Introduction

Voltage-gated, proton-permeable ion currents in a large variety of cell types and organisms are produced by the H_V1 gene (HVCN1 in humans), which encodes a membrane protein that is a member of the superfamily of voltage-sensing domains (VSDs) (1, 2). These VSDs are encountered in voltage-sensitive phosphatases (VSP) and voltage-gated ion channels (VGIC), and their principal function is to detect the membrane potential difference and translate it into a conformational change that activates VSP and opens VGIC (3).

H_V1 is thought to form ion channels activated by voltage and employing a mechanism of activation similar to the VSDs of canonical voltage-gated potassium, sodium and calcium channels (4). The functions of H_V1 channels are diverse, including intracellular pH regulation (5), charge compensation during immune response (6), modulation of flagellar beating in spermatozoa (7), bioluminescence (8), and possible roles in calcification processes in marine organisms (9, 10). Also, H_V1 is involved in different pathologies such as B cell malignancy (11), breast cancer (12), and post-ischemic brain injury (13, 14); consequently, in recent years H_V1 has emerged as a possible pharmacological target (15, 16).

Among all voltage-gated proton channels sequenced, the human orthologue, hH_V1, is the most widely studied (17). This channel is thought to be a functional dimer formed by two subunits comprising an intracellular N-terminus, a bundle of four transmembrane helices (TMH, S1-S4) in the VSD fold, and a long intracellular alpha helix that mediates a coiled-coil interaction, mainly responsible for dimerization and cooperative activation (18–20). The fourth alpha helix, S4, contains positive charges represented by three arginine residues (R205, R208, and R211) in the characteristic VSD repeats and is thought to undergo an outward displacement and rotation in response to depolarization, mostly in accordance to the helical screw rotation and displacement model of voltage-gating (21). Unlike canonical VGIC, H_V1 lacks the two TMH that make up the pore region in canonical ion channels (S5 and S6); therefore, the VSD of H_V1 has a double function: it is responsible for detecting the electrical potential across the membrane and forming the pathway through which the protons will move once the channel is activated.

A characteristic of H_V1 activation is its dependence on the pH gradient or ΔpH (pH_o - pH_i). Native proton channels were first shown to open at more negative voltages when the proton gradient points in the outward direction (22). It was shown that for every unit of ΔpH, the voltage of mid activation shifts ~40 mV. Subsequently, all native and cloned proton channels have been found to approximately follow this rule (1, 2, 10, 17, 23, 24). Recent experiments suggest that the proton gradient produces this effect by acting on the voltage sensor and not only affecting a close-to-open transition, since gating currents and channel opening are similarly modulated (25, 26). However, the molecular mechanisms through which protons modulate voltage-sensor function are not known. Due to technical difficulties, such as not having a pore structure separate from VSD or the impossibility of patch-clamping without protons in the experiments, gating current recordings of H_V1 channels have been obtained from mutants of the Ciona H_V1 (ciH_V1) orthologue (26, 27) or mutants of human H_V1 (28).

Patch-clamp Fluorometry (PCF) is a powerful tool that allows investigation of electrically silent conformational changes in VSDs associated with channel gating (29). In ciH_V1, this technique has been used to obtain evidence of cooperative gating, S1 movement, and the pH sensitivity S4 movement (25, 30, 31). Here, we assess the conformational changes of the S4 from the human voltage-gated proton channel in response to voltage activation and pH modulation with a genetically encoded fluorescent non-canonical amino acid, Anap, of similar size to aromatic amino acids, and a fluorescent signal with strong stability to acidity changes. We find our measurements can resolve a transition during the deactivation process that is strongly modulated by pH. Furthermore, we find that the changes in Anap fluorescence could be produced by the presence of aromatic amino acids within H_V1.

Results

Incorporation of Anap into hH_V1

To study voltage-dependent transitions in a voltage sensor using patch-clamp fluorometry, it’s desirable that the introduced fluorescent probe does not produce a major structural perturbation of the target protein. The relatively recently developed probe Anap (3-(6-acetylnaphtalen-2-ylamino)-2-aminopropanoic acid) is a small non-canonical fluorescent amino acid (Figure 1B) which has been shown to be easily genetically-encoded in proteins expressed in eukaryotic cells (32, 33) and useful as a reporter of voltage-dependent conformational changes (34) and as a FRET pair (35) to probe ion channel dynamics. In this study Anap was inserted into specific positions of the S4 segment of the human H_V1 proton channel (hH_V1) sequence (Figure 1A), with the purpose of examining its voltage and pH-dependent dynamics.

Figure 1.

We selected the S4 helix as insertion target, since this region of the channel is proposed to undergo a voltage-dependent outward displacement that has been previously studied with different approaches, including voltage-clamp fluorometry (13–15).

Anap efficiently rescued expression of channels containing an amber stop codon (TAG) in the indicated position, as judged both by appearance of red fluorescence produced by the mCherry fluorescent protein appended in the C-terminus or the appearance of proton currents recorded from HEK 293 cells in the whole-cell patch-clamp configuration (Figure 1C and Supplementary figure 1).

We were able to successfully substitute amino acids by Anap at positions Q191, A197, L198, G199, L200, L201 and I202 in the S4. The substituted channels gave rise to voltage-activated currents, with similar range of activation to WT as judged by their conductance vs. voltage (G-V) curves (except I202Anap channels) (Figure 1D and Supplementary figure 2 and Supplementary Table I). The observed fluorescence emission spectrum of the Anap signal present in the membrane (Figure 1C), which presumably originates mostly from Anap incorporated into channels, shows that there are no major or systematic variations on the peak emission wavelength (Figure 1E). The peak emission for all positions is between 480-487 nm, except the most C-terminal and presumably deepest position, I202Anap, which is 479 nm. This result suggests that the local environment of Anap in these positions is very similar and consistent with a polar environment since the peak emission of Anap in aqueous solution is ~486 nm (Supplementary Figure 3C and D). Figure 1E also shows that the emission spectra of positions other than Q191Anap exhibit a small extra peak near 610 nm that corresponds to the peak emission of mCherry. Since the 405 nm laser used to excite Anap does not excite mCherry, these results suggest that channel-incorporated Anap is able to undergo resonant energy-transfer (FRET) with mCherry. We did not attempt to quantify FRET between Anap and mCherry.

Insensitivity of Anap to pH

In order to use Anap as a reporter of conformational changes in H_V1 proton channels and given that these channels are able to change the pH of the surrounding solution (36, 37), we first wanted to validate if this fluorophore is insensitive to pH changes. The amino acid form of Anap that we use is the methyl-ester, which contains amino and carboxy groups, whose protonation could alter the electron distribution and thus its fluorescence, as a function of pH. In fact, the fluorescence of the free form of Anap in solution is highly sensitive to pH, with changes in both the peak emission wavelength and the intensity, presumably reflecting the pKa’s of amino and carboxy groups (Supplementary Figure 3E). We reasoned that the best assay to test the pH dependence of Anap fluorescence is to use already incorporated Anap. For this purpose, we used the mutant Q191Anap, which incorporates Anap in the S3-S4 loop which faces the extracellular solution, even in the deactivated state of the channel (Figure 1A). The emission spectrum of Q191Anap channels was measured from transfected HEK cells in which the pH of the extracellular solution was changed over the range 3 to 9.

These measurements showed that the fluorescence of Anap is not significantly changed in intensity or shape of the emission spectrum (Figure 2A and B), indicating that this fluorophore is insensitive to pH and that Anap fluorescence should not be affected by pH changes, which might be produced as a consequence of proton currents.

Figure 2.

As a further test of our data showing pH-insensitivity of channel-incorporated Anap and to validate the use of Anap in proton channels, we incorporated the amino acid in a position at the N-terminus of the channel, V62Anap. This amino acid is located in the intracellular part of the channel and should be subject to changes in local internal pH during channel activation (36) but not show changes in fluorescence as a function of voltage-dependent conformational changes. As expected, we did not detect Anap fluorescence changes, although the amino acid was incorporated into functional channels, as judged from proton currents recorded simultaneously with fluorescence (Figure 2C and D). This result further supports the use of Anap in voltage-gated proton channels to measure conformational changes.

Voltage-dependent changes of Anap fluorescence

Previous experiments in which other dyes like tetra-methyl-rhodamine maleimide (TMRM) were used to label cysteine residues in the S4 segment of voltage-sensing domains, including the Ciona and human H_V1 channels (25, 30, 31), usually result in fluorescence signals that are reduced upon depolarization by a voltage-dependent quenching process (38, 39). In contrast, when we incorporate Anap at position A197, located towards the extracellular end of the S4, depolarization induced an increase of the fluorescence, along with proton currents. The fluorescence increase saturates at depolarized voltages, suggesting that it is produced by a saturable process such as voltage-sensor activation. The direction of this fluorescence change is the same when measured at a ΔpH of 0 or 2, suggesting the same conformational change in the S4 voltage sensor occurs at different pH gradients, albeit over a different range of voltages (Figure 3A and B). The time course of the fluorescence change is very similar to the time course of proton current activation. The values of the time constants of activation (tau) are very similar, with the fluorescence signal slightly slower, but with comparable voltage dependence (Figure 3C).

Figure 3.

When F-V and G-V curves are plotted together, it is evident that sensor movement paralleled the activation of the proton conductance. At three different values of the pH gradient (ΔpH 0, 1 and 2), both the F-V and G-V curves are almost superimposable and shift along the voltage axis by the same amount of ~40 mV/pH unit (Figure 3D), which is expected of H_V1 channels (22). Only at ΔpH = 2 the fluorescence signal is shifted to slightly more negative voltages than the conductance and only at voltages at which channel activation begins. The observed voltage shift of the G-V is ~31 mV from ΔpH 2 to 1 and ~43 mV from ΔpH 1 to 0 and is very similar for the F-V curves. This result indicates that the ΔpH-dependence of gating is preserved in channels with incorporated Anap and that the voltage sensor movement occurs in the same voltage range as the formation of the proton permeation pathway.

To understand the origin of the increased fluorescence observed during activation, we measured the emission spectra of A197Anap in voltage-clamped cells at different voltages. Figure 3E plots the normalized spectra obtained at voltages ranging from −100 to 140 mV and it shows that the spectra superimpose well, indicating that the increase in fluorescence is not due to depolarization-driven wavelength shifts of the emission spectra. We interpret this result as an indication that Anap incorporated at position A197 remains in a polar environment at all voltages or that small changes in polarity change the quantum yield of Anap but not the emission spectrum.

H_V1-197Anap is quenched by a phenylalanine in the S2

The increase of the Anap fluorescence at position 197 in the S4 seen with depolarization could be interpreted as a reflection of an outward movement of the S4 and exposure of Anap to a more polar environment, which in principle will produce a red shift of the emission spectrum and an increase of the fluorescence that is detected. However, as shown in Figure 3E, the shape of emission spectrum of Anap remains unchanged at all voltages and with a constant emission peak at ~480 nm, indicating that the fluorophore remains in a polar environment in the closed and open states and thus, a change in local polarity is not the cause of dequenching.

Anap was derived from Prodan (32) and although the photophysical behavior of Anap is not well characterized, Prodan’s fluorescence is known to be quenched by mechanisms such as π-stacking or photoinduced electron transfer with aromatic amino acids (40), both mechanisms which require close proximity. For these reasons, it is conceivable that the Anap quencher in H_V1, could be an aromatic residue that is located near the introduced fluorophore in the closed state and upon S4 movement, increases its distance, generating the observed dequenching. We used a structural model of H_V1 derived from experimental data (41) and replaced A197 with Anap. Figure 4A shows Anap in salmon-colored spheres and highlights aromatic residues within the transmembrane domains of a monomer as dotted spheres. A possible candidate for an Anap quencher is F150 (yellow spheres), because this residue is the closest aromatic to Anap that is not in the S4 and F150 will presumably remain in its position as 197Anap undergoes an outward displacement with depolarization. In contrast, other aromatic residues which are closer to 197Anap and are part of the S4, will presumably move with all the S4 as a rigid body. Incidentally, an equivalent phenylalanine to F150 has been identified as the charge transfer center in canonical voltage-gated potassium channels and in H_V1 (42, 43).

Figure 4.

To test this hypothesis, we made the double mutant F150A-A197Anap and estimated the relative amount of basal Anap quenching, by comparing the emission spectra of both Anap and mCherry in the same membrane region. Figures 4B and C shows that the double mutant displays a significantly increased Anap fluorescence (~60 %) relative to mCherry, when compared to A197Anap alone, suggesting that indeed, phenylalanine 150 is capable of quenching Anap in the closed state (at the resting potential of HEK cells of −20 to −40 mV (44) and at the employed ΔpH ~ −0.2 (pH_o = 7) most channels should be in the closed state). Despite having removed the quenching group, F150A-A197Anap channels still show voltage-dependent fluorescence changes (Figure 4D), suggesting the presence of additional quenchers or that in the absence of F150, Anap at 197 becomes sensitive to polarity changes.

The voltage dependence of the fluorescence signals from F150A-A197Anap channels shows significant differences from those of A197Anap alone. At values of ΔpH of 0 and 1, fluorescence precedes the increase in conductance, indicating that the conformational change of the S4 segment occurs at more negative voltages than the formation of the proton-permeable pathway. This effect is more pronounced at ΔpH = 1. Interestingly, the difference of V_0.5 of the F-V curve between ΔpH= 0 and 1 is 57.8 mV, similar for A197Anap, which is 45.7 mV.

A distinct gating transition detected by Anap fluorescence

The fluorescence time course of the F150A-A197Anap channels shows an interesting characteristic, which is that the OFF signals (F_off) that are produced at the return of the voltage to the holding potential and represent the return of the voltage sensor to the resting states, show a two-component kinetic behavior. This is particularly evident at ΔpH = 0 (Figure 4D), where F_off shows a very rapid quenching followed by a much slower component, suggesting that the voltage sensor can move back to its resting position at varying rates.

To explore the kinetics of fluorescence signals, especially during repolarization, and since this double mutant removes a quenching group, we used the hH_V1-201Anap channels. We reasoned that this mutant channel, which has Anap in a deeper position in the S4 and presumably closer to F150 in the closed state, might be a better reporter of the kinetics of S4 movement.

Figures 5A, B and C show simultaneous current and fluorescence recordings from hH_V1-L201Anap channels at three different ΔpH values of 0, 1 and 2. As with the hH_V1-A197Anap construct, the voltage-dependence of the conductance and fluorescence are almost superimposable and shows a large shift of >40 mV/pH unit (Figure 5D).

Figure 5.

The most salient feature of these fluorescence traces is that, at ΔpH = 0, the OFF signal during repolarization (F_off) has two distinct kinetic components. The deactivation tail currents at −60 mV decay exponentially, with a time constant of 141 ±55 ms, while the F_off can be fit to a sum of two exponentials with time constants of 129 ± 68 ms and 8.6 ± 0.74 s. (Supplementary figure 4). The presence of the two components in F_off suggest that the return of the voltage sensor to its resting state can occur at varying rates. In particular, the slow component is consistent with the immobilization of the off-gating charge observed in monomeric Ciona H_V1 channels (27). The slow off-component is also present at ΔpH = 1 and 2, although its amplitude is smaller. The slow return of the voltage sensor at ΔpH = 0 is also consistent with the more positive voltages needed to activate the channel at this pH gradient.

To qualitatively understand the kinetics of the fluorescence signals, we used a simplified kinetic model of channel activation (Scheme I), similar to a model that was previously used to study the voltage-dependent kinetics of hH_V1 (45).

In this model, one of the backward transitions (k₂₁) between closed states is set to be much slower than the open to closed transition (O₃->C₂), resulting in a large difference between the closing kinetics of the ionic current, mostly determined by k₃₂, and the fluorescence signals. This kinetic difference can account for the biphasic behavior of the F_off signal, and especially the slow component of its time course (Figure 5D). The model also indicates that when the internal pH is lower than the external pH (ΔpH=2), this slow rate constant is more affected than any other, indicating a conformational step that is especially sensitive to pH.

While the simple model in Scheme I can account qualitatively for the observed kinetics of 201Anap channels, the experimental F-V relationship is positively shifted by ~10 mV with respect to the G-V curve, which is not a feature of Scheme I and is reminiscent of channels that can open to multiple open states, without the need of full voltage sensor activation (46). This observation suggests that hH_V1 channels operate via a more complicated mechanism that the one illustrated by Scheme I, which might include channel opening before complete voltage-sensor movement. We tested a simple version of such an allosteric model and show that it can account, at least qualitatively, for current and fluorescence kinetics and for the relationships between G-V and F-V curves at varying ΔpH (Supplementary figure 5). Interestingly, in this model the slow deactivation rate constant is also the step with the most sensitivity to pH (Supplementary Table III).

Absolute pH values are determinants of voltage sensor movement

One of the most intriguing characteristics of H_V1 channel gating, is its steep modulation by the pH gradient. While it has been shown that this modulation depends on the value of ΔpH, regardless of how it is set up (22), there is evidence that the absolute value of pH can also exert an effect on gating (47). In most of our experiments, the pH gradient was set up with a low value of intracellular pH, between 5.5 and 6. To test the effect of absolute pH, we carried out experiments with the same ΔpH of 0, with symmetric low (5.5/5.5) or high (7/7) intra/extracellular pH. The expectation was that, if pH gating of hH_V1 depends only on the pH gradient, the voltage sensor should move with essentially the same characteristics. Surprisingly, the fluorescence signals display important differences, as do the proton currents. Our results in Figure 6 show that when compared to ΔpH of 0 (5.5/5.5), the fluorescence in symmetric pH_o and pH_i of 7 has a rapid return of the F_off signal (Figure 6A and B). Interestingly, the voltage dependence of the F-V relationship is very similar for (5.5/5.5) or (7/7) conditions, while in (7/7) the proton current appears at more positive voltages than the bulk of the fluorescence (Figure 6C).

Figure 6.

These results suggest that the voltage range of movement of the voltage sensor is dependent on the ΔpH, while the opening of the proton conduction pathway can occur after only a fraction of the voltage sensor movement has occurred and this coupling between voltage sensing and channel opening can be regulated by low pH.

Discussion

In the experiments described here we have implemented patch-clamp fluorometry in combination with incorporation of a fluorescent non-canonical amino acid (NCAA) to study voltage-dependent gating in hH_V1 proton channels. Although voltage-clamp fluorometry (VCF) has been used previously to study H_V1 channels (25, 30, 31), employing the fluorescent Anap NCAA has the advantages of being a smaller size probe and improving the specificity of fluorescence signals, since it is genetically encoded. We were able to incorporate Anap into functional channels in several sites along the S4 and the S3-S4 loop. Since Anap was developed as an environmental sensitive probe, the fact that the emission spectrum of Anap in these sites is very similar to that of Anap in water, suggests that these residues are solvated in the native H_V1. The only position that shows a blue-shifted Anap spectrum is I202 which is the most C-terminal residue explored and might burry the Anap R-group in a more hydrophobic environment.

Since the activity of H_V1 proton channels can change the local concentration of protons near the conduction site and fluorescence probes have been used to detect these proton fluxes (36, 37), we addressed whether Anap could change its fluorescence as a function of pH. We show that Anap is highly insensitive to pH in the range 4 to 8 and it does not change fluorescence in conditions in which high outward fluxes can change the local intracellular pH. Our experiments confirm that Anap can be used without interference from local changes in proton concentration.

When incorporated at position 197, Anap produced fluorescence signals that indicate an increase in intensity with depolarization and saturated in magnitude at positive potentials. This behavior indicates dequenching of Anap as the S4 segments undergoes an outward movement during the activation conformational change. Anap has been incorporated in other membrane proteins, including the Shaker potassium channel, in which Anap was incorporated in the S4-S5 linker and displayed fluorescence quenching upon depolarization (48). As far as we know, Anap has been incorporated in the S4 of only the hyperpolarization-activated cyclic nucleotide-gated (HCN) channel (49), where it is quenched or dequenched upon hyperpolarization in a position-dependent manner. The direction of the fluorescence changes due to S4 motion are difficult to predict, since, as we have shown, Anap’s fluorescence can be affected by both the local environment’s polarity and interaction with specific quenching groups that are part of the channel sequence.

The fluorescence changes we observe in 197Anap channels indicate that the G-V and F-V relationships have almost the same voltage-dependence at the ΔpH values tested, suggesting that S4 movement closely follows channel opening, and that S4 movement and activation of the proton conductance are equally affected by the proton gradient. A similar conclusion has been reached in studies measuring S4 movements of hH_V1 by fluorescence (25) or in Ciona H_V1 by gating current recordings (26). Interestingly, these changes in fluorescence as a function of voltage, are not accompanied by changes in the emission spectrum of Anap, suggesting that the probe remains in a solvated environment regardless of the state of the channel. This is in accordance with the finding that the VSD that forms hH_V1 channels has a large extracellular cavity able to contain many water molecules (50). Since Anap remains solvated in the closed and open state, what is the origin of the reported fluorescence changes? As a Prodan-derived probe, we hypothesized that an aromatic residue could act as an Anap quencher and thus found that the only aromatic outside the S4 close enough to have this function is F150. Mutation F150A in the 197Anap background produced an increased Anap/mCherry fluorescence intensity as compared to 197Anap alone, indicating reduced quenching. This result suggests that 197Anap moves away from F150 as the S4 segment moves outward during channel activation.

Substitution of L201 for Anap allowed us to uncover a slow step in the activation pathway. The fluorescence signal observed upon channel closure by repolarization at ΔpH = 0 shows two components, one which is much slower than channel closing as reported by the tail current. The fact that tail current is faster than the slow component of the deactivation fluorescence signal indicates that the latter is produced by a slow intermediate transition, as is also recapitulated by a simple kinetic model. Interestingly, recordings of gating currents in mutant Ciona H_V1 channels show that the charge return after depolarization can be very slow, producing gating charge immobilization (27). This observation of a singular slow transition in hH_V1 activation illustrates the value of fluorescence recording with a small probe such as Anap.

Our data thus far indicates that a fraction of S4 movement, as reported by the F-V relation, occurs before the increase of the proton conductance, and that S4 movement can continue after channel activation. Comparison of the V_0.5 values of Q-V and G-V curves in hH_V1 channels (28) indicates that charge moves at slightly more negative values than conductance, but not at all ΔpH values. Fluorescence changes depend on all the conformational states in which the fluorophore has distinct fluorescence values, while gating currents are produced during transitions between conformations with state-dependent charge distributions (51). For these reasons, F-V and Q-V curves of multistate channels are not expected to be identical or contain the same information.

Our fluorescence data are consistent with recent experiments that have shown that the characteristic gating effect of the proton gradient on voltage-gated proton channels comes about by a conformational change that affects voltage sensor movement and not a channel opening transition. Furthermore, our modeling suggests that all transitions in the activation pathway, including a characteristic slow transition detected by fluorescence are modulated by ΔpH.

The mechanism of ΔpH modulation is still unknown. It has been proposed that the energy stored in the pH gradient is directly coupled to S4 movement to produce ΔpH-dependent gating (26). We have previously proposed an allosteric model in which both extracellular and intracellular protons can affect local electrostatic networks and bring about ΔpH-gating (10). This class of models predicts the existence of multiple open states, which is supported by the observation that S4 movement can happen after channel opening.

A mechanism in which the proton gradient energy is coupled to S4 movement predicts that the absolute value of pH should not influence gating. Interestingly, we have observed that the absolute pH values used to set up a ΔpH = 0 do affect gating. When pH_o=5.5/pH_i=5.5, G-V and F-V are almost superimposed and the F_off signal has a fast and slow component; in contrast, when pH_o=7/pH_i=7, the F-V curve has almost the same voltage-dependence, but conductance can be observed at more negative voltages and the F_off signal only contains the fast component. These results suggest that the absolute pH in the extracellular side of the channel is a determinant of the steady-state gating, presumably modulating the slow rate constant in the activation pathway.

Materials and Methods

Molecular biology and HEK cell expression

A plasmid containing the human voltage-gated proton channel (hH_V1) was a gift from Dr. Ian Scott Ramsey (Virginia Commonwealth University, Richmond, VA). We used the fluorescent protein mCherry as a reporter to verify L-Anap incorporation. The construct hH_V1-mCherry was made by the PCR overlap technique, adding the sequence of fluorescent protein mCherry after the C-terminus of hH_V1 with the following linker sequence: (Gly-Gly-Ser)₃. This construct was subcloned into the pCDNA3.1 vector. For all hH_V1-TAG mutants, an amber codon (TAG) was introduced using appropriate mutagenic oligonucleotides and a protocol for whole plasmid site-directed mutagenesis employing KOD polymerase (Merck Millipore, Germany) as detailed in manufacturer’s instructions and previous work (52, 53). The bacterial methylated DNA templates were digested with the DpnI restriction enzyme, and the mutant plasmids were confirmed by automatic sequencing at the Instituto de Fisiología Celular, UNAM.

We used HEK293 cells for channel expression and L-Anap incorporation experiments. The HEK cells used in this study were found free of mycoplasma infection (Sigma-Aldrich mycoplasma detection kit). These cells were cotransfected with 0.1 – 1 μg of mutant hH_V1-TAG plasmid and 0.7 μg of pAnap plasmid (a gift from Dr. Sharona Gordon, University of Washington, Seattle, WA) using the transfection reagent JetPei (Polyplus-transfection).

The pANAP plasmid contains the orthogonal pair tRNA/aminoacyl tRNA synthetase specific to L-Anap. The Methyl ester form of L-Anap; L-Anap-Me (AsisChem Inc.) was added to the medium of cells in 35 mm culture dishes from a storing stock solution of 10 mM to a final concentration of 10-20 μM. Through the text, we will refer to L-Anap as Anap for simplicity. Cells were incubated during 12-48 hours before experiments in Dulbecco’s Modified Eagle Medium (DMEM, Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen, USA) and penicillin-streptomycin (100 units/ml — 100 μg/ml, Invitrogen, USA) at 37 °C in a 5% CO₂ atmosphere. Around 4 hours before electrophysiological recordings, HEK293 cells were treated with 0.05% trypsin-ethylenediaminetetraacetic acid (Trypsin-EDTA) to obtain rounded cells, which were then re-platted in 35 mm glass-bottom dishes (World Precision Instruments, USA) and used for experiments within 3-6 hrs. All the experiments were performed at room temperature (~25°C).

Electrophysiology

Recordings of proton currents were performed in the whole-cell patch-clamp configuration using fire-polished borosilicate micropipettes (Sutter Instruments, USA). Currents were recorded by an Axoclamp 200B amplifier (Axon Instruments, USA) and acquired with an ITC-18 AD/DA converter (HEKA Elektronik, Germany), both controlled with Patchmaster software (HEKA Elektronik, Germany). Currents were low-passed filtered at 5 kHz and sampled at 20 kHz. The extracellular solution contained (in mM): 100 tetramethylammonium hydroxide and methanesulfonic acid (TMAOH-HMESO₃), 100 buffer ((2-(N-morpholino)ethanesulfonicacid (MES) for pH 5.5, and 6.0; 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) for pH 7.0 and 7.5), 2 CaCl₂, 2 MgCl₂, 8 HCl and pH-adjusted with TMAOH and HMESO. The intracellular solution contained (in mM): 80 (TMAOH-HMESO₃) 100 buffer (MES for pH 5.5 and 6.0; HEPES for pH 7.0 and 7.5), 10 ethylene glycol-bis(β-aminoethyl ether)-N,N,N’,N’-tetraacetic acid (EGTA), 10 MgCl₂, and 4 HC1 and pH-adjusted with TMAOH and HMESO. With these solutions, patch pipettes had a resistance of 2-5 MΩ. Since cells used in these experiments were round to improve space-clamp and currents were relatively small, no series-resistance compensation was employed. The voltage-clamp protocols varied depending on the value of ΔpH and are indicated in the figure legends. The interval between each test pulse was 45 s at the holding potential to facilitate return of slow fluorescence signals and minimize the effects of proton depletion.

Fluorescence measurements

Fluorescence measurements in whole-cell fluorometry (WCF) experiments were made using a TE-2000U (Nikon, Japan) inverted epifluorescence microscope with a 60x oil immersion objective (numerical aperture 1.4). A 405 nm solid-state laser (Compass 405-50 CW, COHERENT, USA) and a filter cube containing a 405/20-nm excitation filter, a 405-nm long pass dichroic mirror, and a 425-nm long-pass emission filter were used for Anap fluorescence excitation. For mCherry fluorescence, measurements were performed using an Ar-Ion laser (Spectra-Physics, Germany) and a filter cube with a 514/10-nm excitation filter, a 514-nm long pass dichroic mirror, and a 530-nm long-pass emission filter (Chroma, USA). Both lasers were through-air coupled, collimated using an optical cage system and appropriate optics (Thorlabs, USA) and then focused into the objective’s back focal plane through the microscope’s rear port. Imaging was performed using Luca or Ixon Ultra EMCCD cameras (Andor, Oxford instruments, Ireland) controlled by Micromanager software (54). The fluorescence of a region without cells was measured with the same ROI employed with cells and this background was subtracted from Anap fluorescence images. Image stacks from cells were recorded at 10-25 Hz (100 – 40 ms of exposure, respectively). To improve signal-to-noise ratio, 4×4 or 8×8 pixel binning was used. Initially, fluorescence time course was measured from a region of interest (ROI) that included only the membrane of the patched cell. Identical results were obtained by using a ROI encompassing all the cell.

Fluorescence and proton current recording synchronization was achieved through a home-programed Arduino Uno microcontroller board (Arduino, Italy) triggered by a PatchMaster-generated TTL pulse.

For spectral measurements, the light from the microscope was collected by a SpectraPro 2150i imaging spectrograph (Princeton Instruments, USA) mounted between the microscope and EMCCD camera. The spectra of both fluorophores (Anap and mCherry) were recorded by measuring line scans of the spectral image of the cell membrane, and the background fluorescence from a region of the image without cells was subtracted.

Data analysis

IgorPro (Wavemetrics) and ImageJ (NIH) software were used to analyze the data. For the G-V relationships, conductance (G) was calculated from proton currents according to:

Where V_rev is the proton current reversal potential, measured from the current-voltage relation. Then, G was normalized and fit to a Boltzmann function as follows:

Where q is the apparent gating charge (in elementary charges, e₀), V is the membrane potential, V_1/5 is the potential at which half of the maximal activation is reached, K_B is the Boltzmann constant and T is the temperature in Kelvin.

The time course of fluorescence in WCF experiments, was obtained from all the background-subtracted images in a stack (F_i), and the changes through time were normalized to the first image (F₀) as follows:

Then, this normalization was multiplied by 100 to obtain the percent change of fluorescence. This procedure was carried out for each stack at each voltage. The voltage-dependence of the fluorescence was estimated from F-V relationships. The value of the fluorescence at the end of the volage step was normalized and fit to a Boltzmann function:

Where F is the percent of fluorescence change at V potential and F_max is the maximum fluorescence percent change in each experiment at V potential. The meaning of q, V, V_0.5 and K_BT is the same as in equation 1. All data are presented as the mean ± standard error of the mean (s.e.m.).

The time constants activation of proton currents and fluorescence signals were obtained by fitting of the second half of each trace to the equation:

Where A_ss is the amplitude of the fluorescent signal (F) or current (I) at steady state, τ is the time constant, and t_o is the time of start if the voltage pulse, both in ms. The voltage dependence of τ was calculated from a fit to equation:

Where τ(0) is the activation time constant at 0 mV, q is the partial charge in e₀ units and V, K_B and T have the same meaning as in equation 1.

Modelling

Modelling of current and fluorescence was carried out using custom-written programs in IgorPro (Wavemetrics). The occupancy of each discreet state in the models was calculating by numerically solving the differential equations describing the transitions between states. The occupancy of each discreet state i is P_i and it was calculated by numerically solving the differential equations described by a master equation:

The rate constants k_ij or k_ji are given by:

Where k_ij(0) is the value of the rate constant at 0 mV, z_ij is the partial charge associated with the transition and K_BT have the same meaning as in Eq.1.

The current as a function of time t and voltage V was calculated as:

γ_ch is the single proton channel conductance, N is the number of channels, V_rev is the reversal potential and P_o is the probability of the open state. γ_leak is the leak conductance and is the reversal potential of the leak currents.

The fluorescence was calculated as:

f_i is the fluorescence of the i-th state in arbitrary units.

Supporting information

Supplementary data

Acknowledgements

We thank Eduardo Guevara for measurements of Anap spectra in different solvents and Manuel Hernández for excellent technical support. We thank Dr. Sebastian Brauchi for the loan of the 405 nm laser. This work was supported by DGAPA-PAPIIT-UNAM grant No. IN215621. E. S-D is a doctoral student from Programa de Doctorado en Ciencias Bioquímicas-UNAM and was supported by a doctoral thesis scholarship from CONACyT No. 463819 (grantee number: 576613).