Single motoneurons and pairs of a presynaptic reticulospinal axon and a postsynaptic motoneuron were recorded in the isolated lamprey spinal cord, to investigate the role of calcium-dependent K+ channels (KCa) during the afterhyperpolarization following the action potential (AHP), and glutamatergic synaptic transmission on the dendritic level. The AHP consists of a fast phase due to transient K+ channels (fAHP) and a slower phase lasting 100–200 ms (sAHP), being the main determinant of spike frequency regulation. We now present evidence that the sAHP has two components. The larger part, around 80%, is abolished by superfusion of Cd2+ (blocker of voltage-dependent Ca2+ channels), by intracellular injection of 1,2-bis-(2-aminophenoxy)-ethane-N,N,N′,N′-tetraacetic acid (BAPTA; fast Ca2+ chelator), and by apamin (selective toxin for KCa channels of the SK subtype). While 80% of the sAHP is thus due to KCa channels, the remaining 20% is not mediated by Ca2+, either entering through voltage-dependent Ca2+ channels or released from intracellular Ca2+ stores. This Ca2+-independent sAHP component has a similar time course as the KCa portion and is not due to a Cl− conductance. It may be caused by Na+-activated K+ channels. Glutamatergic excitatory postsynaptic potentials (EPSPs) evoked by single reticulospinal axons give rise to a local Ca2+ increase in the postsynaptic dendrite, mediated in part by N-methyl-d-aspartate (NMDA) receptors. The Ca2+ levels remain elevated for several hundred milliseconds and could be expected to activate KCa channels. If so, this activation should cause a local conductance increase in the dendrite that would shunt EPSPs following the first EPSP in a spike train. We have tested this in reticulospinal/motoneuronal pairs, by stimulating the presynaptic axon with spike trains at different frequencies. We compared the first EPSP and the following EPSPs in the control and after blockade with apamin. No difference was observed in EPSP amplitude or shape before and after apamin, either in normal Ringer or in Mg2+-free Ringer removing the voltage-dependent block of NMDA receptors. In conclusion, the local Ca2+ entry during reticulospinal EPSPs does not cause an activation of KCa channels sufficient to affect the efficacy of synaptic transmission. Thus the integration of synaptic signals at the dendritic level in motoneurons appears simpler than would otherwise have been the case.
In cortical pyramidal cells and cerebellar Purkinje neurons, the dendritic trees with their spines participate extensively in the spatial and temporal integration of synaptic input (Takagi 2000). Rather than acting as passive cables, dendrites serve as sophisticated signal processors (Johnston et al. 2000). To appreciate the importance of dendritic processing, it should be studied in a functional context. Here we explore a network in the brain stem–spinal cord of the lamprey, which has provided extensive information on the cellular and molecular organization of central pattern generators (CPGs) for locomotion (see Grillner et al. 2000,2001). The output stage is the motoneuron that has a very extensive dendritic tree (Wallén et al. 1985) but no spines. It integrates phasic input from supraspinal, sensory, and spinal neurons, to generate the rhythmic muscle contractions that propel the animal through the water during swimming.
The synapse between giant reticulospinal (RS) axons and motoneurons has been used extensively in studies of synaptic transmission in the lamprey spinal cord (Brodin et al. 1994,2000; Grillner et al. 2000) since the pre- and postsynaptic neurons are accessible for simultaneous intracellular recordings. Their role in locomotion is well established (Brodin et al. 1988; Deliagina et al. 2000; Rovainen 1974; Ullén et al. 1998; Zelenin et al. 2000). Activation of this glutamatergic synapse leads to a local increase of Ca2+ in the dendrites of target motoneurons (Bacskai et al. 1995) that is due in part to synaptic activation of N-methyl-d-aspartate (NMDA) receptors. While the excitatory postsynaptic potential (EPSP) lasts for around 30 ms, the Ca2+ level remains elevated for 0.5–1.5 s (66% left at 0.5 s) (Bacskai et al. 1995). The same authors reported that during fictive locomotor activity the Ca2+ levels in motoneuronal dendrites also vary in phase with ventral root bursting. In this case the synaptic drive originates from glutamatergic network interneurons, active during the ipsilateral burst.
These observations raise the question of which effects the Ca2+ entry might exert on dendritic processing in the postsynaptic neuron. Ca2+-activated K+ channels (KCa) are present in spinal neurons in lamprey (El Manira et al. 1994; Hill et al. 1985, 1992;Matsushima et al. 1993; Meer and Buchanan 1992; Parker et al. 1996). They become activated by the Ca2+ entering through N- and P/Q-type channels during the action potential (Wikström and El Manira 1998). Most of these KCa channels belong to the SK subclass of KCa, since they are blocked by apamin (Grunnet et al. 2001; Hill et al. 1992; Meer and Buchanan 1992). They are, at least partially, responsible for the slow afterhyperpolarization (sAHP), which is the main determinant of spike frequency adaptation in these neurons (El Manira et al. 1994; Meer and Buchanan 1992; Wallén et al. 1989). At the concentration of apamin used in these studies, which should have been sufficient to fully block KCa channels, a complete blockade of the sAHP was not achieved, raising the possibility of another apamin-insensitive KCa component (Hill et al. 1992). The nature of the different components underlying the sAHP is therefore further investigated here.
The synaptically evoked Ca2+ entry might be expected to activate dendritic KCa channels located near the synapse. If so, the Ca2+increase, which outlasts the EPSP by several hundred milliseconds, should lead to a long-lasting potassium conductance increase. This in turn would be expected to locally shunt subsequent dendritic EPSPs, which thus would contribute less to the depolarization of the cell soma and initial segment than the first EPSP. This form of activity-dependent synaptic depression could have profound effects on the synaptic efficacy of weak and distally located EPSPs. To test this hypothesis we studied the effect of blocking the KCa channels expressed in motoneurons with apamin to investigate whether a KCa channel blockade affects the amplitude of the EPSPs during a train of EPSPs generated by presynaptic stimulation of single reticulospinal axons.
Our experiments employed a total of 35 adult lampreys (31Lampetra fluviatilis and 4 Ichthyomyzon unicuspis). As in previous studies, results did not differ between species (Brodin et al. 1990). The animals were anesthetized by immersion in tricaine methane sulfonate (MS-222, Sigma; 200 mg/l) and decapitated caudal to the gills. The spinal cord was dissected in cooled physiological solution. The Ringer solution used with Lampetra (in mM: 138 NaCl, 2.1 KCl, 1.8 CaCl2, 1.2 MgCl2, 4 glucose, 2 HEPES, and 0.5 l-glutamine) was bubbled with O2 and adjusted to pH 7.4 with NaOH. In the case of Ichthyomyzon, the solution had a slightly different composition (in mM: 91 NaCl, 2.1 KCl, 2.6 CaCl2, 1.8 MgCl2, 4 glucose, and 20 NaHCO3) and was buffered to pH 7.4 by bubbling with 95% O2-5% CO2.
The spinal cord preparations, as a rule between 10 and 20 segments long, were dissected from the region between the gills and the first dorsal fin, together with the dorsal half of the notochord as a mechanical support. After pinning them down in a silicone elastomer (Sylgard)–lined chamber, the meninges were peeled off from the dorsal surface of the cord. The preparations were perfused with cooled physiological solution (5–10°C). In some experiments (9 of 35 preparations) collagenase (0.5 mg/ml, Sigma) was applied to the whole spinal cord for 1–2 min to ease microelectrode penetration and stabilize recordings. In one control experiment, continuous perfusion of 2 mg/ml collagenase for 2 h did not alter the frequency of NMDA-induced fictive locomotion (Cohen and Wallén 1980; Grillner et al. 1981), indicating that the collagenase did not affect synaptic interaction within the spinal cord. With the aid of a stereoscopic microscope andtrans-illumination, individual larger cells and giant axons could be viewed. A glass suction electrode was placed on a ventral root (VR) to allow for identification of impaled motoneurons and for their antidromic activation (Fig.1 A). Cells were impaled with microelectrodes and motoneurons identified by recording a one-to-one correspondence between the spikes elicited in the soma and at the ventral root.
Apamin, BAPTA, and the sAHP in motoneurons
Microelectrodes were filled with 3 M potassium chloride (KCl). Action potentials were elicited by a positive current pulse, and the sAHP was monitored according to the procedure described below (seeresults). Apamin (Sigma) was dissolved in phosphate-buffered saline (PBS) and frozen in stock vials. Just before use, it was thawed and mixed with physiological solution to the desired final concentration. Bovine serum albumin (0.5 mg/ml) was used to prevent unspecific binding of the peptide. In cases where more than one apamin concentration was tested on a particular cell, lower concentrations preceded higher ones (range 0.2–20 μM). The initial control recording and each drug application lasted for a minimum of 30 min.
To test whether high concentrations of apamin could possibly exert a direct effect on Ca2+ channels, Ca2+ spikes were elicited in sensory dorsal cells by intracellular stimulation during perfusion of selective Na+ and K+ channel blockers [14 mM TEA replacing an equal concentration of NaCl; 500 μM 4-aminopyridine (4-AP), 0.5 μM TTX]. Apamin was used as described above. These spikes were abolished by the application of 200 μM cadmium chloride, a broad range Ca2+ channel blocker (Fig. 3 A4).
The fast Ca2+ chelator 1,2-bis-(2-aminophenoxy)-ethane-N,N,N′,N′-tetraacetic acid (BAPTA tetrapotassium salt, Sigma) was dissolved in 3 M KCl or potassium acetate (KAc), to a final concentration of 200 mM. BAPTA was administered into the cell by passive diffusion from the pipette tip. This was confirmed in all cases by a progressive depression of the sAHP, being especially rapid during the first 10–20 min (seeZhang and Krnjevic 1988).
Effect of apamin on synaptic transmission at RS axon-motoneuron synapses
Motoneurons were impaled with microelectrodes filled with 3 M KAc and 0.1 M KCl (final resistance 40–100 MΩ). After identifying a motoneuron, a second microelectrode, filled with 3 M KCl, was inserted in the cord rostrally and ipsilaterally to the first one until an axon was impaled that, if stimulated, evoked an excitatory monosynaptic response in the motoneuron (Fig. 1 A). A minimum distance of 6 mm (∼2 segments) was kept between microelectrodes to avoid perturbation of synaptic transmission with the presynaptic micropipette (Brodin et al. 1994). Criteria for accepting a pair were the following: axonal conduction velocity above 2 m/s identifying it as a large reticulospinal axon (Ohta and Grillner 1989;Rovainen 1978); the presence of a clear chemical component in the EPSP; and a one-to-one appearance of an EPSP on high-frequency presynaptic stimulation (>20 Hz). These axons were found just lateral to the cellular layer (Mauthner or PRRN neuron axons) or ventromedially (Müller neuron axons). In all cases the tip of the presynaptic microelectrode coincided with a large axon or group of axons clearly visible through the microscope.
Most of the pairs studied displayed an initial and long-lasting depression in synaptic efficacy at stimulation rates above 20 EPSPs/min (Fig. 1 B) (cf. Brodin et al. 1994). To ensure that this did not influence our results, the axonal stimulation protocol was allowed to run for 10–20 min until a steady EPSP amplitude was reached. Only then did we apply apamin (2 μM). To confirm that apamin blocked KCa channels, action potentials were elicited at regular intervals in the motoneurons to monitor the reduction of the sAHP.
Acquisition and data analysis
The electrical signal recorded from the VR was sent to a differential AC amplifier (A-M Systems). Intracellular signals were fed to an Axoclamp 2A amplifier (Axon Instruments) working in bridge mode and low-pass filtered at 1 kHz. Output from these amplifiers was acquired at a sampling rate of 10 kHz per channel on a PC equipped with a Digidata 1200 A/D converter, and running Clampex 6 software (both from Axon Instruments). Data were analyzed using custom scripts run under Axograph 4.6 software (Axon Instruments) on a Macintosh G4 computer. Student's t-test was used to compare data sets.
The following drugs were used: collagenase (Sigma), ZD7288 (Tocris), apamin (Sigma), PBS (Sigma), bovine serum albumin (BSA, Sigma), TEA (Sigma), 4-AP (Sigma), tetrodotoxin (TTX, Sigma), cadmium chloride (Sigma), nickel chloride (Merck), and BAPTA (Sigma).
Three types of afterpotentials follow the action potential in motoneurons
In all motoneurons recorded (n = 29), action potentials were followed by a fast afterhyperpolarization (fAHP), an intermediate afterdepolarization (iADP), and finally by a sAHP (Fig.2 A1). When the holding membrane potential was set at different levels by tonic current injection, these afterpotentials changed in amplitude accordingly. Fast and slow AHPs increased in amplitude when the neurons were depolarized, and vanished on hyperpolarization close to the equilibrium potential for K+. The fAHP is due to a high-threshold transient K+ current of the A-type, blocked by catechol (Hess and El Manira 2001), while the sAHP is sensitive to apamin (Hill et al. 1992) and cadmium ions (Fig. 2 A2).
The peak of the iADP increased at hyperpolarized potentials, but the iADP still appeared as a hump between the fAHP and the sAHP at depolarized potentials (above −62 mV; Fig. 2 A1). Since the resting membrane potential of motoneurons ranges between −70 and −75 mV, these cells display a sizeable iADP when not depolarized for instance by locomotor network input.
The iADP was still observed under the following conditions: during perfusion with cadmium ions (Fig. 2 A2, n = 20), which block high-voltage–activated (HVA) and some low-voltage–activated (LVA) Ca2+ channels; perfusion of nickel ions (which block LVA Ca2+channels) in combination with cadmium (Fig. 2 A2.1, n = 2); and intracellular injection of BAPTA (Fig.2 B). These data taken together suggest that the iADP is not due to a Ca2+ current or dependent on intracellular Ca2+ levels. TheI h blocker ZD7288 was also applied at 400 μM without any observable effect on the iADP (n = 2). One possible candidate for generating the iADP could be a persistent Na+ current, which has been shown to underlie similar spike afterdepolarizations in other preparations (Brumberg et al. 2000). Further studies are required to explore this possibility.
Concentration-dependent effect of apamin on the sAHP in motoneurons
KCa channels become activated by Ca2+ entering the cell during the action potential and generate a sAHP (Fig. 2 A1). In lamprey the administration of apamin, a toxin blocking all cloned subtypes of small conductance (SK) KCa channels (SK1–3) (Grunnet et al. 2001), reduces the sAHP in motor- and interneurons (El Manira et al. 1994; Hill et al. 1992; Meer and Buchanan 1992). Apamin at the concentrations used in these studies (0.25–10 μM, mostly in the lower range) did not abolish the entire sAHP (cf. Fig.3 A1) leaving open the possibility that, in addition, apamin-insensitive KCa channels are present in these neurons and that they contribute to the sAHP (Hill et al. 1992).
We thus proceeded to investigate the nature of the components underlying the sAHP. Since the amplitude of the sAHP depends on the membrane potential (V m) via the driving force on K+ ions, it is desirable to take all sAHP measurements at a constantV m. This is difficult to ensure in the course of recordings of a single cell lasting typically 3–4 h, because of the drift in electrode tip potential (Haliwell et al. 1994). We therefore chose an alternative approach in which theV m was set at a level at which the peak of the iADP between the fast and slow AHP was, for convenience, at the same level as the baseline membrane potential (i.e., constant iADP; Fig. 3 A1). This was around −60 to −65 mV (see Fig.2 A1), and as a rule it required active depolarization of the motoneuron by positive current injection. To validate this procedure, we determined empirically the relationship between sAHP measurements obtained at constant V m (where no drift was seen) and those obtained with the method outlined above (Fig.2 C). Cadmium chloride (200 μM) was perfused in the bath causing a progressive decrease in the sAHP (due to blockade of voltage-activated Ca2+ channels), which was continuously monitored with both methods. Their relationship was linear, i.e., a given percentage reduction in Ca2+-dependent sAHP measured with the “constantV m” method, appeared as an identical percentage reduction with a “constant iADP” (Fig. 2 C).
Using this technique, we followed the sAHP (n = 7,Lampetra fluviatilis) first in control, then during increasing concentrations of apamin, and finally under a full blockade of Ca2+ entry from HVA Ca2+channels with 200 μM cadmium (Fig. 3 A1). In Fig.3 A2, we show the effect of apamin expressed as the percentage remaining of the Ca2+-dependent sAHP component (subtracting the Cd2+-insensitive component). An apamin concentration of 2 μM abolished 71 ± 11% (mean ± SD, n = 6) of this part, while a full blockade was obtained only at 20 μM (Fig. 3, A1 andA2). The dose-response data plotted in log/linear fashion (Fig. 3 A3) shows how apamin causes a progressive reduction of the sAHP throughout the concentration range tested (0.2–20 μM), with an IC50 of around 0.6 μM. In all cases the final application of cadmium did not completely abolish the sAHP (Fig.3 A1). A small but consistent cadmium-resistant component had a mean amplitude of 0.77 ± 0.35 mV or 20 ± 8% of the control sAHP. This component was also observed with nickel in combination with cadmium, which together block LVA and HVA Ca2+ channels (n = 2,Lampetra fluviatilis; Fig. 2 A2.1).
At the high apamin concentration necessary to block the Ca2+-dependent part of the sAHP, apamin could possibly exert its effect in an unspecific way for instance by directly blocking Ca2+ channels. To exclude this possibility, we investigated the effect of apamin on Ca2+ spikes in sensory dorsal cells. These neurons possess the same Ca2+ channel subtypes as those of motoneurons (El Manira and Bussières 1997) and display broad Ca2+ spikes when sodium and most potassium channels are blocked with TTX, TEA, and 4-AP. We applied apamin at 10 and 20 μM (n = 2,Lampetra fluviatilis) and observed no change in the initial trajectory and peak amplitude of the Ca2+ spikes (Fig. 3 A4). A slight increase in spike duration and a reduction in the postspike afterhyperpolarization were apparent, consistent with a selective blockade of KCachannels by apamin (Kawai and Watanabe 1986) but with no effect on Ca2+ channels.
Effect of BAPTA injection on the sAHP in motoneurons
To verify that the small remaining Cd2+-resistant sAHP was not due to activation of some other type of KCa channel, motoneurons were loaded with the fast Ca2+ chelator BAPTA, present at 200 mM in the recording micropipette (n = 3,Lampetra fluviatilis). In all cases a progressive diffusion of BAPTA from the pipette tip into the neuron was correlated with an exponential decay of the sAHP amplitude with a time constant of around 30 min (Fig. 3, B1 and B2). Perfusion of cadmium at the end of the experiments (n = 2) showed that BAPTA, given sufficient time to diffuse/accumulate in the motoneuron, could abolish the cadmium-sensitive portion of the sAHP. Again we observed a residual component of the sAHP, which is resistant to the simultaneous presence of the fast Ca2+ chelator BAPTA intracellularly and cadmium extracellularly (Fig.3 B1). This component is thus unlikely to be due to a Ca2+-dependent process, driven for example by an increase in Ca2+ from intracellular stores or from HVA or LVA Ca2+ channels, and in any case not due to an activation of KCa channels.
We thus conclude that the large part of the sAHP (∼80%) is due to an activation of apamin-sensitive KCa, but that there is a consistent small part that is not mediated via Ca2+. This non–Ca2+-KCa–dependent component, was not chloride dependent as it still appeared while using 3 M KCl electrodes and could not be reduced or even reversed by increasing the intracellular Cl− levels with current injection (not shown) (cf. Russell and Wallén 1983).
Shunting effect of the sAHP
The opening of K+ conductances during the sAHP will temporarily decrease the input resistance at the soma (Gustafsson 1974), thereby also shunting incoming synaptic potentials. To verify this in lamprey motoneurons, we have elicited EPSPs from RS axons, before or during a postsynaptic sAHP in motoneurons. Figure 4 Acompares the EPSPs elicited before the action potential in a motoneuron and during the sAHP following a brief sequence of action potentials. A marked reduction of the EPSP occurred during the sAHP (n = 2). Measurements of input resistance by both positive and negative current pulses showed that it is decreased during the sAHP, corroborating the notion that a reduction in EPSP amplitude can occur through shunting during the sAHP (Fig. 4 B).
Effect of apamin on trains of EPSPs at 20 Hz
Activation of the RS axon-motoneuron synapse causes a localized Ca2+ increase in the postsynaptic dendrite (Bacskai et al. 1995), outlasting the associated EPSP by several hundred milliseconds. Such elevated Ca2+could trigger the opening of dendritic KCachannels, thereby increasing the local membrane conductance. If so, EPSPs following the first EPSP during repetitive activation of the same synapse would become attenuated, due to shunting, much in the same way that EPSPs are shunted during the sAHP. Blocking KCa channels with apamin should prevent this type of depression.
We first investigated the possible interaction between synaptically triggered postsynaptic Ca2+ transients and KCa channels in the dendrites by repetitively activating a single reticulospinal-motoneuron synapse with brief trains of nine action potentials at 20 Hz, every 15 s. This pattern of reticulospinal activity is in the physiological range achieved by RS neurons in brain stem–spinal cord preparations in vitro (Ullén et al. 1998) as well as in the freely swimming lamprey (Deliagina et al. 2000; T. Deliagina, personal communication). In none of the pairs tested (n= 6) did the EPSP trains display significant temporal summation (Fig.5 A, thick black trace). In all pairs, the EPSPs had an electrotonic component that was revealed after blocking chemical synaptic transmission with 200 μM cadmium (Fig.5 A, thin black trace).
Before and during apamin application, the first EPSP and an average of the following eight EPSPs in each train were monitored for changes in amplitude and time course. In the case of an interaction between Ca2+ ions and KCachannels, one would expect the first EPSP to remain unchanged after apamin application but the later ones to become potentiated due to a reduced shunting effect by KCa channels. In five pairs tested, apamin did not significantly affect the first or the later EPSPs (Fig. 5, A, B1, B2, E1, andE2). For each EPSP, the control trace is superimposed on the EPSPs occurring after apamin, and Cd2+. In Fig.5 B3, the electrical component has been subtracted, and it is clear that the chemical EPSPs before and after apamin superimpose, and thus that KCa channels do not affect the synaptic transmission during the later part of the train.
To increase the influx of Ca2+ via NMDA receptors at the postsynaptic site, we tested a separate pair in magnesium-free solution to relieve the voltage-dependent block of these channels (Dale and Grillner 1986; Nowak et al. 1984). Under apamin, the last eight EPSPs did not show an increase with respect to control (Fig. 5, C2, C3,and E2), but rather a minimal decrease. A similar marginal decrease was present already in the first EPSP of each train (Fig. 5,C1 and E1). In one of the pairs we also used another strategy to enhance Ca2+ entry. The motoneuron was briefly depolarized halfway between resting potential and firing threshold potential for the duration of each train, thus partly relieving the voltage-dependent Mg2+ block of NMDA receptors, as confirmed by a significant increase in the late EPSP component during such depolarizations. Also in this case the amplitude of both the first and the following EPSPs did not increase in apamin (Fig. 5, D1–D3, E1, and E2), which would have occurred if the Ca2+ entry during the first EPSP had induced a KCaactivation. In summary, in all pairs tested (n = 6), a blockade of KCa channels with apamin did not affect either the amplitude or the time course of EPSP trains.
Effect of apamin on trains of EPSPs evoked by high-frequency stimulation (>50 Hz)
In a second set of experiments, we increased the frequency (50–85 Hz) and the number (14–24) of EPSPs in each train, expecting to further increase the peak Ca2+concentration at the postsynaptic site. We also used Mg2+-free Ringer solution to maximize the Ca2+ entry through NMDA receptors. In this frequency range, EPSPs progressively summate to depolarize the membrane potential by several millivolts (Fig.6 A, control). A major component of this depolarization is due to NMDA receptors, since it was reduced by perfusion of the specific antagonistd-2-amino-5-phosphonopentanoic acid (d-AP5; 100 μM; Fig. 6 A, D-AP5). The remaining summation is likely to depend on α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)/kainate receptor activation and possibly also LVA Ca2+ channels, while the gap junction–mediated electrotonic component does not significantly contribute (Fig.6 A, Cd2+).
We monitored the first EPSP and an average of the following monosynaptic EPSPs in each train (Fig. 6, B1 andB2), in control and during perfusion of apamin. Figure 6,B3 and B4, compares the chemical component of the EPSPs before and after apamin. No change is observed. In six pairs (Fig. 6, C1 and C2), neither the first nor the following EPSPs were altered in amplitude or time course after apamin perfusion (in a 7th pair these parameters could not be measured because the stimulus artifact was superimposed on the EPSPs).
In Fig. 6 A it is apparent that the net depolarization level (due partially to summation of EPSPs) is enhanced after apamin. This occurred in three of the seven pairs (Fig. 6 D), while in four of seven pairs the depolarization was unaffected by apamin. KCa channels blockade can thus affect the net depolarization level (due to temporal summation) evoked by a high-frequency EPSP train. In addition to motoneurons, RS axons also excite (usually subthreshold) excitatory and inhibitory premotor interneurons that in turn may excite motoneurons (Brodin et al. 1988; Ohta and Grillner 1989). Such an oligosynaptic pathway could be affected after apamin and contribute to the temporal summation.
An increase in the variability in EPSP train trajectory following the perfusion of apamin was readily apparent in four of our pairs. This train-to-train variability can be visualized by plotting the SD of the mean EPSP train, during control conditions and in apamin (Fig.7 A; exp. d in Fig. 6). The plot shows that apamin markedly increased the variability during the trains and during the following repolarization. Apamin had this effect in five of seven pairs (Fig. 7 B). KCachannels play a role in limiting neuronal excitability and frequency regulation in motoneurons and interneurons (El Manira et al. 1994). Apamin could thus indirectly facilitate the excitability within the spinal cord, with a recruitment of previously inactive oligosynaptic pathways between the stimulated RS axon and the motoneuron. These synaptic potentials would be more variable and be expected to lead to the observed increase in train-to-train variability. In support of this interpretation, oligosynaptic potentials (i.e., from synapses other than the one stimulated) were often superimposed on the repolarization phase that followed each train (Fig. 7 C, top trace). To quantify this background activity the original recordings were first high-pass filtered (Fig.7 C, bottom trace; ∼20-Hz cutoff), thus removing the slow membrane potential depolarization but leaving the faster PSPs practically untouched. Two intervals were chosen, one before and one after the EPSP train, of 50 and 90 ms, respectively. A measure of the occurrence of oligosynaptic PSPs in each of these intervals was obtained by calculating the SD across sample points within the interval in each train, and then averaging it across all trains. This is simply an indicator of the average variability of the trace trajectory in each interval. Apamin caused significant increases of this indicator both before and after the train (Fig. 7, D1 and D2), but the latter were on average threefold higher than the former. In conclusion, the effects observed with high-frequency trains and apamin can most likely be explained by di- or oligosynaptic interneuronal connections recruited after the KCa blockade.
KCa channels and the sAHP
Apamin-sensitive KCa channels in lamprey not only regulate spike frequency adaptation via the sAHP (Figs. 2 and 3) but also affect the NMDA-dependent oscillatory properties of network neurons and the frequency of in vitro locomotor activity (El Manira et al. 1994). Previous studies did not succeed at completely blocking the sAHP with apamin (0.25–10 μM), leading to the suggestion that lamprey neurons may also express apamin-insensitive KCa channels (Hill et al. 1992;Meer and Buchanan 1992). Here we report that in motoneurons of Lampetra fluviatilis the sAHP can be dissected into a Ca2+-dependent component (80%) that is completely abolished by apamin (20 μM) and a smaller component (20%) that does not rely on Ca2+activation. As is clear from our dose-response graphs (Fig. 3,A2 and A3), apamin fully blocks the Ca2+-dependent sAHP component only at a concentration of 20 μM, which is higher than what has previously been used in lamprey. sAHPs due to apamin-insensitive KCa channels are relatively rare in vertebrate nervous systems, and where observed they have considerably slower kinetics than the apamin-sensitive ones (Bond et al. 1999; Vergara et al. 1998). It is thus not surprising to find that the Ca2+-dependent sAHP of lamprey motoneurons, which has a time course similar to that of apamin-sensitive sAHP in other systems, is also abolished by apamin. KCa channels of the SK subtypes underlie the slow AHP while the BK type of KCa (not blocked by apamin), when present, contribute to the fast AHP that repolarizes the action potential (Sah 1996). Recently it was shown that all three cloned mammalian SK-KCa channels (SK1–3) are blocked by apamin with an IC50 in the range of 27 pM to 196 nM, depending on the subtype (room temperature, rat and human) (Grunnet et al. 2001). On the other hand, apamin blocks the sAHP in lamprey (5–10°C) with an IC50 of about 600 nM. This peptide toxin is thus less effective under the present experimental conditions in lamprey than in mammals, probably due to differences in the extracellular environment such as temperature, or in channel protein composition.
As for the nature of the residual Ca2+-insensitive component of the slow AHP, it was present during a simultaneous blockade of HVA Ca2+ channels by cadmium and fast Ca2+ chelation by BAPTA. This led us to exclude the possibility that Ca2+, either entering the cell through ion channels or released from intracellular stores, can be involved. The sAHP is also unaffected by Cl−injection into the cell from KCl electrodes. A distinct remaining possibility is that of an activation of sodium-gated potassium channels (KNa). These channels are present at different levels in the vertebrate nervous system and have been found in frog embryo spinal neurons (Dale 1993) and in rat motoneurons (Safronov and Vogel 1996). In this latter case they activate in response to brief spike trains and generate a sAHP of several millivolts. If present in the lamprey, they could represent an additional mechanism contributing to the sAHP and spike frequency regulation, and thereby also to the operation of the locomotor network.
Synaptic transmission and KCa channels
Although direct evidence for the distribution of apamin-sensitive KCa channels in the dendrites of motoneurons is currently unavailable, dissociated motoneurons of the larval lamprey display Ca2+ currents in voltage clamp (El Manira and Bussières 1997) and Ca2+spikes in current clamp (R. H. Hill, personal communication), but the sAHP following the action potential is very small (D. Hess and A. El Manira, personal communication). As these cells lose most processes during dissociation, their lack of a sAHP has been taken to suggest that apamin-sensitive KCa channels might be located predominantly in proximal and distal dendrites. These channels are apparently not activated in the present motoneurons by the local dendritic Ca2+ increase (Bacskai et al. 1995), which accompanies the physiological input at a single giant RS axon synapse (Fig. 5). In addition, we can also exclude the possibility that apamin-sensitive KCa channels, if present in the presynaptic region, participate in regulating synaptic vesicle release. If this were the case, apamin should have modified the efficacy of synaptic transmission.
Thus KCa channels seem not to play a role in shaping the integration of descending reticulospinal input to motoneurons from the locomotor centers of the brain stem. If this finding holds true also for input to spinal network interneurons, it is implied that the problem of dendritic integration is simplified, in that synaptic transmission will not be compromised by KCa channels (activated via glutamate receptors) shunting synaptic potentials. This would be important for understanding the CPG for locomotion, and how it responds to supraspinal activation and steering commands.
KCa channels have been reported in the dendrites of mammalian neurons (Andreasen and Lambert 1995;Sah and Bekkers 1996; Schwindt and Crill 1997) and have been postulated to play a role in regulating the amplitude of synaptic potentials (Andreasen and Lambert 1995). To date, this has been documented only in an invertebrate preparation (Wessel et al. 1999) where the putative channel, of the apamin-insensitive (BK) type, was shown to regulate the amplitude of mimicked synaptic potentials evoked in the neuropil by puffs of glutamate. It will be interesting to see whether this role of KCa is limited to invertebrate neurons or extends to vertebrate nervous systems.
Reticulospinal synapses are located on motoneuronal dendrites, preferentially on the distal half of the dendritic tree (Wallén et al. 1985). If KCachannels are present in dendrites, why does the long-lasting Ca2+ entry induced by repetitive activation of a single synapse not activate them sufficiently to shunt the EPSP? One possibility is that the dendritic Ca2+concentration might not reach the threshold for KCa activation. One must bear in mind, however, that apamin-sensitive KCa channels activate already at submicromolar concentrations of cytoplasmic Ca2+ (Vergara et al. 1998). A second possibility is that KCa channels are spatially segregated from the area of synaptic contact. The Ca2+ entering the dendrite on synaptic stimulation would then be unable to diffuse and reach a sufficient concentration over that distance. Alternatively they might be distributed throughout the dendritic membrane at a very low density, so that a local activation limited to a small section of the dendrite near the synapse would be insufficient to significantly shunt the EPSP. The simultaneous Ca2+ entry induced in proximal and distal dendrites by backpropagating action potentials (see alsoBacskai et al. 1995) would on the other hand be able to recruit a number of KCa channels, thereby generating the sAHP and the observed shunt of incoming synaptic potentials that occurs during the sAHP (Fig. 4) (in the rat, seeSah and Bekkers 1996). If this were the case (low density of KCa channels), perhaps the summed effect of multiple subthreshold synaptic inputs, as during locomotion, could recruit enough KCa to decrease input resistance, thus affecting synaptic integration. Preliminary data suggest, however, that this is not the case (Cangiano et al. 2000).
We have tested the role of KCa channels in synaptic transmission only after a preliminary period of repetitive activation of the synapse. This was necessary to allow for the EPSP to undergo a certain amount of depression and reach a stable amplitude (Fig. 1 B) before applying apamin. In this protocol, which mimics the state of the synapse during ongoing locomotion, we have shown that KCa channels do not participate in synaptic transmission. It cannot entirely be ruled out that KCa might be recruited when the synapse is activated initially, such as during the first few bouts of swimming following the onset of locomotion.
We have also provided evidence that a KCa channel blockade by apamin can recruit previously inactive di- or oligosynaptic connections (Figs. 6 and 7). This can be explained by the previous demonstration that KCa channels can affect neuronal excitability (El Manira et al. 1994). The interneurons intercalated between the RS axon and the motoneuron may thus be activated more easily under apamin blockade of KCa channels than under control conditions.
In conclusion, the sAHP, which is the main determinant of spike frequency regulation is due to both apamin-sensitive KCa channels (80%), and to non–Ca2+-dependent channels (20%). KCa channels also play an important role for termination of NMDA receptor–activated plateau potentials (El Manira et al. 1994). The long-lasting Ca2+ transients evoked at RS synapses in dendrites do not, however, lead to an activation of KCa channels sufficient to shunt the EPSPs in a spike train. The latter finding suggests that the interaction between different synaptic inputs in a motoneuronal dendrite may be less complex than could have been assumed. This provides important information for modeling at the cell and network level.
We are grateful to Z. Biró and Drs. R. H. Hill and A. El Manira for useful comments on the manuscript and to Dr. T. Deliagina for unpublished information about giant RS axon activity in the freely behaving lamprey. We also express our appreciation to M. Bredmyr and H. Axegren for skillful laboratory assistance.
This project was funded by the Swedish Medical Research Council (Project No. 3026) and the Marianne and Marcus Wallenberg Stiftelsen. L. Cangiano is a recipient of a Marie Curie doctoral fellowship from the European Commission.
Address for reprint requests: S. Grillner, Dept. of Neuroscience, Karolinska Institute, 17177 Stockholm, Sweden (E-mail:).
- Copyright © 2002 The American Physiological Society