Na⁺-activated K⁺ channels in small dorsal root ganglion neurones of rat

Abstract

1. Whole-cell Na⁺-activated K⁺ (K_Na) channel currents and single K_Na channels were studied with the patch-clamp method in small (20-25 μm) dorsal root ganglion (DRG) neurones in slices of rat dorsal root ganglia.

2. The whole-cell K_Na channel current was identified as an additional K⁺-selective leakage current which appeared after cell perfusion with internal solutions containing different [Na⁺]. The concentration for half-maximal activation of K_Na channel current was 39 mm and the Hill coefficient was 3.5. At [Na⁺]_i above 12 mm, K_Na channel current dominated the unspecific leakage current. The ratio of maximum K_Na channel current to unspecific leakage current was 45.

3. K_Na channel current was not activated by internal Li⁺. It was suppressed by external 20 mm Cs⁺ but not by 10 mm tetraethylammonium.

4. Single K_Na channels with a conductance of 142 pS in 155 mm external K⁺ (${\text{K}}_{\text{o}}^{+}$)-85 mm internal K⁺ (${\text{K}}_{\text{i}}^{+}$) solutions were observed at a high density of about 2 channels μm⁻².

5. In two-electrode experiments, a direct correlation was seen between development of whole-cell K_Na channel current and activation of single K_Na channels during perfusion of the neurone with Na⁺-containing internal solution.

6. Under current-clamp conditions, K_Na channels did not contribute to the action potential. However, internal perfusion of the neurone with Na⁺ shifted the resting potential towards the equilibrium potential for K⁺ (E_K). Varying external [K⁺] indicated that in neurones perfused with Na⁺-containing internal solution the resting potential followed the E_K values predicted by the Nernst equation over a broader voltage range than in neurones perfused with Na⁺-free solution.

7. It is concluded that the function of K_Na channels has no links to firing behaviour but that the channels could be involved in setting or stabilizing the resting potential in small DRG neurones.

K⁺ channels activated by internal Na⁺ (K_Na channels) described initially in cardiac myocytes (Kameyama, Kakei, Sato, Shibasaki, Matsuda & Irisawa, 1984) have been found in many types of neuronal membranes (Dryer, 1991; Egan, Dagan, Kupper & Levitan, 1992a, 1992b; Dryer, 1994). They are characterized by a large unitary conductance, frequent appearance of substates and low sensitivity to external tetraethylammonium (TEA). The channels need at least 20 mm internal Na⁺ to become active. The level of intracellular Na⁺ in neurones, however, is lower than 20 mm and is unlikely to increase considerably as a result of Na⁺ influx through voltage-gated Na⁺ channels (Dryer, 1991). Co-localization of K_Na and voltage-activated Na⁺ channels in the narrow nodal region of myelinated amphibian axon (Koh, Jonas & Vogel, 1994) led to the assumption that K_Na channels could be activated due to local Na⁺ accumulation during long high-frequency trains of action potentials. In the somata of rat motoneurones, K_Na channels were found at increased densities in the vicinity of the processes, probably axons, in the region of increased Na⁺ channel density and most probable Na⁺ accumulation (Safronov & Vogel, 1996). Their activation during trains of action potentials resulted in a pronounced slow after-hyperpolarization. Therefore, K_Na channel distribution and their co-localization with Na⁺ channels appear to play a crucial role in defining their function in the neuronal membrane.

During our experiments with small (20-25 μm) sensory neurones in thin slices of dorsal root ganglion (DRG) of newborn rat (Safronov, Bischoff & Vogel, 1996), K_Na channel activity was observed in membrane patches. In contrast to motoneurones and peripheral axons, small DRG neurones with round somata do not possess an anatomical structure such as the axonal hillock or the node of Ranvier with a high density of Na⁺ channels and a large ‘surface-to volume’ ratio which could enhance Na⁺ accumulation. Moreover, small DRG neurones of C-type are not able to generate action potentials at high frequencies (Harper & Lawson, 1985b; authors’ unpublished observations). Therefore, it seems unlikely that activation of K_Na channels has a direct link to firing behaviour in these neurones. On the other hand, small DRG neurones in the slice preparation have a very high input resistance of several gigaohms (Safronov et al. 1996) and activation of even a small additional conductance could thus have an influence on membrane resting potential.

Most studies of K_Na channels have been performed on cultured or dissociated neurones (Dryer 1991; Egan et al. 1992b; Haimann, Magistretti & Pozzi, 1992) and channel densities and membrane resting conductances could differ from those in intact cells. In addition, previous studies were mainly focused on properties of single K_Na channels and only little is known about whole-cell K_Na channel currents and maximum K_Na channel conductance in the neuronal membrane. The purpose of the present study was to describe the whole-cell K_Na channel conductance and single K_Na channels as well as to estimate their contribution to action potentials and resting potentials in intact DRG neurones.

METHODS

Preparation

Experiments were carried out using the patch-clamp technique (Hamill, Marty, Neher, Sakmann & Sigworth, 1981) on 150 μm thin slices prepared from dorsal root ganglia of 4- to 9-day-old rats (Safronov et al. 1996). In brief, rats were rapidly decapitated (in accordance with the German guidelines) and two dorsal root ganglia from lower thoracal and lumbar regions were carefully cut out in ice-cold standard external solution. The ganglia were desheathed using fine forceps and embedded in 2% agar according to the procedure described elsewhere (Edwards, Konnerth, Sakmann & Takahashi, 1989; Takahashi, 1990). After solidification of the agar, small blocks containing the ganglia were cut out. The slices were made using a tissue slicer. The slices were further prepared and kept according to the description given by Takahashi (1990). The experiments were performed on small DRG neurones (Harper & Lawson, 1985a, 1985b), with a diameter of 20-25 μm, which could be directly accessed on the surface of the slice (Safronov et al. 1996). Membrane input resistance of the neurones measured using pipettes filled with a high internal K⁺ (${\text{K}}_{\text{i}}^{+}$) solution was 1.3 ± 0.2 GΩ (19 neurones).

Solutions

The composition of all external and internal solutions used in the present study is given in Table 1. Most external solutions were bubbled with 95% O₂-5% CO₂. The slices were prepared and maintained in a standard external solution. During experiments the slices were perfused with Ca²⁺-free solution. The Ca²⁺-free external solution used for filling the pipettes in experiments with inside-out patches (Ca²⁺-free^o solution) was buffered with Hepes-NaOH. Tetrodotoxin (TTX; 100 nm) was included in the high external K⁺ (${\text{K}}_{\text{o}}^{+}$) solution, in order to block voltage-activated Na⁺ channels. High-K${\text{K}}_{\text{o}}^{+}$-TEA solution contained in addition 1 mm TEA for suppression of delayed rectifier and large conductance Ca²⁺-activated K⁺ channels. TTX, TEA and 4-aminopyridine (4-AP) were directly added to the external solutions. Bovine serum albumin (0.05%) was added to solutions containing dendrotoxin I (DTX-I) and margatoxin (MTX), in order to prevent non-specific binding of the blocker molecules to the surface of the experimental chamber.

Table 1.

Solutions

	A. External solutions
Solution	NaCl (mm)	KCl (mm)	LiCl (mm)	CaCl2 (mm)	MgCl2 (mm)	Glucose (mm)	NaH2PO4 (mm)	NaHCO3 (mm)	TTX (nm)	TEA (mm)	Hepes (mm)	LiOH (mm)	pH
Standard	115	5.6	—	2.2	1	11	1	25	—	—	—	—	7.4 (O2–CO2)
Ca2+ free	115	5.6	—	—	1	11	1	25	—	—	—	—	7.4 (O2–CO2)
4 mm${\text{K}}_{\text{o}}^{+}$	116.6	4	—	—	1	11	1	25	—	—	—	—	7.4 (O2–CO2)
8 mm${\text{K}}_{\text{o}}^{+}$	112.6	8	—	—	1	11	1	25	—	—	—	—	7.4 (O2–CO2)
12 mm${\text{K}}_{\text{o}}^{+}$	108.6	12	—	—	1	11	1	25	—	—	—	—	7.4 (O2–CO2)
Ca2+-freeo	136.4	5.6	—	—	1	11	—	—	—	—	10	—	7.4 (NaOH total 4.6 mm)
High ${\text{K}}_{\text{o}}^{+}$	5	152.5	—	—	1	—	—	—	100	—	5	—	7.4 (KOH total 2.5 mm)
High ${\text{K}}_{\text{o}}^{+}$-TEA	5	152.5	—	—	1	—	—	—	100	1	5	—	7.4 (KOH total 2.5 mm)
141 mm${\text{Li}}_{\text{o}}^{+}$	—	5.6	138	—	1	11	—	—	—	—	—	3	7.4 (O2–CO2)

	B. Internal solutions
Solution	NaCl (mm)	KCl (mm)	LiCl (mm)	Choline chloride (mm)	MgCl2 (mm)	EGTA (mm)	Hepes (mm)	pH (mm)
High ${\text{K}}_{\text{i}}^{\text{+}}$	5	144.4	—	—	1	3	10	7.3 (KOH total 10.6 mm)
x mm${\text{Na}}_{\text{i}}^{\text{+}}$	x	73.6	—	70 –x	1	3	10	7.3 (KOH total 11.4 mm)
(x = 0, 10, 20, 30, 50 and 70 mm)
70 mm${\text{Li}}_{\text{i}}^{\text{+}}$	—	73.6	70	—	1	3	10	7.3 (KOH total 11.4 mm)

Investigation of inside-out patches was performed in an additional small 0.2 ml bath as described previously (Safronov & Vogel, 1995). This procedure allowed us to obtain several inside-out patches from the same slice without its destruction due to perfusion with internal solutions containing high concentrations of K⁺ ions.

Current and voltage recordings

The patch pipettes were pulled from borosilicate glass tubes (GC 150; Clark Electromedical Instruments, Pangbourne, UK). The pipettes used for single-channel recordings had a resistance of 3-13 MΩ. The resistance of the pipettes for whole-cell recordings ranged between 5 and 10 MΩ. All pipettes were fire polished directly before the experiments. The patch-clamp amplifier used in all voltage- and current-clamp experiments was a List EPC-7 (Darmstadt, Germany). An additional EPC-7 amplifier was employed in two-electrode experiments. The data were either directly filtered at 1 kHz and stored in a computer using commercially available software (pCLAMP; Axon Instruments, Foster City, CA, USA) or they were filtered at 10 kHz and first stored in a digital tape recorder (DTR-1202; Biologic, Claix, France). For further analysis, these data were replayed from tape, low-pass filtered at 1 kHz and stored on disk. The frequency of digitization was at least twice as high as that of the filter. For calculation of the open probability (P_o), a channel was considered as open if its amplitude exceeded 50% of the control level. Events shorter than 1 ms were ignored. Offset potentials were nulled directly before formation of the seal. Liquid junction potentials were not corrected. Traces recorded in current-clamp mode were digitized with an interval of 0.1 ms.

All numerical values are given as means ± standard error of the mean (s.e.m.). The data points were fitted by linear or non-linear least-squares procedures. The errors of the fitting parameters are given as ± standard error (s.e.). All experiments were carried out at a room temperature of 22-25°C.

RESULTS

Whole-cell K_Na channel currents

Small DRG neurones in the slice preparation have a high input resistance (Safronov et al. 1996) and, therefore, could be used as a suitable model for the investigation of whole-cell K_Na channel currents. We identified whole-cell K_Na channel currents as leakage currents which appeared during cell perfusion with Na⁺-containing internal solution. The leakage currents were visualized in Ca²⁺-free external solution by 150 ms hyperpolarizing voltage pulses to -120 mV from a holding potential of -80 mV. Recording pipettes were filled with 0, 10, 20, 30, 50 or 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solutions. In these experiments, the unspecific leakage currents were very small (10-70 pA; membrane input resistance, 0.6-4 GΩ) immediately after membrane breakthrough and establishment of the whole-cell recording configuration. In experiments in which the pipettes contained 0 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution, the leakage currents remained unchanged during tens of minutes (Fig. 1A, top trace; see also Fig. 2C). When the pipettes were filled with 10-70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solutions, the small original leakage currents increased considerably within tens of seconds, as the neurone was perfused with the pipette solution. Recordings of the leakage currents made 4-6 min after their saturation are shown in Fig. 1A for 20 and 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solutions (different neurones). Resting potentials in neurones perfused with 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution were close to -68 mV which corresponded to the equilibrium potential for K⁺ (E_K) under the experimental conditions used (5.6 mm ${\text{K}}_{\text{o}}^{+}$-85 mm ${\text{K}}_{\text{i}}^{\text{+}}$). Therefore, the component of leakage current which developed after neurone perfusion with internal Na⁺ was K⁺ selective. In order to study the concentration dependence of the leakage currents, measurements of their amplitudes were performed in five different neurones for each [Na⁺]_i. The neurones used in these experiments were selected on the basis of similar size (diameter, 20-22 μm). The leakage current was 23.4 ± 5.6 pA in 0 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ (5 neurones) and its amplitude increased with increasing [Na⁺]_i (Fig. 1B). The data points were fitted under the assumption that the total leakage current, I_L, consisted of a Na⁺-independent (or unspecific) component, I_L(0), and a Na⁺-activated component, I_L(Na):

I_L(0) was set to 23 pA, according to our measurements of leakage current in 0 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution, and I_L(Na) was described by a standard isotherm:

where I_L(Na),max is the maximum Na⁺-activated current, EC₅₀ is the concentration for half-maximal current activation and n_H is the Hill coefficient. The best fit of the data points was achieved with the following parameters: I_L(Na),max, 1045 ± 56 pA; EC₅₀, 38.7 ± 1.8 mm; and n_H, 3.5. Results of the fit showed that at high Na⁺ concentrations I_L(Na) was larger than the unspecific I_L(0) component, by a factor of about 45. In order to compare the contributions of unspecific and Na⁺-activated conductances to the leakage current at low Na⁺ concentrations (0-20 mm), we plotted the amplitude of the I_L(0) component (dashed line) and the fitted curve for the I_L(Na) component (continuous line) on an enlarged scale (Fig. 1C). The functions crossed each other at a Na⁺ concentration of about 12 mm at which only 2.2% of I_L(Na),max (23 of 1045 pA) was activated. Thus, at 12 mm ${\text{Na}}_{\text{i}}^{\text{+}}$, the contribution of I_L(0) and I_L(Na) to the total leakage current became equal and at [Na⁺]_i higher than 12 mm I_L(Na) dominated I_L(0).

Figure 1. Na⁺-activated leakage currents in small DRG neurones.

A, whole-cell leakage currents activated by 150 ms hyperpolarizing voltage steps to -120 mV from a holding potential of -80 mV. Recordings were made from three different neurones using pipettes filled with 0, 20 and 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solutions (indicated near the traces). B, dependence of the amplitude of the leakage currents on [Na⁺]_i. The leakage currents were measured at the end of the hyperpolarizing pulse. Each point represents the mean leakage current (±s.e.m., where the error bar exceeds the symbol size) measured in five different neurones. It was supposed that the total leakage current, I_L, consisted of a Na⁺-independent (unspecific) component, I_L(0), and a Na⁺-activated component, I_L(Na): I_L = I_L(0)+I_L(Na). The fitting procedure is described in the text. C, I_L(0) component (dashed line) and fitted curve for the I_L(Na) component as a function of [Na⁺]_i. D, leakage currents activated by a 150 ms voltage pulse from -80 to -120 mV in Ca²⁺-free solution (Control) and after addition of external 20 mm Cs⁺ and 10 mm TEA. Recordings were obtained from the same neurone perfused with 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution. E, leakage currents recorded from two neurones using pipettes filled with 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ and 70 mm ${\text{Li}}_{\text{o}}^{+}$ solutions.

Figure 2.

Experiments with double-solution pipetteA, leakage currents activated by voltage pulses from -80 to -120 mV at different times after membrane breakthrough and establishment of the whole-cell recording mode. The tip of the double-solution pipette was filled with 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution and the rest of the pipette with 0 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution. A schematic drawing is shown at the top. B, amplitude of the leakage current as a function of time elapsed after membrane breakthrough. Data points during the first 2 min are given at a higher time resolution. Filled symbols correspond to original recordings shown in A. Kinetics of ‘recovery’ were fitted by a mono-exponential function with a time constant of 15 min. C, leakage currents recorded with a standard pipette filled with 0 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution only. D, development of the leakage current during an experiment with a standard pipette containing 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ only.

In amphibian axons, external 10 mm Cs⁺ almost completely suppressed the single K_Na channels, whereas external 10 mm TEA blocked them only partially (Koh et al. 1994). In the present study, external 20 mm Cs⁺ almost completely blocked the leakage currents which developed in the neurone after perfusion with 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution (Fig. 1D; 4 neurones). Ten millimolar external TEA had only a small effect on the amplitude of the leakage currents (Fig. 1D; 7 neurones). The Na⁺-activated leakage currents were not blocked by 1 mm 4-AP, 100 nm DTX-I and 10 nm MTX (each in 3 neurones).

It has been shown that internal Li⁺ is not able to substitute for Na⁺ in K_Na channel activation (Dryer, Fujii & Martin, 1989; Safronov & Vogel, 1996). In further experiments, we compared the whole-cell leakage currents recorded in small DRG neurones after perfusion with 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ and 70 mm ${\text{Li}}_{\text{i}}^{\text{+}}$ solutions (Fig. 1E; 3 cells). It can be clearly seen that the leakage current in 70 mm ${\text{Li}}_{\text{i}}^{\text{+}}$ solution was as small as that recorded in the neurones perfused with 0 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution (compare Fig. 1E with A).

Thus, the leakage currents recorded in small DRG neurones after perfusion with ${\text{Na}}_{\text{i}}^{\text{+}}$ solutions demonstrated properties typical of Na⁺-activated K⁺ channels. It can be concluded that most of the leakage current was carried through K_Na channels.

Experiments with double-solution pipettes

Under our experimental conditions, it was not possible to make more than one whole-cell recording from each neurone. In order to show the reversibility of the activation of the leakage current by internal Na⁺, a simple method of whole-cell recording with a pipette filled with two solutions (double-solution pipette) was used. A standard patch pipette for whole-cell recording was filled with two different solutions (Fig. 2A). The pipette was filled from the tip with 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution. The shank of the pipette was filled with 0 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution in a standard way by means of a syringe needle. During filling of the biphasic pipette it was important to ensure that the volume of the shank solution was much larger than that of the tip solution. Each double-solution pipette was used immediately after filling, since any time delay increased dilution of the Na⁺ concentration in the tip by diffusion from the shank solution. Under our experimental conditions a tight seal with the neuronal membrane was made within a minute after filling of the double-solution pipette.

At the moment at which the whole-cell configuration with a double-solution pipette was established, only very small leakage currents were observed (Fig. 2A at 0 s). Within the next 1-2 min, the amplitude of the leakage currents increased dramatically (‘development’), reaching values typical for 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ (see Fig. 1) indicating diffusion of the tip solution into the neurone. As diffusion of the 0 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ shank solution into the pipette tip reduced the [Na⁺]_i in the neurone, a slow reduction of the leakage current amplitude (‘recovery’) was observed. The time course of the recovery could be fitted by a mono-exponential with a time constant of 15 min. The leakage current measured 42 min after establishment of the whole-cell configuration was almost as small as the initial current. Similar development and recovery of the leakage currents was seen in all four neurones examined.

The development of the leakage current during 1-2 min was in good agreement with the 15 s exchange time between pipette and cell solution reported for bovine chromaffin cells with a diameter of 12 μm (Fenwick, Marty & Neher, 1982), bearing in mind that the small DRG neurones were twice the size (diameter) and therefore had an 8 times larger internal volume. The time constant of the current development can also be calculated using an equation derived for diffusion of pipette solution into the cell (Marty & Neher, 1995; two-compartment model):

where R is the pipette resistance, V is the cell volume, D is the diffusion coefficient for Na⁺ and ρ is the resistivity of the pipette solution. Assuming that R is 10 MΩ, D is 10⁻⁵ cm² s⁻¹, ρ is 50 Ω cm, and that the cell is spherical with a diameter of 22 μm, τ will be 110 s.

It should be mentioned that the maximum amplitude of the leakage current and the time course of its recovery depended strongly on pipette geometry and relative volumes of tip and shank solutions. For example, the time course of the current recovery was prolonged if narrower pipettes were used. If the amount of tip solution was too small, then the maximum leak current amplitude recorded with a double-solution pipette was only about one-third of that recorded with a normal pipette containing 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution, probably because the 0 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ shank solution diffused into the tip before the neurone was completely perfused with the 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ tip solution. Therefore, an appropriate combination of pipette geometry and filling proportions played a major role in adjusting the amplitude and the time course of the leakage currents.

For comparison, development of the leakage current recorded using a standard pipette filled with either 0 or 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$is shown in Fig. 2C and D, respectively. The currents remained very small during a period of 30 min when the pipette contained 0 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution. With 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution, the leakage current developed within 2 min and then remained constant for tens of minutes.

Single-channel experiments

Single K_Na channels were investigated in inside-out membrane patches in a small additional bath (Safronov & Vogel, 1995). The pipettes contained high-${\text{K}}_{\text{o}}^{+}$ or Ca²⁺-free^o external solutions and the bath was perfused with internal solutions containing different Na⁺ concentrations.

K_Na channels were observed in approximately two-thirds of the total of more than seventy inside-out patches obtained with 10-13 MΩ pipettes. Each active patch contained one to eight K_Na channels. The single-channel currents had large amplitudes and demonstrated typical substates (Fig. 3A). The current-voltage (i-V)curves for the main open state of the K_Na channels are shown in Fig. 3B Least-squares fitting of the data points with a linear regression under the assumption that the E_K is +15 mV in high-${\text{K}}_{\text{o}}^{+}$ solution (155 mm ${\text{K}}_{\text{o}}^{+}$-85 mm ${\text{K}}_{\text{i}}^{\text{+}}$) and -68 mV in Ca²⁺-free^o external solution (5.6 mm ${\text{K}}_{\text{o}}^{+}$-85 mm ${\text{K}}_{\text{i}}^{\text{+}}$) gave conductances of 142.5 ± 1.6 pS (4 patches) for inward currents in high-${\text{K}}_{\text{o}}^{+}$ solution and 28.3 ± 0.5 pS for outward currents in Ca²⁺-free^o external solution (4 patches). The channel conductance for inward currents in Ca²⁺-free^o solution measured in the potential range -80 to -120 mV was 23.0 ± 2.8 pS (4 patches).

Figure 3. Single K_Na channel currents.

A, single-channel currents recorded from inside-out patches using pipettes filled with high-${\text{K}}_{\text{o}}^{+}$ and Ca²⁺-free^o solutions in 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ bath solution. Holding potentials are indicated near the corresponding traces. Part of the recording is shown at a higher time resolution. Several substates are indicated by asterisks. B, i- V plots in high-${\text{K}}_{\text{o}}^{+}$ (•; 4 patches) and Ca²⁺-free^o solutions (^; 4 patches). The arrows indicate the theoretical E_K for both solutions.

The effect of different [Na⁺]_i on P_o of K_Na channels is shown in Fig. 4. The first channel activity was observed in the presence of 20 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ and it was much higher in 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution (Fig. 4A). Channel activation was not complete in 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ (Fig. 4B; 5 patches). The data were fitted with a standard isotherm:

where P_o(max) is the maximum P_o. The best fit of the data points was achieved with the following parameters: P_o(max), 0.37 ± 0.03; EC₅₀, 47.9 ± 2.3 mm; and n_H, 4.4.

Figure 4. Effect of internal [Na⁺] on K_Na channel activation.

A, recordings of single K_Na channels in 0, 20 and 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solutions. Holding potential was -60 mV. Inside-out patch; high-${\text{K}}_{\text{o}}^{+}$ pipette solution. B, open probability (P_o) as a function of [${\text{Na}}_{\text{i}}^{\text{+}}$. P_o was calculated from 1-2 min recordings at a membrane potential of -60 mV. Each point represents the mean (±s.e.m., where the error bar exceeds the symbol size) of measurements from five inside-out patches. The data points were fitted with the equation: P_o = P_o(max)/(1 + (EC₅₀/[Na⁺]_i)^nH), where P_o(max) (maximum P_o) is 0.37 ± 0.03, EC₅₀ is 47.9 ± 2.3 mm and n_H (Hill coefficient) is 4.4.

In order to determine the P_o value in intact neurones at resting potential, K_Na channel activity in cell-attached patches was recorded. The pipettes filled with high-${\text{K}}_{\text{o}}^{+}$-TEA solution had a resistance of 3-9 MΩ and were held at 0 mV with respect to the bath solution. When the cell-attached recordings had been obtained, the patches were excised and transferred into an additional bath filled with 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution, where the maximum number of active K_Na channels was determined. The Na⁺ dependence of the currents was confirmed by their disappearance in 0 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution (3 patches). In twenty-eight experiments the channels were silent in cell-attached patches but became active after patch excision (Fig. 5; top trace). In another fourteen cell-attached patches K_Na channels were active with P_o values ranging from 0.0003 to 0.43 (Fig. 5; middle and bottom traces). Even in one neurone the P_o values varied considerably from patch to patch from 0 to 0.1. The mean P₀ of K_Na channels in forty-two cell-attached patches was 0.019 ± 0.010.

Figure 5. K_Na channel activity in intact neurones.

Cell-attached recordings of channel activity in three different neurones. Recordings from three membrane patches before (cell attached; left) and after their excision (inside out; right). Pipette potential was 0 mV during cell-attached recording and -80 mV in inside-out mode. The pipettes contained high-${\text{K}}_{\text{o}}^{+}$-TEA solution. For recording in the inside-out mode the patches were transferred to a bath containing 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution.

Two-electrode experiments

The purpose of the following experiments was to show a direct correlation between the appearance of Na⁺-activated leakage current and K_Na channel activation. With two electrodes, we simultaneously recorded the changes in leakage currents (whole-cell electrode) and in single K_Na channel activity (cell-attached electrode). The electrode for whole-cell recording was filled with 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution. The neurone was held at -80 mV and the leakage currents activated by 60 ms pulses to -120 mV were monitored each second. The electrode for cell-attached recording contained high-${\text{K}}_{\text{o}}^{+}$-TEA solution and was voltage clamped at 0 mV with respect to the bath solution. Directly after membrane breakthrough (Fig. 6; t = 0 s) the leakage currents were very small and no single K_Na channel activity was seen in the cell-attached patch. Approximately 20 s later (t = 24 s), the amplitude of the leakage current had increased and the first activity of K_Na channels was recorded. Further increases in the amplitude of the leakage current (t = 44 and 96 s) were accompanied by corresponding increases in single K_Na channel activity. The number of active K_Na channels seen in this cell-attached patch during neurone perfusion with 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution was two. The maximum number of K_Na channels in this patch was determined at the end of the experiment when the pipette used for the cell-attached recording was pulled away and an inside-out patch was formed in the main chamber containing Ca²⁺-free solution (141 mm Na⁺). In the excised patch, activity of the same two K_Na channels was observed. A comparison of the channel activity in cell-attached patches during cell perfusion with 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution with that in inside-out patches allowed us to conclude that K_Na channel sensitivity to ${\text{Na}}_{\text{i}}^{\text{+}}$ was not changed after patch excision.

Figure 6. Simultaneous recordings of leakage and single K_Na channel currents.

Simultaneous recordings of Na⁺-activated leakage current (top traces) and single K_Na channel current (bottom traces) with two electrodes. The electrode used in whole-cell mode contained 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution. Holding potential was -80 mV and the leakage currents were activated each second by 60 ms pulses to -120 mV. The time elapsed (t) after membrane breakthrough is indicated above the traces. The resistance of the whole-cell pipette was 6 MΩ. The pipette used for cell-attached recordings was filled with high-${\text{K}}_{\text{o}}^{+}$-TEA solution and had a resistance of 5 MΩ. The potential in the pipette during cell-attached recording was 0 mV with respect to the bath solution.

Similar correlation between leakage currents and K_Na channel activity was recorded in a total of seven experiments with two electrodes.

Current-clamp experiments

In current-clamp experiments we first compared the shapes of single action potentials recorded in external Ca²⁺-free and 141 mm ${\text{Li}}_{\text{o}}^{+}$ solutions (Fig. 7), since it is known that Li⁺ ions readily permeate Na⁺ channels (Hille, 1972) but are not able to activate K_Na channels (see Fig. 1E). The membrane potential in both solutions was kept at -70 mV by an injection of steady-state current through the recording pipette. The replacement of external Na⁺ with Li⁺ did not change the shape of the action potential (Fig. 7, superimposed traces; 15 neurones) leading to the conclusion that K_Na channels do not contribute to the single action potential in small DRG neurones.

Figure 7. Influence of external Li⁺ on the shape of the action potential.

Action potentials evoked by short (12.5 ms) current pulses of 240 pA in external Ca²⁺-free and 141 mm ${\text{Li}}_{\text{o}}^{+}$ solutions. Both action potentials are shown superimposed on the right. The membrane potential was kept at -70 mV in both solutions by injection of inward or outward steady-state currents. The pipette contained high-${\text{K}}_{\text{i}}^{\text{+}}$ solution.

It has been shown for rat motoneurones that activation of K_Na channels by local Na⁺ accumulation during trains of action potentials could result in a slow after-hyperpolarization (Safronov & Vogel, 1996). In contrast to motoneurones, the small DRG neurones studied here were not able to generate action potentials at high frequencies. They usually responded to sustained depolarizing current injection with a single spike only and, therefore, accumulation of intracellular Na⁺ during cell firing seemed to be unlikely.

In further experiments the effect of ${\text{Na}}_{\text{i}}^{\text{+}}$ on the membrane resting potential was studied (Fig. 8A). The slice was superfused with Ca²⁺-free (5.6 mm ${\text{K}}_{\text{o}}^{+}$) external solution and the pipettes were filled with 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ (85 mm ${\text{K}}_{\text{i}}^{\text{+}}$) solution. At the beginning of the experiment, the voltage-clamp mode was established and the original leakage currents in the neurone were recorded. Thereafter, the amplifier was switched to current-clamp mode and resting membrane potentials were measured. In a neurone with a small original unspecific leakage current and resting potential of about -80 mV (Fig. 8A; ^), diffusion of 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ pipette solution into the neurone produced an increase in leakage current accompanied by a membrane depolarization. This depolarization saturated at -68 mV which corresponds to the E_K under these experimental conditions (5.6 mm ${\text{K}}_{\text{o}}^{+}$-85 mm ${\text{K}}_{\text{i}}^{\text{+}}$). Another neurone with a larger original unspecific leakage current and a relatively low resting potential of -50 mV (Fig. 8A; •) demonstrated an increase in the leakage current and membrane hyperpolarization towards -68 mV as the pipette solution diffused into the cell. In a total of five neurones, regardless of their original resting potential, a shift of the membrane potential in the direction of the new E_K during cell perfusion with internal Na⁺ was observed. These experiments indicated that perfusion of the neurone with ${\text{Na}}_{\text{i}}^{\text{+}}$ can activate an additional K⁺ conductance and stabilize the resting potential in small DRG neurones, moving it towards E_K.

Figure 8. Effect of internal Na⁺ on resting potential in small DRG neurones.

A, resting potential in two different neurones shown as a function of time elapsed after membrane breakthrough. One neurone had a small original leakage current and a resting potential of -82 mV (^). Another neurone with a large original unspecific leakage current had a depolarized resting potential of -50 mV (•). The bath contained Ca²⁺-free solution and the pipette solution was 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$. The theoretical E_K for external Ca²⁺-free-internal 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solutions (5.6 mm ${\text{K}}_{\text{o}}^{+}$-85 mm ${\text{K}}_{\text{i}}^{\text{+}}$) of -68 mV is indicated by the dashed line. The leakage currents recorded immediately after membrane breakthrough and several minutes after their saturation are shown in the insets. B, resting potential as a function of external K⁺ concentration measured in neurones perfused with 0 (•), 10 (▿) and 30 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ (^) solutions. The straight line corresponds to the Nernst equation at 25 °C: 25.7 ln([K⁺]_o/[K⁺]_i), where the value given (25.7) is in mV and [K⁺]_i is 85 mm.

The effect of external [K⁺] on the resting potential in small DRG neurones perfused with 0, 10 and 30 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ pipette solutions is shown in Fig. 8B. The resting potential in neurones perfused with 0 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ (•) deviated from the values predicted from the Nernst equation for K⁺ ions (continuous line) at potentials negative to -60 mV where most of the delayed rectifier K⁺ channels are closed. After perfusion of the neurone with 10 (▿) and 30 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ (^) solutions, a good correlation between measured and predicted potentials over a broader voltage range was observed (23 neurones).

DISCUSSION

The small DRG neurones with diameters of 20-25 μm studied in this work could be sensory neurones of types Aδ and C (Harper & Lawson, 1985a, 1985b), which convey information from pain and thermal receptors. According to the shape of their action potentials, these neurones were most probably C-type (Safronov et al. 1996). The high input resistance of small DRG neurones in the slice preparation (up to several gigaohms; Safronov et al. 1996) allowed us to use them as an appropriate model for investigating the macroscopic Na⁺-activated K⁺ currents.

Na⁺-activated K⁺ conductance in small DRG neurones

Up until now, studies of K_Na channels were mainly focused on their single-channel properties (Dryer, 1991; Egan et al. 1992a, 1992b; Dryer, 1994; Koh et al. 1994; Safronov & Vogel, 1996) and little was known about the macroscopic K_Na channel conductance in the neuronal membrane. The lack of knowledge about the macroscopic K_Na channel conductance resulted from difficulties in recording a ‘pure’ whole-cell K_Na channel current. Whole-cell K_Na channel current was usually identified as a transient K⁺ current activated by Na⁺ influx through voltage-gated Na⁺ channels during membrane depolarization (Bader, Bernheim & Bertrand, 1985; Hartung, 1985; Dryer et al. 1989). Unfortunately, the currents recorded in such a way could result from insufficient voltage clamp of the neuronal membrane (Dryer, 1991). Furthermore, such currents could represent only a portion, perhaps a small one, of the total K_Na channel conductance. Estimation of the total K_Na channel conductance in neurones, however, is important for understanding K_Na channel contribution to membrane resting conductance and therefore to the resting potential. In the present paper we have used another approach to study the whole-cell K_Na channel conductance.

Na⁺-activated K⁺ whole-cell leakage currents recorded in the present study appeared only after cell perfusion with Na⁺-containing internal solutions. The dependence of the amplitude of the leakage currents on [Na⁺]_i (EC₅₀ = 38.7 mm and n_H = 3.5) was similar to that observed for P_o of single K_Na channels (EC₅₀ = 47.9 mm and n_H = 4.4). The leakage currents were not activated by 70 mm ${\text{Li}}_{\text{i}}^{\text{+}}$. They were almost completely blocked by externally applied 20 mm Cs⁺ but were only slightly suppressed in the presence of 10 mm TEA. Our experiments with double-solution pipettes showed a reversibility of leakage current activation by internal Na⁺. Furthermore, in two-electrode experiments a direct correlation between the amplitude of the leakage current and the activity of single K_Na channels was observed. All these results allowed us to conclude that the predominant part of the leakage current recorded in the present study was carried through Na⁺-activated K⁺ channels.

Single K_Na channels in small DRG neurones with a conductance of 142 pS in 155 mm ${\text{K}}_{\text{o}}^{+}$-85 mm ${\text{K}}_{\text{i}}^{\text{+}}$ solutions and with their sensitivity to internal [Na⁺] (EC₅₀ = 47.9 mm) resemble those described in several other neuronal preparations (Dryer et al. 1989; Egan et al. 1992a, 1992b; Koh et al. 1994; Safronov & Vogel, 1996). As observed for K_Na channels in chick brainstem neurones (Dryer, 1993) and rat motoneurones (Safronov & Vogel, 1996), the channels studied here did not show a pronounced ‘run-down’ in excised patches and their activity was not significantly reduced during tens of minutes.

The present experiments allowed us to estimate the total number and the density of K_Na channels in intact small DRG neurones. The mean leakage current activated by a hyperpolarizing voltage step of 40 mV in the presence of 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ was 0.96 nA (Fig. 1B), giving a mean total K_Na conductance of 24 nS. The unitary conductance of K_Na channels measured for inward currents in external Ca²⁺-free-85 mm ${\text{K}}_{\text{i}}^{\text{+}}$ solutions was 23 pS (Fig. 3B) and the P_o value at 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ was 0.31 (Fig. 4B). Thus, the total number of K_Na channels in a small DRG neurone could be estimated as 3366. Assuming that the soma is a sphere with a diameter of 22 μm, one can calculate a mean channel density of 2.2 μm⁻².

In contrast to those in amphibian axons (Koh et al. 1994) and somata of rat motoneurones (Safronov & Vogel, 1996), single K_Na channels were found to be evenly distributed at high density in the somatic membrane of small DRG neurones. In thirteen experiments with inside-out patches containing active K_Na channels, the maximum number of channels observed was 3.8 ± 0.6 per patch and the mean pipette resistance was 10.7 ± 0.3 MΩ, corresponding to a membrane patch area of 1.4 μm² (Sakmann & Neher, 1995). Taking into account that active K_Na channels were found in approximately two-thirds of all inside-out patches, the estimated channel density is 1.8 μm⁻². This value is very similar to that obtained from experiments with macroscopic leakage currents.

Inactivating A and delayed rectifier (DR_F, DR₁, DR₂ and DR₃) K⁺ channels have been described recently in small DRG neurones in a slice preparation (Safronov et al. 1996). It is interesting to note that the K_Na channel reported in the present study had the highest frequency of appearance in inside-out patches. To compare the whole-cell conductances of different K⁺ channels, we re-evaluated our recordings of voltage-gated K⁺ currents in small DRG neurones (Safronov et al. 1996). Inactivating and non-inactivating components of K⁺ currents had been separated using a standard voltage protocol with different prepulses. The total delayed rectifier conductance corresponding to DR₁-, DR₂-, DR₃- and, in part, to DR_F-channels was 14.2 ± 3.0 nS (8 neurones). The total inactivating conductance representing A- and DR_F-channels was 5.4 ± 1.8 nS (8 neurones). Therefore, the maximum K_Na conductance of 24 nS measured in 70 mm ${\text{Na}}_{\text{i}}^{\text{+}}$ solution appears to be the largest K⁺ conductance observed in small DRG neurones studied so far.

A possible function of K_Na channels in small DRG neurones

The functions of K_Na channels in the neuronal membrane are not well understood at present. Numerical solution of the diffusion equation indicated that Na⁺ influx through voltage-gated Na⁺ channels cannot produce an increase in local Na⁺ concentration sufficient for activation of K_Na channels even if the two types of channel are co-localized in the membrane (Dryer, 1991). It was also shown that K_Na channels do not contribute to the shape of a single action potential in rat motoneurones (Safronov & Vogel, 1996). Therefore, the functions of K_Na channels in the neuronal membrane have recently been considered in terms of local Na⁺ accumulation in restricted membrane compartments during repetitive firing (Koh et al. 1994; Safronov & Vogel, 1996). These functions can depend on, or can even be completely determined by, the channel distribution as well as the cell morphology and firing ability. In amphibian axons, K_Na channels were shown to be co-localized with voltage-activated Na⁺ channels in the narrow nodal region (Koh et al. 1994). It was assumed that K_Na channels could be activated due to cytoplasmic Na⁺ accumulation during high-frequency trains of action potentials. In the somata of rat motoneurones, K_Na channels were found at increased densities in the vicinity of the cell processes, in the regions of most probable Na⁺ accumulation (Safronov & Vogel, 1996). Firing of action potentials even at a moderate frequency of about 30 Hz could produce a local Na⁺ accumulation leading to K_Na channel activation resulting in a pronounced slow after-hyperpolarization. In contrast, small DRG neurones are known to generate action potentials only at very low frequencies of about 0.5-2 Hz (C-type neurones; Harper & Lawson, 1985b). Furthermore, the somata of DRG neurones do not possess an anatomical structure such as the node of Ranvier or the axon hillock with a high density of Na⁺ channels and large surface-to-volume ratio which could facilitate Na⁺ accumulation. Therefore, a local Na⁺ accumulation during repetitive firing seemed to be unlikely and functions of K_Na channels may have no links to firing behaviour in small DRG neurones.

However, in the present study we show that small DRG neurones which are normally characterized by a high input resistance of several gigaohms (Safronov et al. 1996) have a large K_Na channel conductance. The ratio of unspecific leakage current to the maximum Na⁺-activated K⁺ current was 1 to about 45 (23 pA to 1045 pA). The physiological level of intracellular Na⁺ measured in different neurones was 4-16 mm (Grafe, Rimpel, Reddy & ten-Bruggencate, 1982; Galvan, Dörge, Beck & Rick, 1984). Our experiments with whole-cell K_Na channel currents indicated that, at an [Na⁺]_i of about 10 mm, at which only 2.2% of K_Na channel current is activated, the current can essentially contribute to membrane conductance. By normalizing the P_o of 0.019 measured in cell-attached patches to the P_o(max) of 0.37 obtained by fitting the data from inside-out patches (Fig. 4B), one can estimate that K_Na channels in intact neurones show up to 5% of their maximum P_o. Furthermore, cell perfusion with internal Na⁺ can stabilize the resting potential and can shift it in the direction of the E_K. Therefore, it could be suggested that K_Na channels in small DRG neurones participate in setting the resting potential under physiological conditions as well as contributing to its stabilization under pathophysiological conditions, like hypoxia, leading to a reduction in Na⁺-K⁺ pump activity and therefore to an increase in cytoplasmic Na⁺ concentration.