Through their repetitive discharge, GABAergic neurons of the substantia nigra pars reticulata (SNr) tonically inhibit the target nuclei of the basal ganglia and the dopamine neurons of the midbrain. As the repetitive firing of SNr neurons persists in vitro, perforated, whole-cell and cell-attached patch-clamp recordings were made from rat brain slices to determine the mechanisms underlying this activity. The spontaneous activity of SNr neurons was not perturbed by the blockade of fast synaptic transmission, demonstrating that it was autonomous in nature. A subthreshold, slowly inactivating, voltage-dependent, tetrodotoxin (TTX)-sensitive Na+ current and a TTX-insensitive inward current that was mediated in part by Na+ were responsible for depolarization to action potential (AP) threshold. An apamin-sensitive spike afterhyperpolarization mediated by small-conductance Ca2+-dependent K+ (SK) channels was critical for the precision of autonomous activity. SK channels were activated, in part, by Ca2+ flowing throughω-conotoxin GVIA-sensitive, class 2.2 voltage-dependent Ca2+ channels. Although Cs+/ZD7288 (4-ethylphenylamino-1,2-dimethyl-6-methylaminopyrimidinium chloride)-sensitive hyperpolarization-activated currents were also observed in SNr neurons, they were activated at voltages that were in general more hyperpolarized than those associated with autonomous activity. Simultaneous somatic and dendritic recordings revealed that autonomously generated APs were observed first at the soma before propagating into dendrites up to 120 μm from the somatic recording site. Backpropagation of autonomously generated APs was reliable with no observable incidence of failure. Together, these data suggest that the resting inhibitory output of the basal ganglia relies, in large part, on the intrinsic firing properties of the neurons that convey this signal.
The GABAergic neurons of the substantia nigra pars reticulata (SNr) convey the final output signal of the basal ganglia to the thalamus and superior colliculus (Deniau et al., 1978; Bentivoglio et al., 1979). SNr neurons also innervate the substantia nigra pars compacta (SNc) in which they influence the activity of dopaminergic neurons (Grace and Bunney, 1979; Tepper et al., 1995; Mailly et al., 2003). SNr neurons, like those of other extrastriatal basal ganglia nuclei, are tonically active (Yung et al., 1991; Bevan and Wilson, 1999; Chan et al., 2004). Brief changes in firing, often associated with movement (Sato and Hikosaka, 2002), therefore result in the disinhibition or further inhibition of neurons in the target nuclei of the basal ganglia (Delong, 1990). In addition, an altered pattern of SNr neuronal activity, from single-spike to burst firing, is characteristic of animal models of parkinsonism (Wichmann et al., 1999; Tseng et al., 2000) and epilepsy (Deransart et al., 2003).
In anesthetized rats, the firing rate of SNr neurons is ∼25 Hz (Gernert et al., 2004; Windels and Kiyatkin, 2004). Lesion of the subthalamic nucleus (STN), the principal excitatory input to the SNr (Bevan et al., 1994), only results in a 20% reduction in SNr firing rate (Feger and Robledo, 1991; Zahr et al., 2004). Repetitive firing also persists in the heavily deafferented brain slice preparation (Nakanishi et al., 1987; Yung et al., 1991; Richards et al., 1997). Together, these observations suggest that the repetitive firing of SNr neurons may be generated, in part, autonomously. Our first objective was therefore to determine the relative contributions of intrinsic mechanisms and synaptic inputs to the activity of SNr neurons in vitro.
Several types of autonomously active neuron are depolarized to the threshold for action potential (AP) generation by a tetrodotoxin (TTX)-sensitive, slowly inactivating, voltage-dependent Na+ (NaV) current (Bevan and Wilson, 1999; Raman and Bean, 1999; Taddese and Bean, 2002). This may also be supplemented by other inward currents, such as the hyperpolarization-activated current (Ih) (Maccaferri and McBain, 1996; Bennett et al., 2000; Chan et al., 2004) or a nonspecific cationic current (Raman et al., 2000; Jackson et al., 2004). In contrast, a membrane potential oscillation driven by Ca2+ currents underlies pacemaking in SNc dopamine neurons (Wilson and Callaway, 2000; Nedergaard, 2004). In several pacemakers, small-conductance Ca2+-activated potassium (SK) channels also play an important role in maintaining the precision of spiking (Bennett et al., 2000; Faber and Sah, 2002; Wolfart and Roeper, 2002; Hallworth et al., 2003). Having shown that the spontaneous discharge of SNr neurons in vitro is indeed driven by intrinsic properties, our second objective was to determine which of these ionic currents underlie activity.
Although spontaneous backpropagating APs have been reported in the dendrites of SNr neurons (Hausser et al., 1995), it is unclear whether these APs were generated by synaptic input or autonomous mechanisms. Our final objective was therefore to determine the somatodendritic site of origin of autonomously generated action potentials in SNr neurons and the reliability of AP propagation along their somatodendritic axis.
Materials and Methods
Slice preparation. Brain slices were prepared from 63 14- to 25-d-old male Sprague Dawley rats (Charles River Laboratories, Wilmington, MA). Experiments were performed in accordance with Society for Neuroscience, National Institutes of Health, and institutional policies. Animals were deeply anesthetized with a mixture of ketamine and xylazine before being perfused transcardially with ∼25 ml of ice-cold modified artificial CSF (aCSF), which contained 230 mm sucrose, 2.5 mm KCl, 1.25 mm NaH2PO4·H2O, 0.5 mm CaCl2·2H2O, 10 mm MgSO4·7H2O, 10 mm glucose, and 26 mm NaHCO3 and had been equilibrated with a gaseous mixture of 95% O2 and 5% CO2. The brain was rapidly removed, blocked, glued to the stage of a vibratome (Vibratome 3000 sectioning system; Vibratome Company, St. Louis, MO), and immersed in cooled equilibrated aCSF. Sagittal slices of 300 μm thickness, which contained the SNr, were cut and transferred to a holding chamber, in which they were submerged in aCSF containing 126 mm NaCl, 2.5 mm KCl, 1.25 mm NaH2 PO4·H2O, 2 mm CaCl2·2H2O, 2 mm MgSO4·7H2O, 10 mm glucose, and 26 mm NaHCO3, equilibrated with 95% O2/5% CO2, and maintained at room temperature.
Drugs. All drugs were prepared as concentrated stock solutions and frozen. Drugs were then diluted in aCSF on the day of the experiment and bath applied. For experiments in which fast synaptic transmission was blocked, 50 μm of the NMDA receptor antagonist d-(–)-2-amino-5-phosphonopentanoic acid (APV) (Tocris Cookson, Ellisville, MO), 20 μm of the AMPA receptor antagonist 6,7-dinitroquinoxaline-2,3-dione (DNQX) (Tocris Cookson), and 20 μm of the GABAA receptor antagonist 6-amino-3-(4-methoxyphenyl)-1(6 H)-pyridazinebutanoic acid hydrobromide (GABAzine) (Tocris Cookson; SR 95531 hydrobromide) were continuously applied. TTX, apamin, and ω-conotoxin GVIA were purchased from Sigma (St. Louis, MO); 4-ethylphenylamino-1,2-dimethyl-6-methylaminopyrimidinium chloride (ZD7288) was purchased from Tocris Cookson.
Electrophysiology. Individual slices were transferred to a recording chamber in which they were perfused, at a rate of 2–3 ml/min, with aCSF at 37°C. Cell-attached, whole-cell, or gramicidin-based perforated patch recordings were made using borosilicate glass micropipettes (Warner Instruments, Hamden, CT) prepared with a P-97 Flaming-Brown micropipette puller (Sutter Instruments, Novato, CA). Pipettes for whole-cell and perforated patch recordings were filled with a solution containing 135 mm K-MeSO4, 3.8 mm NaCl, 1 mm MgCl2·6H2O, 10 mm HEPES, 0.1 mm Na4EGTA, 0.4 mm Na3GTP, and 2 mm Mg1.5ATP titrated to a pH of 7.3 with 0.5 m KOH and diluted to an osmolarity of 290 mOsm. The resistance of pipettes filled with this solution was 3–7 MΩ. For perforated recordings, gramicidin was added to the pipette solution at ∼20 μg/ml. Pipettes for dendritic cell-attached recordings were filled with aCSF and had a resistance of 7–12 MΩ. SNr cells were visualized in the slice with a 40× water-immersion objective (Axioscope; Zeiss, Oberkochen, Germany) and an infragradient contrast video microscopy system (Infrapatch Workstation; Luigs and Neumann, Ratingen, Germany). Experiments were executed and data were recorded using a personal computer running Clampex 9 (Axon Instruments, Union City, CA) connected to a Multiclamp 200B amplifier (Axon Instruments) via a Digidata 1322A digitizer (Axon Instruments). Data were low-pass filtered at 10 kHz and sampled at 50 kHz. Fast capacitive transients were compensated on-line; the calculated series resistance was compensated off-line. A liquid junction potential correction was subtracted off-line from all recordings (Neher, 1992). For the recordings using normal aCSF in the whole-cell configuration, this junction potential was 9 mV [experimentally measured junction potential, 8.7 mV; calculated junction potential (JPCalc; Barry, 1994), 9.2 mV]. In the perforated configuration, a second junction potential exists between the pipette solution and the neuronal cytoplasm. This junction potential cannot be directly measured, but it should disappear soon after the deliberate or accidental establishment of the whole-cell configuration. We observed that there was indeed a depolarization of ∼5 mV under these conditions. Therefore, using this depolarization as an estimate of the junction potential between the pipette and the cell, we estimated the sum of the two junction potentials in the perforated configuration to be 4 mV, and this correction was made off-line. For experiments in which Na+ substitution was required, a HEPES-buffered aCSF was used. This contained 150 mm NaCl, 2.5 mm KCl, 10 mm HEPES, 10 mm glucose, 2 mm MgCl2·6H2O, and 2 mm CaCl2 titrated to pH 7.4 with NaOH (resulting in a final Na+ concentration of ∼154.5 mm) and diluted to 300–310 mOsm. The low Na+ solution was designed so that the Na+ equilibrium potential would be 25 mV: approximately halfway between the equilibrium potential in control conditions and the average resting potential of SNr cells in TTX. This was achieved by the substitution of 136.2 mm N-methyl-d-glucamine (NMDG) for NaCl, giving a final Na+ concentration of 13.8 mm. This solution was titrated with HCl to a pH of 7.4 and diluted to 300–310 mOsm. The calculated junction potentials for these HEPES-based solutions of 11 mV (regular Na+) and 17 mV (low Na+) were subtracted off-line.
Most voltage-clamp recordings were made in the whole-cell patch configuration, because this configuration is associated with a lower series resistance and therefore a more effective clamp of the membrane potential. However, in the whole-cell configuration, the replacement of the cytoplasmic contents with the pipette solution can disrupt intracellular Ca2+ dynamics and block intrinsic ion channels (Hallworth et al., 2003). Therefore, results acquired using whole-cell recordings were verified with perforated patch recordings. There were no gross differences in the data acquired using these two techniques, so, unless otherwise stated, results were collated for the purposes of statistical analysis.
Data analysis. Data were analyzed using the Clampfit 9 (Axon Instruments) and Origin 7 (Microcal Software, Northampton, MA) software packages. The firing characteristics of the cells [interspike interval, coefficient of variation (CV), and AP frequency and threshold] were calculated from trains of 101 spontaneous APs. AP threshold (APth) was defined as the point at which the derivative of the membrane potential (dV/dt) deviated from its mean during the interspike interval by greater than 2 SDs; this was detected using a custom programmed algorithm. Numerical data are presented as mean ± SD, and the distributions of data are represented graphically using box plots. The central line of the box plots represents the median, the edges of the box show the interquartile range, and the edges of the “whiskers” show the full extent of the overall distribution. The mean of the data are also represented on the box plots with a small square. The use of small sample sizes precluded statistical determination of the distribution of the data; therefore, unless stated otherwise, the nonparametric Wilcoxon's signed-rank test was used for all statistical comparisons. An α value of 0.05 was used as the criterion for determining statistically significant differences.
GABAergic SNr neurons are spontaneously active
Data collected from 95 cells were selected for additional analysis. Only cells that exhibited action potentials that crossed 0 mV after junction potential correction were used in this study. Criteria for selection were also based on the electrophysiological comparisons of identified GABAergic and dopaminergic SNr neurons given by Yung et al. (1991) and Richards et al. (1997). Cells with properties that were closer to those described for dopamine neurons were excluded from the study. Specifically, putative SNr GABA neurons exhibited spontaneous action potentials in the single-spike mode with a width at threshold of <2 ms, they exhibited little or no “sag” in response to hyperpolarizing current injections, with no burst firing during release of this hyperpolarization, and they could maintain firing at frequencies above 50 Hz when driven by depolarizing current injections. Spontaneous activity recorded from 68 such neurons had a mean discharge rate of 10.78 ± 6.00 Hz.
Comparison of data acquired in the whole-cell and perforated configurations revealed no differences in the rate and pattern of autonomous activity. The firing rate of SNr neurons in the perforated configuration was 12.54 ± 7.45 Hz, whereas in the whole-cell configuration, it was 9.30 ± 3.98 (n = 31 and 37, respectively; p = 0.09, Mann–Whitney U test). The coefficient of variation was similarly unaffected (perforated patch CV, 0.06 ± 0.03; whole-cell CV, 0.09 ± 0.02; p = 0.12; Mann–Whitney U test).
Synaptic inputs are not required for the generation of the spontaneous activity of SNr neurons in vitro
To test the hypothesis that the spontaneous firing seen in vitro is generated by a mechanism that is intrinsic to SNr neurons, APV, DNQX, and GABAzine were applied to spontaneously active SNr neurons in rat brain slices (Fig. 1). Under control conditions, nine SNr neurons that were recorded in the perforated patch configuration discharged at 11.35 ± 3.27 Hz with aCV of 0.044 ± 0.022, and APth was –44.28 ± 6.20 mV. Bath application of 50 μm APV, 20 μm DNQX, and 20 μm GABAzine did not affect these firing properties (firing frequency, 12.57 ± 4.27 Hz, p = 0.30; CV, 0.041 ± 0.012, p = 0.91; APth, –43.32 ± 5.89 mV, p = 0.30).
The absence of effect of the blockade of fast synaptic transmission on the spontaneous activity of SNr neurons demonstrated that, in vitro, this activity is generated by intrinsic mechanisms, but, to ensure that only these intrinsic mechanisms were studied, all subsequent experiments were performed in the presence of APV, DNQX, and GABAzine.
NaV channels are essential for the generation of rhythmic activity
Unlike the dopaminergic neurons of the SNc (Shepard and Bunney, 1991; Kang and Kitai, 1993), SNr GABA neurons were simply slowed by hyperpolarization and exhibited a stable membrane potential when hyperpolarization was just sufficient to stop firing (Fig. 2 A). We therefore hypothesized that autonomous rhythmic activity in SNr GABA neurons was dependent on APs. To test this hypothesis, AP generation was blocked by the application of the selective NaV channel blocker TTX (1 μm). Figure 2 B1–B3 shows an example of the action of 1 μm TTX on an SNr GABA neuron. In both perforated patch (n = 4) and whole-cell (n = 11) experiments, the resting membrane potential after the complete abolition of AP firing was stable, with no remaining oscillation. Under these conditions, neither depolarizing currents nor the release from hyperpolarizing currents could induce oscillations in the membrane potential.
The resting membrane potential is set by a TTX-insensitive inward current
In the perforated patch configuration, APth was –45.86 ± 6.26 mV, and the membrane potential in the presence of TTX was –43.64 ± 3.92 mV (n = 4). In the whole-cell configuration, APth was –47.12 ± 4.36 mV, and the membrane potential in the presence of TTX was –46.53 ± 8.91 mV (n = 11). Together, the membrane potential in the presence of TTX was not significantly different from the previously measured APth (Fig. 2C,D) (APth in control conditions, –46.79 ± 4.72 mV; resting membrane potential in TTX, –45.76 ± 7.86 mV; n = 15; p = 0.52). Figure 2E illustrates the difference between the resting membrane potential in TTX and APth plotted against the control firing frequency for each cell. For 7 of 15 cells, this difference was negative: the resting membrane potential was hyperpolarized to APth, suggesting that TTX-sensitive currents were necessary to depolarize these cells to APth. To test this hypothesis, voltage-clamp experiments were performed in which neurons were held at –80 mV and presented with a family of 1 s depolarizing steps (Fig. 2F1 shows an example of one such step). I–V curves plotted from the steady-state current at the end of each of these steps revealed a TTX-sensitive inward current that activated at voltages depolarized to approximately –62.5 mV (Fig. 2F2,G) and reached a peak of –72.02 ± 32.72 pA (n = 9), just below APth (–50 mV).
However, 8 of 15 of the cells plotted in Figure 2E had a resting potential that was depolarized to APth. Interestingly, the difference between the resting membrane potential in TTX and APth was correlated with the spontaneous firing frequency (Fig. 2E) (r = 0.53; p = 0.04, Student's t test). Also, during autonomous activity, the average peak of the afterhyperpolarization (AHP) seen after each AP was –68.49 ± 4.93 mV, ∼6 mV hyperpolarized to the voltage range in which Nav channels are activated. As such, we hypothesized that an unidentified, TTX-insensitive current assisted the depolarization of GABAergic SNr neurons to APth. To test this possibility, SNr neurons were recorded in current-clamp mode (in the whole-cell configuration) while the bathing solution was switched to a low Na+ aCSF (Fig. 3A). Low Na+ aCSF abolished autonomous activity and produced a stable resting membrane potential with no subthreshold oscillation (Fig. 3A2). This resting membrane potential was markedly more hyperpolarized than the resting potential that was seen previously in TTX (–64.82 ± 5.97 mV, n = 6 vs –45.76 ± 7.86 mV, n = 15; p = 0.011, Mann–Whitney U test). Returning to normal aCSF resulted in the partial restoration of autonomous action potential generation (Fig. 3A3). To compare the resting membrane potential of SNr neurons in TTX with the resting potential in low Na+ aCSF, experiments were performed in which 1 μm TTX was applied first in normal aCSF and then, subsequently, in low Na+ aCSF. An example recording from such an experiment is shown in Figure 3B. In the perforated patch configuration, the resting membrane potential of SNr GABA neurons in TTX of –41.38 ± 2.67 mV was hyperpolarized to –62.58 ± 2.06 mV by the reduction of external Na+ (n = 2). In whole-cell experiments the resting membrane potential in TTX of –44.86 ± 7.87 mV was hyperpolarized to –61.21 ± 7.92 mV by the reduction of external Na+ (n = 7). Overall, the resting membrane potential in TTX of SNr GABA neurons in this experiment was –44.09 ± 7.05 mV. Changing to low Na+ aCSF resulted in a significant hyperpolarization of the resting membrane potential to –61.51 ± 6.92 mV (Fig. 3C,D)(n = 9; p = 0.0039). Returning the bathing solution to normal aCSF reversed the hyperpolarization in all seven of the cells in which a wash was attempted. Voltage-clamp experiments in the presence of 1 μm TTX (Fig. 3E,F) revealed a negative shift in the reversal potential of whole-cell current from –56.28 ± 1.46 mV in control conditions to –71.43 ± 1.63 mV in low Na+ (n = 3). The discrepancy between the resting membrane potentials in TTX and the whole-cell reversal potential is likely to be a reflection of the greater disruption of intrinsic conductances caused by the more complete perfusion of neurons by the lower resistance electrodes used in the voltage-clamp experiments.
Hyperpolarization-activated currents do not contribute to autonomous firing
Immunohistochemical (Notomi and Shigemoto, 2004) and in situ hybridization (Monteggia et al., 2000; Santoro et al., 2000) studies have indicated that all known hyperpolarization-activated cyclic nucleotide-gated (HCN) channel subunits, although primarily HCN2, are expressed within the SNr. This expression may be attributable, in part, to the presence of dopaminergic neurons within the SNr, but there may also be some expression of functional HCN channels in the GABAergic neurons (Stanford and Lacey, 1996). We therefore hypothesized that Ih may be, in part, responsible for the depolarization toward APth that was observed in SNr neurons. This hypothesis was investigated by applying one of two relatively specific blockers of HCN channels, 50 μm ZD7288 (Harris and Constanti, 1995) or 2 mm CsCl, to SNr neurons. Figure 4 shows the effects of blockade of HCN channels on the AP frequency and pattern (as measured by CV) of SNr neurons. In the perforated patch configuration, the control firing frequency of 12.86 ± 4.78 and CV of 0.031 ± 0.004 were unchanged by the application of 50 μm ZD7288 (firing frequency in ZD7288, 11.33 ± 4.23, n = 6, p = 0.31; CV in ZD7288, 0.034 ± 0.007, p = 0.84). In separate whole-cell experiments, this result was replicated with the application of 2 mm CsCl (control firing frequency, 10.33 ± 3.82; firing frequency in CsCl, 10.25 ± 4.06, n = 6, p = 0.69; control CV, 0.062 ± 0.016; CV in CsCl, 0.048 ± 0.017, p = 0.16).
Analysis of the response of SNr cells to hyperpolarizing current injections revealed that there was an increase in steady-state input resistance at voltages below approximately –60 mV when 2 mm CsCl was present, resulting in a more profound hyperpolarization in response to the same current step (Fig. 4E,F). Voltage-clamp experiments were used to further investigate this change in input resistance. Using families of steps from a holding of –50 to –110 mV, a slowly developing inward current was revealed at voltages negative to approximately –70 mV in all five neurons tested. This current could be blocked by the application of 2 mm CsCl (Fig. 4G,H) and activated with an average rate constant of 581.75 ± 297.74 ms. This rate and voltage range of activation is consistent with subthalamic neurons (Do and Bean, 2003) in which HCN2 expression predominates (Notomi and Shigemoto, 2004). In the whole-cell configuration, it is possible that dialysis of the cell may alter the intracellular cAMP concentration, which could alter the voltage range of activation of HCN channels. Therefore, additional voltage-clamp experiments were performed in the perforated configuration. The inward current revealed in these experiments was similar to that seen in the whole-cell experiments and was also sensitive to a selective HCN channel blocker, in this case 20 μm ZD7288 (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). Together, these data show that, although Ih is present in SNr neurons, it is not involved in the generation of the autonomous activity because the voltage range at which it is activated is more hyperpolarized than that traversed during autonomous activity.
Apamin-sensitive channels are critical for precise, rhythmic autonomous activity
In several types of basal ganglia neuron, an apamin-sensitive outward current mediated by SK channels is responsible for the so-called medium AHP, which has been shown to be important for the frequency and precision of autonomous activity (Shepard and Bunney, 1991; Wolfart et al., 2001; Wolfart and Roeper, 2002; Hallworth et al., 2003; Wilson et al., 2004). In situ hybridization has also shown that SK channels are strongly expressed in putative SNr GABA neurons (Stocker and Pedarzani, 2000; Tacconi et al., 2001). Indeed, in our recordings from GABAergic SNr neurons, the hyperpolarization after each AP often consisted of two distinct elements (Fig. 3A1): a rapid hyperpolarization followed by a slower hyperpolarization similar to the medium AHP that is blocked by apamin in other neuronal types (Bond et al., 2004; Villalobos et al., 2004). Therefore, we hypothesized that apamin-sensitive SK channels also play an important role in the maintenance of the frequency and precision of single-spike firing in these neurons. Figure 5A1–A3 summarizes the range of effects that were observed during the application of 100 nm apamin to SNr GABA neurons. Apamin application resulted in the disappearance of the medium component of the AHP (Fig. 5C), with little apparent effect on the rapid component of AP repolarization (Fig. 5D). In one-third (4 of 12, 1 perforated patch and 3 whole-cell) of the neurons tested, the loss of the AHP was followed within 10 min by a strong depolarization of the cell that resulted in the cessation of firing (as shown in the example in Fig. 5A3). In these cases, firing could often be restored by the injection of hyperpolarizing current (data not shown). Of the remaining neurons, the control firing rate in perforated patch recordings (n = 4) was 14.30 ± 7.75 Hz, and the firing rate in apamin was 18.11 ± 10.72 Hz. The control CV in perforated patch was 0.050 ± 0.013, and the CV in apamin was 0.150 ± 0.066. The APth in control conditions was –43.13 ± 4.82 mV, and in apamin it was –39.85 ± 5.12 mV. The control firing rate in whole-cell recordings (n = 4) was 10.86 ± 4.31, and the firing rate in apamin was 15.53 ± 9.61 Hz. The control CV in whole-cell recordings was 0.059 ± 0.028, and the CV in apamin was 0.174 ± 0.001. The APth in control conditions was –46.84 ± 4.45 mV, and in apamin it was –45.22 ± 4.80 mV. Overall, these eight cells showed a small but nonsignificant increase in firing rate (Fig. 5E) (control, 12.59 ± 6.09 Hz; apamin, 16.82 ± 9.53 Hz; p = 0.20), a disturbed firing pattern (Fig. 5E) (control CV, 0.054 ± 0.021; apamin CV, 0.162 ± 0.045; p = 0.0078), and a depolarized APth (Fig. 5B,D,E) (control APth, –44.98 ± 4.73 mV; apamin APth, –42.53 ± 5.41 mV; p = 0.016).
Class 2.2 voltage-dependent Ca2+ (CaV2.2) channels regulate the frequency and precision of autonomous activity through their functional coupling to SK channels
The strong effects of SK channel blockade with apamin are suggestive that voltage-dependent Ca2+ channels are active during autonomous spiking. Voltage-dependent Ca2+ currents in SNr GABA neurons have not been extensively characterized. Evidence from immunohistochemical studies points to a strong expression of CaV2.2 (N-type) channels in dendrites and a weaker expression in the cell bodies of SNr neurons (Westenbroek et al., 1992), whereas there is little evidence for the expression of other voltage-dependent Ca2+ channel subtypes in these cells. Therefore, we hypothesized that activation of SK channels is coupled to the entry of Ca2+ ions into SNr neurons through CaV2.2 channels. Figure 6 shows the effects of the blockade of CaV2.2 with 1 μm ω-conotoxin GVIA on SNr GABA neurons. As for apamin, ω-conotoxin GVIA application resulted in the loss of the medium component of the AHP (Fig. 6C,D). In the perforated patch configuration (n = 3), the firing rate in control conditions was 11.60 ± 4.38 Hz, and the CV was 0.073 ± 0.014; the firing rate in ω-conotoxin GVIA was 15.39 ± 5.31 Hz, and the CV 0.113 ± 0.029. In whole-cell recordings (n = 3), the firing rate in control conditions was 10.84 ± 5.27 Hz, and the CV was 0.088 ± 0.042; the firing rate in ω-conotoxin GVIA was 12.55 ± 4.92 Hz, and the CV 0.166 ± 0.115. Overall, there was a statistically significant increase in the firing rate (Fig. 6E) (control, 11.22 ± 4.35 Hz; ω-conotoxin GVIA, 13.97 ± 4.84 Hz; n = 6; p = 0.031) and a disruption in the firing pattern (Fig. 6E) (control CV, 0.080 ± 0.029; ω-conotoxin GVIA CV, 0.140 ± 0.080; p = 0.031). Although the precision of autonomous activity was significantly reduced after the blockade of Cav2.2 channels, the disruption was not as profound as that observed after the blockade of SK channels. Furthermore, the frequency of autonomous activity was significantly increased by the blockade of Cav2.2 channels in contrast to the effects observed after apamin application. Although supporting the hypothesis that Ca2+ flowing through Cav2.2 channels is coupled to the activation SK channels, these data leave open the possibility that other Ca2+ sources may also be important.
Autonomously generated APs reliably backpropagate along the dendrites of SNr neurons
It has been reported previously (Hausser et al., 1995) that APs observed at the soma of SNr GABA neurons backpropagate into their dendrites. Thus, single, spontaneous, or driven APs (elicited by somatic or dendritic depolarization) backpropagated up to ∼100 μm from the soma. However, this study did not distinguish between spontaneous APs that were generated by autonomous mechanisms or spontaneous synaptic inputs. Similarly, the reliability of backpropagation of continuously, autonomously generated APs into the dendrites of SNr GABA neurons was not studied. To address these issues further, we used dual somatic and dendritic cell-attached recordings of SNr GABA neurons in the presence of APV, DNQX, and GABAzine (Fig. 7). These recordings revealed an autonomous rhythmic discharge in both the somata and dendrites of all 11 SNr GABA neurons tested (Fig. 7). Analysis of the relative timing of the peaks of action currents demonstrated that, in 10 of 11 of these cases, the peak of the dendritic AP closely followed the peak of the somatic AP (Fig. 7A3,B). The delay of the dendritic AP relative to the somatic AP was also well correlated with the distance of the dendritic electrode from the somatic recording site (Fig. 7C)(r = 0.76; p = 0.018, Student's t test), suggesting that the site of AP generation was closer to the soma than the dendritic recording site. In the 10 of 11 neurons in which the AP was first recorded at the soma, each autonomously generated AP faithfully propagated to the dendritic recording site. Together, these data suggest that, in SNr GABA neurons, autonomously generated APs are generated close to the soma and then propagate with high reliability into the dendritic tree.
Intrinsic inward currents drive spontaneous firing
Intrinsic currents were responsible for the spontaneous discharge of rat SNr neurons in vitro because this activity persisted in the absence of fast synaptic transmission. In an autonomously active neuron, a net inward current is required for depolarization to APth because K+ channels and the Na+/K+ pump combine to hyperpolarize the neuron toward the equilibrium potential for K+ (EK). Indeed, the threshold at which NaV currents were regenerative and triggered APs in SNr neurons was approximately –45 mV, which was over 50 mV positive to EK.
Many types of neuron possess a slowly inactivating NaV current in the subthreshold voltage range. Often called the persistent Na+ current (INaP), this TTX-sensitive current has also been found to support the spontaneous activity of principal neurons in, for example, the STN (Bevan and Wilson, 1999; Beurrier et al., 2000; Do and Bean, 2003), tuberomammillary nucleus (Taddese and Bean, 2002), and cerebellum (Raman and Bean, 1999). Indeed, voltage-clamp experiments in SNr neurons revealed the presence of a slowly inactivating TTX-sensitive inward current that activated at voltages more depolarized than approximately –62.5 mV. However, this value is often up to 7.5 mV more depolarized than the foot of the interspike interval. In addition, the resting membrane potential of SNr neurons after TTX application was close to APth. Together, these data indicate that a TTX-sensitive Na+ current, including INaP, is unlikely to be the only inward current that depolarizes SNr neurons to APth. We also noted that the difference between APth and the resting membrane potential in TTX was correlated with the firing frequency of SNr neurons, suggesting that tonic depolarization set by a TTX-insensitive inward current greatly influences the autonomous firing rate.
Neurons express additional voltage-dependent channels that are activated in the subthreshold voltage range, mediate inward currents, and thus assist depolarization toward APth. One such inward current is Ih: examples of neurons in which both INaP and Ih have been found to contribute to the generation of spontaneous activity include hippocampal stratum oriens-alveus interneurons (Maccaferri and McBain, 1996), striatal cholinergic interneurons (Bennett et al., 2000; Wilson, 2005), and neurons in the external globus pallidus (Chan et al., 2004). In each case, the blockade of HCN channels with Cs+ or ZD7288 slowed but did not prevent spontaneous firing. However, despite the strong expression of HCN2 mRNA in the SNr (Notomi and Shigemoto, 2004), the spontaneous discharge of SNr GABA neurons was unaffected by the application of ZD7288 or Cs+. Moreover, although analysis of current-clamp and voltage-clamp recordings revealed activation of Ih at voltages negative to –70 mV, this value was typically more hyperpolarized than the voltage range traversed during the interspike interval. Thus, under resting conditions, Ih does not contribute to pacemaking in SNr GABAergic neurons.
For each of the neuronal types described above, autonomous activity is dependent on APs: abolition of APs with hyperpolarization or TTX results in a stable membrane potential. There are, however, other cell types in which this is not the case: in these cells, after spiking is abolished, a membrane potential oscillation remains. An example of this are the dopaminergic cells of the SNc (Shepard and Bunney, 1991; Kang and Kitai, 1993). The depolarizing phase of the oscillation in these cells is primarily attributable to the activation at subthreshold potentials of nimodipine-sensitive CaV1.2–1.3 (L-type) channels (Nedergaard et al., 1993; Mercuri et al., 1994; Durante et al., 2004). Similar subthreshold nimodipine-sensitive oscillations have been observed in dorsomedial suprachiasmatic nucleus neurons (Pennartz et al., 2002; Jackson et al., 2004). Conversely, in inferior olive neurons (Llinas and Yarom, 1981a,b), lateral habenula nucleus neurons (Wilcox et al., 1988), and thalamocortical relay neurons (McCormick and Pape, 1990), CaV3 channels underlie a T-type current that is involved in the generation of subthreshold membrane potential oscillations. Our experiments revealed no evidence for such Ca2+-dependent oscillations in SNr GABA neurons because, when spiking was abolished, either with hyperpolarizing current injection or through the application of TTX, subthreshold membrane potential oscillations were not observed.
A final, and less well understood, inward current that is found to be responsible for subthreshold depolarization in some cells is a voltage-independent and TTX-insensitive “background” conductance. Neurons of the cerebellar nuclei were found to have a depolarized resting membrane potential in the presence of TTX that was not hyperpolarized by the application of CsCl or the replacement of Ca2+ with Co2+ (Raman et al., 2000). Only replacement of Na+ with NMDG led to a significant hyperpolarization of the membrane potential. The authors concluded that the neurons of the cerebellar nuclei expressed a tonic cationic flux, carried in part by Na+ ions, which they compared with the nonselective background cation currents of cardiac cells. Similar TTX-insensitive background Na+ currents have been reported in rat locus ceruleus neurons (Alreja and Aghajanian, 1993), pre-Bötzinger complex neurons (Pena and Ramirez, 2004), and dorsomedial suprachiasmatic nucleus neurons (Jackson et al., 2004). The similarity of our findings with those in the cerebellar nuclei is striking. SNr neurons rested at a relatively depolarized membrane potential when exposed to TTX, and their spontaneous firing was unaffected by blockade of HCN channels. Recording in a reduced Na+ medium arrested AP generation and produced a profound hyperpolarization of the resting membrane potential in TTX. We therefore conclude that SNr neurons express a background cationic conductance that depolarizes the neurons into the voltage range at which APs can be generated. The exact nature of this conductance is unknown, but it is carried, at least in part, by Na+ ions. Candidates that may underlie the background conductance include a low-threshold TTX-insensitive Na+ channel (for review, see Delmas et al., 1997) and the Na+/Ca2+ exchanger. In addition, the possible contribution of a tonic Ca2+ flux cannot be ruled out.
SK channels are important for the precision of autonomous activity
Intrinsically generated spontaneous activity is influenced by apamin-sensitive SK channels in several of the neuron types described above. In dopaminergic neurons of the SNc (Kang and Kitai, 1993; Sarpal et al., 2004) and thalamocortical relay neurons (McCormick and Pape, 1990), these channels activate as the cells depolarize and are responsible for the descending phase of the subthreshold oscillation. In STN neurons (Hallworth et al., 2003) and striatal cholinergic interneurons (Bennett et al., 2000), each AP is rapidly followed by a pronounced AHP, the slow component of which is mediated principally by SK channels. In these cases, Ca2+ enters the cell through “high” voltage-activated Ca2+ channels that are activated by APs (Sah, 1996; Xia et al., 1998). As in STN neurons, we found that the single-spike AHP in SNr GABA neurons was reduced by the application of apamin, suggesting that SK channels are also activated after APs in SNr neurons. Indeed, the activation of SK channels in both STN and SNr neurons was coupled, at least in part, to Ca2+ entry through CaV2.2 channels, which are the major (high-voltage) Cav channels in these neurons (Westenbroek et al., 1992; Song et al., 2000) and are presumably activated at suprathreshold voltages associated with APs.
Reduction of the AHP also elevated APth and disrupted the pace and precision of firing, which is symptomatic of a reduction in Nav channel availability. Thus, a fundamental role of the SK channels in the SNr, and other autonomously active neurons, is to deepen and lengthen the AP AHP, which may enable Nav channels to recover from the inactivation that accumulated in the preceding AP(s).
Autonomously generated APs reliably backpropagate into the dendrites of SNr GABA neurons
Backpropagating APs have been observed in a diverse range of neuronal types (for review, see Waters et al., 2005). Our recordings revealed perfectly faithful backpropagation of autonomously generated APs into the dendrites of SNr GABA neurons. In 10 of 11 of these neurons, the APs in the dendrite were observed to always follow the respective APs in the soma, suggesting that the site of AP generation is either axonal or somatic, but generally not dendritic. The conduction velocity of backpropagation was also comparable with previously reported values for hippocampal CA1 neurons and neocortical pyramidal neurons (Spruston et al., 1995; Stuart et al., 1997a).
The significance of backpropagating autonomous activity for synaptic integration/plasticity in SNr neurons is presently unclear. It has been proposed that, when APs evoked by synaptic input backpropagate, this acts as a retrograde signal of neuronal output to the dendrites, promoting changes in the efficacy of those synapses that led to the generation of an AP (Stuart et al., 1997b; Hausser et al., 2000). In SNr neurons, the situation is complicated by the generation of APs without synaptic input. For spike-timing-dependent plasticity to operate effectively in these neurons, both autonomously and synaptically generated APs must presumably backpropagate without failure.
This research was supported by National Institutes of Health–National Institute of Neurological Disorders and Stroke Grants NS020702, NS041280, and NS047085. We are appreciative of Dr. Steve Kitai for his support and scientific contribution to this study.
Correspondence should be addressed to Dr. Mark D. Bevan, Department of Physiology, Feinberg School of Medicine, Northwestern University, 303 East Chicago Avenue, Chicago, IL 60611. E-mail:.
Copyright © 2005 Society for Neuroscience 0270-6474/05/258272-10$15.00/0