Corresponding author G. Boehmer: Department of Physiology and Pathophysiology, Johannes Gutenberg-University, Saarstrasse 21, D-55099 Mainz, Germany. Email: firstname.lastname@example.org
1The pharmacological sensitivity of hyperpolarizing components of spike train after-potentials was examined in sixty-one magnocellular neurones of the rat supraoptic nucleus using intracellular recording techniques in a brain slice preparation.
2In 26 % of all neurones a slow after-hyperpolarization (AHP) was observed in addition to a fast AHP. In 31 % of all neurones a depolarizing after-potential (DAP) was observed.
3The fast AHP was blocked by apamin whereas the slow AHP was blocked by charybdotoxin (ChTX). The DAP was enhanced by ChTX or a DAP was unmasked if not present during the control period.
4Low concentrations of TEA (0.15–1.5 mm) induced effects on the slow AHP and the DAP essentially resembling those of ChTX. The same was true for the effects of CoCl2 (1 mm).
5Spike train after-potentials were not affected by either iberiotoxin (IbTX), a selective high-conductance potassium (BK) channel antagonist, or margatoxin (MgTX), a Kv1.3 α-subunit antagonist.
6Kv1.3 α-subunit immunohistochemistry revealed that these units are not expressed in the somato-dendritic region of supraoptic neurones.
7The effects of ChTX, IbTX, MgTX, TEA, CoCl2 and CdCl2 on spike train after-potentials are interpreted in terms of an induction of the slow AHP by the activation of calcium-dependent potassium channels of intermediate single channel conductance (IK channels).
8The results suggest that at least the majority of supraoptic magnocellular neurones share the capability of generating both a slow AHP and a DAP. The slow AHP may act to control the expression of the DAP, thus regulating the excitability of magnocellular neurones. The interaction of the slow AHP and the DAP may be important for the control of phasic discharge.
The release of the peptide hormones vasopressin and oxytocin strongly depends on the pattern of neuronal activity of magnocellular neurosecretory neurones located in the supraoptic and paraventricular nuclei (see Poulain & Wakerley, 1982, for review). During physiological stimulation of these magnocellular neurones an increase of the discharge rate and, in a subpopulation of neurones, the transition from continuous to phasic activity was observed. The excitability and the discharge pattern of magnocellular neurones of the supraoptic nucleus (SON) strongly depend on depolarizing and hyperpolarizing after-potentials (Andrew & Dudek, 1984a,b). When a train of action potentials is evoked by the injection of a depolarizing direct current, an after-hyperpolarization (AHP) or a sequence of an AHP and a depolarizing after-potential (DAP) follows the termination of the train of spikes. The DAP was reported to be more frequently observed in vasopressin neurones than in oxytocin neurones (Armstrong et al. 1994). Sufficient summation of DAPs induces a plateau potential that gives rise to repetitive neuronal discharge (Andrew & Dudek, 1984a; for review, see Legendre & Poulain, 1992). Both the AHP (Andrew & Dudek, 1984b) and the DAP (Bourque, 1986; Andrew, 1987; Li et al. 1995) are generated by the activation of calcium-dependent mechanisms. During neuronal discharge, the intracellular calcium concentration is increased by the activation of voltage-gated calcium channels. T-, N-, L-, and P-type calcium channels, as well as two novel types of voltage-activated calcium channels, are of functional importance in magnocellular SON neurones (Fisher & Bourque, 1995). The ionic mechanisms involved in the generation of the DAP are still obscure. A reduction of an outward potassium current is suggested to be involved in the induction of the DAP (Li & Hatton, 1997b). The AHP was demonstrated to be evoked by an activation of calcium-dependent potassium channels (Andrew & Dudek, 1984b; Bourque et al. 1985; Bourque & Brown, 1987; Kirkpatrick & Bourque, 1996). Due to differing pharmacological properties, two major types of calcium-dependent potassium channel can be discriminated (for review, see Rudy, 1988; Storm, 1990; Brown et al. 1990). Calcium-dependent potassium channels with a small single channel conductance (SK channels) are selectively blocked by the bee venom toxin apamin, are insensitive to tetraethylammonium (TEA) and have little or no voltage dependence. Calcium-dependent potassium channels with a large single channel conductance (BK or maxi-K channels) are blocked by the scorpion venom peptide charybdotoxin (ChTX; Leiurus quinquestriatus; for review, see Garcia et al. 1995), are very sensitive to TEA and are strongly voltage dependent. BK channels have been demonstrated to be very selectively blocked by the scorpion venom peptide iberiotoxin (IbTX; Butus tamulus; see Garcia et al. 1991). When compared with the action of IbTX, the action of ChTX is less specifically directed against BK channels, since it also blocks calcium-dependent potassium channels with an intermediate single channel conductance (IK channels) which are also sensitive to TEA but not to apamin or IbTX (for review, see McManus, 1991). In addition to its action on calcium-activated potassium channels, ChTX binds to the α-subunit Kv1.3 of the Shaker-related subfamily of rat Kv channels and thus blocks voltage-gated potassium channels containing the Kv1.3 α-subunit (for review, see Garcia et al. 1995). Kv1.3 α-subunit-containing potassium channels are selectively blocked by the scorpion venom peptide margatoxin (MgTX; Centruroides margaritatus;Garcia-Calvo et al. 1993).
In supraoptic neurones, a major portion of the AHP was demonstrated to be blocked by apamin (Bourque & Brown, 1987; Armstrong et al. 1994), suggesting at least an important contribution by the SK type of calcium-dependent potassium channels to the genesis of AHPs. In the present study the contribution of ChTX-, IbTX- and MgTX-sensitive potassium channels to the generation of AHPs in magnocellular neurones of the rat SON was examined in brain slices. The effects of these toxins on spike train after-potentials were compared with the effects of apamin. This pharmacological approach allows the selective dissection of AHP components induced by the activation of a broad spectrum of calcium-activated potassium channels. Preliminary accounts of this study have appeared in abstract form (Greffrath & Boehmer, 1997).
Adult male Sprague-Dawley rats (120–250 g; Charles River, Germany) were transferred to a glass container fitted with a venting connector and were deeply anaesthetized with diethyl ether. The concentration of the evaporated diethyl ether in the container was gradually elevated. During deep anaesthesia animals were killed by rapid decapitation; this method is in accordance with the German national law for animal protection as well as with the rules of the local ethical committees of the district of Rheinland-Pfalz and the J. Gutenberg University, Mainz. The whole brain was rapidly removed from the skull and chilled in artificial cerebrospinal fluid (ACSF) at 4°C. After trimming a tissue block containing the hypothalamus, coronal brain slices (350–500 μm) were cut on a vibratome (Vibroslice, Campden Instruments, UK). During the cutting procedure, the tissue was covered with chilled ACSF. The slices were maintained in an incubation chamber at room temperature for at least 2 h before recording. For recording, a single slice was transferred to a Haas-type recording chamber and was superfused with ACSF at 32 ± 1°C. The flow rate of the superfusate was 2–3 ml min−1. Throughout the entire experiment including the preparation procedures, ACSF was saturated with carbogen gas (95 % O2, 5 % CO2). The membrane potential of magnocellular supraoptic neurones was intracellularly recorded with glass microelectrodes filled with 2–3 m potassium acetate (resistance 140–190 MΩ). Microelectrodes were fabricated from borosilicate glass capillaries (o.d. 1.2 mm, i.d. 0.6 mm, Hilgenberg, Germany) using a horizontal micropipette puller (P-87, Sutter, USA). Neuronal potentials were recorded with an Axoclamp 2A amplifier (Axon Instruments) in bridge mode. Penetration of the neuronal membrane with the tip of the microelectrode was facilitated by a new type of stepper motor-driven micro-positioner (developed by E. Martin). Cells included in this study had a resting membrane potential of at least −50 mV and discharged trains of overshooting action potentials during prolonged depolarizing direct current pulses. Neurones not displaying activity-dependent spike broadening (Bourque & Renaud, 1985) were excluded from further evaluation. Signals were recorded on videotape using an AD/DA converter (VR-10B, Instrutech, USA) and a conventional VCR. For off-line analysis, signals were AD/DA converted (Labmaster TL 1–125, Axon Instruments) and fed to a personal computer. Data acquisition and analysis was performed with Axotape and pCLAMP software (Axon Instruments). Time constants of spike train after-potentials were determined for the decay from the peak potential to 63 % of the range between the peak potential and the resting membrane potential. Since the peak potential of the slow AHP was shunted by the fast AHP, the decay from the apparent onset of the slow AHP was determined instead.
Ionic conductances involved in the generation of the AHPs following a current-induced spike-train were pharmacologically characterized by application of specific agents to the bath: apamin (50–500 nm, Alomone Labs, Jerusalem, Israel), charybdotoxin (5– 15 nm, Alomone), iberiotoxin (50–200 nm, Alomone), margatoxin (100–200 nm, Alomone), tetraethylammonium (0.15–4.5 mm; Sigma), CdCl2 (0.3–1 mm, Merck), CoCl2 (1 mm, Merck). The superfusion medium contained (mm): NaCl, 124; KCl, 2; KH2PO4, 1.25; MgSO4, 1.3; CaCl2, 2.5; NaHCO3, 26; glucose, 10. For the application of CdCl2 or CoCl2, KH2PO4 and MgSO4 were replaced by KCl and MgCl2, respectively. The osmolality of ACSF was adjusted to 295 mosmol kg−1, and pH was 7.4 ± 0.1. For the application of charybdotoxin, iberiotoxin and margatoxin the superfusate contained 0.01 % bovine serum albumin (BSA). The standard ACSF used for superfusion of brain slices during the control and washout periods also contained 0.01 % BSA.
For immunohistochemistry, adult male Sprague-Dawley rats were used. They stemmed from the same source as those used for electrophysiological experiments and were held under constant conditions (light/dark 12 h:12 h; food and water ad libitum). In the middle of the light period, animals were killed as described above and perfused transcardially with a solution containing 4 % paraformaldehyde, 1.4 % lysine and 0.2 % periodate in 0.1 m phosphate-buffered saline (PBS) according to McLean & Nakane (1974). The brain was removed, cryoprotected in sucrose solution and cut at 40 μm thickness in the frontal plane on a freezing microtome. Sections were rinsed in PBS and incubated for 24 h at room temperature (20–22°C) in rabbit-raised anti-Kv1.3 serum (Alomone) diluted 1:50 in PBS to which 1 % normal serum and 0.2 % Triton X-100 was added. The immunoreactions were visualized by using a biotinylated anti-rabbit IgG (Amersham, Braunschweig, Germany; 1:100), followed by incubation in streptavidin-Cy3 (Jackson Immunoresearch, West Grove, PA, USA; 1:200), each in PBS for 90 min at room temperature. Specificity controls were performed by pre-incubating the antiserum with the antigen (1:30 in PBS; Alomone) overnight and then processing the sections as described above.
The contribution of ChTX-, IbTX-, MgTX- and apamin-sensitive potassium channels to the generation of AHPs was examined in sixty-one magnocellular supraoptic neurones. The resting membrane potential of these neurones was −54.4 ± 8.42 mV (mean ±s.d.), and their input resistance, as calculated from the amplitude of hyperpolarization induced by direct current injection, was 221.8 ± 118.76 MΩ. The time constant τ, as determined from the initial potential trajectory induced by a hyperpolarizing direct current pulse, was 14.3 ± 6.59 ms. All neurones examined in the present study were inactive during uninfluenced control periods, i.e. no spontaneous discharge activity was generated.
Components of spike train after-potentials
In all magnocellular supraoptic neurones examined in the present study, the train of spikes evoked by an injection of depolarizing direct current was followed by after-potentials. In twenty-six neurones (42.6 % of all cells) only a fast (early and short-lived) spike train AHP was observed (Fig. 1A). Characteristically, the fast AHP reached its peak amplitude within 10–30 ms of termination of the spike train and rapidly decayed thereafter. The time constant of the decay was 256.1 ± 118.46 ms (mean ±s.d., n= 19). Within 2 s the resting membrane potential was reached again. In sixteen neurones (26.2 % of all cells), the AHP was composed of a fast component and an additional slow component (slow decay; Fig. 1B). In two of these neurones the peak amplitude of the slow AHP was reached 384 and 372 ms, respectively, after the termination of the spike train. However, in the remaining cases the time to peak of the slow AHP could not be determined, since the early phase of the slow AHP was shunted by the fast AHP. In some cases, the transition from the fast to the slow AHP could be recognized by an abrupt deceleration of the decay 200–300 ms after the termination of current injection (Fig. 1C). The time constant of the decay of the slow AHP was 1010.3 ± 311.30 ms (n= 10). The duration of the slow AHP was up to 6 s. In nineteen neurones (31.2 % of all cells) a sequence of a fast AHP and a DAP was induced after the termination of the train of spikes (Fig. 1D). The DAP reached its peak amplitude 700.2 ± 252.19 ms after the termination of current injection (n= 10) and slowly decayed thereafter. The time constant of the decay was 1732.4 ± 237.10 ms.
In ninety-five neurones (including 34 neurones not examined for the pharmacological properties of AHP components) the examination of the distribution of resting membrane potentials within groups of neurones expressing different forms of after-potentials, i.e. a fast AHP, a sequence of a fast and a slow AHP, or a sequence of a fast AHP and a DAP, revealed that the expression of the various after-potential components did not depend on the resting membrane potential. Furthermore, there was neither a correlation between the amplitude of the fast AHP and the resting membrane potential (r= 0.177) nor between the amplitude of the DAP and the resting membrane potential (r=−0.273; n= 27). However, current-induced generation of spike trains from membrane potentials set to between about − 50 mV and about − 90 mV by continuous injection of hyperpolarizing currents demonstrated that the expression of the fast AHP and the DAP is altered with increasing hyperpolarization (n= 8; results not demonstrated). During increasing hyperpolarization, the DAP decreased in amplitude while the fast AHP first increased and then strongly decreased. At about −75 mV the DAP was no longer detectable while a small fraction of the fast AHP was still present even at about −90 mV. The latter results are in accordance with the findings of recent studies (Andrew, 1987; Li & Hatton, 1997a,b). Also in accordance with these studies, the activation of neurones from depolarized membrane potentials resulted in an increase in amplitude of the DAP, in the generation of an active DAP and finally in a transformation of the DAP to a plateau potential.
Pharmacological characterization of AHP components
In most cases, pharmacological characterization of AHP components was performed in neurones recorded from separate brain slices. Successive recording of two neurones was performed in four brain slices. In two of these slices only the responses to TEA were examined, in one slice only the responses to apamin and in the remaining slice the responses to ChTX and CoCl2. Multiple applications of potassium channel antagonists were restricted to a minimum of cases. A single application of antagonists was performed for ChTX in 3 of 8 neurones, for apamin in 8 of 15 neurones, for TEA in 7 of 16 neurones and for IbTX and MgTX in all neurones examined. In no case was more than one peptide potassium channel antagonist applied to supraoptic neurones. Single applications of CoCl2 and CdCl2 were performed in 2 of 8 and 4 of 7 neurones, respectively. When multiple application of drugs was performed, TEA, CoCl2 or CdCl2 were, with one exception, applied after the application of one of the peptide toxins. In one case TEA was applied prior to the application of apamin. Triple applications were performed in three cases only. The effects of TEA at the concentrations used were obviously fully and rapidly reversible. In accordance with the results of recent studies (Armstrong et al. 1994; Kirkpatrick & Bourque, 1996) the effects of apamin were partially reversible after an extended washout period (up to 2 h), as was observed when the effects of apamin were examined in two neurones subsequently recorded from the same brain slice. Essentially the same was true for ChTX after up to 1.5 h washout, as was indicated by the effects of a subsequent application of TEA or CoCl2.
In all neurones examined, superfusion of the slice with ACSF containing CdCl2 resulted in a strong attenuation or block of both the AHP (n= 7) and the DAP (n= 6; Fig. 2A). This result suggests that all components of spike train after-potentials at least strongly depend on activation of calcium-dependent mechanisms. However, different components of the sequence of after-potentials were blocked with different delays after the onset of superfusion with ACSF containing CdCl2. During the early period of application of CdCl2, the AHP was strongly attenuated (Fig. 2A), presumably due to an effective reduction of calcium influx as indicated by the reduction of spike broadening. The ratios of the sixth to the first spike width were 1.43 and 1.29, respectively, for the control and the early wash-in periods. Moreover, the single spike hyperpolarizing after-potentials were reduced. In contrast to the fast AHP, the DAP was not attenuated or was slightly enhanced during the early period of CdCl2 application. During the further course of application, the DAP was attenuated and finally blocked (Fig. 2A, lower two panels). During the late wash-in period, the ratio of the sixth to the first spike width was 1.16, indicating that a significant block of calcium influx may have been induced by cadmium. Thus, the differing effects of CdCl2 on the AHP and the DAP may be due to a higher sensitivity of the DAP than of the AHP components to the rise in the intracellular Ca2+ concentration. However, since the actual concentration of cadmium at the recording site is unknown, an action of cadmium on potassium channels cannot be excluded. A reduction in potassium conductance has been demonstrated to underlie the generation of the DAP (Li & Hatton, 1997b). Thus, the effects of cadmium on the DAP during the early period of application may not be due only to the effects of cadmium on different types of calcium conductance; an additional contribution due to a reduction in potassium conductance has to be considered.
However, the importance of calcium-dependent mechanisms is accentuated by the dependence of the expression of the slow AHP and the DAP upon the activity during the spike train. In a recent study, the expression of the AHP was demonstrated to depend on the number of action potentials in a train, regardless of frequency (Kirkpatrick & Bourque, 1996). The AHP reached more than 90 % of its maximal amplitude before the occurrence of the fifteenth action potential in a train. The results of the present study also indicate that the composition of the after-potentials depends on the number of spikes in a train of action potentials (Fig. 2B). In neurones expressing a sequence of a fast AHP and a DAP after fifteen action potentials (Fig. 2B, top traces), a slow AHP was induced in addition to the fast AHP when the number of evoked action potentials was increased to twenty and thirty-one spikes, by prolonging the depolarizing current pulse (Fig. 2B, centre and bottom traces). In the latter case, the DAP was strongly attenuated and its onset was delayed due to the expression of a slow AHP. Hence, the expression of the DAP may depend on the magnitude of the induced slow AHP. When the depolarizing current was augmented, the number of action potentials and the discharge frequency was increased (Fig. 2C). Moderate depolarization resulted in the generation of a DAP sufficient to initiate discharge activity (Fig. 2C, top traces). With increasing depolarization, the onset of the active DAP was delayed (Fig. 2C, centre traces) and it finally became subthreshold (Fig. 2C, bottom traces). Comparable effects have also been demonstrated in recent studies (e.g. Andrew & Dudek, 1984b). This result may be interpreted to show a suppression of the DAP by a slow AHP that was enhanced with increasing discharge rate. Together, these results indicate that the slow AHP has a low sensitivity to calcium, compared with that of the DAP and also of the fast AHP.
A sustained calcium-dependent potassium current recorded in neurones isolated from the rat supraoptic nucleus area was reported to be suppressed when 2 mm Co2+ was added to the medium containing 2 mm Ca2+ (Cobbett et al. 1989). The action of CoCl2 on AHPs in supraoptic neurones was also examined in the present study. The end-concentration of CoCl2 added to the standard ACSF was 1 mm. When applied for 5 min (3 neurones), no effects of Co2+ on spike train after-potentials could be observed. When applied for 10–15 min (8 neurones), Co2+ blocked the slow AHP after more than 5 min and finally unmasked a DAP in all neurones lacking this potential during the control period (n= 2; Fig. 3A). If a DAP was already present during the control period (n= 6), its amplitude was increased by applying 1 mm CoCl2 (Fig. 3B). The time course of the Co2+-induced enhancement of the DAP closely resembled that of the attenuation of the slow AHP. In four neurones examined, the enhancement of the DAP resulted in the generation of an active plateau potential. The fast component of the AHP was slightly to moderately reduced during the application of CoCl2.
The action of the bee toxin apamin on the spike train AHP was examined in fifteen neurones. Application of 50–100 nm apamin (in two cases up to 500 nm) resulted in a strong attenuation or a full block of the fast AHP in thirteen neurones (Fig. 4A). In eight of these neurones a small DAP was unmasked or, if present during the control period, was slightly enhanced by apamin. In two neurones showing AHPs which lasted longer than 3 s (duration 4 and 6 s, respectively), the amplitude and duration of the long-lasting AHPs were also strongly increased (Fig. 4B). The latter effect of apamin in these neurones may be due to its blocking action on spike frequency adaptation and the concomitant increase of Ca2+ entry. Furthermore, the slow AHP in these neurones was not blocked by 1.5 mm TEA. Thus, the properties of the long-lasting AHP in these neurones obviously differ from the properties of both the fast and the slow AHP.
The action of the scorpion toxin ChTX on the spike train AHP was examined in eight neurones. In contrast to apamin, ChTX affected slow components of the spike train after-potentials, while the fast AHP essentially remained unaffected (only minor decreases in the amplitude of the fast AHP were observed). In neurones already showing a slow AHP during the control period (n= 5), the slow AHP was strongly attenuated or blocked during and after the application of 5–15 nm ChTX (Fig. 5A). As a result, the abrupt transition from the fast to the slow AHP component, as observed in four neurones, could no longer be recognized, suggesting a complete block of the slow AHP. In all of these neurones, ChTX induced an alteration in the pattern of after-potentials not only by blocking the slow AHP but also by unmasking a DAP not present during the control period. The latter was also true for three other neurones showing only a fast AHP during the control period (Fig. 5B). Thus, in all neurones examined the existence of a DAP was unmasked by ChTX. In two cases, depolarization by the unmasked DAP was strong enough to induce an after-discharge of action potentials.
The ChTX-sensitive calcium-activated potassium channels were also reported to be sensitive to low concentrations of TEA (for review, see Brown et al. 1990). Consequently, the effects of TEA on the AHP of supraoptic neurones were also examined in the present study and compared with the effects of ChTX. In sixteen neurones TEA was applied in concentrations of 150 μm to 1.5 mm. In four additional neurones, the results of TEA applications at concentrations of 2–4.5 mm were considered, since the effects on after-potentials were consistent with those of lower concentrations. In 4 of these 20 neurones the sequence of after-potentials was not affected by TEA up to 750 μm. In nine neurones examined, the slow AHP observed during the control period was strongly attenuated or blocked by TEA (Fig. 5C). In neurones showing an abrupt transition from the fast to the slow AHP component, this transition could no longer be recognized, suggesting a complete block of the slow AHP by TEA. In 8 of these 9 neurones a DAP was unmasked during the application of TEA. A DAP was also unmasked in four neurones showing only a fast AHP during the control period. In neurones showing both a fast AHP and a DAP during the control period (n= 3), the peak amplitude of the DAP was increased (Fig. 5D). Thus, in 15 of 16 neurones affected by TEA, a DAP was unmasked or enhanced during the application of TEA. In four of these neurones the TEA-induced enhancement of the DAP was strong enough to induce an after-discharge of action potentials (data not shown). In all neurones examined, the fast AHP was increased to variable extents. The latter effect was most probably due to the TEA-induced weak or moderate delay of spike repolarization which was observed even with low concentrations of TEA (data not shown). This effect may have induced a concomitant increase in calcium entry.
The effects of the BK channel-specific scorpion toxin IbTX on spike train after-potentials were examined in fourteen supraoptic neurones. In none of these neurones did the application of 50–100 nm IbTX induce a significant effect on the time course or the amplitude of the fast AHP. In neurones showing a sequence of a fast and a slow AHP during the control period (n– 4), the slow AHP was also unaffected by IbTX. In four neurones showing a sequence of a fast AHP and a DAP during the control period, the DAP was not affected by IbTX (Fig. 6A). By analogy with our interpretation of the effects of TEA and Co2+ on the DAP, the lack of an effect of IbTX on the DAP was interpreted as showing that the toxin does not affect the slow AHP in these neurones. Furthermore, increasing the concentration of IbTX up to 200 nm did not result in any effect on the sequence of after-potentials.
The contribution of voltage-gated potassium channels containing the Kv1.3 α-subunit to spike train AHPs was examined in eight supraoptic neurones by the application of the Kv1.3 α-subunit-specific scorpion toxin MgTX. In all neurones examined, neither the amplitude nor the time course of the fast AHP and the slow AHP were affected by MgTX (Fig. 6B). The same was true when a DAP was present during the control period; neither the peak amplitude nor the time course of the DAP was affected by MgTX. By analogy with our interpretation of the effects of TEA, Co2+and ChTX, the lack of an MgTX-induced effect on the DAP was interpreted as showing that the toxin does not affect the slow AHP. That the lack of effect was not due to an insufficient concentration of MgTX was confirmed by the observation that neither of the spike train after-potentials was affected when the concentration of MgTX was increased to 200 nm (n= 2).
The potassium antagonists used in this study exerted only minor effects on the resting membrane potential of neurones. TEA (0.3–1.5 mm) exerted no significant effects on the resting membrane potential in 8 of 20 neurones examined. In the remaining neurones, application of TEA elicited either an enhancement of the DAP, resulting in an after-discharge generated by a plateau potential or by a slow depolarization. In the latter neurones the effects of TEA on the DAP were evaluated during the initial phase of TEA action. During this period of application, TEA was without any significant effect on the resting membrane potential. The above sequence of effects on spike train after-potentials was also observed in 2 of 8 neurones examined for the effects of ChTX. Thus, the effects of ChTX also were evaluated during the initial phase of its action. In the remaining six neurones the resting membrane potential essentially remained unaffected by ChTX. The same was true for 14 of 15 neurones examined for the effects of apamin. In the remaining neurone, a slow depolarization of about 5 mV occurred. No significant effects on the resting membrane potential were observed during the application of IbTX, MgTX and CoCl2.
Immunohistochemical demonstration of Kv1.3 α-subunits
The immunohistochemical investigation of rat brain sections with an antibody directed against the Kv1.3 α-subunit of voltage-gated potassium channels did not result in specific staining of the somatodendritic region of SON neurones (Fig. 7A and B). In some cases, a few weak reaction products were seen that were most probably located at presynaptic terminal sites. In contrast, we observed distinct staining of neuronal perikarya and processes in many brain regions including the cerebral cortex and the hippocampal dentate gyrus. In the latter, granular cells and interneurones were labelled (Fig. 7C and D), indicating the presence of Kv1.3 α-subunits. Pre-incubation of the antibody with the appropriate antigen abolished the immunoreaction, indicating the specificity of the reaction. These results showing the localization of Kv1.3 channel subunits in the cerebral cortex and the hippocampus are in accordance with the results of studies using immunohistochemical techniques to locate Kv1 channel subunits (Veh et al. 1995) or in situ hybridization to detect the regional distribution of subunit-specific messenger RNAs (Kues & Wunder, 1992).
Spike trains induced by constant current injection in magnocellular neurones of the supraoptic nucleus are followed by a sequence of after-potentials. A fast AHP could be observed in all neurones examined in the present study. The AHP was demonstrated to be induced by an activation of calcium-dependent potassium channels (Andrew & Dudek, 1984b; Bourque et al. 1985). In a subpopulation of SON neurones, the fast AHP is followed by a DAP (Andrew & Dudek, 1984a), which has also been demonstrated to be calcium dependent (Andrew 1987; Bourque, 1986; Li & Hatton, 1997a,b). In the present study a slow AHP was observed in a subpopulation of supraoptic neurones during the control period. This slow AHP was induced in addition to the fast AHP when a train of action potentials was evoked by a depolarizing current pulse. Similarly, it was reported for a minority of supraoptic neurones recorded from hypothalamic explants that the decay of the AHP can be better fitted with two time constants (Armstrong et al. 1994). The calcium dependence of the fast AHP and of the DAP was confirmed in the present study by the blocking action of cadmium. The slow AHP is most probably also induced by a calcium-dependent mechanism, since (1) it was not generated when the fast AHP and the DAP were blocked by Cd2+, and (2) the expression of the slow AHP depended on the number of spikes in a train of action potentials, as has been reported for the fast AHP (Kirkpatrick & Bourque, 1996). The results of the present study also confirm that the fast component of the spike train AHP in supraoptic magnocellular neurones is blocked by apamin (Bourque & Brown, 1987; Armstrong et al. 1994; Kirkpatrick & Bourque, 1996), suggesting the involvement of calcium-activated potassium channels of the SK type in the generation of the fast AHP. As demonstrated in the present study, the slow AHP in supraoptic magnocellular neurones of the rat is blocked by ChTX. This result suggests that the slow AHP is induced by an activation of ChTX-sensitive potassium channels. For the characterization of the type of membrane channel contributing to the induction of the slow AHP, the spectrum of action of ChTX has to be considered. ChTX blocks calcium-activated potassium channels of the BK type (Miller et al. 1985). However, ChTX does not selectively act on BK channels; it additionally acts on calcium-activated potassium channels of the IK type (Reinhart et al. 1989). In addition, ChTX has been demonstrated to act on voltage-dependent potassium channels containing α-subunits of the Kv1.3 type (Grissmer et al. 1994).
Due to their slow inactivation and slow recovery, delayed rectifier channels containing the Kv1.3 α-subunit may be involved in the regulation of excitability and switching between tonic and phasic firing (Turrigiano et al. 1996). However, the involvement of voltage-gated channels containing Kv1.3-type α-subunits in the generation of the ChTX-sensitive slow component of the AHP in supraoptic neurones is improbable, since the application of the Kv1.3-selective scorpion venom peptide MgTX did not exert any effect on the fast and slow AHPs or on the DAP. This assumption is further confirmed if a previous report that MgTX is the reversible ligand with highest affinity for membrane-bound ion channels (Knaus et al. 1995) is taken into consideration. Furthermore, the apparent lack of Kv1.3 immunoreaction in the somatodendritic region of supraoptic neurones makes the involvement of these channels in the action of ChTX unlikely. In contrast to supraoptic neurones, granular neurones and interneurones of the dentate gyrus were shown to contain Kv1.3 α-subunits, a result that is in accordance with the demonstration of the subunit genes in these neurones by in situ hybridization (Kues & Wunder, 1992). Thus, both electrophysiological and immunohistochemical results provide evidence for the lack of Kv1.3 α-subunits in the neurones examined. These results suggest that the action of ChTX on the slow AHP in magnocellular supraoptic neurones is most probably due to its interaction with calcium-dependent potassium channels.
In isolated supraoptic neurones, a calcium-activated potassium current was reported to be blocked by 2 mm Co2+ added to the superfusion medium (Cobbet et al. 1989). Our observation that the effects on the slow AHP induced by 1 mm Co2+ added to the ACSF closely resembled the effects of ChTX on the slow AHP (i.e. the slow AHP was blocked and replaced by a DAP) again favours an involvement of calcium-activated potassium channels in the induction of the slow AHP. The results of the present study suggest that the slow AHP has a low sensitivity to calcium compared with the fast AHP and the DAP (e.g. in Fig. 2). The selective blocking action of Co2+ on the slow AHP may be due to a moderate reduction of Ca2+ influx, resulting from a partial block of Ca2+ channels by Co2+ added to ACSF containing a normal Ca2+ concentration (Andrew, 1987). Our observation that the pattern of TEA-induced effects on the time course of the after-potentials also closely resembled those of ChTX and Co2+ suggests that either BK or IK channels may be involved, since it has been reported that these channels are sensitive to both ChTX and low concentrations of TEA (McManus, 1991).
The action of IbTX on the repolarization of calcium spikes in supraoptic neurones (Stern & Armstrong, 1997) suggests the presence of BK channels in supraoptic neurones. However, BK channels are most probably not involved in the induction of the slow AHP since IbTX failed to exert any effect on the sequence of spike train after-potentials. Moreover, the voltage dependence and the kinetic features of BK channels demonstrated in hippocampal neurones (Yoshida et al. 1991) suggest that BK channels are unlikely to be involved in the induction of the slow AHP in supraoptic neurones. In hippocampal neurones, BK channels are activated during an action potential and the deactivation time course is consistent with the fast spike AHP. Further confirmation of this suggestion comes from the demonstration that binding of ChTX to synaptic plasma membranes from rat brain is not affected by IbTX (Vázques et al. 1990). Thus, the effects of ChTX on spike train after-potentials are most probably due to its antagonistic action on calcium-activated potassium channels of the IK type, resulting in an attenuation of the slow AHP.
In a subpopulation of supraoptic neurones characterized by a sequence of after-potentials composed of a fast AHP and a DAP, the application of ChTX resulted in an enhancement of the DAP amplitude. Considering the specificity of ChTX, this result suggests that a slow AHP is also present in these neurones. In addition, the observation that, in all neurones examined, ChTX unmasked a DAP not present during the control period confirms this suggestion. The findings that the application of Co2+ and TEA also resulted in unmasking of a DAP or in an enhancement of the DAP amplitude in almost all neurones examined are also in accordance with the latter results for ChTX. This profile of the pharmacological sensitivities of the slow AHP is consistent with the assumption that IK channels were present in all magnocellular supraoptic neurones examined.
In order to interpret the TEA-induced effects, two mechanisms must be considered. First, due to a delay in spike repolarization the entry of Ca2+ may have been increased, resulting in a stronger activation of calcium-dependent channels. Second, since the IK type of calcium-activated potassium channel is as sensitive to low concentrations of TEA as the BK type (McManus, 1991), the application of TEA may have resulted in a block of the slow AHP. Indeed, both mechanisms are likely to be involved, since (1) the amplitude of the fast AHP was increased by low concentrations of TEA in all neurones examined, and (2) the slow AHP was blocked by TEA in spite of an increased calcium influx.
The ramp-shaped decay of the slow AHP observed in some neurones (e.g. Fig. 1C) suggests a possible contribution by a delay current to the slow AHP. An A-current could have been activated during the decay of the fast AHP. However, the contribution of an A-current to the slow decay of the AHP is improbable, since A-current-induced ramps in supraoptic neurones recorded under the same experimental conditions as in the present study were not blocked by 1 mm CoCl2 and the time constant of these ramps was very different from that of the slow AHP (W. Greffrath & G. Boehmer, unpublished observations). The time constant of 4-aminopyridine-sensitive ramps observed after relaxation from hyperpolarization was 64.7 ± 30.53 ms (n= 6). This time constant is small compared with that of the slow AHP, which was 1010.3 ± 311.30 ms.
The functional relevance of the results presented in this study is most obvious in neurones expressing a DAP during the control period or when the slow AHP was blocked. In the former neurones application of ChTX, TEA, or CoCl2 induced an enhancement of the DAP amplitude or even a transformation of a DAP to an active plateau potential, while in the latter neurones DAPs not observed during the control period were unmasked. These effects of TEA, ChTX and CoCl2 on the late phase of the sequence of after-potentials suggest that most if not all magnocellular supraoptic neurones examined in the present study share the capability of generating a DAP. Although the neurones examined in the present study were not immunohistochemically characterized, it may be assumed that the capability of generating a DAP is not restricted to either of the main populations of magnocellular neurones, i.e. to vasopressin or oxytocin neurones. This assumption is in accordance with observations in immunohistochemically identified magnocellular neurones (Armstrong et al. 1994). In this study it was demonstrated that DAPs are expressed in a majority of vasopressin neurones but also in a minority of oxytocin neurones. Hence, as suggested by the results of the present study, the expression of the DAP primarily depends on the expression of the slow AHP. This assumption further suggests that the temporal coincidence of the DAP and the slow AHP, as well as the resulting interaction of these mechanisms, may be important for the regulation of excitability in supraoptic magnocellular neurones. The activation of mechanisms inducing the slow AHP (most probably IK channels) simultaneous with mechanisms inducing the DAP may result in attenuating or even shunting the DAP. The results of recent studies demonstrate an important function of the DAP in the generation of rhythmic phasic discharge by magnocellular supraoptic neurones (Andrew & Dudek, 1984a,b; Bourque et al. 1985; Andrew, 1987; Armstrong et al. 1994; Li et al. 1995; for review, see Armstrong, 1995). In this context, the results of the present study may be of functional importance. These results suggest that the temporal conditions for rhythmic discharge, i.e. the induction, maintenance and termination of a plateau potential or of a burst of action potentials, are regulated by the balance between the DAP and the slow AHP. The balance between the DAP and the slow AHP on the other hand may be regulated by differing sensitivities to calcium of the ionic channels involved. The observed sequence of the blockade of after-potentials by Cd2+, as well as the suppression of the active DAP by increasing the number of action potentials during a train, indicates that the DAP is more sensitive to an increase in the intracellular calcium concentration than the fast and slow AHPs. Results of recent studies demonstrate that the intracellular concentration of calcium is regulated by an activity-dependent activation of voltage-gated calcium channels (Fisher & Bourque, 1995), as well as by a release of calcium from intracellular stores (Li & Hatton, 1997a). The latter source of calcium is probably of great importance for the generation of DAPs by a calcium-activated reduction in potassium currents (Li & Hatton, 1997b). The functional importance of the control of the calcium concentration is further supported by the observation that an increase in calcium buffering by an injection of the calcium binding protein calbindin results in suppression of DAPs and rhythmic phasic discharge in all phasically discharging neurones examined (Li et al. 1995). In contrast, the introduction of anti-calbindin antiserum unmasked DAPs and converted continuous into phasic discharge in all continuously firing neurones examined. Our observation that all neurones examined share the capability of expressing a DAP, as indicated by the unmasking action of ChTX, TEA and Co2+, is in accordance with the latter results. Consequently, taking into account the lower Ca2+ sensitivity of the slow AHP compared with the DAP, the contribution of the slow AHP to the regulation of phasic discharge by reduced calcium buffering may be considered. Furthermore, since the type of after-potentials generated during control conditions was obviously independent of the individual resting membrane potential it may be speculated that the balance between the DAP and the slow AHP primarily depends on the expression of calcium buffering systems, e.g. of calbindin (Li et al. 1995).
Little is known about the properties of IK channels, the activation of which most probably results in the induction of the slow AHP. However, the calcium sensitivity of IK channels seems to be comparable to that of BK channels (McManus, 1991), whereas the calcium sensitivity of SK channels, the activation of which results in the induction of the fast AHP, is tenfold higher than that of BK channels (Rudy, 1988). Blocking SK channels by an application of apamin has also been reported to enhance or unmask DAPs evoked by trains of action potentials (Bourque & Brown, 1987; Armstrong et al. 1994; Kirkpatrick & Bourque, 1996). This effect was also observed in about 50 % of magnocellular neurones examined in the present study, a frequency that is in accordance with that reported by others (Armstrong et al. 1994). However, since apamin not only blocks the fast AHP but also blocks spike frequency adaptation in magnocellular neurones (Bourque & Brown, 1987), this effect may be due to the concomitant increase in calcium influx, resulting in a stronger activation of mechanisms involved in the generation of the DAP. Interestingly, unmasking of a small DAP during the action of apamin was observed in vasopressin neurones but not in oxytocin neurones (Armstrong et al. 1994). These results suggest that the expression of the DAP may be controlled not only by the balance between the slow AHP and the late phase of the DAP but also by the interaction between the fast AHP and the early phase of the DAP. Thus, a delicate homeostasis of the intracellular calcium concentration seems to be essential not only for the balance between the DAP and the slow AHP but also for the balance between the DAP and the fast AHP. However, the latter interaction may be of minor functional importance, since it was demonstrated that apamin resulted in an earlier onset of the DAP but not in an increase in the DAP amplitude (Armstrong et al. 1994).
In conclusion, the effects of the potassium channel blockers ChTX, IbTX and MgTX on the sequence of after-potentials in supraoptic magnocellular neurones suggest that the after-hyperpolarization in these neurones is composed of an apamin-sensitive fast AHP and a ChTX-sensitive slow AHP. The latter is induced by an activation of calcium-dependent potassium channels of the IK type, while the former is induced by an activation of SK channels. The lack of effect of IbTX in magnocellular neurones indicates that the contribution of BK channels to the generation of the slow AHP is at least of low importance. The same holds true for ChTX-sensitive voltage-gated potassium channels containing the Kv1.3 α-subunit, since MgTX exerted no effects on the sequence of after-potentials in the magnocellular neurones examined. The latter assumption is confirmed by the negative result of the immunohistochemical test for Kv1.3 α-subunits in the somatodendritic region of supraoptic neurones. The interaction of the slow AHP with the DAP may be of functional importance for the regulation of excitability and the generation of phasic discharge in magnocellular neurones. The involvement of the slow AHP in the regulation of phasic discharge by magnocellular supraoptic neurones remains to be demonstrated.
We thank Mrs U. Disque-Kaiser for her excellent technical assistance. This work was supported by the Naturwissenschaftlich-Medizinisches Forschungszentrum (NMFZ, Rheinland-Pfalz, Germany). Parts of the results are the subject of a doctoral thesis (W.G.).