Distinct mechanisms for activation of Cl⁻ and K⁺ currents by Ca²⁺ from different sources in mouse sympathetic neurones

Abstract

1. We have investigated the roles of different voltage-dependent Ca²⁺ channels in the activation of the Cl⁻ and K⁺ channels responsible for the afterdepolarization (ADP) and slow afterhyperpolarization (AHP) in sympathetic neurones of the isolated mouse superior cervical ganglion in vitro.

2. The ADP and its associated Ca²⁺-activated Cl⁻ current were markedly decreased by ω-agatoxin IVA (40–200 nm) and nifedipine (1–10 μm), but not by ω-conotoxin GVIA (300 nm).

3. In contrast, the AHP and the apamin-sensitive Ca²⁺-activated K⁺ current that underlies this potential were blocked by ω-conotoxin GVIA, but were not affected by ω-agatoxin IVA and were only slightly reduced by nifedipine.

4. Ryanodine (20 μm) reduced the Ca²⁺-activated Cl⁻ current following an action potential by 75 % but on average did not affect the Ca²⁺-activated K⁺ current.

5. Evidence that R-type channels provide a proportion of the Ca²⁺ activating both types of Ca²⁺-dependent channel was obtained.

6. We conclude that Ca²⁺ entering through L- and P-type Ca²⁺ channels preferentially activates the Cl⁻ current responsible for the ADP in mouse sympathetic neurones, predominantly via Ca²⁺-induced Ca²⁺ release, whereas the Ca²⁺ that activates the K⁺ channels responsible for the AHP enters predominantly through N-type channels. The data can be explained by the selective association of each type of Ca²⁺ channel with particular intracellular mechanisms for activating other membrane channels, one indirect and the other direct, probably located at discrete sites on the soma and dendrites.

In mammalian sympathetic ganglion cells, Ca²⁺ entry during action potential firing activates K⁺ and Cl⁻ conductances which contribute to slow afterpotentials (McAfee & Yarowsky, 1979; Sacchi et al. 1995; De Castro et al. 1997). In rat superior cervical ganglion (SCG) cells, which exhibit a long afterhyperpolarization (AHP), Ca²⁺ entry via N-type channels activates apamin-sensitive Ca²⁺-activated K⁺ channels (SK-type), which are responsible for most of the underlying conductance change (Davies et al. 1996). This conductance has been named g_K,Ca1 (Cassell & McLachlan, 1987a), to distinguish it from the slower Ca²⁺-dependent conductance g_K,Ca2, which underlies a prolonged AHP in one subclass of sympathetic neurone, and from I_AHP, which differs in kinetic and pharmacological characteristics between several types of central neurone (Sah, 1996). On the other hand, in rat SCG cells, large conductance Ca²⁺-activated K⁺ channels (BK-type) are in part activated by Ca²⁺ entering through L-type channels, which has no effect on g_K,Ca1 (Davies et al. 1996). These observations suggest that Ca²⁺ entering through a specific channel type during the action potential (AP) is directed predominantly to a particular type of K⁺ channel. Recently, N-type and BK-type channels in some cases, and L-type and SK-type channels in others, have been identified in single patches isolated from hippocampal somata, providing direct evidence for colocalization of pairs of functionally linked channels (Marrion & Tavalin, 1998). While the relative locations of Ca²⁺ and K⁺ channels must be crucial in determining the electrical behaviour of neurones, these data imply that particular associations between channels are not necessarily the same for all neurone types. Data from sympathetic neurones have confirmed that, even for this one type of neurone, there are species differences and differences between phenotypically distinct classes of neurone in the way Ca²⁺ entry is linked to activation of BK and SK channels (Ireland et al. 1998; Davies et al. 1999).

In mouse SCG cells, Ca²⁺ entry during the AP activates, in addition to g_K,Ca1, a Cl⁻ conductance (g_Cl,Ca) that produces an afterdepolarization (ADP; De Castro et al. 1997). When the cell is clamped at −55 mV, this is recorded as an inward current. About 45 % of the high-voltage-activated Ca²⁺ current evoked by a step depolarization in dissociated mouse SCG somata flows through P/Q-type channels, 40 % through N-type channels, 10 % through L-type channels and a small proportion through channels resistant to known selective organic antagonists (here called R-type channels; Namkung et al. 1998). The question then arises as to how Ca²⁺ entering through these diverse types of voltage-dependent Ca²⁺ channel during an AP activates the Cl⁻ and K⁺ conductances.

In this study, we have used pharmacological antagonists to distinguish the origin of the Ca²⁺ responsible for activation of Ca²⁺-dependent afterpotentials in mouse sympathetic neurones. The experiments have utilized recordings from intact mouse SCG neurones, as the distribution of channels on the soma and dendrites is not uniform (e.g. Westenbroek et al. 1998) so that responses from dissociated somata will not necessarily reflect physiological events. Our data indicate that, as in rat SCG cells, the K⁺ channels responsible for the slow AHP are activated mainly by Ca²⁺ entering through N-type channels, whereas Cl⁻ channel activation results from Ca²⁺ entering through L- and P-type Ca²⁺ channels. Ca²⁺-induced Ca²⁺ release from intracellular stores is a major contributor to Cl⁻ channel activation but participates in K⁺ channel activation in only a subgroup of cells. Some of the results have been presented in abstract form (Martínez-Pinna et al. 1998).

Methods

The methods used for intracellular and voltage-clamp recording have been described previously (De Castro et al. 1997). Briefly, 5- to 7-week-old male mice (Swiss OF-1) were deeply anaesthetized by an i.p. injection of sodium pentobarbitone (40 mg kg⁻¹) and perfused through the heart with cold oxygenated saline. The superior cervical ganglion was taken from the animal and pinned to the Sylgard (Dow-Corning, Midland, MI, USA) bottom of a chamber continuously perfused with saline (mM: NaCl, 128; KCl, 5; CaCl₂, 2; NaH₂PO₄, 1; NaHCO₃, 16; MgSO₄, 1.2; and glucose, 5.5) equilibrated with 95 % O₂-5 % CO₂ (pH 7.4) at room temperature (22–25°C). The ganglion was illuminated from one side by a fine fibre optic light source, allowing the cells on the surface to be seen using a microscope (× 375–600). The experimental procedures were in accordance with the guidelines of the ethics committee of the Instituto de Neurociencias de la Universidad Miguel Hernández-CSIC.

Potentials were measured with respect to a Ag-AgCl pellet submerged in the outflow section of the bath. Cells were impaled with microelectrodes filled with 3 M KCl (50–90 MΩ) and data were collected only if the neurone generated APs of at least 70 mV in amplitude at the resting potential. The sampling frequency for discontinuous single-electrode current and voltage clamp was 2–6 kHz, with a duty cycle of 30/70 (Axoclamp 2B, Axon Instruments Inc.). Capacitance compensation was continuously monitored and adjusted to ensure head stage settling. Data were digitized and stored on a computer for subsequent analysis using commercial software (pCLAMP6, Axon Instruments Inc.). After impaling a cell, we waited for the membrane potential to stabilize and then measured the basic electrical properties at the resting membrane potential, including the amplitude and duration of the afterpotentials. Action potential half-width was measured as the duration at half the peak amplitude. Input resistance was calculated from the linear part of the I-V relationship determined near the end of 200 ms hyperpolarizing current pulses. APs were initiated after a 5 ms depolarizing current pulse or by a train, usually of six pulses at 40 Hz, and the ensuing afterpotentials were examined. To test the effect of the different drugs, we adjusted the membrane potential to −55 mV by passing current in discontinuous current-clamp mode before making measurements of APs and tail currents. Generally, we averaged five sweeps when a single AP was evoked and three when recording a train of discharges. To enable the degree of block to be quantified, cells were selected with relatively large outward tail currents (measures of g_K,Ca1) or inward tail currents (measures of g_Cl,Ca) so that their responses are not necessarily representative of those of all SCG neurones.

Tail currents underlying the afterpotentials were measured using a hybrid clamp protocol in which APs were evoked with depolarizing current pulses and the amplifier was switched to voltage clamp mode, either after repolarization of a single AP, or 10 ms after the end of a brief train at 40 Hz. Tail currents were recorded under voltage clamp at −55 mV after filtering at 0.3 kHz and were analysed using commercial software (Origin, Microcal Software Inc., Northampton, MA, USA). After digital smoothing using a 9-point running average, the peak values were measured and the currents integrated from 20 ms after the peak of the AP in order to derive the amount of charge transferred. To test the effects of increasing Ca²⁺ loads on the current underlying the ADP, we could not use trains of > 10 APs because large ADPs were generated that reduced AP amplitude by inactivating Na⁺ currents. Therefore, for these experiments, the cells were voltage clamped in the presence of 1 μm TTX and 100 nm apamin and trains of 5 ms pulses (to 0 mV from a holding potential of −55 mV) of varying duration at 40 Hz were applied. All values are expressed as means ±s.e.m. Statistical comparisons were made using Student's paired t test (Sigmastat, Jandel Scientific Software, Erkrath, Germany).

All drugs were obtained from Sigma, except nifedipine (RBI, Natick, MA, USA) and ω-agatoxin IVA (Peptides International, KY, USA), and were applied dissolved in the perfusing solution. Cytochrome c (0.1 mg ml⁻¹) was added to the toxin stock solutions to saturate non-specific peptide binding to the apparatus. Care was taken when using nifedipine to protect it from breakdown by light. The antagonists were used at concentrations that are relatively selective for different channel types. ω-Conotoxin GVIA blocks N-type channels quite selectively with an IC₅₀ of 1–40 nm (Regan et al. 1991; Boland et al. 1994). ω-Agatoxin IVA blocks both P- and Q-type channels but with very different IC₅₀ values of 1–20 nm and ∼100 nm, respectively (Mintz et al. 1992; Randall & Tsien, 1995). Dihydropyridines block neuronal L-type channels with an IC₅₀ of 0.3–2 μm (Jones & Jacobs, 1990; Regan et al. 1991). We usually applied a relatively high concentration of 10 μm, which is commonly used in intact neurones to ensure full blockade. This concentration has no non-specific actions in rat SCG cells (Davies et al. 1996). It should be noted, however, that 1 and 10 μm nifedipine had similar effects in the present experiments. Application of the drugs or toxins had significant effects on the input resistance, time constant, the holding current required to maintain the resting potential at −55 mV or the amplitude and duration of the AP only in the cases that are noted specifically.

Results

Properties of SCG neurones of the mouse

The electrical properties of 71 mouse SCG neurones measured in control solution were: resting membrane potential, −51 ± 1.0 mV; AP amplitude, 99 ± 1.2 mV; AP half-width, 2.4 ± 0.1 ms; input resistance, 54 ± 3.1 MΩ; capacitance, 171 ± 10 pF; and time constant, 8.6 ± 0.6 ms.

Depolarizing and hyperpolarizing afterpotentials following AP firing varied greatly between neurones. In all cells, the fast AHP after repolarization of the AP was followed by a slow AHP (Fig. 1B). The slow AHP was separated from the fast AHP by a depolarizing inflection in 22 % of the cells (Fig. 1B, arrow) including those cells (6 %) in which the peak of this inflection was positive to the resting membrane potential, i.e. an ADP (Fig. 5Aa; see also De Castro et al. 1997). In all but six of the remaining cells, the transition from fast to slow AHP was marked by a change in the rate of depolarization (Fig. 2B a, arrow). A train of six APs at 40 Hz generated an ADP followed by a slow AHP in 46 % of cells (Fig. 1C; see also De Castro et al. 1997). In another 31 % of cells, the membrane potential at the end of the train was more positive than the initial fast AHP (Fig. 3Ab), indicating that a depolarizing conductance had been activated during the train. In the remaining cells, the train generated a slow AHP that built up from the fast AHP. Thus, the relative magnitude of the conductances underlying the AHP and ADP differed between cells.

Figure 1. Ca²⁺-dependent afterpotentials in mouse sympathetic ganglion cells.

Records from a neurone in the superior cervical ganglion. A, addition of 500 μm Cd²⁺ reduced action potential (AP) half-width without affecting its amplitude. B, addition of Cd²⁺ abolished the depolarizing inflection following AP repolarization (arrow) and markedly reduced the afterhyperpolarization (AHP) following a single AP. C, addition of Cd²⁺ abolished the afterdepolarization (ADP) and reduced the AHP after a train of six APs at 40 Hz. APs in B and C are truncated (as in subsequent figures) and traces in control and test solutions have been superimposed with labels placed closest to the relevant trace. In these, and in the current-clamp records in subsequent figures, the dashed line indicates the baseline membrane potential, which was maintained close to −55 mV by passing current through the microelectrode.

Figure 5. L-type Ca²⁺ channels contribute to activation of the ADP but have little effect on the AHP.

A, addition of 10 μm nifedipine abolished the ADP revealing an AHP after one AP (a) and reduced the ADP without affecting AHP amplitude after a train of 13 APs (b). B, averaged inward currents from nine cells in apamin were reduced when nifedipine was added. C, averaged outward currents from 11 cells in the presence of 9AC showed, when nifedipine was added, a small decrease after a single AP (a) and a change in the time course after a train of APs (b).

Figure 2. Apamin and 9AC reveal the Cl⁻ and K⁺ currents underlying afterpotentials.

Responses following a single AP are shown on the left (a) and after a train of six APs at 40 Hz on the right (b) in this and subsequent figures. A, in one cell, addition of apamin (100 nm in this and subsequent figures) abolished the slow AHP, revealing an ADP after one AP (a) and enhancing the ADP after a train of six APs at 40 Hz (b). In the same neurone, outward tail currents following a single AP (c) or a train of APs (d) were blocked by apamin, revealing inward tail currents. B, in another cell, addition of 9AC (2 mm in this and subsequent figures) reduced the depolarizing inflexion in the AHP after a single AP (a) and the small ADP following a train of APs (b), enhancing the AHP. Arrow points to the transition from fast to slow AHP. In the same neurone, 9AC blocked the inward currents after a single AP (c) and after a train of APs (d), enhancing the outward tail currents. In these, and in voltage-clamp records in the following figures, the dashed line indicates the holding current needed to keep the membrane potential at −55 mV. In this, and in the following figures except Fig. 5Ab, the trains consisted of six APs at 40 Hz.

Figure 3. N-type Ca²⁺ channels contribute to activation of the AHP but not the ADP.

A, in one cell, addition of 300 nmωCgtx GVIA reduced the AHP, therefore increasing the ADP. B, averaged inward currents from six cells in the presence of apamin showed little variation when ωCgtx GVIA was added. C, averaged outward currents from five cells in the presence of 9AC were reduced when ωCgtx GVIA was added.

Afterpotentials depend on Ca²⁺ entry

As in other mammalian SCG cells (Adams & Harper, 1995), blockade of Ca²⁺ entry by addition of Cd²⁺ (100–500 μm) to the perfusate (n = 9) abolished both of the slower afterpotentials (when present) after a single AP (Fig. 1B) or a train of APs (Fig. 1C). Cd²⁺ also produced a 32 % decrease in AP half-width (Fig. 1A; control, 2.2 ± 0.20 ms; Cd²⁺, 1.5 ± 0.10 ms; P < 0.01).

Blockade of SK channels with apamin reveals g_Cl,Ca

As in the rat SCG (Kawai & Watanabe, 1986), apamin (100 nm) largely abolished the slow AHP (Fig. 2A). In 52 % of 37 cells, addition of apamin to the control solution revealed an ADP after a single AP (Fig. 2Aa) and the same occurred after a train of APs in all but one cell, including those neurones in which only an AHP was detected in control solution. Apamin increased the amplitude and duration of ADPs after a train (Fig. 2Ab), generating repetitive discharges in 11 cells. In contrast, very rarely one or two APs occurred during a large ADP in control solution. These data indicate that, in most cells that lacked an ADP in control solution, activation of the Cl⁻ conductance after a single discharge was obscured by the K⁺ conductance underlying the AHP.

When the cells were voltage clamped in control solution, the currents recorded at −55 mV after an AP had different shapes due to the variable amounts of inward Cl⁻ and outward K⁺ conductances. There was usually an early inward inflection or plateau in the outward current after a single AP (Fig. 2Ac and Bc) but, after a train, a small net inward current of variable amplitude was generated and was followed by an outward current (Fig. 2Ad and Bd). In the presence of apamin (100 nm), the outward tail currents were largely blocked, revealing a clear inward current after one AP (Fig. 2Ac) and increasing markedly the inward current that followed a train (Fig. 2Ad). The enhancement of the inward current after a train resulted in as much as 40 times the total charge transfer as that following a single AP (Table 1).

Table 1.

Effect of ωCgtx GVIA, ωAga IVA, nifedipine and ryanodine on g_Cl,Ca in the presence of apamin

	Single action potential		Action potential train
	Current peak (nA)	Total charge (pC)	Current peak (nA)	Total charge (pC)
Apamin (n = 6)	−0.08 ± 0.03	−6.7 ± 3.7	−1.7 ± 0.4	−279 ± 69
Apamin +ωCgtx GVIA	−0.09 ± 0.04	−8.6 ± 5.1	−1.5 ± 0.4	−255 ± 80
Apamin (n = 10)	−0.11 ± 0.01	−8.4 ± 0.9	−2.0 ± 0.3	−334 ± 58
ZApamin +ωAga IVA	−0.06 ± 0.01 **	−3.5 ± 0.9 **	−1.2 ± 0.2 **	−165 ± 36 **
Apamin (n = 9)	−0.22 ± 0.06	−20.9 ± 7.6	−3.0 ± 0.6	−490 ± 119
Apamin + nifedipine	−0.09 ± 0.03 **	−6.6 ± 2.6 *	−1.5 ± 0.4 **	−184 ± 53 **
Apamin (n = 5)	−0.24 ± 0.05	−30.9 ± 7.9	−2.8 ± 0.9	−481 ± 158
Apamin + ryanodine	−0.09 ± 0.05 *	−7.6 ± 4.4 *	−2.2 ± 0.8 *	−319 ± 131 *

Tail currents after a single action potential or after a train of six action potentials at 40 Hz were recorded in voltage clamp at a holding potential of −55 mV. Inward Ca²⁺activated Cl⁻ currents were isolated by blocking g_K,Ca1 with 100 nm apamin. Changes after adding nifedipine (10 μm), ωCgtx GVIA (300 nm),ωAga IVA (40 nm) or ryanodine (20 μm) were compared using Student's paired t test

^*P < 0.05

^**P < 0.01.

Thus application of 100 nm apamin reveals the inward Cl⁻ current, providing a measure of the Ca²⁺-activated Cl⁻ conductance, g_Cl,Ca (see Introduction). In all the voltage-clamp experiments on g_Cl,Ca described below, 100 nm apamin was present throughout.

In 14 cells in which electrical properties were measured in control solution, apamin decreased the holding current needed to keep the resting membrane potential at −55 mV (from −0.43 ± 0.10 to −0.29 ± 0.09 nA, P < 0.05) and also increased input resistance (from 43 ± 4 to 60 ± 8 MΩ, P < 0.05). Apamin also increased AP amplitude (from 94 ± 2 to 101 ± 2 mV, P < 0.01) but reduced AP duration (from 2.6 ± 0.16 to 2.3 ± 0.13 ms, P < 0.01). These results suggest that apamin blocks a K⁺ conductance that is active at −55 mV. Such a conductance was not detected in rat (Davies et al. 1996) or guinea-pig (Ireland et al. 1998) sympathetic neurones.

Blockade of Cl⁻ channels with 9AC reveals g_K,Ca1

As reported previously (De Castro et al. 1997), addition of the Cl⁻ channel blocker anthracene-9-carboxylic acid (9AC; 2 mm, n = 21) substantially reduced the amplitude of the ADP, increasing the amplitude and duration of the AHP (Fig. 2Ba and b) and the outward current carried by g_K,Ca1 (Fig. 2Bc and d). In general, the degree of block produced by 9AC was more marked after a single AP than after a train. The enhancement of the outward current after the train resulted in about 4 times as much total charge transfer as that following a single AP (Table 2).

Table 2.

Effects of ωCgtx GVIA, ωAga IVA, nifedipine and ryanodine on g_K,Ca1 in the presence of 9AC

	Single action potential		Action potential train
	Current peak (nA)	Total charge(pC)	Current peak(nA)	Total charge(pC)
9AC (n = 5)	0.32 ± 0.06	41.9 ± 12.4	0.72 ± 0.08	190 ± 31
9AC +ωCgtx GVIA	0.14 ± 0.03 **	12.1 ± 4.2 *	0.46 ± 0.08 *	111 ± 31 **
9AC (n = 5)	0.47 ± 0.06	75.5 ± 16.7	0.80 ± 0.08	329 ± 48
9AC +ωAga IVA	0.46 ± 0.05	74.0 ± 11.4	0.90 ± 0.08	371 ± 36
9AC (n = 11)	0.44 ± 0.03	62.2 ± 4.8	0.87 ± 0.05	290 ± 24
9AC + nifedipine	0.41 ± 0.03	50.0 ± 5.3 *	0.94 ± 0.07	277 ± 21
9AC (n = 7)	0.39 ± 0.03	82.7 ± 17.7	0.66 ± 0.07	352 ± 64
9AC + ryanodine	0.33 ± 0.03	60.6 ± 8.0	0.63 ± 0.09	298 ± 47

Tail currents after a single action potential or after a train of six action potentils at 40 Hz were recorded in voltage clamp at a holding potential of −55 mV. Outward Ca²⁺activated K⁺ currents were isolated by blocking g_Cl,Ca with 2 nm. Changes after adding nifedipine (10 μm), ωCgtx GVIA (300 nm), ωAga IVA (40 nm) or ryanodine (20 μm) were compared using Student's paired t test

^*P < 0.05

^**P < 0.01.

Thus, application of 2 mm 9AC reveals the outward K⁺ current, providing a measure of the Ca²⁺-activated K⁺ conductance g_K,Ca1 (see Introduction). In all the voltage-clamp experiments on g_K,Ca1 described below, 2 mm 9AC was present throughout.

N-type Ca²⁺ channel block has no effect on g_Cl,Ca but reduces g_K,Ca1

The N-type channel antagonist ω-conotoxin GVIA (ωCgtx GVIA, 300 nm) decreased the AHP in all cells tested (n = 9,Fig. 3Aa) and increased the ADP following a train of APs (Fig. 3Ab). On average, the membrane voltage measured at the end of the train shifted from −2.4 ± 6.1 mV to 4.2 ± 5.6 mV (P < 0.001). These results suggest that blockade of Ca²⁺ entry through N-type channels decreased activation of g_K,Ca1, as occurs in rat SCG neurones (Davies et al. 1996), allowing for a greater effect of g_Cl,Ca in depolarizing the membrane following AP discharge.

N-type channels were not involved in the activation of g_Cl,Ca because 300 nmωCgtx GVIA had no significant effect on the mean inward current evoked after a single AP or a train of APs (n = 6; Fig. 3B and Table 1; see also Fig. 11A).

Figure 11. Percentage block of Ca²⁺-activated Cl⁻ and K⁺ conductances following different treatments suggests the mechanisms of their activation by APs.

Changes in g_Cl,Ca (A) and changes in g_K,Ca1 (B) (measured as total charge transfer) produced by ωCgtx GVIA, ωAga IVA, nifedipine and ryanodine (see also Tables 1 and 2). □, effect after a single AP; , effect following a train. Data were obtained in the presence of 100 nm apamin (A) or 2 mm 9AC (B). Error bars indicate s.e.m. Statistically significant comparisons between different groups are indicated. All changes were significantly different from control (in the absence of any Ca²⁺ current blocker) except those marked * (see Tables 1 and 2). g_Cl,Ca was blocked to a similar extent by all treatments except ωCgtx GVIA. The data suggest that Ca²⁺ from P- and L-type channels must act together to release Ca²⁺ from intracellular stores that, in turn, activates Cl⁻ channels underlying the ADP. In contrast, ωCgtx GVIA markedly reduced g_K,Ca1, suggesting a direct link between Ca²⁺ entering through N-type channels and SK channels underlying the AHP. A small amount of Ca²⁺ entry through L-type channels also contributes to activation of SK channels in type II cells. See text for further discussion.

In contrast, ω-Cgtx GVIA reduced g_K,Ca1, the effect being larger on the outward tail current evoked by a single AP (Fig. 3Ca) than on that following a train (Fig. 3Cb), when measured as either charge transfer or mean peak outward current (n = 5; Table 2; see also Fig. 11B). Thus, Ca²⁺ entering through N-type channels activates g_K,Ca1.

P-type Ca²⁺ channel block reduces g_Cl,Ca but has no effect on g_K,Ca1

Relatively selective blockade of P-type channels can be achieved by using ω-agatoxin IVA (ωAga IVA) at concentrations < 50 nm, as the IC₅₀ for Q-type channels is ∼100 nm (Randall & Tsien, 1995). Application of ωAga IVA (40 nm, n = 3) always enhanced the AHP, particularly after a train of APs (Fig. 4A), suggesting that P-type channel block reduces the activation of g_Cl,Ca without having much effect on g_K,Ca1.

Figure 4. P-type Ca²⁺ channels contribute to activation of the ADP but not the AHP.

A, in one cell, addition of ωAga IVA (40 nm) decreased the ADP, therefore increasing the AHP. B, averaged inward currents from 10 cells in the presence of apamin were partially blocked by ωAga IVA. C, averaged outward currents from five cells in the presence of 9AC showed little variation when ωAga IVA was added.

In 10 cells, 40 nmωAga IVA decreased peak inward current by 45 % and charge transfer by about 60 % after a single AP, the effect being slightly smaller after a train (Fig. 4B and Table 1; see also Fig. 11A). The peak inward current after a single AP was not affected by 20 nmωAga IVA whereas peak current was reduced by 53 % and charge transfer by 61 % in 200 nmωAga IVA (n = 3). This latter change was not significantly different from the effect of 40 nmωAga IVA, suggesting that the toxin's action was more likely to have occurred by blockade of P-type than Q-type channels.

In contrast outward currents were not decreased by 40 nmωAga IVA in any of the five cells studied (Fig. 4C and Table 2; see also Fig. 11B). These data indicate that Ca²⁺ entering through P-type channels has no effect on g_K,Ca1.

L-type Ca²⁺ channel block reduces g_Cl,Ca but has little effect on g_K,Ca1

In 12 cells, the L-type Ca²⁺ channel antagonist nifedipine (10 μm) reduced or abolished the ADP that followed APs (Fig. 5A). In contrast, the peak amplitude of the slow AHP increased, consistent with the majority of the Ca²⁺ for activation of the slow AHP entering through channels other than L-type ones.

The amplitude and duration of g_Cl,Ca were substantially reduced by 10 μm nifedipine (n = 9), both after a single AP and after a train (Fig. 5B and Table 1; see also Fig. 11A). These results suggest that, during an AP, the great majority of Ca²⁺ entering through L-type channels is directed towards activation of Cl⁻ channels with little effect on g_K,Ca1.

In agreement with this idea, nifedipine (n = 11) had no significant effect on the peak amplitude of the outward current after a single AP, although it produced a small significant reduction in charge transfer (Fig. 5Ca and Table 2; see also Fig. 11B). There was no significant effect after a train of APs (Table 2). However, the responses to nifedipine differed between cells, which could be separated into two distinct groups on the basis of the time course of decay of g_K,Ca1 (Fig. 6). In six of the 11 cells (type I), outward currents initially fell either above (Fig. 6Aa) or on the exponential fitted to the later decay of g_K,Ca1. In the remaining five cells (type II), there was a plateau during the early part of the outward current, so that the data fell below the extrapolated exponential (Fig. 6Ba; see also Fig. 8Ca). In type I cells, nifedipine had no significant effect on g_K,Ca1 after one AP and produced a small increase in the peak outward current after a train (from 0.87 ± 0.09 to 1.00 ± 0.08 nA, P < 0.05). This increase could be due to concurrent block of residual g_Cl,Ca (due to incomplete block by 9AC). In type II cells, nifedipine decreased g_K,Ca1 after one AP (Fig. 6B), blocking 35 % of the charge transfer (from 69 ± 6 to 45 ± 10 pC, P < 0.05) and 17 % of the peak current (from 0.42 ± 0.04 to 0.35 ± 0.06 nA, P < 0.05), but had no significant effect on g_K,Ca1 after a train. This suggests that, with low Ca²⁺ loads, such as those generated by a single AP, L-type channels are involved in the activation of g_K,Ca1 only in type II SCG neurones but not in type I neurones.

Figure 6. L-type Ca²⁺ channels contribute to activation of g_K,Ca1 in a subgroup of neurones.

Two groups of cells were identified on the basis of the time course of the outward current recorded in the presence of 9AC. A, response of type I cells. When an exponential function was fitted to the outward current, starting 150 ms after the AP (continuous line), and extrapolated to the initial part (dotted line in this and subsequent figures), the outward current of type I cells either fitted the exponential well or initially fell above it (a). When 10 μm nifedipine was added, the outward current from a type I cell was not affected after one AP (a) and showed a small increase in peak current after a train (b). B, response of type II cells. The outward current of type II cells initially fell below the extrapolated exponential (a). In one type II cell, nifedipine markedly reduced g_K,Ca1 after one AP (a) and had a smaller effect on the latter part of the current after a train (b).

Figure 8. Ca²⁺-induced Ca²⁺ release contributes to activation of g_K,Ca1 only in type II cells.

A, ryanodine (20 μm), applied after 9AC, had no significant effect on the averaged outward currents from seven cells. When the cells were separated into two groups as in Fig. 6, ryanodine did not significantly affect the averaged outward currents from four type I cells (B) but decreased g_K,Ca1 in three type II cells (C) after a single AP (a) and after a train (b).

Ryanodine reduces g_Cl,Ca but has little effect on g_K,Ca1

To investigate the contribution of Ca²⁺-induced Ca²⁺ release (CICR) in the activation of g_Cl,Ca and g_K,Ca1, we applied ryanodine (20 μm), which blocks Ca²⁺ release from intracellular stores (Meissner, 1994).

Ryanodine (n = 5) markedly decreased the inward current following an AP and to a lesser extent that following a train (Figs 7A and 11A, Table 1). This suggests that, after a single AP, g_Cl,Ca is activated predominantly by CICR but, as intracellular Ca²⁺ concentration increases, the proportion of Cl⁻ channels activated directly by Ca²⁺ entering the cell also increases. When the cells were challenged with pulse trains of increasing duration, the proportion of the peak inward current blocked by ryanodine decreased from about 75 % for a single pulse to about 35 % for the longer trains (n = 5; Fig. 7B). Thus, CICR plays an important role in activating g_Cl,Ca when intracellular Ca²⁺ concentration is low (i.e. in the physiological range).

Figure 7. Ca²⁺-induced Ca²⁺ release contributes to activation of g_Cl,Ca.

A, averaged inward currents from five cells in the presence of apamin were markedly reduced after one AP when 20 μm ryanodine was added (a) but less so after a train (b). B, mean inward currents from five cells following 5 ms depolarizing pulses (to 0 mV from −55 mV) applied at 40 Hz, plotted against the number of pulses in the train. The reduction in peak current was larger (75 %) when one or two pulses were applied than when the train was prolonged (35 %). Note the logarithmic ordinate. Error bars show s.e.m.

In contrast, ryanodine did not significantly affect the averaged outward currents (n = 7) recorded in 9AC (Fig. 8A and Table 2; see also Fig. 11B). However, if the above criteria were used to separate the cells on the basis of the decay of g_K,Ca1, the responses to ryanodine differed between type I and type II cells. In four type I cells, ryanodine had no significant effect (Fig. 8B) but in three type II cells (Fig. 8C) ryanodine clearly decreased the outward current after one AP (charge transfer, 46 ± 1.1 % block, P < 0.01) and to a smaller degree after a train (32 ± 7.0 % block, P < 0.05). This suggests that CICR contributes to activation of g_K,Ca1 only in type II cells, in which it delays the decline in the outward current after an AP, whereas g_K,Ca1 is probably activated directly by Ca²⁺ entering through N-type channels in type I cells.

The effects of ryanodine on g_Cl,Ca and on g_K,Ca1 in type II cells were almost identical to those of nifedipine. In other cell types, activation of CICR commonly occurs via the Ca²⁺ that enters through L-type channels (Shoshan-Barmatz & Ashley, 1998). We therefore tested the effects of adding ryanodine on the residual g_Cl,Ca and g_K,Ca1 that remained after application of 10 μm nifedipine. In four cells, nifedipine blocked g_Cl,Ca as described above, but subsequent application of ryanodine produced no significant change in g_Cl,Ca (Fig. 9A). Similarly, nifedipine decreased g_K,Ca1 in four type II cells, but ryanodine did not reduce the current further (Fig. 9B). Thus, Ca²⁺ entry through L-type channels triggers the ryanodine-sensitive mechanism that activates g_Cl,Ca and, in type II cells, g_K,Ca1.

Figure 9. Ca²⁺-induced Ca²⁺ release is activated by Ca²⁺ entering through L-type channels.

A, averaged inward tail currents from four cells in the presence of apamin were decreased by 10 μm nifedipine, but showed no significant reduction when 20 μm ryanodine was added. B, averaged outward tail currents from four type II cells were decreased by nifedipine, but not by subsequent addition of ryanodine.

Resistant Ca²⁺ channels contribute to both g_Cl,Ca and g_K,Ca1

Because the degree of block of g_Cl,Ca was similar (about 65 %) when either P- or L-type channels were blocked, we examined whether R-type channels were involved in the activation of g_Cl,Ca by applying both ωAga IVA (40–200 nm) and nifedipine (1–10 μm) in the presence of 100 nm apamin (n = 4). After a single AP, charge transfer was blocked by about 70 % and peak current by about 65 %, whereas after a train, block was about 70 and 60 %, respectively (Fig 10A and Fig 11A). Given the lack of effect of N-type channel blockers on g_Cl,Ca, this result indicates that R-type channels contributed to activation of the inward current.

Figure 10. R-type channels contribute to activation of the currents underlying afterpotentials.

A, averaged inward tail currents from four cells in the presence of apamin were markedly reduced, but not completely abolished, in the presence of 40 nmωAga IVA and 10 μm nifedipine. B, averaged outward tail currents from three cells in the presence of 9AC were markedly reduced, but not completely blocked, in the presence of 10 μm nifedipine and 300 nmωCgtx GVIA. Subsequent addition of 300 μm Cd²⁺ blocked the remaining outward currents.

To test whether Ca²⁺ entering through R-type channels also activates g_K,Ca1, we applied nifedipine (10 μm) and ωCgtx GVIA (300 nm) in the presence of 2 mm 9AC (n = 3). Overall block of the outward current was much greater after a single AP (charge transfer about 85 %, peak current about 75 %) than after a train (charge transfer about 50 %, peak current 30 %; Fig 10B and Fig 11B), but there was clearly a fraction of g_K,Ca1 resistant to both drugs, even after a single AP, because 300 μm Cd²⁺ blocked the remaining current. Since P-type channel block did not affect g_K,Ca1, this result suggests that Ca²⁺ entry through R-type channels contributes to activation of g_K,Ca1.

Discussion

Our main conclusion is that activation of the Cl⁻ conductance (g_Cl,Ca) responsible for the ADP following APs in mouse SCG neurones involves Ca²⁺ entering through P- and L-type channels and acting via Ca²⁺ release from ryanodine-sensitive intracellular stores. In contrast, the K⁺ conductance responsible for the slow AHP (g_K,Ca1) is activated predominantly by Ca²⁺ entering through N-type channels. Assuming that our pharmacological tools provide selective and complete blockade, we can conclude that a smaller proportion of both currents is activated directly by Ca²⁺ entering through R-type channels.

These conclusions rest on the assumption that nifedipine, ωCgtx GVIA and ωAga IVA did not directly block Cl⁻ or K⁺ channels. None of these drugs had effects on AP amplitude or passive properties, indicating that they do not interfere with Na⁺ currents or most K⁺ currents as previously shown in rat SCG neurones (Davies et al. 1996). This is in agreement with the known specificity of ωCgtx GVIA for N-type Ca²⁺ channels (Olivera et al. 1994) and of ωAga IVA for P/Q-type channels (Randall & Tsien, 1995). Also, ryanodine did not directly affect Ca²⁺, K⁺ or Cl⁻ channels because AP shape and input resistance were not modified. Furthermore, the currents generated by g_K,Ca1 in type I cells were not reduced in the presence of ryanodine.

Activation of g_Cl,Ca

Our results after blocking g_K,Ca1 with apamin demonstrate that most, if not all, normal mouse SCG neurones express g_Cl,Ca, as suggested by De Castro et al. (1997). Also, when 9AC was applied to cells that apparently lacked an ADP, the AHP became larger, indicating that blockade of g_Cl,Ca allowed g_K,Ca1 to hyperpolarize the membrane further.

The function of g_Cl,Ca in sympathetic neurones is not clear (e.g. Sánchez-Vives & Gallego, 1994). When SK channels were blocked with apamin, the ADP often generated APs, which is very rare in control solution (Sánchez-Vives & Gallego, 1994; De Castro et al. 1997). This suggests that decreases in the AHP, either indirectly following inhibition of N-type channels by e.g. catecholamines and neuropeptides (Toth & Miller, 1995), or directly following inhibition of SK-type channels by e.g. muscarinic agonists (Cassell & McLachlan, 1987b), might lead to repetitive discharge. Thus the Cl⁻ conductance underlying the ADP is expressed to a greater or lesser extent in all mouse SCG neurones where it could play a role in generating bursts of activity. However, this mechanism to increase firing appears to be absent from normal sympathetic neurones of other species (Cassell & McLachlan, 1987a; Sánchez-Vives & Gallego, 1994).

Activation of g_Cl,Ca was dependent on Ca²⁺ entering through L-, P- and R-type channels (Fig. 11A). Functional coupling between L-type channels and Cl⁻ channels in sensory neurones has been previously implied by the potentiation of Cl⁻ tail currents by BayK8644, an L-type channel agonist (Scott et al. 1988). Although L- and N-type channels have been implicated in the activation of Ca²⁺-dependent Cl⁻ currents in rabbit parasympathetic neurones (Akasu et al. 1990), ωCgtx GVIA did not affect g_Cl,Ca in mouse SCG neurones.

After one AP, g_Cl,Ca was blocked by about 70 % by ryanodine (Fig. 11A), indicating that CICR plays an important role in its activation. The fact that either L- or P-type channel blockade reduced g_Cl,Ca by approximately the same amount (Fig. 11A) suggests that Ca²⁺ entering through these channels activates CICR. The lack of effect of ryanodine on g_Cl,Ca and g_K,Ca1 when applied after nifedipine confirms this for L-type channels, as has been shown in muscle cells (Shoshan-Barmatz & Ashley, 1998) and some neurones (Osmanovic & Shefner, 1993; Tanabe et al. 1998; Cordoba-Rodriguez et al. 1999). It appears, however, that both L- and P-type channels are needed simultaneously to trigger CICR in mouse SCG neurones, because a similar degree of block (70 %) was produced when the two drugs were applied together (Fig. 11A). As the amounts of Ca²⁺ entering through each type of channel during an AP are quite different (J. Martínez-Pinna & R. Gallego, unpublished observations), they presumably act co-operatively but non-linearly in activating CICR. As the amount of g_Cl,Ca (30 %) remaining after combined blockade of L- and P-type channels (Fig. 11A) was similar to the ryanodine-insensitive component, Ca²⁺ entry through R-type channels probably activates g_Cl,Ca directly.

The fact that the degree of block of g_Cl,Ca by nifedipine and ωAga IVA was similar after a single AP and after a train of six APs (when the peak inward current was about 16 times larger) implies that ‘spill over’ from channels other than L- and P-type was relatively ineffective in activating g_Cl,Ca. This suggests that CICR activation of g_Cl,Ca is spatially restricted (Berridge, 1998). This arrangement would allow CICR to deliver the concentration of Ca²⁺ necessary for activation of g_Cl,Ca because the Ca²⁺ sensitivity of the Cl⁻ channels is relatively low (EC₅₀∼25 μm; Gomez-Hernandez et al. 1997; Hallani et al. 1998). In contrast, after a train of six APs, ryanodine produced a smaller degree of block (40 %, Fig. 11A) than that produced by nifedipine and ωAga IVA (65 %), suggesting that Ca²⁺ entering through L- and P-type channels, as well as through R-type channels, can activate g_Cl,Ca directly when the neurone is overloaded with Ca²⁺ (see Fig. 7). Since sympathetic ganglion cells fire in vivo at low frequencies (McLachlan et al. 1997), the physiological activation of g_Cl,Ca probably depends primarily on CICR mechanisms. Recent evidence in dorsal root ganglion neurones confirms that Cl⁻ channels can be activated by Ca²⁺ released from intracellular stores during one AP (Ayar & Scott, 1999).

Activation of g_K,Ca1

Activation of g_K,Ca1 after a single AP is clearly dependent on Ca²⁺ entry through N-type channels (Fig. 11B). This dependence is in agreement with findings in rat hypoglossal (Viana et al. 1993), vagal (Sah, 1995), sympathetic (Davies et al. 1996) and trigeminal (Kobayashi et al. 1997) neurones and guinea-pig sympathetic neurones (Ireland et al. 1998), as well as lamprey spinal neurones (Wikström & El Manira, 1998). Because g_K,Ca1 primarily involves SK channels, the functional link between these and N-type Ca²⁺ channels is therefore common to many neurones (cf. Marrion & Tavalin, 1998). On the other hand, R-type channels also contribute to g_K,Ca1 activation because the combined application of nifedipine and ωCgtx GVIA did not abolish the outward current (Fig. 11B).

Ryanodine had different effects in two subtypes of mouse SCG neurone that could be separated on the basis of the time course of g_K,Ca1. In type II cells, in which g_K,Ca1 showed a plateau before it declined, CICR activated about half of the current after one AP, whereas in type I cells in which g_K,Ca1 declined exponentially, CICR played no significant role. A small contribution of CICR (30 %) to g_K,Ca1 was previously reported in guinea-pig prevertebral sympathetic neurones (Jobling et al. 1993). Because nifedipine inhibited g_K,Ca1 and blocked the effect of ryanodine only in type II cells, its action is most likely to have been to prevent activation of CICR. According to this interpretation, g_K,Ca1 activation in type I neurones depends entirely on Ca²⁺ entering through N- and R-type channels, whereas in type II neurones, Ca²⁺ release induced by Ca²⁺ entering through L-type channels is also involved. L-type channels, CICR and Ca²⁺-activated K⁺ (SK) channels are functionally linked in vascular smooth muscle (Jaggar et al. 1998) and hippocampal neurones (Tanabe et al. 1998).

The contribution of R-type channels to g_K,Ca1 activation was substantially larger after a train than after one AP (Fig. 11B), consistent with ‘spill over’ of accumulated Ca²⁺. Under these conditions, there was only slight inhibition of g_K,Ca1 by nifedipine or ryanodine in type II cells (Figs 6Bb and 8Cb), consistent with the saturation of CICR by Ca²⁺ influx after a few APs.

Distinct intracellular mechanisms for activation of Cl⁻ and K⁺ channels

The physical bases for the functional links between Ca²⁺ and K⁺/Cl⁻ channels in intact neurones are not known. The present results could be explained if, in mouse SCG neurones, L- and P-type channels are linked to an intracellular pathway, perhaps the store in the subsurface cisternae (Berridge, 1998), from which Ca²⁺ release directly activates membrane Cl⁻ channels, whereas N-type channels are closely juxtaposed to SK channels. R-type channels provide Ca²⁺ that activates both Cl⁻ and SK channels, to a much greater extent following a train of APs, and so they may be more widely distributed. Colocalization of Ca²⁺ and K⁺ channels has been demonstrated directly at the neuromuscular junction, where BK channels are located adjacent to N-type channels in the presynaptic terminal (Robitaille et al. 1993), and in membrane patches from hippocampal somata (Marrion & Tavalin, 1998). In the latter case, N-type, SK and BK channels were present in the same somatic patch but SK channels were not activated by Ca²⁺ entering through the N-type channels, suggesting that some form of direct linkage between channels can exist at the molecular level.

Another possibility is that the different channels are located in separate regions of the cell (Davies et al. 1996). Although a dendritic location for high-threshold Ca²⁺ channels in sympathetic neurones has been proposed (Hirst & McLachlan, 1986), no data are available about the distribution of the different channel types in the somatodendritic membrane. In central neurones, L-type channels, although present in distal dendrites, are more common in the cell body and proximal dendrites, whereas N-type channels are more or less equally expressed in both cellular compartments (Westenbroek et al. 1998). Based on indirect evidence, it has been proposed that the Cl⁻ channels responsible for the ADP are located in the dendrites (Sánchez-Vives & Gallego, 1994; De Castro et al. 1997) which, in mouse SCG neurones, are relatively few in number and short. If this were the case, L- and P-type Ca²⁺ channels may also be located on the dendrites. There might even be two distinct populations of Cl⁻ channels in mouse SCG cells: a larger one that is activated via intracellular pathways dependent on Ca²⁺ entering via L- and P-type channels and another population that is activated by Ca²⁺ derived from R-type channels. While the second population of Cl⁻ channels might be activated either directly or indirectly, it is likely to be spatially distinct from the Cl⁻ channels dependent on Ca²⁺ entering via L- and P-type channels.

In conclusion, our results provide the first evidence for distinct mechanisms linking the physiological entry of Ca²⁺ through particular channel types during AP firing with the activation of Cl⁻ and K⁺ membrane conductances in the same neurone. The data confirm that general intracellular Ca²⁺ levels are unlikely to reflect the localized consequences of Ca²⁺ influx through specific channels during physiological events.