFisher Scientific) was used as a molecular weight marker.

First, the oPCR products were purified using the PCR purification kit (Qiagen), and then they were site-directly cloned into the BamHI/HindIII restriction sites from the pGEM-3 vector. After identifying Hv1 transcripts by Sanger nucleotide sequencing, selected clones were linearized with the BamHI restriction enzyme to perform in vitro transcription and obtain cRNA, following the instructions of the T7 Message Machine Kit (Thermo Scientific). Finally, the cRNA was quantified and visualized by electrophoresis using agarose. cRNA was stored at −80°C until use. For mHv1 isoforms, 50 nL of cRNA at concentrations ranging from 0.5–4 μg/μL were injected into X. laevis oocytes. The oocytes were incubated at 12–25°C in ND96 or SOS medium, supplemented with antibiotics and 2–5 mM pyruvate, for approximately 24–80 h post-injections before recording.

Borosilicate glass (World Precision Instruments, Inc.) used as recording micropipettes was pulled using a P-97 Flaming/Brown automatic pipette puller (Sutter Instruments, Co.) and polished in an MF-830 Microforge (Narishige, Co.). Micropipettes used for inside-out patch clamp recordings in oocytes had diameters ranging from 3 to 15 μm, with resistances between 1 and 8 MΩ, while micropipettes used for whole-cell patch clamp recording in MDSCs had a resistance of 7–12 MΩ. The Ag-AgCl reference electrode was connected to the bath solution through an agar bridge with 3M KCl solution and ground. The registered currents were obtained using an Axopatch 200B amplifier (Axon Instruments) by applying voltage protocols (pCLAMP, version 10.3) acquired at 250 kHz, low-pass-filtered at 10 kHz (Frequency Devices), and analog-to-digital-converted using a Digidata 1550 (Axon instruments). Solutions used for proton current recordings in MDSCs for both bath and pipette were as follows: 100 mM MES buffer, 30 mM tetraethylammonium hydroxide (TEAOH), 2 mM magnesium chloride (MgCl₂), 1 mM ethylene glycol-bis (b-aminoethyl) N, N, N′, N″- tetra acetic acid (EGTA), and 160 mM N-methyl-D-glucamine (NMDG)—methane sulfonic acid (MeSO₃) to reach an osmolarity of ∼300 mOsm for solutions at pH 5.5 and 100 mM HEPES buffer, 30 mM TEAOH, 2 mM MgCl₂, 2 mM MgCl2, 1 mM EGTA, and 160 mM NMDG-MeSO₃ (or solutions at pH 7.5). The solution used for pharmacological perfusion during proton current recordings contained 10 mM ZnCl₂, 100 mM HEPES buffer, 30 mM TEAOH, 2 mM MgCl_2, and 160 mM NMDG-MeSO₃ at pH 7.5. Solutions used for proton current recordings in oocytes injected with cloned transcripts include 110 mM HEPES, 1 mM EGTA, 2 mM MgCl₂, and 110 mM NMDG + MeSO₃ at pH 5, 5, 7, and 5, respectively. All electrophysiology experiments were performed at room temperature (∼21°C).

Macroscopic currents of mHv1 isoforms were obtained at the end of each depolarization pulse, which can be expressed as conductance according to Ohm’s law: I H + = G H + V − V H + , (1) where G H + represents the macroscopic conductance of protons and V − V H + represents the electrical driving force, with V the voltage and V H + the reversal potential of protons. To build current–voltage (IV) plots, the current values were plotted against the potential (I–V curve) and fitted to a normalized

Boltzmann function: I H + = I max 1 + exp z δ F V − V 0.5 R T . (2)

Here, I_max represents the maximum current, zδ represents the voltage dependence, F is the Faraday constant, V is the voltage, V_0.5 is the voltage at which 50% of the channels are open, R is the universal gas constant, and T is the temperature in Kelvin.

For nonstationary noise analysis, current ensembles (ranging from 11 to 256) from isoforms of Hv1, expressed in membranes of X. laevis oocytes, were used to calculate the mean current and variance. These calculations followed procedures described elsewhere (Alvarez et al., 2002) and were fitted using the following parabolic equation: σ 2 t = i < I t > − < I t > 2 N , (3) where σ 2 t represents the variance, i is the unitary current, < I t > is the mean current, and N is the number of channels in the membrane patch.

For the measurement of reversal potential, it is generally defined as follows: V r e v = ∑ E i g i ∑ g i . (4)

In the previous equation, E_i represents the reversal potential of the ion i, and g_i is the conductance of the ion i. Using solutions with no sodium and potassium chloride, Eq. 4 transforms into the reduced form of the reversal potential equation as follows: V r e v = V H + , (5)

where V H + can be described by the following Nernst potential for protons: V H + = R T z F ln H e x + H i n + , (6) leading to the conclusion that the membrane is exclusively selective for the permeation of protons.

The experimental determination of the reversal potential can be achieved by observing the intersection of fast ramp currents. The intersection of these currents can be represented by the expanded form of the total current as follows: I = I H + + I L e a k + I C a p a c i t i v e = G H + V − V H + + G L e a k V − V L e a k + C d V d t , (7) where the macroscopic total current (I) is equal to the sum of the proton current I H + (described in Eq. 1), the leak current I L e a k , and the capacitive current, I C a p a c i t i v e , with G L e a k as the leak conductance, V L e a k as the equilibrium potential of the leak, and C as the capacitance of the membrane. It is of utmost importance to note that capacitive currents were considerably lowered on-line by using the amplifier circuitry, leading I C a p a c i t i v e to a diminished, but constant, remnant capacitive current, and that leak currents correspond to an artifact that may come from an imperfect seal that favors the leak of ions. As a consequence, G_leak can be very large or very small depending on the patch nature, but such values were corrected off-line by leak subtraction, leading I L e a k to transform into a constant remnant leak current, I c o n s t . These corrections transform Eq. 7 into I ≈ I H + + I c o n s t . (8)

When the voltage ramp matches V H + , V − V H + is 0, as does I H + , leading Eq. 8 to become I ≈ I c o n s t . (9)

Therefore, at V = V H + , all the currents induced by the voltage ramp will cross each other at the same point. I c o n s t V i ≈ I c o n s t V i + 1 . (10)

Macroscopic measurements of mHv1 isoform currents were performed by using variable-duration pulse protocols, applied to diminish the proton depletion (Carmona et al., 2021; Zhao and Tombola, 2021; Alvear-Arias et al., 2022). All current traces were followed by a fast ramp. The variance of the currents elicited by the fast ramp was obtained to determine the time at which the currents were reversed. The minimum value of variance indicates V = V H + .

For the limiting slope method, slow and symmetrical voltage ramps (0.5 mV/s–1 mV/s) were used. Currents were leak-subtracted off-line and filtered to 10 Hz when external noise was contaminating the currents. The data points were decimated (average of five points) and mean-binned (100 bins).

To estimate the effective charge coupled to the channel opening, a log-linear plot of the conductance vs. voltage was built, and the data were fitted to the equation at low open conductance values (at hyperpolarized potentials) as follows: G V = G 0 exp − z δ F V R T . (11)

In Eq. 7, G₀ represents a pre-exponential conductance factor. Subsequently, zδ was plotted against voltage using the following relationship: z δ V = k T e 0   d d V l n G V . (12)

In Eq. 8, k is the Boltzmann’s constant and e₀ is the elementary charge.

For calculation of proton conductances, tail currents were transformed into conductance values measured from isochrones taken at the tail of the currents, where I H + = G H + . (13)

All the conductance values were plotted against the potential and fitted to a normalized Boltzmann function. G H + = G max 1 + exp z δ F V − V 0.5 R T . (14)

G_max represents the maximum conductance. In a two-state channel model, the time constants for activation α (V) and deactivation β (V) can be represented as follows: τ V = 1 α V + β V , (15) where α (V) can be isolated from α (V) +β (V) at depolarized potentials, transforming into τ a c t V = 1 α V . (16) mHv1 isoform proton currents were generated by shortened-depolarizing voltage protocols coupled with a fast ramp and evaluated to determine if there was depletion of protons by checking the reversal potential. The time constant for mid-activation, τ_act, was obtained by fitting the previous macroscopic ON currents to a standard exponential function: I t = α 0 exp − τ a c t t + C . (17)

In the aforementioned equation, α (0) is the time constant at 0 mV, t represents the time, and C is a constant value. Assuming that MDSCs express the three mHv1 isoforms and that, at 0 mV, all these isoforms exhibit the same activation kinetic differences, then MDSC Hv1 currents at 0 mV were reproduced by the sum of three exponential functions: I t = α 1 0 exp − τ a c t , 1 t + α 2 0 exp − τ a c t , 2 t + α 3 0 exp − τ a c t , 3 t + C . (18)

Each exponential returned three time constants, representing the presence of each mHv1 isoform. All the data were analyzed using Clampfit 10.7, GraphPad Prism 6.0, and custom-made Python 3.8 scripts. Statistics are presented as mean ± standard error of the mean (SEM), unless otherwise reported.