Regulation of Na⁺ channels by luminal Na⁺ in rat cortical collecting tubule

Abstract

1. The idea that luminal Na⁺ can regulate epithelial Na⁺ channels was tested in the cortical collecting tubule of the rat using whole-cell and single-channel recordings. Here we report results consistent with the idea of Na⁺ self-inhibition.

2. Macroscopic amiloride-sensitive currents (I_Na) were measured by conventional whole-cell clamp. I_Na was a saturable function of external Na⁺ concentration ([Na⁺]_o) with an apparent K_m of 9 mm. Single channel currents (i_Na) were measured in cell-attached patches. i_Na increased with pipette Na⁺ concentration with an apparent K_m of 48 mm. Since I_Na= (i_Na)NP_o, the different K_m values imply that the channel density (N) and/or open probability (P_o) increase as [Na⁺]_o decreases. Reduction of [Na⁺]_o after increasing intracellular Na⁺ concentration also increased the outward amiloride-sensitive conductance, consistent with activation of the Na⁺ channels.

3. The underlying mechanism was studied by changing pipette Na⁺ concentration while recording from cell-attached patches. No increase in NP_o was observed, suggesting that the effect is not a direct interaction between [Na⁺]_o and the channel.

4. [Na⁺]_o was varied outside the patch-clamp pipette while recording from cell-attached patches. When amiloride was in the bath to prevent Na⁺ entry, no change in NP_o was observed.

5. Activation of the channels by hyperpolarization was observed with 140 mm Na⁺_o but not with 14 mm Na_o ⁺.

6. The results are consistent with the concept of self-inhibition of Na⁺ channels by luminal Na⁺. Activation of the channels by lowering [Na⁺]_o is not additive with that achieved by hyperpolarization.

The transport of Na⁺ by amiloride-sensitive epithelia is a saturating function of the mucosal or luminal Na⁺ concentration (Garty & Benos, 1988; Palmer, 1991). This self-limiting process has been attributed to three different mechanisms. First, the current through individual Na⁺ channels is known to saturate with increasing Na⁺ concentration (Olans, Sariban-Sohraby & Benos, 1984; Palmer & Frindt, 1988; Oh & Benos, 1993). However, noise analysis of Na⁺ currents in frog skin (Van Driessche & Lindemann, 1979) indicated that at Na⁺ activities below 60 mm the macroscopic current saturated while the single-channel current was a linear function of concentration. This implies that the number or open probability of the channels decreased with increasing mucosal Na⁺. This regulation in channel activity has in turn been attributed to two phenomena, as defined by Lindemann (Lindemann, 1984). ‘Feedback inhibition’ of the channels is defined as a downregulation due to the transport of Na⁺ across the apical membrane. ‘Self-inhibition’ is defined as a modulation of the channel activity by the apical Na⁺ concentration per se.

Evidence for feedback inhibition has been accumulating in recent years, mainly through electrophysiological analysis of Na⁺ channels in frog skin (Abramcheck, Van Driessche & Helman, 1985; Helman & Baxendale, 1990), epithelial A6 cells (Ling & Eaton, 1989) and rat cortical collecting tubule (Breyer, 1990; Breyer, 1991; Silver, Frindt, Windhager & Palmer, 1993; Frindt, Silver, Windhager & Palmer, 1993, 1995; Frindt, Palmer & Windhager, 1996). Proposed mediating factors have included the apical membrane voltage (Frindt et al. 1993), intracellular Ca²⁺ and Ca²⁺-dependent protein kinase activity (Ling & Eaton, 1989; Breyer, 1990; Breyer, 1991; Silver et al. 1993; Frindt et al. 1996), and cellular metabolism (Frindt et al. 1995).

The most direct evidence for self-inhibition has come from experiments in frog skin in which the mucosal Na⁺ concentration was increased rapidly (Fuchs, Hviid-Larsen & Lindemann, 1977). Under these conditions the macroscopic current increased rapidly to a maximum which is proportional to the Na⁺ concentration before relaxing over a period of about 1 s to a lower steady-state level. This relaxation process occurred before significant changes in intracellular Na⁺ concentration could be measured. This behaviour of the channels has not been confirmed by single-channel analysis. In the rat cortical collecting tubule (CCT), no correlation could be found between open probability of channels in cell-attached patches and the concentration of Na⁺ or Li⁺ in the patch-clamp pipette (Palmer & Frindt, 1988). In channels reconstituted into planar lipid bilayers, there was an increase, rather than a decrease in P_o with increasing Na⁺ concentration on the presumed extracellular side of the channels (Ismailov, Berdiev & Benos, 1995).

We have reinvestigated this issue using the whole-cell clamp approach to measure saturation of macroscopic Na⁺ currents, and cell-attached patches to measure saturation of single channels under similar conditions. The results are consistent with a downregulation of Na⁺ channels with increasing extracellular Na⁺, as predicted by the self-inhibition hypothesis.

METHODS

Biological preparations

Sprague-Dawley rats of either sex (100-150 g), raised free of viral infections (Charles River Laboratories, Kingston, NY, USA), were fed with a low Na⁺ diet (ICN, Cleveland, OH, USA) for at least 1 week to enhance Na⁺ channel activity. Animals were killed by cervical dislocation, the kidneys removed, and CCTs dissected free and opened manually to expose the luminal surface. Under these conditions the tissues retain their epithelial structure and the cells are presumed to remain polarized. The split tubules were attached to a small plastic rectangle coated with Cell-Tak (Collaborative Research, Bedford, MA, USA) and placed in a perfusion chamber mounted on an inverted microscope. The chamber was continuously perfused with solution which was prewarmed to 37°C by a miniature water-jacketed glass coil (Radnotti Glass Technology, Monrovia, CA, USA). Principal cells of the tubule were identified visually.

Solutions

For whole-cell recordings, the tubules were superfused with a solution consisting of (mm): 135 sodium methanesulphonate, 2 CaCl₂, 1 MgCl₂, 2 glucose, 5 BaCl₂ and 10 Hepes, adjusted to pH 7.4 with NaOH. The Na⁺ concentration was reduced by substitution with N-methyl-D-glucamine (NMDG). To measure the currents attributable to Na⁺ channels (I_Na), currents at voltages between -120 and +80 mV (cell relative to bath) were measured in the presence and absence of 10 μm amiloride (Research Biochemicals International). The basic pipette solution contained (mm): 7 KCl, 123 aspartic acid, 20 CsOH, 20 TEA-OH, 5 EGTA, 10 Hepes, 3 adenosine nucleotide (ATP or ADP) and 0.3 guanosine nucleotide (GTP, GTPγS, GDP or GDPβS), with the pH adjusted to 7.4 with KOH. In some experiments NaOH was used to adjust the pH, so that Na⁺ was the major cation in the pipette solution.

For cell-attached patch recordings, tubules were superfused with a high K⁺ solution to control the cell membrane potential. This solution consisted of (mm): 140 potassium methanesulphonate, 2 CaCl₂, 1 MgCl₂, 2 glucose and 10 Hepes, adjusted to pH 7.4 with KOH. The patch-clamp pipettes were filled with solutions identical to the extracellular solutions used for the whole-cell recordings. In some experiments the tubules were superfused with a high Na⁺ solution containing (mm): 140 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 2 glucose and 10 Hepes, adjusted to pH 7.4 with NaOH. The Na⁺ concentration of this solution was reduced by substitution with NMDG. The measured osmolarities of all solutions were 290-295 mosmol l⁻¹.

In some experiments, the pipette solution was exchanged using a technique that minimized mechanical vibrations (Gu & Sackin, 1995). In this method, the test solution was stored in PE-10 tubing attached to a quartz capillary that was inserted into the patch pipette to within 70 μm of the tip. Initially, equal negative pressures were applied to the patch pipette and the inner quartz capillary. This permitted formation of a gigaohm seal on the cell without movement of the test solution relative to the solution within the shaft of the patch pipette. Using remotely connected valves and negative pressures of about 15 mmHg, the solution at the tip of the patch pipette could be replaced within 5 s by the test solution stored in the PE-10 tubing. The initial solution could also be returned to the tip of the pipette by resetting the valves and applying negative pressure to the back of the PE-10 tubing. Use of remote valves and thin PE-10 tubing allowed these manoeuvres to be performed without disruption of the high resistance seal between the pipette and the cell. Electrical noise associated with the electrolyte solution stored in the PE-10 tubing could be effectively eliminated by temporarily shielding the front of the Faraday cage.

Electrical recordings

Basic patch-clamp methods were as described previously (Frindt et al. 1993; Silver et al. 1993; Frindt & Palmer, 1996). Pipettes were fabricated from haematocrit glass using a three-stage pulling process and were coated with Sylgard. Pipette resistances were 2-4 MΩ for whole-cell recording and pipette-perfusion experiments, and 3-6 MΩ for other cell-attached recordings. Seals were made on the apical surface of the cells, and were normally between 2 and 10 GΩ. Whole-cell clamp conditions were obtained by applying suction (usually 20-50 mmHg) to the pipette until an abrupt decrease in resistance and increase in capacitance were observed. Recording of currents and analysis of data were carried out with an Atari 1040 ST computer equipped with interface and data acquisition software (Instrutech, Mineola, NY, USA). Current records were stored on video tape using a pulse-code modulator (Instrutech). For analysis, records were filtered at 1 kHz for whole-cell recordings, and 500 Hz for single-channel recordings. Single-channel currents were measured from ten to twenty clear transitions in the current records. Computation of the mean number of open channels (NP_o), open probability (P_o) and mean open and closed times was carried out using the TAC program (Instrutech).

Under whole-cell conditions, currents through Na⁺ channels were estimated under each experimental condition by subtraction of currents measured under identical conditions except for the addition of 10 μm amiloride. In most instances the amiloride-sensitive currents were measured immediately after each measurement of the total current. When measurements at different Na⁺ concentrations were obtained from the same cell (e.g. Fig. 3), the total current was first measured at each Na⁺ concentration, and then the amiloride-insensitive current was measured, again at each Na⁺ concentration. In some cases the leak currents were measured both before and after the total currents, and did not change substantially over this time period.

Figure 3. Effect of [Na⁺]_o on I_Na.

A, currents in the absence (top) and presence (bottom) of 10 μm amiloride. The voltage was held at 0 mV and pulsed to values between -120 and +60 mV in 10 mV steps. After each pulse the voltage was returned to 0 mV for 100 ms. Current traces from all voltage steps are superimposed on the same time scale. [Na⁺]_o was 140 mm. B, I_Na-V relationships. Data are from a single cell with 140 (▪), 14 (•) and 3.5 mm Na⁺₀ (▴). Currents were measured 10-20 ms after the voltage shift, and leak corrected by subtracting currents measured in the same solutions in the presence of 10 μm amiloride. C, [Na⁺]_o dependence of the macroscopic current at -120 mV, 10-20 ms after the voltage shift. The data are normalized to the values obtained with 140 mm Na⁺₀ and represent means ±s.e.m. of 5-7 cells at each concentration. Data were fitted to the Michaelis-Menten equation using non-linear least squares analysis. The best fits were obtained with I_max= 1.07 and K_m= 9 mm.

RESULTS

Effects of adenosine and guanosine nucleotides

Previous measurements of whole-cell currents in principal cells of the CCT were made with 3 mm ATP and 0.3 mm GTP in the pipette solution. Currents under these conditions tended to run down, declining by about 60-70 % after 15 min. In preliminary experiments designed to optimize the whole-cell recordings, we studied the effects of replacing ATP and GTP with other nucleotides. As shown in Fig. 1, replacing ATP with ADP had no discernible effects on the time course of run-down of I_Na. Thus, it is unlikely that the channels are regulated directly by intracellular ATP. Figure 2 shows the effects of replacing GTP with either GTPγS, which should activate G-proteins continuously, or GDP or GDPβS, both of which should stabilize the inactive state of G-proteins. Of these, only GDPβS had a marked influence on the channels. With this compound in the pipette, I_Na was stable over a period of at least 10 min. It is not clear why this effect was specific for GDPβS and not observed with GDP (see Discussion). In the experiments described below, however, we made use of the stabilizing effect by including GDPβS in the pipette.

Figure 1. Effect of ATP and ADP on I_Na.

Currents were measured at a cell potential of -60 mV under whole-cell clamp conditions with 3 mm ATP or ADP in the pipette solution. In both cases the pipette contained 0.3 mm GTP. Currents measured in the presence of 10 μm amiloride were subtracted from the total current to obtain I_Na. Measurements were made within 1 min of formation of the whole-cell clamp, and other measurements, made from 5 to 15 min later, were normalized to this value. Data represent means ±s.e.m. of 6 measurements.

Figure 2. Effect of GTP and GDP on I_Na.

Currents were measured at a cell potential of -60 mV under whole-cell clamp conditions with 0.3 mm of either GTP, GTPγS, GDP or GDPβS in the pipette solution. In all cases the pipette solution contained 3 mm ATP. Currents measured in the presence of 10 μm amiloride were subtracted from the total current to obtain I_Na. Measurements were made within 1 min of formation of the whole-cell clamp, and other measurements, made from 5 to 15 min later, were normalized to this value. A, comparison of GTP and GDP. Data represent means ±s.e.m. of 5 measurements under each condition on the same set of tubules. B, comparison of no guanine nucleotides (Control), GTPγS and GDPβS. Data represent means ±s.e.m. of 6 (control), 12 (GTPγS) and 11 (GDPβS) measurements under each condition on the same set of tubules.

Macroscopic K_m for [Na⁺]_o

The concentration dependence of the macroscopic current was measured under whole-cell clamp conditions. We presume that the intracellular composition is determined primarily by that of the pipette solution, and in particular that Ca²⁺ and pH are strongly buffered. Therefore, changes in the Na⁺ concentration of the extracellular fluid ([Na⁺]_o) will not alter the intracellular environment. When [Na⁺]_o was reduced, the amiloride-sensitive Na⁺ current decreased rapidly, within the time required to exchange the bath solution (approximately 30 s). The changes in current were stable and reversible. In Fig. 3A we show superimposed current traces from a typical experiment with [Na⁺]_o= 140 mm in which the voltage was held at 0 mV and jumped to values of between -120 and +60 mV for 100 ms intervals. Large inward currents (downward deflections) were observed which were abolished by 10 μm amiloride (lower traces). At the most negative voltages there was a small increase in inward current over the 100 ms time period. Figure 3B shows I_Na-V curves for several different [Na⁺]_o in the same cell. Currents were measured 10-20 ms after the voltage switch, after the capacitative current transients had decayed. In each case, currents measured with the same [Na⁺]_o in the presence of 10 μm amiloride were subtracted to obtain I_Na. The absence of outward currents over the voltage range tested with 140 mm Na⁺ in the bath indicates that intracellular Na⁺ concentration ([Na⁺]_i) was less than 14 mm under these conditions. Small outward currents were observed at positive voltages when [Na⁺]_o was reduced to 14 mm or below. To analyse the concentration dependence we selected a voltage of -120 mV at which the signal-to-background ratio was high for all [Na⁺]_o. Currents were normalized to values at 140 mm[Na⁺]_o and plotted in Fig. 3C. The data could be well described by the equation:

where K_Na, the apparent Michaelis-Menten constant, has a value of 9 mm. Similar results were obtained when currents at -60 mV were analysed (K_Na= 9 mm). A slightly higher value of K_Na (12 mm) was estimated when currents were measured 100 ms after the beginning of the voltage shifts to -120 mV.

Single-channel K_m for [Na⁺]_o

The concentration dependence of the single-channel currents was measured in cell-attached patches. A high-K⁺ bath solution was used in these experiments in order to depolarize the cell potential so that the voltage across the patch would be determined by the clamping potential. The solutions used in the pipette were identical to the extracellular solutions used in the whole-cell experiments shown in Fig. 3. Figure 4A shows single-channel records at the same voltage (-120 mV) for different patches with various Na⁺ concentrations in the pipette ([Na⁺]_o). Channel openings and closings could be well resolved in each case. Figure 4B shows a plot of the single-channel current as a function of [Na⁺]_o. As for the macroscopic currents, the single-channel currents (at the same voltage) are fairly well described by the equation:

However, the estimated value of k_Na, the apparent Michaelis-Menten constant for the single-channel current, was 48 mm, considerably larger than that for the whole-cell currents. This suggests that as the Na⁺ concentration is decreased over the physiological range from 140 to < 10 mm, the macroscopic current decreases less than the single-channel current.

Figure 4. Effect of [Na⁺]_o on i_Na.

A, current tracings with 140, 70, 35 and 14 mm Na⁺ in the pipette. The pipette potential was 120 mV. Bars on the right indicate closed state of channels. B, [Na⁺]_o dependence of the single-channel current. Data represent means ±s.e.m. for 8-9 patches at each concentration. Data were fitted to the Michaelis-Menten equation using non-linear least squares analysis. Best fits were obtained with i_max= 1.29 pA and K_m= 48 mm.

This effect can be quantified from the relationship:

where N and P_o are the channel density and the open probability, respectively. Dividing the normalized value of I_Na by the value of i_Na at each Na⁺ concentration provides an estimate of NP_o, the mean number of open channels, as a function of [Na⁺]_o. This plot is shown, together with superimposed values of I_Na and i_Na, in Fig. 5. NP_o is a decreasing function of [Na⁺]_o, with the highest directly measured value (at [Na⁺]_o= 14 mm) about twice that at 140 mm Na⁺₀. This conclusion does not depend on the assumptions that I_Na and i_Na are hyperbolic functions of [Na⁺]_o, since the values for NP_o were obtained directly from the measured values of I_Na and i_Na, except for the point at the lowest [Na⁺]_o, which was estimated using the fitted Michaelis-Menten parameters.

Figure 5. Effect of [Na⁺]_o on NP_o.

Data for I_Na and i_Na are replotted from Figs 3B and 4B, respectively, with the i_Na values as well as the I_Na values normalized to those obtained at 140 mm Na⁺₀. Data for NP_o were obtained by dividing I_Na by i_Na at each [Na⁺]_o. The value at [Na⁺]_o= 7 mm was estimated using the best-fit Michaelis-Menten parameters for the two data sets.

Effects of [Na⁺]_i

In principle, the effects observed could be a direct consequence of changes in [Na⁺]_o or an indirect effect mediated by changes in [Na⁺]_i. To evaluate the effects of possible changes in [Na⁺]_i, we examined the effects of including a high Na⁺ concentration in the pipette on the whole-cell amiloride-sensitive conductance. Figure 6A illustrates I_Na-V relationships from two cells of the same tubule, one with K⁺ as the major cation in the pipette (and cell) and one with Na⁺ as the major cation. In both cases, the currents were measured 5 min after the establishment of the whole-cell clamp to allow the solutes in the pipette and the cell to equilibrate. The major difference in the two curves is the presence of outward current at positive cell potentials when Na⁺ is present in the pipette. The reversal potential obtained by interpolation was +17 mV. In eleven experiments, the mean reversal potential was 16 ± 9 mV. This indicates an [Na⁺]_i concentration close to the membrane of 77 mm, compared with 100 mm in the pipette.

Figure 6. Effect of [Na⁺]_i on I_Na.

A, I_Na-V relationships were obtained from 2 cells from the same tubule with either the normal high K⁺ solution in the pipette (▪), or with Na⁺ replacing K⁺ as the major cation (□). Currents were measured 5-6 min after formation of the whole-cell clamp in the absence and presence of 10 μm amiloride. [Na⁺]_o was 140 mm in both cases. B, amiloride-sensitive conductance (G_Na) with K⁺ (▪) or Na⁺ (□) in the pipette. Whole-cell conductances between -60 and -100 mV were measured immediately after formation of the whole-cell clamp and 5-6 min later. Data represent means ±s.e.m. for 11 cells for each condition.

Figure 6B shows the inward whole-cell amiloride-sensitive conductance (G_Na), measured as the slope of the I_Na-V curve between -100 and -60 mV, for cells with either K⁺ or Na⁺ in the pipette. G_Na was measured immediately after formation of the whole-cell clamp and 5-6 min later. In this case the pipette solution contained GTP, which accounts for the decline in G_Na observed with both Na⁺ and K⁺ in the pipette. There was no significant difference between the conductances even with the large increase in [Na⁺]_i. Thus, it is unlikely that changes in [Na⁺]_i, which are in any case rather small, can account for the observed dependence of NP_o on [Na⁺]_o as shown in Fig. 5.

Effects of [Na⁺]_o on outward Na⁺ currents

If Na⁺ channels are activated by lowering [Na⁺]_o, then the permeability to Na⁺ in the outward direction should also be increased under these conditions. To test this prediction, we loaded cells with Na⁺ through the patch-clamp pipette as in Fig. 6 and measured outward currents with both high and low Na⁺ in the bath. The result of one experiment is shown in Fig. 7A. With 140 mm Na⁺ in the bath, outward amiloride-sensitive currents and slope conductances were low. Reduction of [Na⁺]_o to 7 mm shifted the reversal potential to negative voltages, lowered the inward conductance and increased outward currents at potentials > 0, as would be expected from the increased driving force for outward Na⁺ movements. The outward slope conductance measured between 50 and 100 mV also increased. This increase is not expected if the permeability to Na⁺ is constant, but is consistent with the idea that the channels are activated under conditions of low [Na⁺]_o.

Figure 7. Effect of [Na⁺]_o on outward G_Na.

A, I_Na-V relationships were obtained with high Na⁺ solution in the pipette and either 140 or 7 mm Na⁺ in the bath. Currents with 140 mm Na⁺ were measured both before (▴) and after (▿) the currents with 7 mm Na⁺ were obtained. The slope conductances for outward currents were measured between +50 and +100 mV. B, the ratio of outward G_Na in 7 mm Na⁺₀ to the mean G_Na with 140 mm Na⁺₀ is plotted as a function of the absolute outward G_Na in 140 mm Na⁺₀. Each point represents a single experiment.

Results of eight similar experiments are shown in Fig. 7B. In every cell studied, the outward Na⁺ conductance was increased by reducing [Na⁺]_o. The magnitude of the increase, however, was variable and correlated with the initial outward current; larger activations were observed when the initial conductance was small. This suggests that the channels may already be maximally activated when the outward currents are large, and is consistent with the idea that activation can occur through different pathways (see Discussion).

Local effects of [Na⁺]_o

We next tested whether the effect of [Na⁺]_o on NP_o could be mediated by local effects of [Na⁺]_o at the outer membrane surface, by changing the composition of the fluid in the pipette during recording from cell-attached patches. In these experiments, seals were formed with 140 mm Na⁺ in the pipette. Na⁺ channel activity was recorded for approximately 60 s, after which time the pipette was perfused with a solution containing 14 mm Na⁺. After the fluid exchange was complete, as judged from the decline in the amplitude of the single-channel current, the activity was measured for an additional 60 s. In some cases the seals were sufficiently stable to reverse the exchange, so that activity could be monitored again with the original 140 mm Na⁺ solution. Previous work has shown that channel activity is generally stable over this time period (Silver et al. 1993). Figure 8A illustrates a typical recording, showing channel opening and closing with 140 mm Na⁺₀, and in the same patch after reduction of [Na⁺]_o to 14 mm. The amplitude of the single-channel currents is decreased as expected, but there is no clear change in the activity of the channels measured as NP_o. In Fig. 8B the results of eight successful experiments are reported. Although there were sometimes minor changes in NP_o after changing the pipette solution, there was no consistent trend, and the mean change in NP_o was not significant; the mean ratio of NP_o in 14 mm Na⁺ to that in 140 mm Na⁺ was 0.95 ± 0.12.

Figure 8. Effect of changing pipette [Na⁺] on single-channel events.

A, current records obtained in the same patch with 140 and 14 mm Na⁺ in the pipette about 60 s after exchange of the pipette solution. The pipette voltage was -60 mV. Numbers 1-3 indicate number of open channels. The level with all channels closed is not seen in these segments of the recording. B, NP_o of patches with 140 and 14 mm Na⁺ in the pipette. Points representing the same patch before and after exchange of the pipette solution are joined by lines.

Global effects of [Na⁺]_o

It is possible that the reason that changing the pipette solution had no discernible effect on channel activity is that the channel is controlled not just by the local Na⁺ concentration but by that experienced by the entire apical cell membrane. For example, if a receptor for extracellular Na⁺ was coupled to a second messenger system rather than directly to the channel, channels within a patch could be influenced by changes in [Na⁺]_o outside the patch. This possibility was tested by first forming a cell-attached patch and then changing [Na⁺]_o while recording channel activity. This protocol was complicated, however, by two previously observed phenomena. First, reduction in Na⁺ entry across the apical membrane by amiloride was shown to activate the channels, acting primarily through hyperpolarization of the membrane voltage (Frindt et al. 1993). Reducing [Na⁺]_o is expected to have similar effects which would be independent of the self-inhibition process. To eliminate this effect, amiloride (1 μm) was included in the superfusate of both high and low Na⁺ solutions. Second, reducing [Na⁺]_o, presumably at the basolateral surface, was shown to increase intracellular [Ca²⁺] ([Ca²⁺]_i) by altering the driving force for Na⁺-Ca²⁺ exchange (Breyer, 1991). These changes in [Ca²⁺]_i are known to affect channel activity (Silver et al. 1993). To prevent this process, we used an extracellular solution with 0.1 mm EGTA and no added Ca²⁺ to reduce the dependence of [Ca²⁺]_i on [Na⁺]_o.

Cell-attached patches were formed with high [Na⁺]_o and channel activity was measured for about 60 s. The superfusate was then switched to a low [Na⁺]_o (7 mm) solution. After a period of 30 s to allow for exchange of the bath solution, channel activity was measured for an additional 1-3 min. In some cases, the bath was switched back to the high [Na⁺]_o solution. There was no clear change in either the current amplitude or the activity of the channels. Figure 9 shows the results of twelve experiments. There was no significant change in NP_o; the mean ratio of values in 140 mm and 7 mm Na⁺₀ was 1.02 ± 0.10. There was only one patch in which there was a clear, reversible increase in NP_o, and in this case there was a concomitant increase in i_Na, suggesting that the cell voltage had hyperpolarized, which might account for the change in NP_o.

Figure 9. Effect of changing bath [Na⁺] on single-channel events.

NP_o of the same patches with 140 mm Na⁺ in the pipette and 140 or 7 mm Na⁺ in the bath. Points representing the same patch before and after exchange of the bath solution are joined by lines.

Interaction between voltage and [Na⁺]_o

One possible explanation for the failure to observe activation at the single-channel level by reducing [Na⁺] in the patch pipette is that the channels were already in a relatively activated state. In order to measure channel currents and kinetics with low pipette [Na⁺] it was necessary to maintain the patch at large transmembrane voltages. In the experiment shown in Fig. 8A the single-channel currents were 0.97 pA in the presence of 140 mm Na⁺₀, indicating a transpatch voltage of about -120 mV. This large voltage was necessary to resolve currents clearly when the Na⁺ concentration was reduced. We have shown previously, using high concentrations of Li⁺ in the pipette, that hyperpolarization of cell-attached patches leads to a slow activation of the channels (Frindt et al. 1993; Palmer & Frindt, 1996). It is possible that this activation is not additive with the putative activation by low [Na⁺]_o.

If this is the case then, conversely, channels activated in the presence of low [Na⁺]_o should not be activated further by hyperpolarization. To test this idea, we measured the slow time-dependent activation of I_Na by hyperpolarizing the cells under whole-cell clamp conditions. After establishing the whole-cell clamp, the cells were held at 0 mV for at least 30 s and then hyperpolarized to -100 mV for 15 s. Amiloride-sensitive currents were normalized to values obtained 10 ms after the voltage switch, after the decay of the capacitative current transient. As shown in Fig. 10, there was a biphasic activation of I_Na by hyperpolarization. An initial increase, complete within 1 s, was followed by a slower increase, which in many cells was still apparent after 15 s. In the same cells with 14 mm Na⁺₀, the rapid activation was blunted and the slow activation was abolished. This result would be expected if the channels were activated by low [Na⁺]_o and if this activation process were not additive with that of voltage.

Figure 10. Activation of I_Na by hyperpolarization.

I_Na was measured with high K⁺ solution in the pipette and either 140 or 14 mm Na⁺ in the bath. After holding the cell voltage at 0 mV for 30 s or more, the voltage was switched to -100 mV. Currents were measured for 15 s, normalized to the value obtained 10 ms after the onset of the voltage pulse, and corrected for currents measured in the presence of 10 μm amiloride. Data are the means ±s.e.m. for 11 experiments in which the time-dependent activation was measured with both [Na⁺]_o.

DISCUSSION

Effects of nucleotides on Na⁺ channels

Under whole-cell clamp conditions, the Na⁺ channels in the CCT were stabilized, if not activated, in the presence of GDPβS. This non-hydrolysable analogue of GDP would be expected to stabilize G-proteins in an inactive state. If this is the mechanism by which the nucleotide is acting in this case, it would suggest the presence of a G-protein which can inhibit the channels. This interpretation is complicated, however, by two findings. First, GTPγS, which should maximally activate G-proteins, was not more effective than GTP. Second, GDP, which should also stabilize the inactive state of the G-protein, did not reproduce the effect of GDPβS. We have no good explanations for these findings.

Effects of G-proteins on Na⁺ channels from other tissues have been variable. Activation of G-proteins inhibited highly selective channels in excised patches from A6 cells (Ohara, Matsunaga & Eaton, 1993), but stimulated Na⁺ channels in toad bladder membrane vesicles (Garty, Yeger, Yanovsky & Asher, 1989) as well as a less selective channel in A6 cells (Cantiello, Patenaude & Ausiello, 1989). The results of the present study hardly resolve this issue. We were able, however, to make use of the stabilizing effect of GDPβS to study whole-cell currents under conditions of different [Na⁺]_o without the confounding effect of rapid channel run-down.

Na⁺ self-inhibition

The concept of Na⁺ self-inhibition was defined as a downregulation of Na⁺ channels by Na⁺ acting on the exterior surface of the apical membrane (Lindemann, 1984). This was distinguished from ‘feedback inhibition’ processes, which resulted from Na⁺ entry into the epithelial cell. The clearest evidence for self-inhibition comes from experiments with frog skin in which the Na⁺ concentration at the outer surface of the epithelium was rapidly increased (Fuchs et al. 1977). After the current reached a peak, it relaxed with a time constant of around 1 s, presumably reflecting self-inhibition. The possibility that the relaxation might be due to Na⁺ entry into the cell, and hence feedback inhibition, was discarded since current-voltage analysis indicated that [Na⁺]_i did not change over this time period.

Fluctuation analysis of the frog skin provided data that were consistent with this hypothesis (Van Driessche & Lindemann, 1979). The number of open Na⁺ channels decreased with increasing mucosal Na⁺ concentrations. In this case, however, a concomitant increase in [Na⁺]_i could not be ruled out.

Finally, a number of compounds including sulfhydryl reagents such as p-chloromercuribenzenesulphonic acid (PCMBS; Fuchs et al. 1977), guanidinium derivatives such as benzimidazolyl guanidine (BIG; Li & Lindemann, 1983), and detergents (Li, Zuzack & Kan, 1986) were all found to stimulate Na⁺ transport in frog skin and/or toad urinary bladder through a mechanism that apparently involved interference with Na⁺ self-inhibition.

Our measurements of the effects of [Na⁺]_o on macroscopic and single-channel currents suggest that Na⁺ channels in the CCT are also regulated through a self-inhibition mechanism. The key finding in this regard is that the single-channel currents saturated at higher Na⁺ concentrations than did the macroscopic currents. The inference that this was an effect of extracellular Na⁺ per se rests on the assumption that under whole-cell conditions no significant changes in intracellular composition occurred when [Na⁺]_o was changed. Although the low resistance pathway between the cell and the patch pipette is likely to keep the solute composition of the cell similar to that of the pipette, it is possible that changes, especially in the Na⁺ concentration near the membrane, could occur which could confound the interpretation of the results.

Two results make this possibility unlikely. First, the reversal potential for amiloride-sensitive current was in most cases more positive than +80 mV. This implies that even with 140 mm Na⁺ in the bath, the Na⁺ concentration just inside the apical membrane was less than 7 mm. Second, even large changes in [Na⁺]_i, elicited by including Na⁺ in the patch pipette, had minimal effects on the inward Na⁺ conductance. In this regard, the CCT appears to differ from the salivary duct, in which a G-protein-dependent downregulation of Na⁺ channels by high [Na⁺]_i was recently described (Komwatana, Dinudom, Young & Cook, 1996).

It is, in principle, possible that changes in [Na⁺]_o could also affect intracellular Ca²⁺ or H⁺ through changes in the Na⁺ gradient which would alter the driving force for Na⁺-Ca²⁺ or Na⁺-H⁺ exchange. Such effects are also unlikely in our preparation, since the intracellular Ca²⁺ and H⁺ are strongly buffered by EGTA and Hepes, respectively. In addition, any putative effects of raised [Ca²⁺] or [H⁺] would be to inhibit the Na⁺ channels when [Na⁺]_o was low. This would, if anything, tend to blunt the observed increase in NP_o.

Mechanism of self-inhibition

We were unable to demonstrate the self-inhibition effect in cell-attached recordings either by changing the solution in the pipette (with constant [Na⁺] outside the pipette) or by changing the solution outside the pipette (with constant [Na⁺] in the pipette). Apparently, the single-channel recording conditions used interfere with the effect. One possibility is that the large transmembrane voltage applied to the patch activates the channels (Frindt et al. 1993; Palmer & Frindt, 1996), and that they cannot be further activated by reducing [Na⁺]_o. The results of Fig. 10 are consistent with this idea. In the presence of high [Na⁺]_o the channels can be activated over a 15 s time course by hyperpolarization of the cell membrane from 0 to -100 mV. In the presence of low (14 mm) [Na⁺]_o, this activation is not observed. We suggest that this is because the channels are already activated at low [Na⁺]_o even when the cell voltage is zero. This would be the converse of the situation in Fig. 8, in which reducing [Na⁺]_o failed to activate channels that were presumably already activated by the hyperpolarized cell potential. A direct test of this explanation would require measurements of channel activity at low voltages and low [Na⁺]_o. These measurements are technically difficult due to the low conductance of these channels. It is also possible that tests for ‘global’ effects of [Na⁺]_o were negative because the apical membrane would have been hyperpolarized by the use of amiloride in the superfusate to prevent changes in membrane potential and [Na⁺]_i.

The effect of voltage was previously interpreted by us in terms of a switch in gating mode (Palmer & Frindt, 1996). According to this model, hyperpolarization favours a high P_o mode, whereas depolarization stabilizes a low P_o mode. We now further suggest that the high P_o mode is stabilized by low [Na⁺]_o (Fig. 11). Channels already in the high P_o mode as a result of hyperpolarization are unaffected by low [Na⁺]_o, and conversely channels already in the high P_o mode as a result of low [Na⁺]_o are unaffected by hyperpolarization.

Figure 11.

Schematic representation of gating modes of the Na⁺ channel.

Physiological importance

Self-inhibition, like feedback inhibition, is a means of limiting Na⁺ transport by any one cell. This will have the effect of distributing Na⁺ transport more evenly along the collecting duct. One benefit to the transporting cells is that the tendency to overload the pump or metabolic capacity during periods of high Na⁺ delivery would be blunted. In addition, other transport processes such as K⁺ and H⁺ secretion, which depend in part on the transepithelial voltage established by the Na⁺ transport system, will also be more evenly distributed. Some of these implications of Na⁺ dependence of the apical Na⁺ permeability have been examined quantitatively in a mathematical model of the collecting tubule (Strieter, Stephenson, Giebisch & Weinstein, 1992).

It is difficult to assess quantitatively the relative importance of the feedback and self-inhibition processes to regulation of transport in vivo. Self-inhibition will affect the channels directly. Together with the regulation by the transmembrane voltage, it will provide the first line of defence against overloading of the cell with Na⁺ under conditions of an increase in delivery of Na⁺ to the distal nephron.