Author's present address F. Pouille: Brain Research Institute, University of Zurich, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland.
1The role of Ca2+ entry in determining the electrical properties of cerebellar Purkinje cell (PC) dendrites and somata was investigated in cerebellar slice cultures. Immunohistofluorescence demonstrated the presence of at least three distinct types of Ca2+ channel proteins in PCs: the α1A subunit (P/Q type Ca2+ channel), the α1G subunit (T type) and the α1E subunit (R type).
2In PC dendrites, the response started in 66 % of cases with a slow depolarization (50 ± 15 ms) triggering one or two fast (∼1 ms) action potentials (APs). The slow depolarization was identified as a low-threshold non-P/Q Ca2+ AP initiated, most probably, in the dendrites. In 16 % of cases, this response propagated to the soma to elicit an initial burst of fast APs.
3Somatic recordings revealed three modes of discharge. In mode 1, PCs display a single or a short burst of fast APs. In contrast, PCs fire repetitively in mode 2 and 3, with a sustained discharge of APs in mode 2, and bursts of APs in mode 3. Removal of external Ca2+ or bath applications of a membrane-permeable Ca2+ chelator abolished repetitive firing.
4Tetraethylammonium (TEA) prolonged dendritic and somatic fast APs by a depolarizing plateau sensitive to Cd2+ and to ω-conotoxin MVII C or ω-agatoxin TK. Therefore, the role of Ca2+ channels in determining somatic PC firing has been investigated. Cd2+ or P/Q type Ca2+ channel-specific toxins reduced the duration of the discharge and occasionallyinduced the appearance of oscillations in the membrane potential associated with bursts of APs.
5In summary, we demonstrate that Ca2+ entry through low-voltage gated Ca2+ channels, not yet identified, underlies a dendritic AP rarelyeliciting a somatic burst of APs whereas Ca2+ entry through P/Q type Ca2+ channels allowed a repetitive firing mainly by inducing a Ca2+-dependent hyperpolarization.
Synaptic activity in response to transmitter release on dendritic and somatic post-synaptic sites is generally transduced into trains of action potentials (APs) at the axon hillock. The firing pattern coding the input-output relationship in a given neuronal population is determined mainly by the intrinsic properties of each neuron, based on sets of voltage-dependent channels with specific membrane distributions.
Furthermore, the axo-somatic compartment communicates with the dendrites via back-propagating Na+ APs, which modulate the impact of synaptic inputs on the dendritic membrane potential (see Magee et al. 1998). Whether or not the dendrites communicate actively with the axo-somatic compartment by propagating APs, such as Ca2+ APs probably initiated in the dendrites (see Yuste & Tank, 1996) is not yet clear, nor is the nature of the channels involved.
Purkinje cells (PCs) of the cerebellar cortex represent a unique model to study the role of ionic channels in determining the firing pattern generated by various synaptic inputs as well as, owing to their large dendritic arborization, the electrophysiology of the dendro-somatic interactions.
These neurons occupy a central position in the cerebellar circuitry: they integrate excitatory post-synaptic potentials from climbing and parallel fibres into, respectively, ‘complex’ or ‘simple’ spikes (Eccles et al. 1967). Furthermore, PCs display spontaneous and rhythmic activity characterized by prolonged periods of bursting followed by periods of electrical quiescence (Llinás & Sugimori, 1980a). These neurons express a large number of voltage-gated ionic channels and based on the results of voltage-clamp studies of PCs ionic conductances, a computational model of PCs has been developed. This reproduces many features of the discharge pattern of PCs in response to current injections (De Schutter & Bower, 1994a) and generates appropriate responses to climbing and parallel fibre activation (De Schutter & Bower, 1994b).
A compartmental model of PCs with Na+ APs restricted to the soma and Ca2+ APs to the dendrites had been proposed as early as 1980 (Llinás & Sugimori, 1980a,b). Na+ APs have been shown to be initiated in the PC axon close to the soma and to spread passively into the dendrites where the density of Na+ channels decreases with the distance from the soma (Stuart & Häusser, 1994). Ca2+ APs are generated in dendrites (Llinás & Sugimori, 1980b). Among voltage-gated neuronal Ca2+ channels, PCs in organotypic slice cultures as well as PCs in acute slices (Usowicz et al. 1992) expressed P/Q type Ca2+ channels and a high density of low-threshold transient type Ca2+ channels mainly in the dendrites (Mouginot et al. 1997). We describe here the role of these Ca2+ conductances in determining the dendritic and somatic firing behaviour of PCs in this culture model. This preparation appears to be an advantageous model to characterize the interactions between ionic currents in relation to the establishment of somatic and dendritic firing patterns. Somatic and dendritic compartments are accessible for patch-clamp recordings to determine the localization and the biophysical properties of various channels types. The relation between the firing pattern and a given channel type could be easily determined by using specific pharmacological and/or molecular tools. Some of the present results have been published in a preliminary communication (Bossu et al. 1998).
Organotypic cerebellar cultures were prepared from rats using the roller tube technique as described previously by Gähwiler (1981). Briefly, cerebella were removed aseptically from newborn rats (0–1 day old) that had been killed by decapitation. Parasagittal slices of 425 μm thickness were cut using a McIlwain tissue chopper. Individual slices were attached to glass coverslips in a film of clotted chicken plasma (Cocalico, Reamstown, PA, USA) and placed in culture tubes containing 750 μl of culture medium (50 % Eagle's basal medium, 25 % balanced salt solution containing either Hanks' or Earle's salts, 25 % heat-inactivated horse serum, 33.3 mm D-glucose and 1 mm glutamine). The tubes were placed in a roller drum inside an incubator at 36°C. Cultures were fed once a week with fresh culture medium. Media were purchased from Life Technologies, Cergy Pontoise, France.
After a period of at least 3 weeks in culture, cerebellar slices were transferred to a recording chamber mounted on the stage of a Nikon Optiphot2 microscope (Tokyo, Japan) equipped with differential interference contrast optics. The slice cultures were perfused with an external solution containing (mm): NaCl, 130; KCl, 2; CaCl2, 2.8; MgCl2, 2 (or CaCl2, 0; MgCl2, 5), Hepes/Tris, 10; glucose, 5.6. 6-Cyano-7-nitroquinoxaline-2,3-dione (CNQX) and bicuculline were added in the external solution at 10−5 M to abolish excitatory and inhibitory synaptic activities. The pH was adjusted to 7.4 with Tris/OH.
Recordings were carried out under current clamp in the whole-cell recording (WCR) configuration using an Axopatch 200A amplifier (Axon Instruments, Foster City, CA, USA). Purkinje cells were identified by their typical morphology (a cell body with a diameter of at least 20 μm and extensive dendritic arborization) and by their localization in the slice culture (peripheral fields of cultures).
Electrodes were pulled with a horizontal micropipette puller (BB-CH-PC, Mecanex, Nyon, Switzerland) from borosilicate glass capillaries (Clark Electromedical Instruments, Pangbourne, UK) and filled with a solution containing (mm): potassium gluconate, 132; EGTA-KOH, 1; MgCl2, 2; NaCl, 2; Hepes-KOH, 10; plus MgATP, 2 and GTP, 0.5 to prevent rundown of the P/Q type Ca2+ current. The pH was adjusted to 7.2 with TrisOH. Electrodes had a final tip resistance of 4 and 5 MΩ for somatic and dendritic recordings, respectively. Gigaohm seals between the pipette and the cell were obtained and pipette capacitance transients were minimized; the remaining capacitance transients had an amplitude of 50–100 pA. In some cases electrodes were coated with Sylgard to reduce the pipette capacitance transients to 0–25 pA.
After 2 min in the whole-cell recording configuration (mean series resistance of 54 ± 10 MΩ, n= 10 for dendritic recordings and of about 21 MΩ, see Table 1, for somatic recordings), the membrane potential was maintained at −80 mV by injecting a steady-state current ranging between 120 and 600 pA to remove inactivation of Ca2+ low-threshold transient type channels. Cells were depolarized or hyperpolarized by injecting steps of current (duration of 2–3 s) using a custom-built stimulator. The resulting voltage traces were digitized at 47.2 kHz using a digital data recorder (VR-10B, Instrutech, Great Neck, NY, USA) before storage on a Panasonic video recorder (Matsushita Electric Industrial, Osaka, Japan) for off-line analysis using the pCLAMP6 software (Axon Instruments).
Table 1. General electrophysiological properties of PCs stimulated and recorded from the soma: three main modes of discharge are depicted
Selected voltage traces were imported into Sigma plot 4.0 (SPSS, Erkrath, Germany) for illustration.
Bicuculline methiodide and tetrodotoxin (TTX) (Sigma, St Louis, USA) were prepared, respectively, as 10−2 M and 5 × 10−5 M stock solutions in distilled water, and CNQX (Tocris Cookson, Bristol, UK) as a 10−2 M stock solution in DMSO. BAPTA AM was prepared as a 5 × 10−3 M stock solution in DMSO. ω-Conotoxin-MVIIC (Latoxan, Rosans, France) and ω-agatoxin TK (Alomone, Jerusalem, Israel) were prepared, respectively, as 10−4 M and 2 × 10−5 M stock solutions in distilled water and stored at −20°C for up to 2 weeks. Drugs were diluted to their final concentration in the external solution for perfusion into the recording chamber.
Organotypic cerebellar slice cultures were fixed overnight in 4 % paraformaldehyde, washed with 0.3 % BSA in PB (phosphate buffer) and finally blocked and permeabilized in PB containing 0.1 % Triton X-100 and 2 % goat serum. The cultures were incubated overnight at 4°C in PB containing 0.1 % Triton X-100, 2 % goat serum and rabbit polyclonal antibodies raised against the α1G subtype (Craig et al. 1999) diluted at 1/40 or against the α1E subtype (Alomone, Jerusalem, Israel or Volsen et al. 1995) diluted at 1/50, or against the α1A subtype (Alomone) diluted at 1/50. After extensive washing, the cultures were incubated for 24 h at 4°C in the presence of an Alexa goat anti-rabbit IgG conjugate (Molecular Probes, Oregon, USA) diluted 1/400 as secondary antibody. The cultures were coverslipped in Vectashield (Vector, Burlingam, USA) and were examined and photographed using a Zeiss Axioskop fluorescence microscope (Oberkochen, Germany). A negative control was achieved by omitting the anti-α1 primary antibody in the incubation medium.
Dendritic evoked electrical activity
The evoked electrical activity was recorded after 3–4 weeks in vitro from 73 PC dendrites (see Fig. 1E) in control conditions. In 66 % of the recordings (Fig. 1) the evoked dendritic AP discharge started with a slow transient depolarization (duration 50 ± 15 ms, n= 37, amplitude ∼ 5 mV, see Fig. 1A–C) which triggered one (Fig. 1A) or several (Fig. 1B and C) fast (1–2 ms) APs. In 27 % of the cases, this was followed by a sustained discharge of fast APs (Fig. 1B), the frequency of which increased with the intensity of the injected current up to 60 Hz (n= 5, not illustrated). Bursting activity of slow transient depolarizations was observed in a few cases (6 %, see Fig. 1C). In the remaining case, only a sustained discharge of fast APs was evoked (Fig. 1D) which also increased in frequency with the intensity of the injected current (up to 80 Hz, n= 3, not illustrated).
The next series of experiments were performed to identify the initial slow depolarization that typified the dendritic response.
The slow transient depolarization is a low-threshold Ca2+ spike
The ionic basis as well as the nature of the channel underlying the initial slow transient depolarization typically recorded from the dendrite were determined. In the presence of TTX, fast APs were blocked and the slow transient depolarization could be recorded in isolation (Fig. 2A, n = 8). However, this slow initial depolarization was reversibly abolished by removal of the external Ca2+ (Fig. 2B, n = 3), whereas fast APs were maintained (not illustrated). Thus, fast APs are Na+ dependent, TTX-sensitive APs, whereas the slow initial transient depolarizations are Ca2+ dependent. To assign an already identified Ca2+ channel to this slow transient depolarization, the distinct pharmacological and biophysical profiles of P/Q type and low-threshold transient type Ca2+ channels (see Mouginot et al. 1997) were used as identification criteria. The slow transient depolarization was insensitive to bath application of 2 × 10−7 M ω-agatoxin TK (Fig. 2C, n = 2), a selective peptide antagonist of P/Q type Ca2+ channels. Similarly, the slow transient depolarization could still be evoked after treatment of the cerebellar slice cultures for 15–30 min with either this toxin (2 × 10−7 M, n= 14, not illustrated) or with ω-conotoxin-MVIIC at 5 × 10−6 M (Fig. 2D, n = 2), another peptide that blocks P/Q type Ca2+ channels (see Vilchis et al. 2000 for references concerning the use and specificity of ω-agatoxin TK and ω-conotoxin-MVIIC). Accordingly, the slow Ca2+-dependent transient depolarization is not due to the activation of P/Q type Ca2+ channels. Furthermore, cadmium (Cd2+), bath applied at 10−5 M, slightly reduced (by about 30 %) the duration of the initial slow transient depolarization (not illustrated, n= 2). The voltage sensitivity of the activation and inactivation of the slow transient Ca2+-dependent depolarization was determined (Fig. 3). In the presence of TTX, the slow transient depolarization was triggered when membrane potential reached −50 mV (mean value −53 ± 6 mV, n= 7, see Fig. 3A). Furthermore, these experiments demonstrated the all-or-none nature of this depolarization. Indeed, at threshold the slow transient depolarization had a maximal amplitude, and increasing the intensity of the current only reduced the latency of onset. From these observations we can conclude the following: first, that the slow transient Ca2+-dependent depolarization is an AP, and second, that it exhibits a low threshold for initiation. The amplitudes of APs evoked from various holding potentials were measured and plotted as a function of membrane potential (Fig. 3B). The membrane potential dependency of inactivation varied among slow APs. Membrane potentials where inactivation was observed ranged between −90 and −30 mV and 50 % of inactivation was observed at −65 ± 10 mV (n= 6).
TEA application reveals P/Q type Ca2+ channel induction of a Ca2+ component in fast dendritic APs
To reveal the Ca2+ component of dendritic fast APs we applied TEA (5 × 10−3 M) to the external solution (see Hagiwara, 1981). TEA decreased the firing frequency and prolonged the duration of APs due to a progressive blockade of K+ conductances (Fig. 4, upper panel). After 2–3 min of TEA, APs were triggered with a low frequency and were followed by a long-duration plateau component (see the left trace in Fig. 4, lower panel). This plateau component was suppressed by external Ca2+ removal (data not shown) or by bath application of ω-agatoxin TK (2 × 10−7 M) that blocks P/Q type Ca2+ channels (Fig. 4 lower panel, n= 3). From these observations we can conclude that P/Q type Ca2+ channel activation could produce a sustained depolarization of the membrane.
Dendritic initiation of the non-P/Q type Ca2+ spike
The somatic or dendritic initiation site of the non-P/Q type Ca2+ spike and its active propagation in the dendrite were investigated by comparing the evolution of the amplitude of the Na+ spike (Fig. 5A) and the non-P/Q type Ca2+ spike (Fig. 5B) with respect to the distance of the dendritic recording site from the soma. The amplitude of the Na+ spike (Fig. 5A left panel) decreased from 54 ± 5 mV (n= 20) at the soma to 18 ± 11 mV (n= 11) at the dendrite at 140–150 μm from the soma. The Na+ spike of PCs in slice cultures and in acute cerebellar slices behaves similarly. Indeed, a decrease in dendritic Na+ spike amplitude with increasing distance from the soma has been described also for PCs in acute cerebellar slices where it is initiated in the axon close to the soma (Stuart & Häusser, 1994). In contrast, no relationship between Ca2+ spike amplitude (as measured in Fig. 5B, left panel) and distance from the soma (Fig. 5B, right panel) was detected (Fig. 5B, right panel). The Ca2+ spike amplitude varied from one cell to another (range 5–40 mV) and did not depend on the recording site, having a mean amplitude of 13 ± 8 mV (n= 13) when recorded at the soma and a mean amplitude of 14 ± 6 mV (n= 5) when recorded from the dendrite at 140–150 μm from the soma. To further test whether the Na+ and Ca2+ spikes have distinct behaviour, we performed successive recordings (2–3) on the same cell (n= 10). The Na+ spike was completely attenuated at 120 μm from the soma (Fig. 6A) whereas the Ca2+ spike had a larger amplitude (19.5 mV compared with 10 mV). Analysis of recordings from 10 PCs (Fig. 6B) confirmed that only Na+ spikes were attenuated in the dendrites. In nine cases out of ten, the amplitude of the Na+ spike decreased with distance from the soma but the amplitude of the Ca2+ spike increased in five cases out of ten (or did not change significantly). These data suggest that Na+ spikes are most probably initiated in the soma and back propagate passively in the dendrite whereas Ca2+ spikes may be initiated in the dendrites. To strengthen this hypothesis, we mechanically separated the dendrite from the soma prior to recording. In this case a Ca2+ spike could still be evoked intradendritically (Fig. 6A, right panel, n= 2).
Surprisingly, the multi-recordings revealed that in some cases the Ca2+ spike was not detected when recorded close to the soma (10–25 μm, Fig. 6B filled symbols, n= 3 out of 10) whereas it was elicited in dendrites. Consequently the Ca2+ spike may be initiated at the dendrite and its conduction to the soma may be blocked at proximal dendrites. To investigate this possibility we analysed those recordings displaying an evoked Ca2+ spike as a function of the distance of the recording site from the soma (Fig. 6C). Whereas the percentage of recordings with an initial Ca2+ spike (see Fig. 8) was low (16 %, see Table 1) at the somatic level or on proximal dendrites at 10–25 μm from the soma, it increased up to 80 % for dendro-somatic distances greater than 25 μm. This strongly suggests a dendritic origin of the Ca2+ spike, and that a block of conduction may occur at 10–25 μm from the soma.
Somatic evoked electrical activity
The next series of experiments were performed to establish the role of Ca2+ entry, low-threshold Ca2+ spike and P/Q type Ca2+ channels in determining the firing pattern of PCs recorded at the somatic level. For this purpose, we have first characterized extensively the evoked firing pattern recorded at the somatic level and second analysed the Ca2+ dependency of somatic electrogenesis.
Somatic current-clamp recordings were obtained from a population of 206 PCs and their responses to somatic square-pulse current injections were analysed. Under our experimental conditions, with the cells held at −80 mV, three modes of response were characterized on the basis of elicited firing pattern (Table 1). Mode 1 observed in 29 % of PCs shows a single AP or a short burst of APs (2–5) at the beginning of the current step, followed by a plateau of depolarization. In modes 2 and 3, PCs fire repetitively. Mode 2 was recorded in 41 % of PCs and consisted of a sustained and regular discharge of APs (with a mean frequency between 23 and 67 Hz, depending on the intensity of the depolarizing injected current) during the entire duration of the step. As the current intensity was further increased, however, we observed a decreased duration in spikes discharge (i.e. accommodation) due to the development of a plateau of depolarization that inactivates APs. In mode 3, cells fired at between 20 and 56 Hz, but a time-dependent decline of the rate of discharge (time-dependent adaptation) was observed and bursts of APs were elicited when the intensity of the current was further increased. These three modes of discharge were also observed when the evoked activity was elicited from a potential close to the resting membrane potential (in between −45 and −50 mV) with the following proportions: mode 1: 16 %, mode 2: 63 % and mode 3: 21 % (n= 19). The classification of the evoked responses into modes was justified by the fact that PCs can change their responses according to the experimental conditions. For example a PC with a mode 1 response fired repetitively (mode 2 or 3) when changing the holding potential from −80 mV to −50 mV (n= 8 out of 18) or after bath-application of 2 to 5 × 10−4 M TEA (n= 5 out of 9). A PC with a mode 2 response fired in mode 3 after bath application of 2 × 10−4 M TEA (n= 3 out of 3). And finally PCs with modes 2 or 3 responses switched to a mode 1 response when external Ca2+was removed (see Fig. 7).
Independently of the mode of discharge, the soma-evoked response of PCs started in 16 % of cases, with a short burst of APs at high frequency elicited by an initial transient depolarization with a mean amplitude of 17 mV and a mean duration of 27 ms as described in the dendrite (see Table 1, right column). In a few cases (n= 7 out of 206) bursting activity superimposed on slow transient depolarizations was also observed.
The basic electrophysiological properties of PCs have been determined and compared according to their modes of response (see Table 1). No relationship was detected between the mode of response and the resting membrane potential (about −45 mV) or the series resistance (about 21 MΩ) or the Na+ spike amplitude (about 65 mV) and duration (about 1 ms) or the fast after-hyperpolarization (AHP) amplitude (about 20 mV) that followed the spike.
The input resistance of the PC soma was determined using low amplitude depolarizing and hyperpolarizing pulses injected through the recording pipette (−200 to + 300 pA). The time course and amplitude of the membrane potential changes induced by the current pulses are presented in Table 1 (first column). The subthreshold response of PCs to current pulses exhibited time and voltage-dependent characteristics reflecting inward rectification in the hyperpolarizing, but also depolarizing directions. During the response, membrane potential deflection reached an early peak value (filled circle) and then decreased to a steady-state level (open circle). An overshoot of the membrane potential following the termination of the hyperpolarizing stimulus could also be observed. This overshoot was larger when cells were recorded at −50 mV and in some cases was able to trigger a burst of APs (not illustrated).
The amplitude of the changes in potential at the early peak and the plateau were used to determine the peak input resistance (Rpeak) and the plateau input resistance (Rplateau) from the current-voltage relationships (not illustrated). Comparison of the mean values (see Table 1) of the peak and plateau input resistances for PCs according to their mode of response revealed the following sequence: 96 and 77 MΩ, 143 and 110 MΩ, 177 and 145 MΩ in modes 1, 2 and 3, respectively.
Ionic basis of somatic PC firing pattern: role of Ca2+ entry
To investigate the role of Ca2+ entry in spike generation, we removed external Ca2+ and analysed the consequences on PC firing (Fig. 7). The effects produced by Ca2+ removal on mode 2 (Fig. 7A, n= 4) or mode 3 (Fig. 7B, n= 3) firing were similar. In both cases repetitive firing was abolished: only a single (in A) or a short burst (in B) of APs could be elicited at the onset of the step, after which the potential reached a plateau level as observed for a mode 1 response. After removal of external Ca2+ we observed that, for the same current injection, the response stabilized at a more depolarized potential (−40 mV in A and −5 mV in B, see arrows) when compared to the mean potential reached between APs during the firing in control conditions (−50 mV in A and −15 mV in B, see arrows). This effect was clearly seen when we compared responses to current injections of increasing intensity in control conditions and during perfusion with Ca2+-free solution (not illustrated). Whereas the resting input resistance was not affected (n= 2) or slightly affected (±10 % change, n= 5) during the perfusion with Ca2+-free medium, responses to subthreshold current injection in control conditions elicited APs whereas suprathreshold current injections induced cell firing with increased frequency and decreased duration because of inactivation of fast APs. This 48 ± 27 % (measured on 7 cells) increase in the level of depolarization induced by Ca2+ removal may result from an increase in membrane resistance, or from an inhibition of hyperpolarizing Ca2+-dependent conductances, due to the suppression of Ca2+ influx through Ca2+ channels. Figure 7C shows that a relative hyperpolarization sensitive to Ca2+ removal controls the firing of PCs. Indeed, in this case (Fig. 7C, left panel) current injections of 800 pA triggered a burst of APs under control conditions (thin traces), suppressed by Ca2+ removal (thick traces), with no change in the resistance (the potential reached at the end of the step is the same in both conditions). In addition, transient oscillations of potential preceded each burst and were also suppressed by external Ca2+removal (Fig. 7C, right panel).
We propose that each transient depolarization could be an attenuated Ca2+ spike, followed by a transient hyperpolarization due to the activation of Ca2+-dependent conductances. If the hyperpolarization is large enough, a fast AP is triggered because of recovery from inactivation of Na+ channels. This first AP elicits Ca2+ entry, amplifying the subsequent activation of these hyperpolarizing conductances. As a consequence, the membrane potential is maintained below the potential for inactivation of Na+ channels. Depending on the duration of the activation of these hyperpolarizing conductances, leading to what we call a hyperpolarizing envelope (indicated by a dotted line in Fig. 7C, right panel), either bursts or a sustained discharge of APs are evoked. To test the hypothesis of a role for Ca2+-dependent hyperpolarizing conductances in establishing repetitive firing we have studied the effects of bath application of the membrane-permeable Ca2+ chelator, BAPTA AM, on PC firing. As illustrated in Fig. 7D, the effects of external Ca2+ removal were reproduced by a bath application of BAPTA AM (n= 9, this effect was not associated with a consistent change in resting input resistance, and was seen in 5 cells firing in mode 2 and 4 cells firing in mode 3). Indeed, as with removal of external Ca2+, BAPTA abolished the repetitive firing of fast APs after 6 to 15 min. Furthermore, for the same current injection, the response stabilized at a more depolarized potential (−5 mV, see arrow) when compared to the mean potential reached between APs during the firing in control conditions (−20 mV, see arrow). As a consequence of this increase in depolarization during the perfusion of BAPTA AM of 61 ± 32 % (n= 9), injecting currents which were subthreshold in control conditions-elicited APs and injected suprathreshold current-induced cell fire with increased frequency (not illustrated). In some cases, however, before discharge was abolished by BAPTA AM application, oscillations of the membrane potential and bursting activity were induced instead of a sustained discharge of APs (n= 4, data not shown). Furthermore, during BAPTA AM application, we observed an 18 ± 8 % (n= 5) increase in AP duration associated with an apparent decrease in the fast AHP amplitude (data not quantified). This indicates that chelation of Ca2+ inhibits a K+ conductance participating in AP repolarization.
The dendritic low-threshold Ca2+ spike triggers an initial burst of APs at the somatic level
In 16 % of PCs, the response started with a short burst of three to six fast APs riding on a slow transient depolarization (Table 1) that we identified as the low-threshold Ca2+ spike, previously characterized in the dendrites. Thus, the slow initial depolarization and the associated burst of fast APs were suppressed by removal of external Ca2+ (Fig. 8A, n= 4) and were inactivated when holding the membrane potential at −40 mV (Fig. 8B, n= 5). External application of TTX (5 × 10−7 M) blocked fast APs and the slow initial depolarization could be recorded in isolation (Fig. 8C, n= 4). Furthermore, we demonstrated that the initial Ca2+ spike was recorded in the presence of toxins that block P/Q type Ca2+ channels (data not shown) and was relatively insensitive to Cd2+ applied at 5 × 10−5 M. Indeed, Cd2+ had no effect on the amplitude of the initial Ca2+ spike (n= 5), but in three cases Cd2+ reduced its duration by 30 ± 15 % (see an example in Fig. 8D).
TEA application reveals P/Q type Ca2+ channel induction of a Ca2+ component in fast somatic APs
As for dendritic recordings, a Ca2+ component of somatic fast APs was revealed after bath application of TEA (2–5 × 10−3 M). After 30 s to 2 min of TEA, APs were followed by a long duration plateau component (see Fig. 9, n= 21 out of 23) which was totally suppressed by external Ca2+ removal (n= 3, data not shown) or by application of Cd2+ (5 × 10−5 M, delay of 30 s to 4 min, Fig. 9A, n= 5). The plateau component was also suppressed with bath application of ω-agatoxin TK (2 × 10−7 M, delay of 6 to 10 min, Fig. 9B, n= 4).
This demonstrates that, as in the dendrite, P/Q type Ca2+ channels can underlie a depolarizing plateau if K+ conductances are blocked. We further demonstrate, that this plateau is also sensitive to Cd2+.
P/Q type Ca2+ conductances and PC firing pattern
To better establish the role of P/Q type Ca2+ channels in PC firing (mode 2 or 3) we have bath applied either 5 × 10−5 M Cd2+ or 5 × 10−6 M ω-conotoxin-MVII C or 2 × 10−7 M ω-agatoxin TK. Treatment by any one of these three abolished the TEA-induced Ca2+ component of the fast AP but did not suppress the low-threshold Ca2+ spike. Similarly to what we observed with external Ca2+ removal, Cd2+ produced an increase in firing frequency followed by a shortening of the discharge duration (Fig. 10A, upper panel, n= 7). During Cd2+ application, for the same current injection, the cell became more depolarized (from – 35 to −10 mV, in the case illustrated in Fig. 10A, upper panel). This effect started with a delay of 56 ± 38 s and was maximal (59 ± 19 % of increase) after 97 ± 65 s and was associated with an increase of the resting input resistance of 32 ± 15 %.
ω-Conotoxin-MVII C or ω-agatoxin TK were bath applied and the induced effects on mode 2 and mode 3 PC firing were examined. A block of P/Q type Ca2+ channels by the toxins also produced a progressive shortening of the discharge duration (Fig. 10A, lower panel, n= 12 out of 20). This effect was not associated with a consistent change of resting input resistance. In some cases (n= 2) a complete abolition of the repetitive firing was observed. The reduction of the duration of discharge was also linked to an increase in the plateau depolarization leading to inactivation of Na+ conductances driving APs. For example after 10 min of ω-conotoxin-MVII C application, the response stabilized at a plateau value of −30 mV (see arrow) compared to the mean potential of −45 mV (see arrow) observed between APs during the firing in control conditions (on average the potential increased by 25 ± 16 %). Furthermore, during Cd2+ or toxin application the AP duration increased by 12 ± 6 % (n= 5) and 14 ± 18 %, respectively, and the fast AHP amplitude decreased (data not quantified).
However, P/Q type Ca2+ channel blockade produced distinct effects on PC firing, when compared to those induced by external Ca2+ removal. First, whereas the instantaneous frequency was relatively constant during all steps for mode 2 cells in control conditions, a time-dependent adaptation in AP firing progressively occurred during toxin applications and was followed by a reduction of the frequency of discharge (see Fig. 10A, lower panel). Similar affects on the PCs firing were obtained (not illustrated, n= 3) when cells were recorded using an internal pipette solution with no ATP and GTP, a condition leading to progressive rundown of the P/Q type Ca2+ current (Mouginot et al. 1997).
In addition, during Cd2+ (Fig. 10B, upper panel, n= 5) or toxin (Fig. 10B, lower panel, n= 7) application, oscillations of the membrane potential (as indicated by asterisks) appeared on the depolarizing plateau. When the amplitude of these oscillations was sufficiently large, they induced APs and PCs displayed an irregular pattern of discharge (see Fig. 10C, lower panel, n= 7).
We conclude that Ca2+ entry via Cd2+-sensitive Ca2+ channels mostly mediated by P/Q type Ca2+ channels contributes to the increase in internal Ca2+ that most probably activates a hyperpolarizing conductance allowing PCs to fire repetitively in mode 2 or 3. In addition, P/Q type Ca2+ channels also control the firing frequency, the time-dependent firing adaptation, and the firing pattern of PCs. These effects are not described in detail in this report and the underlying mechanisms have to be further investigated.
α1 Subunit expression of Ca2+ channels in PCs from slice cultures
The cloning of the pore-forming α1 subunits corresponding to the different members of the voltage Ca2+ channels family (L, N, P/Q, R and T) has enabled the production of specific antibodies that can be used to determine the cellular and subcellular localization of each type of Ca2+ channel. We used specific antibodies against α1A (corresponding to P/Q type Ca2+ channels, Westenbroek et al. 1995), α1E (corresponding to the R type Ca2+ channel, Yokoyama et al. 1995), and α1G (corresponding to one subtype of the T type Ca2+ channels, Perez-Reyes et al. 1998) to determine their pattern of expression in PCs from slice cultures. We find that α1A (Fig. 11A) as well as α1G (Fig. 11B) and α1E (Fig. 11C) were expressed, not only over the dendrite (arrow heads), but also on the soma (arrows) of PCs in slice cultures. α1E (Fig. 11D) were also expressed on cerebellar granular cells, mainly on the soma but also on the neurites.
The dendritic low-threshold Ca2+ spike
We demonstrate the existence of a dendritic non-P/Q type Ca2+ spike with modulated dendro-somatic propagation. This represents new insights into the complex physiology of electrical and Ca2+ signalling in PCs. Patch-clamp experiments as well as optical imaging have shown that low-threshold transient type Ca2+ channels are present on the dendrites of hippocampal neurons (Karst et al. 1993; Christie et al. 1995; Magee & Johnston, 1995; Kavalali et al. 1997), thalamic relay cells (Munsch et al. 1997; Zhou et al. 1997) and cerebellar Purkinje cells (Bossu et al. 1989; Mouginot et al. 1997). Computational modelling techniques (Antal et al. 1997; Destexhe et al. 1998) demonstrate that a dendritic T current is needed to reproduce the low-threshold burst of thalamocortical neurons. However, such a dendritic action potential has not been characterized yet. Our study demonstrates that the activation of dendritic low-threshold transient type Ca2+ channels is able to generate a low-threshold spike (l.t.s.) in the dendrite of PCs and rarely in the soma. A l.t.s. has been observed at the somatic level of thalamic neurons (Deschênes et al. 1984; Jahnsen & Llinás, 1984), inferior olivary (Llinás & Yarom, 1981), and pontine reticular neurons (Greene et al. 1986). The kinetics and voltage dependency of inactivation of the l.t.s. are similar to those of the dendritic Ca2+ spike in PCs. In thalamic neurons, the somatic l.t.s. was attributed to T type current activation (Coulter et al. 1989) and most likely to the α1G subtype (see Huguenard, 1998; Talley et al. 1999). α1G Ca2+-channel protein was detected immunohistochemically in the dendrites of PCs in organotypic slice cultures and this subtype of T channels could be responsible for the Ca2+ spike. However, several observations indicate that more than one subtype of T channel and/or R type Ca2+channel could participate in spike generation. Firstly, the Ca2+ spike was observed to decay in some cases with a fast and a slow component (see Fig. 3A and B) indicating that at least two different channel populations are involved. Secondly, the duration of the low-threshold Ca2+ spike was affected by Cd2+ in some cases. Thirdly, the voltage-dependence range of inactivation varies from one spike to another (see Fig. 3). Finally, on PCs dendrites we detected an α1E immunoreactivity corresponding probably to the R type Ca2+ channel which is a transient current activated around −30 mV (see Randall & Tsien, 1997), and which consequently may contribute to the non P/Q type Ca2+ spike we describe here. The concept that more than one type of low-threshold transient type Ca2+ channel is involved in the generation of the AP is fundamental for the physiology of neurons, taking into account the fact that the different channel types and subtypes are differentially regulated. The composite nature of the PC Ca2+ spike requires new pharmacological tools to be further elucidated.
We provide indirect evidence that the non P/Q type, low-threshold Ca2+ spike of PCs initiated in dendrites rarely propagates to the soma. This observation supports the compartmental model proposed by the Llinás group (reviewed in Llinás & Sugimori, 1992, De Schutter, 1994). Ca2+ APs restricted to the distal parts of the apical dendrites have also been identified in neocortical pyramidal neurons (Schiller et al. 1997). Our study shows that block of propagation of the non-P/Q type, low-threshold Ca2+ spike occurs in a restricted zone of the proximal dendrite. Such inhibition of conduction may be not specific to PCs and consequently may explain why a somatic low-threshold Ca2+ spike is not detected in some neurons, such as hippocampal pyramidal cells, despite the presence of dendritic low-threshold transient type Ca2+ channels (Kavalali et al. 1997). We have preliminary observations indicating that P/Q type Ca2+ channels may play a role in this block of propagation although the exact ionic mechanism is not known.
Intradendritic recordings from PCs in acute slices have revealed the presence of two distinct types of Ca2+ conductances, a plateau-generating Ca2+ conductance and a spike-generating Ca2+ conductance (Llinás & Sugimori, 1980b). The Ca2+ spikes recorded in the presence of TTX have a slow decay (about 50 ms) and display also a faster component (see Fig. 5D in Llinás & Sugimori, 1980b). These characteristics are similar to the non-P/Q type Ca2+ spike we describe in PCs from slice cultures (compare with Fig. 3A and B). From our study, P/Q type Ca2+ channels are not able to induce a spike but could give rise to a plateau component of the AP when K+ channels are blocked by TEA. This suggests that P/Q type Ca2+ channels might be responsible for the Ca2+-dependent plateau-generating conductance described by Llinás & Sugimori (1980b).
Heterogeneity of somatic PC-elicited responses: a model based on the role of Ca2+ and Ca2+-dependent hyperpolarizing conductances
PCs in slice cultures display three modes of somatic discharge in response to positive current injection steps from a membrane potential of −80 mV: single APs, sustained discharge, and bursts. We propose a model in which a balance between depolarizing conductances and Ca2+-dependent hyperpolarizing conductances activated by Ca2+ entry through voltage-gated Ca2+ channels determines the firing mode of PCs. According to this model, a depolarizing conductance, such as a sustained Na+ conductance, is activated by a current injection, allowing the cell membrane potential to reach the activation threshold for the fast-inactivating Na+ conductance driving fast APs. In the absence of external Ca2+ or in the presence of internal Ca2+ chelator, the membrane potential rapidly reaches the potential for inactivation of fast-inactivating Na+ channels: only a single or a short burst of APs are elicited, which are then followed by a sustained plateau of depolarization (as we observed for mode 1 responses). In the presence of external Ca2+, Ca2+ entry during the first AP are thought to activate Ca2+-dependent hyperpolarizing conductances which maintain the membrane potential between the activation and inactivation potential thresholds of fast-inactivating Na+ channels: subsequent spikes can then be initiated. If the activation of these hyperpolarizing conductances is sustained, a discharge of APs is expected to occur during the entire step, as observed for the mode 2 response. For the mode 3 response, activation of these hyperpolarizing conductances may be transient, so that a burst of APs is elicited each time the membrane potential of the cell lies between the potential for activation and inactivation of Na+-fast conductances. In addition, rhythmic interactions between Ca2+ conductances and these hyperpolarising conductances may generate the repetitive burst firing pattern observed in mode 3. We have preliminary evidence (J. L. Bossu, unpublished observations) showing that an apamine-sensitive K+ conductance participates in the Ca2+-dependent hyperpolarization that controls PC firing.
Our model is in agreement with data obtained by Llinás & Sugimori (1980a) from PCs recorded in acute slices and with the active membrane model of PCs developed by De Schutter & Bower (1994a). In these models, the plateau potential represents an unstable equilibrium state involving at least three simultaneously occurring voltage-dependent conductances: a non-inactivating Na+ conductance, a non-inactivating K+ conductance, and a dominant Ca2+-dependent component. Similarly, it has been postulated that tonic and phasic neurons in bronchial ganglia represent a single population with different states of K+-channel activation or inactivation (Myers, 1998).
We propose that P/Q type Ca2+ channels are important for eliciting repetitive firing. Their activation is probably able to increase the axonal [Ca2+]i close to the site for fast AP initiation (Callewaert et al. 1996) where Ca2+-dependent hyperpolarizing conductances might be activated. Interestingly, during P/Q type Ca2+ channel blockade, a time-dependent adaptation in firing frequency was induced, indicating a modification in activity of Ca2+-dependent K+ channels responsible for time-dependent adaptation (Storm 1990, for a review). In neocortical neurons, P type Ca2+ channels were shown to be closely associated with K+ channels eliciting medium and slow AHPs (Pineda et al. 1998). Conversely to what we have observed in PCs from slices cultures, blockade of P type channels in neocortical neurons induced a reduction in spike-frequency adaptation. Finally, during P/Q type Ca2+ channel blockade, membrane oscillations were induced or amplified, allowing PCs to fire in a bursting mode. This effect, to be further investigated, may be relevant to the suppression of a P/Q type Ca2+ channel-mediated shunt of conduction that we suspect normally prevents propagation of low-threshold Ca2+ spikes between dendrites and soma of PCs in slice cultures.
It is not yet clear whether or not Ca2+ entry through low-threshold Ca2+ channels responsible for the Ca2+ spike activates hyperpolarizing conductances that control somatic PCs firing. Indeed, in our hands (data not illustrated in this report), amiloride, which affects the low-threshold Ca2+ current (Mouginot et al. 1997) in PCs from slice cultures, had variable effects on the low-threshold Ca2+ spike but efficiently abolished repetitive firing of PCs. On the other hand, mibefradil, which blocks the low-threshold Ca2+ current of isolated PCs, (McDonough & Bean, 1998) had no effect on the low-threshold Ca2+ spike and on firing (see also Raman & Bean, 1999). A large diversity of low-threshold Ca2+ channels emerges from recent molecular biology investigations (Perez-Reyes et al. 1999; Cribbs et al. 2000). Pharmacological tools allowing their electrophysiological characterization, as well as their physiological implications in neural electrogenesis, are not yet available.
PCs in organotypic culture present a heterogeneity in their AP firing pattern. We propose that this heterogeneity is linked to different levels of activation of some depolarizing and Ca2+-dependent hyperpolarizing conductances. Future studies will investigate how activation of these conductances is modulated under physiological conditions in order to modify the integrative properties of PCs. Ca2+-dependent hyperpolarization is produced by activation of several different types of Ca2+-activated K+ channels (Sah, 1996 for a review), which remains to be characterized, as well as their coupling to P/Q type Ca2+ channels and possibly to low- threshold Ca2+ channels.
PCs in organotypic cultures provide a model to study interactions between different populations of ionic channels in the control of cellular excitability, according to their relative density, their distribution, and the morphological development of a cell. It would be advantageous to investigate the interactions between different types of depolarizing Ca2+ conductances and different types of hyperpolarizing Ca2+-activated conductances in order to establish a firing pattern. Furthermore, increases of (Ca2+)i are required for both LTD at the parallel fibre input (see Daniel et al. 1998) and for activity-induced changes in inhibitory inputs (Llano et al. 1991; Kano et al. 1992). It would be of great interest to investigate the role of the dendritic low-threshold Ca2+ spike and P/Q type Ca2+ channels on these many Ca2+-dependent processes.
The authors thank Professors F. Crepel, B. Gähwiler and Drs Y. Bailly, A. Feltz, U. Gerber and A. Marty for their helpful comments and suggestions on this manuscript.