Characterization of a proton-activated, outwardly rectifying anion channel


Corresponding author J. Oberwinkler: Experimentelle und klinische Pharmakologie und Toxikologie, Gebäude 46, Uniklinikum des Saarlandes, 66421 Homburg, Germany. Email:


Anion channels are present in every mammalian cell and serve many different functions, including cell volume regulation, ion transport across epithelia, regulation of membrane potential and vesicular acidification. Here we characterize a proton-activated, outwardly rectifying current endogenously expressed in HEK293 cells. Binding of three to four protons activated the anion permeable channels at external pH below 5.5 (50% activation at pH 5.1). The proton-activated current is strongly outwardly rectifying, due to an outwardly rectifying single channel conductance and an additional voltage dependent facilitation at depolarized membrane potentials. The anion channel blocker 4,4′-diisothiocyanostilbene-2,2′-disulphonic acid (DIDS) rapidly and potently inhibited the channel (IC50: 2.9 μm). Flufenamic acid blocked this channel only slowly, while mibefradil and amiloride at high concentrations had no effect. As determined from reversal potential measurements under bi-ionic conditions, the relative permeability sequence of this channel was SCN> I> NO3> Br> Cl. None of the previously characterized anion channel matches the properties of the proton-activated, outwardly rectifying channel. Specifically, the proton-activated and the volume-regulated anion channels are two distinct and separable populations of ion channels, each having its own set of biophysical and pharmacological properties. We also demonstrate endogenous proton-activated currents in primary cultured hippocampal astrocytes. The proton-activated current in astrocytes is also carried by anions, strongly outwardly rectifying, voltage dependent and inhibited by DIDS. Proton-activated, outwardly rectifying anion channels therefore may be a broadly expressed part of the anionic channel repertoire of mammalian cells.

Channels permeable to anions are extremely widespread and believed to be present in every mammalian cell. In native tissues, Cl ions are the most important anions transported by these channels. Depending on their expression profile, their intracellular targeting and their biophysical properties, these channels serve a wide variety of cellular functions. They have been implicated in cellular volume regulation, ionic homeostasis, transepithelial ion transport, the control of electrical excitability and vesicular acidification (for reviews, see Nilius et al. 1997a; Frings et al. 1999; Jentsch et al. 2002). Meanwhile, many genes from several families have been identified that encode bona fide chloride channels. On the other hand, there still exist electrophysiologically prominent and well characterized anion channels that have not yet been identified at the molecular level (Jentsch et al. 2002; Nilius & Droogmans, 2003).

Biophysically, anion channels are very diverse. The shape of reported current–voltage relationships ranges from inwardly rectifying (CLC-2; Jordt & Jentsch, 1997) to steeply outwardly rectifying (ClC-4 and ClC-5; Friedrich et al. 1999) with many intermediate cases (e.g. the volume-regulated anion channel; Nilius et al. 1997a). Some anion channels are not voltage dependent (e.g. the cystic fibrosis transport regulator, CFTR), while others show voltage-dependent facilitation upon depolarization to positive membrane potentials (e.g. ClC-4, ClC-5; Friedrich et al. 1999). The volume-regulated anion channels (VRAC), however, deactivate under the same conditions (Nilius et al. 1997a). In Ca2+-acitvated Cl channels, the amount of voltage dependence can be regulated by the intracellular Ca2+ concentration (Evans & Marty, 1986).

Channels permeable to anions generally discriminate much less between different anionic species compared to cationic channels (Hille, 1992). Typically all halide anions can pass any given anion channel, but differences exist in the rank order of permeabilities for halides between different anion channels. For example, all members of the ClC family so far characterized are more permeable to Cl than to I, while the reverse is true for VRAC, Ca2+-activated Cl channels, GABAA receptors, glycine receptors, and chloride channels associated with excitatory amino acid transporters (summarized by Jentsch et al. 2002). Equally, in general the available pharmacological tools discriminate poorly between different anion channels. Nevertheless, under some circumstances, pharmacology can be employed to discriminate between different anion channels. For example, ClC-5 has been reported to be unaffected by 1 mm 4,4′-diisothiocyanostilbene-2,2′-disulphonic acid (DIDS; Steinmeyer et al. 1995), while the outward current of VRAC is inhibited by this substance with an IC50 of 26 μm (Hélix et al. 2003).

In the present paper, we characterize a current that is activated by very acidic extracellular conditions (pH below 5.5). In the initial experiments, we established that this current was carried by anions and that it is caused by the activation of a proton-activated, anion permeable channel. In order to test whether this channel corresponds to any of the known anion channel, we analysed the voltage-dependent behaviour, the pharmacology and ionic selectivity of this channel. We thereby established that this current is strongly outwardly rectifying, voltage dependent and unusually sensitive to DIDS and displays an ionic permeability sequence similar to Ca2+-activated and volume-regulated anion channels, but different from the characterized members of the ClC family. This profile clearly distinguishes the proton-activated, outwardly rectifying anion channel from the previously characterized channels. Nevertheless, Nobles et al. (2004) recently argued that the proton-activated, outwardly rectifying anion current is mediated by the same channels as the volume-regulated anion channel. Here, we report several independent lines of experimental evidence that demonstrate that proton-activated and volume-regulated anion channels are two distinct populations of ion channels that can be, under appropriate experimental conditions, simultaneously active in the same cell. Importantly, however, they then can still be separated from each other biophysically. We also establish that a proton-activated, outwardly rectifying anion current exists endogenously in hippocampal astrocytes.


Cell culture

HEK293 cells were grown in minimal essential medium (Invitrogen, Karlsruhe, Germany) supplemented with 10% fetal calf serum (FCS) in a humidified atmosphere at 37°C containing 5% CO2. Cells were passaged three times a week (using trypsin/EDTA, Sigma, Taufkirchen, Germany) until they reached passage 30. For experimental usage, they were passaged and plated at low densities in 35 mm diameter plastic culture dishes at least 12 h before the experiment. Immediately before the experiments, the culture medium was removed and cells were washed at least 3 times with the relevant bath solution. Experiments were carried out at room temperature and only single cells not connecting to neighbouring cells were used for experiments.

Hippocampal astrocytes were prepared from neonatal wild-type mice (C57BL6/N and 129/SvJ). Neonatal mice (postnatal day 1–2) were killed by decapitation, and the brains quickly removed from the skull and placed in ice-cold Hanks' balanced salt solution (Invitrogen). All procedures were in accordance with the German legislation (Deutsches Tierschutzgesetz). Hippocampi were dissected free and mechanically dissociated by gently pushing the tissue through a cell strainer (100 μm, BD Biosciences, Heidelberg, Germany) into a 50 ml plastic tube containing Dulbecco's modified Eagle's medium (DMEM, Invitrogen). The resulting suspension was centrifuged and the cell pellet resuspended in 1 ml DMEM. The cell density of the resulting suspension was determined and cells plated at a density of 0.5–1 × 106 cells per 75 cm2 collagen (Typ I, rat tail, BD Biosciences) coated culture flask. They were maintained in an incubator at 37°C in a humid atmosphere containing 5% CO2. The culture medium consisted of DMEM supplemented with 10% FCS, Pen/Strep (Invitrogen) and 1% MITO (BD Biosciences). The culture medium was replaced twice weekly until the adherent cell layer reached 70–80% confluence. Subsequently, cells were washed with medium, trypsinated and the resulting cell suspension plated on collagen coated coverslips at low density. Cells were allowed to adhere to the coverslip for at least 48 h before experimentation and only single, isolated cells of polygonal shape were used.


Standard tight-seal, whole-cell or excised (inside-out) patch recordings were performed with HEKA EPC-9 or EPC-10 amplifiers under the control of Pulse or Patchmaster software (HEKA, Lambrecht, Germany). Electrodes were pulled from thin-walled borosilicate glass and had a resistance of 2–5 MΩ when filled with standard intracellular solution and immersed in bath solution.

Electrodes were coated with Sigmacote (Sigma), in order to reduce capacitative transients. In whole-cell experiments, capacitative transients arising from the capacitance of the cell were compensated just before each measurement (voltage ramp or voltage step) with the built-in automatic compensation feature of the recording software. The series resistance then was routinely compensated for 80%. In the initial experiments, the reference electrode consisted of a carefully chlorided silver wire that we replaced with an agar bridge (made with bath solution) in later experiments, including all those experiments in which we determined the reversal potentials. In excised patch recordings we used the chlorided silverwire as reference electrode.

HEK293 cells were voltage clamped to a membrane potential between −16 mV and 0 mV (depending on liquid junction potential, see below), while astrocytes were held at −70 to −80 mV. At a frequency of 0.5–1 Hz, the current–voltage relationship was probed with voltage ramps consisting of an initial jump of the membrane potential to −115 mV for 20 ms and a subsequent increase to +85 mV at a rate of 1 mV ms−1. Alternatively, the membrane potential was changed in various pulse protocols as indicated. The extracellular solution was rapidly exchanged (within 2–3 s) by placing the outlet of a custom-built manifold with small dead-volume at a distance of less than 200 μm from the recorded cells.

Composition of solutions

Table 1 shows the composition of the intracellular solutions, Table 2 lists the extracellular solutions used with HEK293 cells, while Table 3 contains the extracellular solutions used when working with astrocytes. In Table 4 we detail the composition of solutions used in experiments where we exposed the cells to hypotonic extracellular conditions. For the adjustments of pH we used HCl or the hydroxide salt of the main cation. The osmolality was adjusted with H2O or d-glucose. Values in Tables 1–4 indicate final concentrations after these adjustments. For excised patch recordings, the solution in the recording dish was solution iII (Table 1) and the recording pipette contained solution eII at pH 5.0 or pH 6.0 (Table 2).

Table 1.  Composition of solutions used inside the patch pipette
  1. All values indicate the final concentrations after adjustment of pH and osmolality. Solutions were adjusted to a pH of 7.2 and an osmolality of 305–315 mosmol kg−1. All solutions additionally contained 10 mm HEPES to buffer the pH, except solution iIX that contained 100 mm HEPES. †Indicated are free divalent concentrations as calculated with MaxChelator ( ‡The addition of 4 mm Na2ATP also causes a final concentration of 8 mm Na+ in the solution. Abbreviations: LJP: liquid junction potential between solution in the patch pipette and standard bath solution (eI, Table 2); Asp: Aspartate.

iI1627 201301016.1 
iVI14810010 1484104.7
iVII16210010 149106.2
iVIII15710 145104.2
iIX1197 80106.1
iX 887 80106.4
Table 2.  Ionic composition of extracellular solutions used with HEK293 cells
Other anions
  1. Indicated are the final concentrations after adjustment of pH and osmolality. All solutions were adjusted to 320–330 mosmol kg−1, except solution eVIII, which was intentionally made hypertonic (578 mosmol kg−1). Solutions eI and eIV were buffered with 10 mm HEPES, all other solutions with 5 mm citrate. However, solution eV contained 141 mm citrate (as indicated) and no other buffer. †Indicated is the free divalent concentration. The total amount of the divalent cations needed to achieve the indicated free concentration was calculated with MaxChelator ( and added to the solution. Abbreviations: Cit: citrate; Asp: aspartate; Glu: glutamate; LJP: liquid junction potential between the solution used as bath solution (eI) and the relevant other extracellular solution.

eV21022218141 Cit4.1−2.3
eVI1392263145 Asp4.8−10.8
eVII2602263145 Glu4.8−17.4
eVIII284221513145 Asp4.8−3.2
eXIII150224.62.2145 SCN4.0−2.6 
eXIV1622263150 J4.80
eXV1622263150 NO34.8−0.9 
eXVI1662263150 Br4.80
Table 3.  Ionic composition of extracellular solutions used with hippocampal astrocytes
Other anions
  1. Indicated are the final concentrations after adjustment of pH and osmolality. All solutions were adjusted to 345–355 mosmol kg−1. Solution eXVIII was buffered with 10 mm HEPES, all other solutions with 5 mm citrate. †Indicated is the free divalent concentration. Abbreviations: LJP: liquid junction potential between the solution used as bath solution (eXVIII) and the relevant other extracellular solution.

eXX159120 citrate4.0−4.2  
eXXII307150 SO42−4.0−6.5  
Table 4.  Ionic composition of extracellular solutions used in experiments to induce cell swelling by hypotonicity
(mosmol kg−1)
  1. Indicated are the final concentrations after adjustment of pH and osmolality. Solutions eXXIII and eXXIV were buffered with 10 mm HEPES, the other solutions with 5 mm citrate. The liquid junction potential between pairs of these solutions was calculated to be less than 1 mV. †Indicated is the free divalent concentration.

eXXIII 94310221117.2320–330
eXXIV 9422 987.2195–205

Stock solutions of 4,4′-diisothiocyanatostilbene-2,2′-disulphonic acid disodium salt (DIDS) and mibefradil were prepared in H2O (10 mm). The stock solution of flufenamic acid (FFA) was prepared by dissolving it in ethanol (50 mm). Amiloride was added directly to the acidic extracellular solution (eII) at the final concentration of 0.5 mm. All chemicals were obtained from Sigma and were of the highest grade available.

Correction of liquid junction potentials and data analysis

We routinely corrected for the liquid junction potential arising between the pipette solutions and the standard bath solution. The values for the liquid junction potentials were calculated with Clampex 8.1 (Axon Instruments, Union City, CA, USA) and are indicated in Table 1. In experiments where we determined the reversal potential (Figs 5 and 11), we also corrected for the liquid junction potential arising between the bath solution and the solution used for local perfusion. The values used for this correction are given in Tables 2 and 3. Current sizes at membrane potentials of −80 and +80 mV and the reversal potentials were extracted off-line from the current recorded during voltage ramps. Leak subtraction was performed where indicated (in the figure legends). To achieve leak subtraction, current values just before applying the test solutions were subtracted from the current values obtained under the various experimental conditions. Results are presented as mean and s.e.m. Where appropriate, the number of cells analysed is indicated in the figures or figure legends. Statistical tests were performed with Student's two-tailed, unpaired t test or a χ2-test. The number of asterisks in the figures indicates the P values (*P < 0.05; **P < 0.01; ***P < 0.001). Curve fitting was performed with GraphPad Prism software version 3 (GraphPad Software, San Diego, CA, USA).

Figure 5.

The reversal potential of the outwardly rectifying current induced by low extracellular pH strongly depends on the extracellular NaCl concentration
A, individual current–voltage relationships obtained with voltage ramps at different concentrations of NaCl in acidic extracellular solution (solutions eIX–eXII) and with standard bath solution (solution eI). The intracellular solution (iII) contained 151 mm Cl. B, same recordings as in A plotted at a larger scale. C, statistical analysis (n= 6) of the reversal potentials obtained during experiments identical to those shown in A and B. The squares denote values that were obtained without leak subtraction (arrowheads in B), while circles denote values obtained after leak subtraction (arrows in B). The dashed line shows the theoretical Nernst potential for chloride.

Figure 11.

Ionic selectivity of the outwardly rectifying current induced by low pH
A–D, current–voltage relationships obtained with voltage ramps during superfusion of the recorded cells with standard bath solution (pH 7.2; solution eI), acidic bath solution with chloride as main anion (solution eXVII at pH 4.0 in A or pH 4.8 otherwise) or acidic solution containing the sodium salt of the indicated anion (solutions eXIII – eXVI). E, statistical analysis of the reversal potential obtained without background subtraction from recordings similar to those shown in A–D. The reversal potentials obtained with the different anions are all significantly different from each other (P < 0.05).


An outwardly rectifying current activated at low, external pH

Lowering the extracellular pH values to 4.0 quickly activated a large current at positive holding potentials during recordings of whole-cell currents from HEK293 cells (Fig. 1A). The development of inward currents was more complex: immediately after the start of the superfusion with acidic saline, a brief and transient inward current was observed. Most likely, this transient current was due to the activation of channels belonging to the ASIC family as it could be largely reduced by coapplication of 500 μm amiloride (Fig. 8A) or by replacement of extracellular cations with NMDG+ (Fig. 3B). ASIC channels have previously been described as endogenously expressed in HEK293 cells (Gunthorpe et al. 2001). After the transient inward current had subsided, however, a small inward current remained (Fig. 1A). Consequently, the current–voltage relationship at that time showed strong outward rectification (Fig. 1B). All these currents could be elicited repeatedly in the same cell by alternately exposing it to neutral (pH 7.2) and acidic (pH 4.0) saline (Fig. 1A). The absolute size of the currents, however, was very variable from cell to cell, and even more between batches of cells. We observed absolute current sizes of acid-induced outward currents (at +80 mV) ranging from 200 pA to more than 2 nA. The cause of this large variation is not known, and we therefore attempted, wherever possible, to normalize cell responses to a standard stimulus (see below).

Figure 1.

Lowering the pH of the extracellular solution to 4.0 induces an outwardly rectifying current in HEK293 cells
A, current recording of a cell at membrane potentials of +80 mV and –80 mV obtained from voltage ramps during application of acidic extracellular solution (eII, as detailed in Table 2). The activation and deactivation of the current rapidly and repeatedly follows the change of the extracellular solution. Arrowheads point to ASIC-like inward currents. B, corresponding current–voltage relationships at times indicated in A obtained with voltage ramps. C, pH dependence of the outward current at a membrane potential of +80 mV. The current size obtained at various pH values was normalized to the current size obtained with a solution of pH 4.0 in the same cell. The number of cells analysed for each pH value is indicated in the figures. The continuous line was obtained by fitting a Hill equation to the data at pH values ≥ 4.5. The fit yielded the following parameters: Maximum: 1.28, EC50: pH 5.1, Hill coefficient: 3.6.

Figure 8.

Pharmacological interference with the outward current induced by low extracellular pH
A, typical current traces obtained during the application of acidic conditions (solution eII at pH 4.0) with or without an added pharmacological substance. Pharmacological substances used were DIDS (100 μm), flufenamic acid (FFA, 100 μm), mibefradil (Mib, 100 μm) and amiloride (Amil, 500 μm). Pharmacological substances were added to the acidic bath solution (eII at pH 4.0). B, current–voltage relationships during application of the indicated concentration of DIDS (in μm). Intracellular solution contained 130 mm aspartate and 20 mm Cl (solution iI). C, percentage of inhibition of the outward current (at +80 mV) activated at pH 4.0 achieved by the application of various concentrations of DIDS (n= 4–11). The continuous line was obtained by fitting a Hill equation to the data (maximal inhibition: 95%, IC50: 2.9 μm, Hill coefficient: 1.0). D, statistical analysis of experiments similar to the recordings shown in A. The size of the current 2–3 s after applying the pharmacological substance was analysed relative to the current size in the absence of the inhibitor (the number of cells analysed is indicated). E, FFA slowly blocks the current activated by low external pH. The reduction of the outward current after a 30 s application was assessed during application of the bath solution at pH 4.0 without or with 100 μm FFA. The difference is statistically significant (P < 0.001, n= 7).

Figure 3.

The inward currents activated at low external pH depend on the intracellular chloride concentration, but not extracellular cations
A and B, current recording at a membrane potential of –80 mV measured before, during and after exposure of the recorded cell to acidic bath solution (solution eII at pH 4.0). Arrowheads denote the transient activation of an ASIC-like current. In A, the pipette (intracellular) solution contained only 20 mm Cl (solution iI), while in B it contained 151 mm Cl (solution iII). The trace shown in A is from a single cell, while the trace in B is the average of identical recordings from 11 cells. In B, all extracellular cations were replaced with NMDG+ as indicated (solutions eIII and eIV). C–E, current–voltage relationships obtained from a single cell measured with voltage ramps at times indicated in A and B. F, statistical analysis of the inward current (at −80 mV) obtained in experiments similar to those shown in A and C. The difference between the two conditions is not statistically significant. G and H, statistical analysis of the inward current (at −80 mV) obtained in the experiment shown in B, D and E. The data in G were obtained with Na+ containing extracellular solutions (eI and eII at pH 4.0), while the data in H were obtained with NMDG+ as the only extracellular cation. The difference between neutral and acidic extracellular conditions was statistically significant as indicated.

We recorded the outward current at +80 mV during the application of bath solutions adjusted to different pH values. The outward current was maximally activated at pH values between 4.0 and 4.5. In Fig. 1C the current sizes recorded at different pH values are shown normalized to the current at pH 4.0 (which was recorded for every cell analysed). At pH values higher than 4.8, the current size dropped steeply. Fitting a Hill equation to the data obtained at pH values ≥ 4.5 (continuous line in Fig. 1C), we obtained an apparent KD for protons of 7.9 μm (95% confidence interval: 7.7–8.2 μm), corresponding to a pH of 5.1. The best fit was obtained using a Hill coefficient of 3.6 (95% confidence interval: 3.1–4.1), indicating that certainly more than one proton (and possibly three to four protons) needed to bind independently in order to activate the current. At pH values lower than 4.0, the size of the outward current also decreased strongly, albeit in log-linear fashion (Fig. 1C). After exposure to such very acidic solutions, we often found at reapplication of a pH 4.0 solution that the decrease in current amplitude was irreversible. This suggests that very acidic conditions can damage components of the plasma membrane and we therefore avoided using solutions with pH values below 4.0 in all subsequent experiments.

During prolonged application of acidic extracellular solution (pH 4.0) the current induced by acidic extracellular conditions inactivated only slowly (as seen, e.g. in Figs 2A and B and 8A). We observed that inactivation of the current induced under acidic conditions was virtually absent at pH values of 4.5 and above, while it was more pronounced at pH 4.0 and very strong at even more acidic pH values. This might have contributed to the smaller size of the current under these very acidic conditions (Fig. 1C). Upon reapplication of neutral bath solution (pH 7.2), however, the outwardly rectifying current always disappeared rapidly (e.g. Fig. 1A).

Figure 2.

The outward currents activated at low extracellular pH are carried by anions
A and B, current traces at a membrane potential of +80 mV obtained from repeated voltage ramps. In A, the intracellular solution contained 158 mm Cs+ and 7 mm free Mg2+ (iII, Table 1), while in B, all intracellular cations were replaced with NMDG+ (iIII). The outward current was repeatedly activated by superfusing the recorded cell with acidic (pH 4.0) solutions, in which the main anion was varied. Extracellular solutions eI and eII (at pH 4.0) and eV were used (Table 2). C and D, current–voltage relationships obtained from voltage ramps at the time points indicated in A and B. E, statistical analysis of experiments similar to those shown in A and C. F, statistical analysis of similar experiments in which the extracellular anions were replaced with glutamate and aspartate instead of citrate (solutions eVI and eVII, at pH 4.8, n= 6). G, statistical analysis of similar experiments in which 145 mm aspartate was added to the normal extracellular solution to test for blocking effects of aspartate (solution eVIII, n= 7). H, statistical analysis of experiments similar to that shown in B and D. The only intracellular cation was NMDG+ (solution iIII, n= 5). Note that it is not possible to compare the absolute amplitude of outward current densities between panels E–H, since the use of different batches of cells resulted in differences of current amplitudes even under control conditions (left hand columns in panels E–H). The differences between the current densities with extracellular Cl present compared to those without (panels E, F and H) are statistically significant (P < 0.05).

The outward current activated at low external pH is carried by Cl ions

To determine which ions contribute to the current observed under acidic conditions, we substituted components of the extracellular or intracellular solutions with large ions of equivalent charge that are typically not or very little permeant for ionic conductances. When we reduced the concentration of extracellular Cl ions to 20 mm by replacing it with citrate, only a transient outward current was seen that declined rapidly to baseline (Fig. 2A and B). Upon re-administration of a Cl containing solution at pH 7.2, we also observed a transient outward current that quickly subsided. We attribute both transients to a temporary situation of incomplete exchange of solutions, in which intermediate concentrations of Cl and H+ were present at the plasma membrane of the recorded cell. This interpretation is supported by experiments in which we applied a neutral solution (at pH 7.2) with equally reduced Cl concentration before challenging the cells with acidic solution with reduced Cl content. Under these conditions, we no longer observed the transient current at the onset of the superfusion with acidic solutions (data not shown). We obtained similar results when we used aspartate or glutamate instead of citrate to replace the Cl ions (Fig. 2F). When the extracellular Cl concentration was largely reduced (Fig. 2E and F), the size of the outward current under acidic conditions was indistinguishable from the size of the outward current at pH 7.2, suggesting that this current is carried by Cl ions. Alternatively, however, it is conceivable that these large anions act as blockers of a cationic outward current.

We tested this hypothesis in two independent ways. First, we applied an acidic (and hypertonic) extracellular solution containing both 145 mm aspartate and 151 mm Cl. We obtained currents of the same size as with the acidic bath solution that contains 151 mm Cl and no aspartate (Fig. 2G). This experiment shows that aspartate does not block the outward currents under these conditions. Second, we repeated the above experiments using a pipette solution in which all mono- and divalent cations (Fig. 2B, D and H) were replaced with NMDG+. In these experiments we allowed several minutes after break-in to completely equilibrate the cytosol with the pipette solution before starting the recordings. We obtained essentially the same results as with the pipette solution containing Cs+ and Mg2+ (Fig. 2EGversusFig. 2H). Outward currents were still observed when all intracellular cations were replaced by NMDG+, but not when additionally the extracellular anions were replaced by citrate (Fig. 2B, D and H). Taken together, the results presented in Fig. 2 establish that the outward currents elicited at low external pH are carried by Cl ions and not by cations.

The inward current activated at low external pH is dependent on the intracellular Cl concentration

Currents through ASIC channels have been shown to be mainly transient, but some ASIC channels have also been reported to support a sustained component during prolonged exposure to acidic conditions (Waldmann et al. 1997; Lingueglia et al. 1997). As HEK293 cells express ASIC channels (albeit not the family members responsible for sustained proton-gated currents; Gunthorpe et al. 2001), we considered it important to determine whether the inward currents during application of acidic bath solutions were also carried by anions. We therefore severely reduced the Cl concentration in the pipette solution. Using such an intracellular solution, we were no longer able to observe significant inward currents when applying an extracellular solution of pH 4.0 (Fig. 3A, C and F). It is possible that under these conditions there still is a very small inward current present since we sometimes observed that acidification – in addition to activating the proton-activated anion current – also reduced the leak component of the currents (see Discussion). This reduction in leak current potentially masks very small inward currents. On the other hand, large outward currents were still elicited under conditions of strongly reduced intracellular Cl concentration (Fig. 3CE), indicating that the intracellular aspartate did not act as a strong blocker of the proton-activated current.

In an alternative approach, we superfused the cell with neutral and acidic extracellular solutions in which NMDG+ was the only cation and measured the resulting inward currents (Fig. 3B, E and H). We were able to observe a robust increase of the inward current upon application of the acidic solution (Fig. 3B and H). Furthermore, we still observed strongly outwardly rectifying, acid-induced currents when no permeant cations were present intra- and extracellularly (replaced by NMDG+; data not shown, n= 5). Altogether these results establish the presence of a proton-activated, outwardly rectifying anionic current (PAORAC). Both, the very small inward current and the much larger outward current appear to be carried by Cl under physiological conditions.

The proton-activated anion current is not influenced by strong intracellular buffering of protons or Ca2+ ions

Superfusing cells with very acidic solutions may lead to changes of the intracellular pH. In order to test whether such presumed intracellular pH changes are involved in the activation of the proton-activated, outwardly rectifying anion current, we compared proton-activated currents using intracellular solutions containing either 10 mm or 100 mm of the pH buffer HEPES. A difficulty in these experiments lies in the aforementioned variability of the size of the proton-activated anion currents. We therefore recorded cells from the same batch that were treated identically during the cell culture process in order to minimize the variability. Proton-activated, outwardly rectifying anion currents readily activated also when perfused intracellularly with 100 mm HEPES and we were unable to find any differences in the recordings between the two HEPES concentrations (Fig. 4A). We thus conclude that intracellular acidification does not play a role in activation of PAORAC.

Figure 4.

The proton-activated, outwardly rectifying anion current (PAORAC) is not affected by strong intracellular buffering of protons or calcium
The size of the proton-activated outward current measured at +80 mV was analysed. A, cells were intracellularly perfused with solutions containing either 10 mm (iX) or 100 mm (iIX) HEPES. The 100 mm HEPES did not prevent the activation of the PAORAC (at pH 4.0) and did not alter the current size (n= 7 for each condition). B, neither strong intracellular Ca2+ buffering (10 mm BAPTA, ‘0’ Ca condition; solution iVIII), nor raising the intracellular Ca2+ concentration to 100 nm (10 mm BAPTA and 2.65 mm total Ca2+; intracellular solution iVII) inhibited PAORAC (elicited by solution eII at pH 4.5, n= 5–6).

Since the proton-activated anion currents were strongly outwardly rectifying and voltage dependent (see below), properties reminiscent of Ca2+-activated Cl channels (e.g. Evans & Marty, 1986), we tested whether the acid-induced anion currents were influenced by manipulating the intracellular Ca2+ concentration. We first changed the intracellular Ca2+ buffer to 10 mm BAPTA in order to better suppress fast, localized Ca2+ signals, but this did not affect the activation of proton-activated outwardly rectifying currents (Fig. 4B). We then clamped the intracellular Ca2+ concentration to 100 nm with a mixture of 10 mm BAPTA and 2.65 mm Ca2+. This also did not influence the activation of PAORAC, and the size of the outward currents was similar with both intracellular Ca2+ concentrations (Fig. 4B). Furthermore, the acid-induced current activated also in the absence of any intra- and extracellular Ca2+ (data not shown). These experiments rule out that the current evoked at low extracellular pH is a Ca2+-activated Cl current.

The Cl current elicited at low external pH is due to a Cl permeable conductance

Two cloned chloride channels belonging to the mammalian ClC family of chloride channels have been reported to be activated by lowering the extracellular pH (ClC-2; Jordt & Jentsch, 1997; and ClC-7; Diewald et al. 2002). However recently, the bacterial protein homologous to the mammalian ClC-family, ClC-ec1, has been shown to function as a H+–Cl cotransporter rather than a proton-activated Cl channel (Accardi & Miller, 2004). In order to test whether the current elicited at low external pH was caused by the activation of a membrane channel, we measured the reversal potential at four different extracellular Cl concentrations. To this end, we used extracellular solutions (at pH 4.0) containing only NaCl at various concentrations and added glucose in order to maintain the osmolality (solutions eIX–eXII, Table 2). We analysed the data in two ways. First, we simply evaluated the membrane potential at which the observed currents were zero (arrowheads in Fig. 5B). From the reversal potentials measured in this way, we obtained a slope of 39 mV per 10-fold change in extracellular Clconcentration (Fig. 5C). The alternative method consisted of evaluating the intersections of the current voltage relationships before and during the application of the acidic test solution (arrows in Fig. 5B, equivalent to a ‘leak subtraction’ procedure). Fitting the values obtained with this method yields a slope of 71 mV per 10-fold change in Cl concentration. Thus, both methods produce results deviating from the theoretically expected 59 mV per 10-fold change of Cl concentration. We interpret these findings as follows: Not leak-subtracting the current–voltage relationships (first method) underestimates the true reversal potential, while leak subtracting (second method) overestimates the reversal potential (see Discussion). It should also be kept in mind that the very small inward currents and the shallow slope of the I–V relationship close to the reversal potential render the estimation of the reversal potential difficult and error-prone. Nevertheless, these results demonstrate that the reversal potential of the current activated at low external pH is strongly dependent on the extracellular Cl concentration, consistent with the idea that the current is caused by a proton-activated Cl permeable membrane channel. The alternative idea, i.e. that this current is caused by an electrogenic transporter, is difficult to reconcile with the data. Under our experimental conditions, the reversal potential for protons, EH, is +186 mV, while the reversal potential for Cl, ECl, is –3.4 mV when the intracellular and extracellular Cl concentrations are 151 mm (solution iII) and 173 mm (solution eIX), respectively. Assuming an obligatorily coupled transport of protons and Cl anions with an unknown coupling ratio R, the coupling ratio R can be obtained as (Blaustein & Lederer, 1999; Accardi & Miller, 2004):


where R=n/m, with n indicating the number of Cl ions coupled to the transport of m protons and Erev is the measured reversal potential. The sign on the right hand side of the equation depends on whether cotransport or antiport of protons and Cl ions is considered. Under the conditions indicated above, our upper estimate of the reversal potential was Erev= 6.0 ± 2.3 mV (Fig. 5C), yielding a very high coupling ratio (R > 10). Assuming a coupling ratio of R= 10, the predicted reversal potential would be 14 mV (for antiport) or −24 mV (for cotransport), both of which are incompatible with the measured values. Coupling ratios of more than 10 have not yet been described for ion-exchangers or cotransporters.

The outwardly rectifying current activated at low external pH is voltage dependent but has an outwardly rectifying instantaneous current–voltage relationship

In order to compare the voltage dependence of the proton-activated anion channels with that of already characterized anion channels, we recorded the time-dependent behaviour of PAORAC upon step-like changes of the membrane potential (Fig. 6). The current showed a clear time-dependent facilitation upon depolarization to values higher than +15 mV (Fig. 6A). Noteably, the outward rectification and the voltage-dependent facilitation were preserved when we recorded the cells under conditions where all cations (intra- and extracellularly) were replaced with NMDG+ (data not shown).

Figure 6.

Voltage dependence and tail currents of the current activated at low extracellular pH
A, representative current recording of a cell from a holding potential of −55 mV to various membrane potentials as indicated in the lower panel. At depolarizing potentials, the current displays a time-dependent facilitation. B, representative current recording used to determine the instantaneous current voltage relationship with the voltage protocol shown in the lower panel. C, enlarged representation of some of the traces shown in A starting 2 ms after the voltage jump back from the activating voltage (indicated) to the holding potential (−55 mV). The resulting tail currents are small and decay rapidly. D, enlarged representation of some of the traces shown in A starting 2 ms after the voltage jump back from the activating voltage (+95 mV) to the potentials indicated. Traces shown in A–D were leak subtracted by running the same voltage protocol two to four times under acidic (pH 4.8) and normal (pH 7.2) conditions. After averaging, the current traces obtained under normal conditions were subtracted from those obtained under acidic conditions. E, statistical analysis (n= 5) of recordings similar to the traces shown in A and C. The current–voltage relationships were obtained at the time points indicated in A and C, i.e. 2 ms after the voltage jumps in order to ensure that the capacitative transients had subsided. F, instantaneous current–voltage relationships obtained at the time point indicated in B and D. The instantaneous current–voltage relationship shows strong outward rectification. Solutions used: intracellular iII, extracellular eII (at pH 4.8).

The current–voltage relationship obtained at the beginning (after 2 ms) of the voltage step was already strongly outwardly rectifying (Fig. 6E, curve 1). Due to the voltage-dependent facilitation, however, the current–voltage relationship obtained at the end (after 500 ms) of the voltage step showed an even steeper outward rectification (Fig. 6E, curve 1 and 2). We also sought to record so-called tail currents. While measurable (Fig. 6C and E, curve 3), these currents were very small (less than 10% of the outward current) and decayed rapidly (within 20–30 ms) to baseline. Equally, when we tried to measure the instantaneous current–voltage relationship (Fig. 6B, D and F), we obtained only very small tail currents at negative potentials (Fig. 6D). The instantaneous current–voltage relationship therefore is outwardly rectifying (Fig. 6F), as has been reported in a preliminary abstract (Bompadre et al. 2001). Due to the leak subtraction performed in this experiment, we probably underestimated the size of the inward current (see Discussion). However, this underestimation does not compromise our conclusion that the instantaneous I–V relationship is outwardly rectifying.

We next investigated the time course of deactivation of previously activated channels by performing double-pulse experiments and varied the time of the intermittent voltage step to −55 mV (Fig. 7). The results indicate that after 2 ms (the time needed for settling of the voltage transients), voltage-dependent gating can only account for about 20% reduction in current size. Deactivation kinetics for the remaining current was still fast under these conditions and essentially complete after 400 ms (Fig. 6). Taken together, these results provide evidence that the channels underlying the current activated at low external pH are voltage-dependently facilitated by depolarization of membrane potentials more positive than +40 mV. The time courses for activation and deactivation were similar and essentially complete after ca 0.5 s.

Figure 7.

Deactivation kinetics of the current induced by low external pH
A, superimposed current traces (upper panel) of a typical recording performed to determine the time course of deactivation. The double-pulse voltage protocol used for these experiments is indicated in the lower panel. B, statistical analysis (n= 5) of recordings similar to that shown in A, showing that the deactivation after hyperpolarizing for 2 ms to −55 mV was only 20%, but was essentially complete after a 400 ms hyperpolarization. Current values were determined 2 ms before the end of the first voltage jump to +95 mV and 2 ms after the beginning of the second voltage jump (also to +95 mV). The current obtained 2 ms after the beginning of the first voltage jump was subtracted from both values, and the quotient between the two was calculated. The percentage of deactivation was obtained as 100 × (1 − quotient). The same solutions as in Fig. 6 were used.

DIDS and FFA block the outward current activated at low external pH, but mibefradil and amiloride do not

To test whether the anionic current activated at low extracellular pH can be modulated by pharmacological interventions we tested three well-established inhibitors of anionic currents, DIDS, FFA and mibefradil (Nilius et al. 1997b). Of these substances, only DIDS (100 μm) was able to rapidly (within 2–3 s) inhibit the outward current activated with an acidic extracellular solution (Fig. 8A and D). The same concentration of FFA produced a slowly developing block of the current. After 30 s, the current recorded in the presence of FFA was only 40% of the current measured without FFA (Fig. 8A and E). Applying 100 μm FFA for longer periods further enhanced the block caused by this substance (Fig. 8A). Mibefradil, on the other hand, was not able to significantly inhibit the current activated at low external pH values at the maximum concentration tested (100 μm). We also tested whether amiloride, an inhibitor of ENaC (epithelial sodium channels) and ASIC channels (Lingueglia et al. 1997; Waldmann et al. 1997), was able to influence the anionic current activated at pH 4.0, but found no inhibitory effect even at high concentrations (500 μm, Fig. 8A and D). The rapid (and reversible, data not shown) effects of DIDS made it possible to study the dose–response curve for this substance (Fig. 7B and C). We found that DIDS inhibits the outward current (at +80 mV) with an IC50 of 2.9 μm (95% confidence interval: 2.1–3.9 μm). The maximum inhibition was more than 95% (at 100 μm). Since we analysed the blocking effect of DIDS without performing leak subtraction, the maximum inhibition is likely to be even higher. The dose–response curve was well fitted with a Hill equation using a Hill coefficient of 1.0 (95% confidence interval: 0.71–1.34), indicating that DIDS likely binds in a 1: 1 stoichiometric relationship to the channels that underlie PAORAC.

The single channel conductance of proton-activated anion channels is outwardly rectifying

The outwardly rectifying shape of the instantaneous I–V relationship (Fig. 6F) can be taken as the first indication that the single channel conductance of channels that underlie PAORAC may also be outwardly rectifying. To obtain more definitive proof for this hypothesis we searched for single channel events of channels underlying PAORAC in excised patches. Using an extracellular solution at pH 5.0 (where the proton-activated anion channels should be activated to > 50%, Fig. 1C) in the recording pipette and holding the voltage across the membrane at +86 mV, we recorded single channel events as shown in Fig. 9A. In 33 of 93 patches analysed under these conditions, we observed such channel activity with an apparent open probability (NPo; N, number of channels; Po, open probability) greater than 0.3. The active patches often contained more than one channel. However, when we repeated the same experiments with the extracellular solution (in the pipette) adjusted to pH 6.0, only 2 of 36 patches showed an NPo value larger than 0.1. This statistically significant difference (P < 0.001) shows that the observed channels were strongly dependent on the extracellular pH. Also, we found that these channels were inhibited by inclusion of 100 μm DIDS to the acidic (pH 5.0) extracellular solution. In an independent set of experiments we found no channel activity similar to the one shown in Fig. 9A in 22 patches assayed with a DIDS containing solution (100 μm), while 17 out of 36 patches recorded with a control solution containing no DIDS showed channel activity (P < 0.001). Next, we tested whether the observed proton-activated channels could account for the strongly outwardly rectifying whole-cell I–V relationship of PAORAC (e.g. Fig. 1). We subjected patches showing active channel activity to a repeated voltage-ramp protocol similar to the one used in whole cell recordings (but with a holding potential of +86 mV; Fig. 9B). In three patches we obtained enough data to attempt averaging the traces that showed no apparent channel activity and subtracting this averaged background trace from the average of all recorded traces (Fig. 9B, middle panel). The resulting I–V relationship of the ensemble average was strongly outwardly rectifying and essentially indistinguishable from whole cell I–V relationships (Fig. 9C). Finally, we investigated whether the observed outward single channel currents (at +86 mV) were due to cations flowing from the cytosolic to the extracellular side of the membrane. When we superfused the excised inside-out patch with an intracellular solution containing no other cations besides NMDG+, we still observed outwardly directed single channel events (Fig. 9D). We remarked, however, that the activity of the inside-out patches was consistently (n= 6) reduced when the NMDG+ containing solution was applied to the cytosolic face of the patch. This reduced activity stems from a reduced amplitude of the single channel currents and a – sometimes strongly – reduced apparent open probability. Although we did not investigate this effect quantitatively, it may partly be explained by the liquid junction potential that arises between the two solutions applied to the cytosolic side of the membrane. This liquid junction potential, which we did not correct for, is predicted to reduce the holding potential by ca 10 mV. Nevertheless, our data indicate that the channels observed with acidic extracellular solution are capable of carrying significant anionic currents.

Figure 9.

Excised patch recordings of proton-activated channels
A, exemplary recording (10 times 5 s) of an excised patch at a holding potential of +86 mV showing the activity of at least two channels. Solution inside the recording pipette (extracellular solution eII) was at pH 5.0 and the solution in the bath was the intracellular solution iII. Traces were digitally smoothed by 500 Hz lowpass filtering for presentation purposes. The dotted lines labelled c, o1 and o2 indicate the closed, one open and two open channels. B, the I–V relationship of the acid-activated channels was determined by ensemble averaging. An excised patch was subjected to repeated voltage ramps (at 1 Hz) as shown in the lower panel. The upper panel shows the 10 superimposed current traces. The middle panel shows the average of 60 traces and the average of traces that did not show any discernible channel activity. The difference between these traces is the ensemble average of the channel activity. C, the ensemble averaged I–V relationship (average of three patches) compared to an I–V relationship measured by whole cell patch clamp recording under similar conditions. Traces were normalized to the current size at +80 mV. D, replacement of the cytosolic cations with NMDG+ does not abolish the single channel events. Shown are six stretches (each 300 ms long) of channel activity recorded from an exemplary, excised inside-out patch (holding potential +86 mV). The recordings in the left and right columns were obtained during superfusion of the cytosolic face of the membrane with standard, CsCl and MgCl2 containing intracellular solution (iII, Table 1). The middle column, however, was recorded with a cytosolic solution in which all cations were replaced with NMDG+. Note that during application of the NMDG+ containing solution, the holding potential is predicted to be reduced by 10 mV (see text for details). The dotted lines labelled c and o indicate closed and opened state of a channel.

Taken together these data demonstrate anionic single channel currents that share the essential properties of the whole-cell proton-activated currents, namely pH dependence, DIDS sensitivity and strongly outwardly rectifying I–V relationship as judged by ensemble averaging. This suggests that the described single channel events underlie the whole-cell PAORAC. The size of the observed single channel currents (> 1 pA at +86 mV) provides an additional, independent argument for the notion that the proton-activated anionic whole-cell currents flow through ion channels, and are not the product of electrogenic ion exchange processes (Gadsby, 2004).

Next we determined the voltage dependence of single proton-activated channels. Figure 10A shows a typical recording of a patch showing proton-activated channel activity (pH 5.0) at various holding potentials. At potentials > 40 mV, single channel events could readily be discerned. Stronger depolarizations revealed that the single channel conductance is strongly outwardly rectifying. On average, the chord conductance of single proton-activated channels was 10 pS at +46 mV, 13 pS at +86 mV and 20 pS at +126 mV (Fig. 10B). At negative holding potentials, only small (< 0.5 pA) single channel currents could be detected, corresponding to single channel conductances of 4–7 pS, depending on holding potential. Events were rare and difficult to resolve at hyperpolarized potentials, asking for considerable caution in their quantitative interpretation. Nevertheless, these data demonstrate that the single channel conductance of proton-activated anion channels is outwardly rectifying.

Figure 10.

The voltage dependence of proton-activated single channel currents is outwardly rectifying
A, recording of an excised patch at different holding potentials (as indicated in mV on the left of the traces). The same solutions as in Fig. 9 were used. Open and closed states are indicated by dotted lines. Traces were digitally low pass filtered at 500 Hz for presentation purposes. The dotted lines labelled c and o indicate closed and opened state of the channel. B, statistical analysis of recordings similar to those shown in A. Number of patches analysed for each data point: 13–29 for depolarized and 7–10 for hyperpolarized holding potentials.

Ionic selectivity of the anion permeable channel that is activated at low external pH

In order to determine the anionic permeability sequence for PAORAC, we measured the reversal potential of the anionic current activated with acidic bath solution under bi-ionic conditions (i.e. one extracellular and one intracellular anion species). The extracellular solution, however, also contained 3.1 mm sulphate (as counter ion for Mg2+) and 5 mm citrate (as buffer). On their own, neither citrate (Fig. 2) nor sulphate (data not shown) causes measurable currents through the channel activated at low extracellular pH, and therefore these anions are not expected to influence the measurements of the reversal potentials. Replacing the extracellular Cl with other anions, we observed statistically significant shifts in the reversal potential for all anionic species tested (thiocyanate, iodide, nitrate, bromide and chloride, Fig. 11). From the determined reversal potentials it can be concluded that the permeability sequence for this channel is SCN> I> NO3> Br> Cl (Fig. 11E). Additionally, we tested whether larger anionic species could also permeate the acid-activated anion channel. We obtained statistically significant shifts of the reversal potential (+20.4 ± 4.2 mV, n= 5) when replacing the extracellular chloride with glutamate, but not when we replaced it with aspartate or sulphate. The outward currents observed with these larger anions were exceedingly small (e.g. Fig. 2F), asking for caution when interpreting these data. In particular, we cannot rule out significant contributions of other endogenous anionic conductances (not activated under acidic extracellular conditions) to the reversal potential when applying the large anions extracellularly. The relative permeability sequence determined for the halide ions, however, allows us to conclude that the anion conductance in HEK cells activated under acidic extracellular conditions is more permeable for I than for Cl. This parameter is useful for the classification of chloride channels (see Discussion).

High intracellular Mg2+ inhibits VRAC, but not PAORAC

Recently, Nobles et al. (2004) proposed that upon extracellular acidification, the channels underlying volume-regulated anion currents (VRAC) change their biophysical and pharmacological characteristics so that they then produce the proton-activated, outwardly rectifying anion currents (PAORAC). Attempting to investigate this question in more detail, we were unable to elicit VRAC when using an intracellular solution containing 10 mm free Mg2+ (Fig. 12), even in cells that were visibly swollen in response to the strong hypotonic stimulus (200 mosm). In agreement with the literature (reviewed in: Nilius et al. 1997a), omission of the intracellular Mg2+ allowed us to record large VRAC in every cell we investigated. Importantly, in the same cells the activation and the current amplitude of PAORAC was not influenced by the intracellular Mg2+ concentration (Fig. 12).

Figure 12.

Volume-regulated anion currents (VRAC), but not proton-activated outwardly rectifying anion currents (PAORAC) are inhibited by 10 mm intracellular Mg2+
Cells were intracellularly perfused with solutions that contained either 0 mm (iV) or 10 mm free Mg2+ concentration (iVI) supplemented with 4 mm Na2ATP and a free Ca2+ concentration of 100 nm.A, the activation of PAORAC (at pH 4.5) and the size of the resulting outward currents (+80 mV) were independent of the intracellular Mg2+ concentration. B, on the contrary, in the same cells activation of the volume-regulated anion current (VRAC) by hypo-osmotic solutions (eXXIV, ca 200 mosmol kg−1) was entirely inhibited by 10 mm free internal Mg2+ (n= 11–12). The difference between the conditions (Mg2+-free or 10 mm Mg2+-containing intracellular solution) is statistically significant, P < 0.01).

Effect of low extracellular pH on VRAC currents

We next investigated the response of VRAC (induced by hypotonic extracellular conditions) to strong acidification (pH 4.5, Fig. 13). Concentrating on steady-state outward currents we failed to detect a significant increase in current amplitude due to the acidification (Fig. 13C, column 2 and 4), as has been reported previously by Nobles et al. (2004). On closer inspection of the time course (Fig. 13A), however, we noticed that the outward current at +80 mV showed a transient decrease immediately following the acidic stimulation. From the concomitant I–V relationships (Fig. 13B) we determined that the temporary decrease of outward current arises from a selective inhibition of outward currents at potentials > +50 mV (Fig. 13B, trace 3). Within a few seconds after the current decrease, the outward currents (but not inward currents) increased again (Fig. 13B, trace 4), consistent with the appearance of an additional, strongly outwardly rectifying current. Statistically, this late increase of outward current was not different in amplitude from the current evoked by acidification before the activation of VRAC (Fig. 13C, columns 1 and 5). One possible interpretation of these experiments is that outward currents through volume-regulated anion channels actually are inhibited by extracellular acidification. This is consistent with previous reports in other cell types (Ackerman et al. 1994; Jackson & Strange, 1995; Voets et al. 1997; Nilius et al. 1998) of accelerated inactivation of VRAC upon depolarization under acidic conditions. Accelerated inactivation is expected to reduce outward currents in the voltage-ramp protocol employed here. The inhibition of outward currents subsequently is reversed by the activation of an additional, outwardly rectifying current. This interpretation therefore assumes the existence of two independent components contributing to the whole-cell currents under hypotonic and acidic extracellular conditions.

Figure 13.

Time course of proton-induced effects before and after hypotonically activated volume-regulated anion currents (VRAC)
A, recordings of proton-activated currents at −80 and +80 mV of a cell before (left-hand side) and during (right-hand side) activation of VRAC. Note the transient decrease of outward current (arrow no. 3). B, current–voltage relationships recorded at times indicated by numbered arrows in A. C, statistical analysis (n= 5) of recordings similar to those shown in A. Current amplitudes were measured at +80 mV at times indicated in A. The last column is the subtraction of the current measured at time points 4 and 3. This difference is not statistically different from the current size of PAORAC before the activation of VRAC (column 1). Solutions used: intracellular iV, extracellular eXXIII (pH 7.2 and isotonic, but reduced Cl content, to match the hypotonic solutions), eXXVI (pH 4.5 and isotonic, but reduced Cl content), eXXIV (pH 7.2, hyptonic) and eXXV (pH 4.5, hypotonic).

Depolarizing voltage pulses allow the separation of VRAC and PAORAC under acidic and hypotonic extracellular conditions

To test more directly if there are indeed two different conductances coexisting under hypotonic and acidic extracellular conditions, we designed experiments that exploited the opposite behaviour of VRAC and PAORAC to a strongly depolarizing stimulus (Fig. 14). VRAC inactivates upon strong depolarization (for review: Nilius et al. 1997a), while PAORAC shows facilitation under these conditions (Fig. 6). We confirmed this by using a protocol that consisted of six voltage pulses from a holding potential of 0 mV to +100 mV with only short repolarizations (Fig. 14A, upper panel). Cells did not show any voltage-dependent behaviour to this protocol as long as they were kept in isotonic and neutral extracellular solutions (Fig. 14A). Upon extracellular acidification, the evoked PAORAC displayed voltage-dependent facilitation to each of the six voltage pulses (Fig. 14B). We then activated VRAC by exposing the cells to an hypotonic (ca 200 mosmol kg−1 at pH 7.2) extracellular solution and waited until the outward currents reached an amplitude that was 1- to 3-fold the amplitude of PAORAC recorded in the same cell. We then subjected the cell again to the six depolarizing voltage pulses, which caused almost complete inhibition of VRAC (Fig. 14D). Importantly, the sixth voltage pulse did not show much inactivation any longer, indicating that the inactivation of VRAC had reached a stable state. We then let VRAC currents recover fully at a holding potential of 0 mV (checked by repeated voltage ramps), changed the extracellular solution to pH 4.5 (still hypotonic) and recorded a fourth time the current response to the six voltage pulses (Fig. 14E). The response to the first voltage pulse was an inhibition of the current (although the amount of inhibition was much smaller than under pH 7.2). Interestingly, however, the response to the sixth voltage pulse (when VRAC was presumably fully inhibited) was facilitatory. Reversing the order of experiments (first testing the acidic condition and then the neutral condition) led to identical results. We quantified these responses by measuring the absolute amount of current increase or decrease to the first and the sixth voltage pulse (ΔII and ΔIVI). The reversal of polarity observed under hypotonic and acidic extracellular conditions was statistically significant (Fig. 14F, P < 0.01). This shows (in disagreement with the report from Nobles et al. 2004) that VRAC, even under acidic extracellular conditions (pH 4.5) still is inhibited by depolarizing voltage pulses. Our experiments suggest that, in order to detect the inhibitory response of VRAC to depolarizations, it is necessary to study relatively large VRAC currents, since otherwise this behaviour might be masked. Possibly, Nobles et al. (2004) did not observe such large VRAC currents, which might be related to the composition of their intracellular solution (no ATP, Ca2+ buffered to values significantly below 100 nm, added Mg2+, all of which are known to inhibit VRAC, Nilius et al. 1997a). Furthermore, our data show that under hypotonic and acidic extracellular conditions two current components are simultaneously active. The first component is inhibited by depolarizations (VRAC-like), while the second component is facilitated (PAORAC-like). The voltage pulse protocol used, differentially affecting the individual components, allows observing both components under identical ionic conditions. Interestingly, the activation and subsequent depolarization induced inhibition of VRAC did not reduce the acidic pH-dependent, depolarization-induced facilitation (right hand columns in Fig. 14C and F). This might indicate that the magnitude of PAORAC was not substantially reduced by activation of sizeable VRAC currents.

Figure 14.

Volume-regulated (VRAC) and proton-activated (PAORAC) currents can be separated by depolarization induced inhibition of VRAC
Whole cell recording (holding potential 0 mV) of a cell that was subjected to the voltage protocol shown in A, upper panel, before (A and B) and during (D and E) hypo-osmotically induced activation of VRAC. Before and during the activation of VRAC, the cell was assayed under neutral (pH 7.2, A and D) and acidic (pH 4.5, B and E) extracellular conditions. A, the depolarizing pulses provoke only small leak currents in conditions that are isotonic and neutral. B, acidification induces voltage-dependently facilitated outward currents (see also Fig. 6). D, in strong contrast, under neutral but hypotonic conditions, VRAC currents show strong inhibition caused by the depolarizing pulses; note that the inhibition is almost complete by the time of the sixth voltage pulse. E, combining hypotonic and acidic extracellular conditions (pH 4.5), an inhibitory response was observed for the first voltage pulse, but a facilitatory response was seen at the sixth voltage pulse. The amount of inhibition or facilitation was quantified for the first (I) and sixth (VI) voltage pulse (ΔII and ΔIVI) as indicated by the dashed lines. C and F provide a statistical analysis of 5 recordings similar to those shown in the other panels. Cells were analysed that exhibited VRAC current sizes ca 1–3 times larger than PAORAC currents (recorded just before the activation of VRAC). C, before the activation of VRAC, an inhibitory current response was never observed in response to the voltage pulses. F, after induction of VRAC, but without acidification, a facilitatory current response was never observed. Acidification then allowed observation of inhibitory (voltage pulse I) and facilitatory (voltage pulse VI) current responses under the same ionic conditions. The reversal from inhibitory (voltage step I) to facilitatory (voltage step VI) current responses was statistically significant (P < 0.01). Solutions used: intracellular iIV, extracellular eI, eXXV (pH 4.5), eXXIV (hypotonic at pH 7.2).

Hippocampal astrocytes in primary culture also express an anionic conductance activated by low extracellular pH

Having characterized the anionic current activated at low external pH in HEK293 cells, we wanted to know whether this current could also be found in native tissues. We therefore investigated whether such a current is expressed in hippocampal astrocytes maintained in primary culture. When applying acidic extracellular solution to hippocampal astrocytes, we indeed found that a current was activated that closely resembled the current found in HEK293 cells in all investigated aspects (Fig. 15). The current activated by low extracellular pH was outwardly rectifying, and not abolished by substituting all extracellular cations with NMDG+. In most of the cells we observed that the proton-activated outward currents were smaller (on average 50% of control) when replacing all extracellular cations with NMDG+ (Fig. 15AC). This may in part be due to the lower Cl concentration of the NMDG+ solution (Table 3), but we cannot rule out other causes, such as a mild block of the current by NMDG+. In contrast to the NMDG+ solution, the outward current was completely abolished when we replaced extracellular Cl with citrate or sulphate (Fig. 15AC). The proton-activated outward current was also sensitive to DIDS (100 μm, Fig. 15AC). These results establish that hippocampal astrocytes endogenously express an anionic current activated at low extracellular pH. Like the current in HEK293 cells, the endogenous current in astrocytes displays a time-dependent facilitation upon step-like changes of the membrane potential to depolarizing voltages (Fig. 15D).

Figure 15.

Primary cultured hippocampal astrocytes possess a current that shares all key characteristics with the current activated at low external pH in HEK293 cells
A, current recording from a cultured hippocampal astrocyte obtained from repeatedly applied voltage ramps (current levels at −80 mV and +80 mV are displayed, holding potential between ramps was −75 mV). Cells were kept in astrocyte bath solution (pH 7.3, solution eXVIII) and acidic solution (pH 4.0, eXIX) was applied as indicated. The acidic solution was either of similar composition to the bath solution, or 100 μm DIDS was added, or all cations were replaced with NMDG+ (eXXI), or the concentration of chloride was reduced (citrate or sulphate as substitute, eXX or eXXII). The composition of the solutions used in these experiments is detailed in Table 3. B, current–voltage relationships obtained with voltage ramps at times indicated in A. C, statistical analysis of experiments similar to those shown in A and B; current densities were obtained from voltage ramps at a membrane potential of +80 mV. Asterisks indicate current densities statistically different (P < 0.01) from current densities obtained while applying the acidic bath solution. D, the current activated at low external pH in astrocytes is voltage dependent when depolarized to positive membrane potentials. The voltage protocol is indicated in the lower panel.


In this paper we characterize a current endogenously expressed in HEK293 cells that is activated by lowering the extracellular pH to values below 5.5. By manipulating the intracellular and extracellular ionic conditions in various ways (Fig. 2–4) we provide evidence that this current is carried by anions in the outward as well as in the inward direction. The existence of a proton-activated anion channel has been previously reported for CHO, 3T3 and Calu cells in a preliminary report (Bompadre et al. 2001) and confirmed without much further characterization for CHO and HEK293 cells (Vanoye & George, 2002; Xu et al. 2004). Recently Nobles et al. (2004) published a report that demonstrated proton-activated anion currents in HEK293 cells that were voltage dependent and inhibited by DIDS, and displayed a permeability sequence I> Br> Cl. Our initial characterization of the proton-activated anion current in HEK293 cells, which we performed before the publication of the report by Nobles et al. (2004), confirms and extends these findings. Specifically, we conclude from our data that the proton-activated outwardly rectifying anion channel in HEK293 cells possesses the following properties. (1) More than one proton, probably 3 or 4, needs to bind to the anion channel in order to activate it, because there is a very steep relationship between the proton concentration and the current size (Fig. 1C). (2) Both inward and outward components can be inhibited by strongly reducing the intra- or extracellular Cl concentration, respectively (Figs 2 and 3). (3) PAORAC activation does not require either protons, or Ca2+ (in agreement with Nobles et al. 2004) or Mg2+ intracellularly and is not inhibited by high intracellular Mg2+ (Figs 4 and 12). (4) PAORAC is voltage-dependently facilitated by depolarization (in agreement with Nobles et al. 2004) with fairly rapid, subsecond, activation and deactivation kinetics (Figs 6 and 7). Furthermore, the proton-activated anion channels possess an outwardly rectifying single channel conductance (Fig. 10), which results in a strongly non-linear instantaneous current–voltage relationship and very small tail currents (Fig. 6). These findings agree with the preliminary report by Bompadre et al. (2001). We find that at +86 mV, the chord conductance of the proton-activated anion channels is 13 pS. Both, the voltage dependence and the outward rectification of the single channel conductance contribute to the very steep outward rectification observed in voltage ramps and in the steady-state current–voltage relationship (Fig. 6). (5) Pharmacologically, the outward current is blocked by DIDS (Xu et al. 2004; Nobles et al. 2004) with an IC50 of 2.9 μm. Because it was well described with a Hill coefficient close to 1, the dose–response curve indicates that one molecule of DIDS bound to the channel is sufficient to block that channel. At 100 μm, DIDS produces an essentially complete block of the outward current with very rapid onset kinetics. FFA, on the other hand, blocks the outward anion current with much slower onset kinetics, while mibefradil (100 μm) and amiloride (500 μm) do not block the proton-activated anion current (Fig. 8). (7) The relative permeability sequence for anions is (determined from reversal potential measurements, Fig. 11): SCN> I> NO3> Br> Cl, in agreement with the data from Nobles et al. (2004). We show that the larger anions citrate, glutamate and aspartate permeate PAORAC only very little, if at all (Fig. 2).

We found that a current similar to PAORAC in HEK293 cells also exists endogenously in mouse hippocampal astrocytes maintained in primary culture. Like the current in HEK293 cells, the endogenous proton-activated outward current in hippocampal astrocytes was dependent on the presence of extracellular Cl, could be inhibited by DIDS, was strongly outwardly rectifying and displayed voltage-dependent facilitation (Fig. 15). Currents similar to PAORAC have recently also been described in Sertoli cells (Auzanneau et al. 2003). As the channels in HEK293 cells, the proton-activated anion current in Sertoli cells is outwardly rectifying, voltage dependent, independent of the presence of intra- or extracellular Ca2+, and sensitive to DIDS (Auzanneau et al. 2003). However, the current in Sertoli cells displayed a quite different permeability sequence (Cl> Br> I; Auzanneau et al. 2003). This raises the possibility that different types of proton-activated anion channels exist that can be differentiated by their halide permeability sequence. Nevertheless, it appears that proton-activated anion channels are broadly expressed in a large variety of cell types and must be expected to exist in many native tissues.

Technical considerations

The I–V relationship of PAORAC is very steep, but typically intersects with the x-axis with a fairly shallow slope. This necessitates accurate background subtraction for measuring the reversal potential of this current. However, we observed occasionally (in cells that showed only very small PAORAC or when attempting to record PAORAC in outside-out patches, data not shown) that also the ‘leak’ currents are inhibited by the strong acidification required to activate PAORAC. This means that background correction performed by subtracting currents measured under neutral extracellular conditions leads to an underestimation of the size of inward currents (Figs 3 and 6) and to considerable inaccuracies when determining reversal potentials (Fig. 3CE, and especially Fig. 5). Not performing background subtraction at all, on the other side, would lead to overestimation of inward currents and to opposite errors in the reversal potentials (Figs 5 and 11). This problem prevents us from more quantitatively interpreting the permeability data for anions (Fig. 11), but still allows constructing a semiquantitative ranking of relative permeabilities. Importantly, however, the large outward currents are at most marginally affected by the uncertainty in background correction.

The proton-activated, outwardly rectifying anion channel does not match the characteristics of cloned Cl channels

Of the cloned chloride channels, several share some of the biophysical and pharmacological properties of PAORAC, but no cloned channel can completely account for the entire feature set. While ClC-2 activity is enhanced by acidic extracellular conditions (Jordt & Jentsch, 1997), its other biophysical characteristics are quite distinct from the channel characterized in this paper. ClC-2 is inwardly rectifying, activated by hyperpolarization and has a Cl > I permeability relationship (summarized in Jentsch et al. 2002). ClC-4 and ClC-5, on the other hand, are strongly outwardly rectifying and voltage dependent at positive membrane potentials (Friedrich et al. 1999). Although ClC-4 was first described as being activated only under acidic conditions (Kawasaki et al. 1999), other studies found that ClC-4 and ClC-5 channels are activated under neutral extracellular conditions and are in fact inhibited by lowering the extracellular pH (Friedrich et al. 1999; Mo et al. 1999; Vanoye & George, 2002). Furthermore, ClC-5 channels have been reported to be insensitive to DIDS (Steinmeyer et al. 1995).

The reported biophysical properties of ClC-7 expressed in Xenopus oocytes, on the other hand, seem to fit the bill of PAORAC (Diewald et al. 2002). It was also reported that 1 mm DIDS blocks this current. However, neither a more detailed pharmacological profile, nor the ionic permeability sequence of ClC-7 channels is available. Another study (Brandt & Jentsch, 1995) has reported failure to functionally express this channel in oocytes, and, additionally, Xenopus oocytes already express an endogenous proton-activated, outwardly rectifying current without being injected with ClC-7 mRNA (Diewald et al. 2002). Thus, the available data do not allow the identification of ClC-7 encoded channels unambiguously as PAORAC. Finally, in mammalian cells, this channel has been reported to be exclusively present in intracellular compartments, specifically in late endosomes and lysosomes (Kornak et al. 2001; Auzanneau et al. 2003). An exclusive intracellular location would be incompatible with the hypothesis that CLC-7 underlies the proton-activated, outwardly rectifying anion current in HEK293 cells or in astrocytes.

Comparison of PAORAC with anion channels that have not been identified molecularly

Ca2+-activated and volume-regulated anion channels are functionally well-characterized and exist in many tissues and cell types, but it has not yet been discovered which genes encode them (reviewed in: Jentsch et al. 2002; Nilius & Droogmans, 2003). In contrast to the electrophysiologically characterized members of the ClC gene family, Ca2+-activated and volume-regulated anion channels generally have a higher permeability for I than for Cl (reviewed in Frings et al. 1999; Jentsch et al. 2002). This permeability sequence, however, is shared by many other Cl channels (including GABAA and glycine receptors, Jentsch et al. 2002). It is also the permeability sequence we found for PAORAC (Fig. 11). Ca2+-activated Cl channels generally are outwardly rectifying and their voltage dependence at least superficially resembles the voltage dependence of PAORAC (e.g. Evans & Marty, 1986; Papassotiriou et al. 2001). However, the other characteristics of these channels are not compatible with the properties of PAORAC reported here. PAORAC also activates when Ca2+ in the intracellular solution is strongly buffered to a very low concentration (10 mm BAPTA and no added Ca2+, Fig. 4B). It can therefore not be considered a Ca2+-activated anion channel. Also the pharmacological profiles of both, Ca2+-activated and volume-regulated anion channels, are quite different from our findings on PAORAC in HEK293 cells. The IC50 of DIDS is lower in the proton-activated, outwardly rectifying anion channel (2.9 μm) than in Ca2+-activated Cl channels (IC50 16–247 μm; Frings et al. 1999) or in volume-regulated anion channels (IC50: 26 μm in HEK293 cells at +100 mV holding potential; Hélix et al. 2003). Also, mibefradil blocks both of the latter channels (Nilius et al. 1997b), but does not seem to affect the proton-activated, outwardly rectifying anion channel in HEK293 cells (Fig. 8). Finally, volume-regulated anion channels inactivate at strongly depolarized membrane potentials (Ackerman et al. 1994; Voets et al. 1997; Nilius et al. 1998; and Fig. 14, see also below), which is the opposite of the behaviour of the proton-activated, outwardly rectifying anion channel (Figs 6 and 14).

In conclusion, there is neither a cloned, nor an otherwise well characterized ion channel described that shares all the characteristics of the proton-activated, outwardly rectifying anion channel. This does not necessarily mean, however, that none of the known anion channels underlies the proton-activated anion currents. It is conceivable that there are channels that change their properties dramatically upon exposure to very acidic (pH 5.5 and below) extracellular conditions. Indeed, it has been reported that the aquaporin AQP6 changes from being essentially impermeable to ions to an anion channel when exposed to such an acidic environment (Yasui et al. 1999). Accepting, however, that plasma membrane proteins can alter their properties depending on extracellular pH, membrane proteins other than classical membrane channels must also be considered as potential candidates for PAORAC. For example, amino-acid transporters (Wadiche et al. 1995) have also been shown to act as anion channels under certain conditions.

PAORAC and VRAC are different conductances

Nobles et al. (2004) recently proposed that VRAC and PAORAC are different manifestations of the same channel. They argued that extracellular acidification changes some of the pharmacological and biophysical characteristics of this channel and that these changes account for all of the differences observed between PAORAC and VRAC: activation kinetics (less than 5 s for PAORAC, more than 1 min for VRAC), I–V relationship (much steeper for PAORAC), voltage dependence (VRAC is inhibited, while PAORAC is facilitated by depolarizations) and pharmacology (VRAC, but not PAORAC was sensitive to tamoxifen; Nobles et al. 2004). In addition to these points, we observed that PAORAC was insensitive to mibefradil (Fig. 8, in contrast to VRAC; Nilius et al. 1997b) and was – in stark contrast to VRAC – not blocked by 10 mm intracellular Mg2+ (Fig. 12). Although these points show clearly that VRAC has different properties compared to PAORAC, they are not incompatible with the model proposed by Nobles et al. (2004). Based on three different lines of evidence, however, we think that the proposal of Nobles et al. (2004) is incorrect. As detailed below, characteristics of single channel events, the effects of extracellular acidification on VRAC and the effects of depolarizing voltage pulses on VRAC under acidic extracellular conditions all provide independent arguments that VRAC and PAORAC are caused by different channels.

We measured single channel events that – as evidenced by the tests performed (pH sensitivity, DIDS sensitivity, and ensemble averaged I–V relationship) – are likely to underlie PAORAC as observed in whole-cell recordings (Fig. 9). Furthermore, we showed that these channels are permeated by anions (Fig. 9D). The voltage dependence of the single channel conductance of these channels (Fig. 10) measured at pH 5.0 is incompatible with the characteristics of single channels underlying VRAC in CPAE cells measured under similar conditions (pH 5.0, depolarized membrane potentials; Sabirov et al. 2000). At +120 mV, Sabirov et al. (2000) reported single channel currents of ca 6 pA for VRAC, while we measured only 2.6 pA for single channel currents through PAORAC (Fig. 10). This strongly suggests that the channels underlying VRAC and PAORAC are not the same. It is, however, necessary to keep in mind that CPAE cells are – in contrast to HEK293 cells, which are of human origin – of bovine origin. We therefore cannot formally exclude the possibility that the channels underlying VRAC in HEK293 cells have different single channel characteristics compared to those in CPAE cells. At neutral extracellular conditions, however, the single channel properties of volume-regulated anion channels in HEK293 cells (Ando-Akatsuka et al. 2002) are very similar to those described in CPAE cells (Sabirov et al. 2000).

We next investigated whether and how the I–V relationship of VRAC currents changed upon extracellular acidification. We found that the outward current showed a biphasic behaviour, first decreasing and then increasing again (Fig. 13). This is of relevance here since this biphasic response is difficult to reconcile with a simple binding reaction of protons to VRAC channels, necessitating adaptations to the original proposal of Nobles et al. (2004). In order to maintain this model, it is necessary to either assume at least two different proton binding sites to the channel, or a biphasic conformational response of the channel upon proton binding. It should also be noted that the pronounced inward currents of VRAC are not inhibited, but rather enhanced, by extracellular acidification (Fig. 13B), i.e. the I–V relationship under acidic and hypotonic extracellular conditions is clearly distinct from the I–V relationship under acidic conditions before the activation of VRAC. In the model of Nobles et al. (2004), this necessitates the assumption that the channel can exist in at least three biophysically distinct open conformations. Alternatively, however, these data are perfectly compatible with an explanation assuming two independent ionic conductances. In this alternative model, acidification activates proton-activated, outwardly rectifying conductances that are not influenced by hypotonicity. Outward VRAC currents activated by hypotonicity, on the other hand, are rapidly but only partially inhibited by extracellular acidification (note that we use the term ‘inhibited’, which can be fast, as opposed to ‘deactivated’, which is a slow process for VRAC). Such an inhibition of VRAC channels is consistent with most of the reports investigating pH regulation of VRAC (Ackerman et al. 1994; Jackson & Strange, 1995; Voets et al. 1997; Nilius et al. 1998; but see also Meyer & Korbmacher, 1996). It is, however, directly opposed to the model of Nobles et al. (2004).

In order to test the model proposing two independent channels, we devised an experimental procedure to differentially affect each of the two channels under identical, steady state ionic conditions (external pH 4.5 and hypotonicity). To achieve this, we took advantage of the fact that VRAC is fairly rapidly (time course in seconds) inhibited (and de-inhibited) by depolarization (and repolarization), while PAORAC very rapidly facilitates after depolarizing voltage jumps (e.g. Figs 6 and 7). Applying this protocol to cells after activation of VRAC (caused by extracellular hypotonicity) and after lowering the pH to 4.5, we consistently observed that outward currents were inhibited during the first voltage pulse. On the other hand, outward currents increased during the last voltage pulse of a train of six voltage pulses (Fig. 14E and F). This shows (in disagreement with Nobles et al. 2004) that VRAC-like inhibition persists even under acidic conditions. Furthermore, this experiment provides strong evidence for the simultaneous existence under these conditions of two independent types of channels, one showing VRAC-like inhibition, the other showing PAORAC-like facilitation in response to depolarizing voltage steps. Interestingly, we found that the activation of VRAC did not diminish the magnitude of PAORAC-like, depolarization-induced facilitation (Fig. 14C and F), indicating that the activation of VRAC did not occur to the expense of PAORAC, further strengthening our conclusion that both current components are independent from each other.

Altogether, the data summarized in this section provide several independent lines of evidence showing that the properties and characteristics of PAORAC are distinct and separable from those of VRAC, even under identical experimental conditions. This demonstrates that the proton-activated anion currents and the volume-regulated anion currents flow across the plasma membrane through physically distinct populations of ion channels. We wish to emphasize, however, that our argument does not necessarily imply that the proteins that provide the pores for VRAC and PAORAC must be different (at present, however, there is no experimental evidence to support or reject such a claim). This question probably cannot be addressed with biophysical techniques alone, but requires the molecular identification of the proteins responsible for VRAC and/or PAORAC.

Influence of PAORAC on the measurements of heterologously expressed ion channels

HEK293 (and CHO) cells are very useful and popular systems for the functional overexpression and analysis of cloned ion channels. Often it is interesting to investigate the pH dependence of these channels, either because a direct physiological relevance of pH changes to the channel function is suspected (e.g. for TRPV1, Tominaga et al. 1998; TRPV4, Suzuki et al. 2003; HCN1 and 4, Stevens et al. 2001), or because the pH-induced changes in biophysical characteristics allow the inference of structural properties of the ion conducting pathway (e.g. ClC-2, Niemeyer et al. 2003). Potentially, the existence of endogenous PAORAC in these popular expression system (Bompadre et al. 2001; Vanoye & George, 2002; Xu et al. 2004; Nobles et al. 2004) constitutes a problem with such studies. Several strategies can be adopted in order to avoid contaminating the currents through heterologously expressed channels with PAORAC: It may be possible to avoid lowering the pH to values below 5.5, effectively eliminating any contamination by PAORAC (Fig. 1). Alternatively, when inward currents are of special interest the small contribution of PAORAC (generally below 50 pA, Fig. 3) at negative holding potentials may be tolerable. When studying cationic conductances, exchanging Cl in the extra- or intracellular solutions for larger anions (e.g. aspartate or glutamate, Fig. 2) might be an experimental option (e.g. Xu et al. 2004). Finally, a pharmacological strategy using DIDS can be considered for eliminating the strong outward currents of PAORAC.

Possible functions of PAORAC

At first glance, the properties of PAORAC are puzzling. The steep pH dependence together with the very acidic KD (equivalent to pH 5.1) means that in native tissues these channels never will be activated as long as they stay on the plasma membrane. Other anion channels have been found primarily in intracellular compartments where they are important in vesicular acidification (ClC-3: Stobrawa et al. 2001; Li et al. 2002; Hara-Chikuma et al. 2005; ClC-5: Günther et al. 2003; ClC-7: Kornak et al. 2001). These channels are believed to provide an electrical shunt for the electrogenic proton pumps in these intracellular compartments (reviewed in Faundez & Hartzell, 2004). However, even in intracellular organelles the pH rarely reaches values below 5.5 (reviewed in: Mellman, 1992; Faundez & Hartzell, 2004) and PAORAC would therefore only be activated in lysosomes and – to a lesser extent – in late endosomes. It is not evident why a channel that is primarily important in lysosomes should ever be trafficked to the plasma membrane, especially since CLC-7, which clearly is targeted to lysosomes, could not be detected in the plasma membrane (Kornak et al. 2001; Auzanneau et al. 2003). While a vesicular, especially lysosomal, function for PAORAC should not be ruled out completely, it is also possible that this channel is capable of changing its properties in response to intra- or extracellular signalling events in a way that it becomes more readily activated under less acidic conditions. Also, the possibility should be entertained that the anionic channel properties of the protein that is responsible for PAORAC activity arises as a by-product of the molecular architecture of a plasma membrane resident protein that primarily serves an entirely different function.



This work was supported by the Deutsche Forschungsgemeinschaft through its Emmy Noether program (OB 177/1). We thank H. Löhr, M. Simon-Thomas, S. Johne and K. Dieckmann for excellent technical assistance and Professor Dr V. Flockerzi for his generous support.