Plateau potentials can be elicited in nigral GABAergic neurons by injection of 500 ms depolarizing current pulses from hyperpolarized holding potentials in whole-cell recordings in vitro. In approximately one-third of these neurons, plateau potentials were observed under control conditions and could be elicited in the remaining neurons after blocking potassium conductances. Application of the L-type calcium channel agonist Bay K 8644 or activation of NMDA receptors enhanced plateau potentials observed under control conditions and caused a plateau to be elicited in neurons not exhibiting it previously. The plateau potential was abolished in calcium-free buffer, as well as by nickel or cadmium. The L-type calcium channel blockers nimodipine and nifedipine abolished the plateau potential observed under control conditions but did not affect plateaus unmasked by tetraethylammonium. Plateau potentials observed under control conditions as well as those observed in the presence of Bay K 8644, NMDA, or tetraethylammonium were abolished in low-sodium buffer and by the calcium-activated nonselective cation conductance blocker flufenamic acid. These data suggest that nigral plateau potentials are mediated by a calcium-activated nonselective cation conductance (ICAN) that is activated by calcium entry predominantly through L-type calcium channels. In many nigral neurons, ICAN is masked by tetraethylammonium-sensitive potassium conductances, but plateaus can be evoked after increasing calcium conductances. The ICAN-mediated plateau potential in nigral GABAergic neurons likely affects the way these neurons integrate input and may represent a mechanism contributing to the rhythmic firing of these neurons seen in pathological conditions such as Parkinson's disease.
- synaptic integration
- synaptic communication
- basal ganglia
GABAergic neurons in the substantia nigra pars reticulata along with the entopeduncular nucleus and internal segment of the globus pallidus make up the major output of the basal ganglia and function to tonically inhibit and phasically disinhibit target neurons in the thalamus and the superior colliculus, thereby allowing movement (Albin et al., 1989; Chevalier and Deniau, 1990; DeLong, 1990). Nigral GABAergic neurons are spontaneously active both in vivo (Wilson et al., 1977; Deniau et al., 1978; Guyenet and Aghajanian, 1978; Celada et al., 1999; Windels and Kiyatkin, 2004) and in vitro (Nakanishi et al., 1987b; Lacey et al., 1989; Yung et al., 1991; Rick and Lacey, 1994; Richards et al., 1997; Atherton and Bevan, 2005), and the brief changes in firing rate that are responsible for movement are caused by afferent input (Lestienne and Caillier, 1986; Schultz, 1986; Chevalier and Deniau, 1990; Windels and Kiyatkin, 2006). Nigral GABAergic neurons receive inhibitory input primarily from the striatum (Bevan et al., 1994), globus pallidus (Smith and Bolam, 1991), and other nigral GABAergic neurons (Deniau et al., 1982; Nitsch and Riesenberg, 1988; Hajos and Greenfield, 1994; Rick and Lacey, 1994; Gulácsi et al., 2003), whereas the majority of the excitatory afferents originate from the subthalamic nucleus (Robledo and Feger, 1990; Bevan et al., 1994). The final output of nigral GABAergic neurons in vivo results from an interaction between the afferent inputs and intrinsic membrane conductances.
In vitro recordings have described the presence of a depolarizing plateau potential in subsets of both nigral and entopeduncular GABAergic neurons in response to depolarizing current pulses or stimulation of the subthalamic nucleus (Matsuda et al., 1987; Nakanishi et al., 1987b, 1990, 1991). The conductances underlying the plateau potential could contribute to increased excitability in nigral GABAergic neurons and perhaps to changes in their firing patterns. Many of the behavioral pathologies of Parkinson's disease, and associated animal models, can be attributed to a shift away from tonic, regular firing in nigral GABAergic neurons to an irregular and bursty firing pattern that is likely caused, at least in part, by an increase in excitatory input from the subthalamic nucleus (Albin et al., 1989; DeLong, 1990; Bergman et al., 1994, 1998; Wichmann et al., 1994; Murer et al., 1997; Hurtado et al., 1999; Degos et al., 2005).
Despite the possible importance of the plateau potential to normal and pathological activity in pars reticulata, the mechanisms underlying this plateau have not been described. Here, we used in vitro visualized whole-cell current-clamp recordings to investigate the cellular mechanisms underlying plateau potential generation in rat nigral GABAergic neurons. We show that the plateau is caused by a calcium-activated nonselective cation conductance, which is activated by calcium entry through both voltage- and transmitter-dependent calcium conductances.
Portions of these data have been presented previously in abstract form (Lee and Tepper, 2005), and the baseline physiological and anatomical properties reported here include measurements from some neurons used previously (Lee and Tepper, 2007).
Materials and Methods
In vitro visualized whole-cell current-clamp recordings were performed on substantia nigra pars reticulata GABAergic neurons in 300 μm coronal slices from 21- to 28-d-old male Sprague Dawley rats. All procedures were performed with the approval of the Rutgers University Institutional Animal Care and Use Committee and in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Animals were deeply anesthetized with 150 mg/kg ketamine and 30 mg/kg xylazine intraperitoneally and transcardially perfused with chilled, modified Ringer's solution containing (in mm) 225 sucrose, 2.5 KCl, 0.5 CaCl2, 7 MgCl2, 28 NaHCO3, 1.25 NaH2PO4, 7 glucose, 1 ascorbate, and 3 pyruvate, which was bubbled with 95% O2 and 5% CO2 until the blood had been replaced with this solution. The brain was then quickly removed and trimmed to a block containing the midbrain. Slices were cut in the same medium using a Vibroslice (model NVSL; World Precision Instruments, Sarasota, FL) and immediately transferred to warmed (34°C) Ringer's solution composed of (in mm) 125 NaCl, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 1 MgCl2, 2 CaCl2, 25 glucose, 1 ascorbate, 3 pyruvate, and 0.4 myo-inositol, continuously bubbled with 95% O2 and 5% CO2 for 1 h before recording and then maintained at room temperature until use. Recordings were made in the same medium, which was heated to 34°C and perfused through the slice chamber at ∼5 ml/min. Neurons were visualized at 40× with a water-immersion objective on an Olympus (Tokyo, Japan) BX50WI fixed-stage microscope with infrared differential interference contrast microscopy and an MTI CCD-300 CCD camera (Dage-MTI, Michigan City, IN).
Pipettes were constructed from 1.5 mm outer diameter borosilicate capillary tubing (World Precision Instruments) using a PP-83 vertical pipette puller (Narishige International USA, East Meadow, NY). The pipette solution contained (in mm) 129 potassium gluconate, 11 KCl, 10 HEPES, 2 MgCl2, 10 EGTA, 3 Na2-ATP, and 0.3 Na3-GTP, into which 0.3% biocytin (Sigma-Aldrich, St. Louis, MO) was dissolved by sonication. Pipettes possessed resistances of 4–6 MΩ. In some experiments, potassium gluconate was replaced by an equimolar concentration of potassium methyl sulfate.
Recordings were amplified with a Neurodata IR-283 amplifier and sampled at 40 kHz using a Micro1401 mk II digitizer (Cambridge Electronic Design, Cambridge, UK) and a personal computer running Signal version 3.05 (Cambridge Electronic Design). Data were analyzed using the same program.
The substantia nigra pars reticulata contains both GABAergic and dopaminergic neurons. The GABAergic neurons differ from the dopaminergic neurons in terms of their higher firing rate, shorter duration action potential, short latency and small amplitude afterhyperpolarization, and a much less pronounced voltage sag in response to hyperpolarizing current pulses (see Results and Table 1) (Matsuda et al., 1987; Nakanishi et al., 1987b; Grace and Onn, 1989; Yung et al., 1991; Lacey et al., 1989; Richards et al., 1997; Iribe et al., 1999; Gulácsi et al., 2003). These criteria were the primary factors in differentiating nigral GABAergic neurons from dopaminergic neurons, and the data presented here come from neurons that were inferred to be GABAergic based on these characteristics.
In cases in which cells were later assessed using immunocytochemistry, slices were immediately placed into chilled 4% paraformaldehyde in 0.1 m phosphate buffer (PB), pH 7.4, after recording where they were stored overnight at 4°C. In these cases, all efforts were made to reduce dilution of intracellular antigens with the pipette solution by minimizing the recording time (<5 min).
Drug preparation and delivery.
Drugs were dissolved in either recording buffer or dimethyl sulfoxide where appropriate and bath applied. Final concentrations of dimethyl sulfoxide were between 0.05 and 0.1%. Tetrodotoxin citrate (TTX) and 4-(N-ethyl-N-phenylamino)-1,2-dimethyl-6-(methylamino)pyridinum chloride (ZD 7288) were purchased from Tocris Bioscience (Ellisville, MO). Bay K 8644, cadmium chloride, choline bicarbonate, choline chloride, flufenamic acid, nickel chloride, nifedipine, nimodipine, NMDA, and tetraethylammonium chloride (TEA) were purchased from Sigma-Aldrich. Calcium-free buffer was made by removing calcium chloride from the recording solution and replacing it with an equimolar concentration of cobalt chloride (Fisher BioReagents, Fair Lawn, NJ). In these cases, sodium phosphate was also omitted. Low-sodium buffer was made by substituting equimolar concentrations of choline chloride and choline bicarbonate for sodium chloride and sodium bicarbonate, respectively.
Immunocytochemistry was performed as described previously (Lee and Tepper, 2007). Briefly, sections were incubated in a solution containing 1:1500 rabbit anti-calretinin polyclonal antibody (#AB5054; Millipore, Billerica, MA) and 1:1500 mouse anti-parvalbumin monoclonal antibody (#P 3088; Sigma-Aldrich) along with 10% normal goat serum, 2% bovine serum albumin, and 0.5% Triton X-100 in PBS, pH 7.4, for 24 h at room temperature. The recorded neuron was revealed with 7-amino-4-methylcoumarin-3-acetic acid conjugated streptavidin, calretinin with cyanine 3, and parvalbumin with Oregon Green. After immunocytochemical identification, sections were incubated with avidin–biotin peroxidase and reacted with 3,3′-diaminobenzidine (0.025%) and H2O2 (0.0008%) in PB with nickel intensification. The sections were then embedded between glass slides and coverslips. Somatic measurements were obtained from neurons photographed at 40× to 100× using a Leica (Nussloch, Germany) DMRE microscope and a Hamamatsu (Bridgewater, NJ) Orca-ER digital camera and Openlab version 4.0.2 (Improvision, Boston, MA). The shape factor is equal to 4A/P2, where A is the area and P is the perimeter of the measured neuron. Neurons were drawn under a 60× oil-immersion objective using a Nikon (Tokyo, Japan) Optiphot microscope fitted with a drawing tube.
To measure the decay time of the plateau potential, plateaus were evoked by a 500 ms depolarizing pulse from neurons hyperpolarized to prevent spontaneous spiking by injection of constant current. The decay time was determined by calculating the time required for the membrane potential of the neurons to return to the baseline voltage recorded 10 ms before onset of the pulse. Only traces that were devoid of spiking after cessation of the depolarizing current pulse were used in these measurements. Traces with depolarizing current pulses of the same amplitude were chosen when comparing the decay time in different experimental conditions. If necessary, the prepulse membrane potential was maintained as close as possible to the potential measured under control conditions using different bias currents. Sag amplitudes were measured from the voltage difference between the most hyperpolarized voltage achieved during a 500 ms, 200 pA hyperpolarizing current pulse delivered from rest and the voltage measured 500 μs before current offset. Action potential and other baseline physiological measurements were obtained as described previously (Lee and Tepper, 2007). Basic action potential parameters were measured from four consecutive spontaneous action potentials and averaged. The input resistance was assessed from single, small-amplitude hyperpolarizing current pulses delivered from rest, which were sufficient to halt spiking and calculated from the difference between the voltage 50 μs before current onset and the most hyperpolarized voltage obtained during the current pulse. The threshold was defined as the membrane potential at the time point reaching 15% of the maximum rising slope achieved during the action potential upstroke, and action potential amplitudes were calculated as the difference between the threshold and the peak of the action potential. The spontaneous action potential width was measured at half-amplitude between the action potential threshold and the peak of the action potential. The afterhyperpolarization amplitude and trough time were determined by calculating the difference between the threshold and the most negative voltage measured during the afterhyperpolarization. The spontaneous firing rate was calculated for each neuron from a 1 s sweep of spontaneous activity.
All data are presented as mean ± SEM. Mean values were compared using paired or unpaired t tests where appropriate, with the criterion for significance set to p < 0.05. Statistics were computed using SAS version 8.02 (SAS Institute, Cary, NC).
Voltages reported have been corrected for the liquid junction potential, which was estimated to be 14.4 mV using JPCalc (Barry, 1994).
Measurements of the basal electrophysiological parameters from a sample of the recorded neurons are summarized in Table 1. All neurons exhibited regular spontaneous spiking at rest (Fig. 1A,D, Table 1) and a variable, although modest voltage sag in response to hyperpolarizing current pulses (mean, 6.21 ± 0.41 mV; range, 0.43–18.02 mV; n = 78). The voltage sag in nigral dopaminergic neurons is caused by the presence of Ih and is abolished by ZD 7288 (Harris and Constanti, 1995; Neuhoff et al., 2002). Similarly, ZD 7288 (30 μm) caused a significant reduction in sag amplitudes observed in nigral GABAergic neurons in response to hyperpolarizing current pulses (6.32 ± 0.83 to 2.29 ± 0.28 mV; t = −4.09; p < 0.001; n = 17), suggesting that the sag seen in these neurons is also caused by Ih (data not shown) (Atherton and Bevan, 2005).
Plateau potentials are present in a subset of nigral GABAergic neurons after depolarization
A subset of nigral GABAergic neurons responded to modest depolarizing current pulses from rest with a protracted increase in firing rate outlasting the depolarizing current pulse (Fig. 1B), whereas this effect was absent in others (Fig. 1E). When depolarized from a hyperpolarized holding potential with a 500 ms depolarizing pulse, neurons that exhibited an increase in spontaneous firing after the pulse also exhibited a depolarizing plateau potential that outlasted the current pulse (Fig. 1C). A plateau potential was observed in 269 of 867 neurons (31%) under control conditions and was frequently of sufficient amplitude to support spiking (Fig. 2B). In contrast, neurons not exhibiting increased firing after the depolarizing pulse from rest lacked the plateau potential when hyperpolarized (Fig. 1F).
The plateau potential could be observed under control conditions regardless of whether the pipette solution was potassium gluconate based or potassium methyl sulfate based (supplemental Fig. 1, available at www.jneurosci.org as supplemental material).
These results, especially the prolonged increase in firing rate after depolarization from rest, suggest that the mechanism underlying the plateau potential operates at membrane potentials seen during the spontaneous activity of the neuron. However, the basal physiological properties did not differ between neurons exhibiting the plateau, and those not exhibiting the plateau, except for the action potential amplitude, which appeared to be significantly smaller in neurons exhibiting the plateau (Table 1). Therefore, the plateau potential is most likely to affect the physiology of the neuron when it is perturbed from its spontaneous activity, as would be the case in response to depolarizing synaptic input.
In some neurons, action potentials occurring during the depolarizing current pulse exhibited acceleration in their firing frequency, suggesting that the plateau may contribute to increased firing rates during the depolarizing current pulse (Figs. 1C, 2B). There was considerable variability in the expression of this firing frequency acceleration (FFA), with some neurons exhibiting the plateau exhibiting little to no FFA (Fig. 3A,C) and other neurons not exhibiting the plateau that did show FFA.
Anatomical properties and calcium-binding protein immunoreactivity of plateau and nonplateau neurons
Like the physiological properties, the morphological properties of neurons exhibiting a plateau after depolarization were similar to those not exhibiting a plateau. The only significant difference found between neurons exhibiting a plateau after depolarization and those that did not was that the neurons exhibiting the plateau tended to have smaller areas. These results are summarized in Table 2.
Most nigral GABAergic neurons express immunoreactivity for either of the calcium-binding proteins parvalbumin or calretinin, with an exceedingly small proportion exhibiting immunoreactivity for both (Gonzalez-Hernandez and Rodriguez, 2000; Lee and Tepper, 2007). Of neurons exhibiting the plateau (n = 18), 11 expressed immunoreactivity for parvalbumin, 4 for calretinin, and 3 failed to express immunoreactivity for either calcium-binding protein. For those not exhibiting the plateau (n = 36), 14 expressed immunoreactivity for parvalbumin, 12 for calretinin, and 10 failed to express immunoreactivity for either calcium-binding protein. Therefore, parvalbumin and calretinin do not show preferential distribution within neurons based on the presence or absence of a plateau (X2 = 2.39; p = 0.30).
Anatomical reconstructions of nigral GABAergic neurons either with or without a plateau did not reveal any obvious differences in their dendritic or axonal arborizations. Neurons exhibiting a plateau and those without were both observed to issue local axon collaterals that arborized within the substantia nigra pars compacta and/or pars reticulata and were indistinguishable from pars reticulata GABAergic projection neurons identified previously in vivo and in vitro (Deniau et al., 1982; Mailly et al., 2001, 2003; Lee and Tepper, 2007).
Ionic mechanisms underlying plateau potentials in nigral GABAergic neurons
The plateau potential was not blocked by TTX in any neuron (2 μm) (Fig. 2D). When the current–voltage relation was compared between neurons exhibiting the plateau under control conditions, and those from which the plateau was absent, the latter were found to exhibit strong outward rectification that was not present in neurons exhibiting a plateau (Fig. 2E,F). However, superfusion of TEA (10–30 mm) in addition to TTX abolished the outward rectification (Fig. 2E,G) and invariably revealed a latent plateau potential in neurons not exhibiting it under control conditions (Fig. 2H). These results suggest that the plateau is resistant to block by TTX, and that most, if not all, nigral GABAergic neurons are capable of generating a plateau potential, but under most conditions, it is masked by TEA-sensitive potassium conductances.
The role of calcium
Replacement of CaCl2 with an equimolar concentration of CoCl2 in the bath led to a complete abolition of the plateau potential observed after depolarization under control conditions, as did the addition of nickel (500 μm to 1 mm) or cadmium (400 μm) (Fig. 3A–C). This was evident as a decrease in the plateau decay time after the depolarizing current pulse after removal of calcium from the recording medium (194.6 ± 24.7 to 83.4 ± 13.6 ms; t = −8.80; p < 0.01; n = 4) or by the addition of nickel (201.4 ± 18.1 to 103.5 ± 17.2 ms; t = −10.00, p < 0.01, n = 4) or cadmium (166.7 ± 19.8 to 77.1 ± 10.1 ms; t = −5.30; p < 0.01; n = 6). Similar results were obtained with the plateau unmasked by TEA in the presence of TTX (Fig. 3D–F). Under these conditions, the plateau decay time was also significantly reduced by removal of calcium (334.5 ± 32.7 to 98.2 ± 15.3 ms; t = −6.96; p < 0.01; n = 5) or the addition of nickel (266.2 ± 29.7 to 114.5 ± 13.8 ms; t = −4.70; p < 0.01; n = 9) or cadmium (285.0 ± 21.6 to 96.8 ± 23.0 ms; t = −11.53; p < 0.0001; n = 6) (Fig. 3G–I). These results suggested that the plateau is dependent on calcium.
To determine the calcium channel subtype involved, the effects of the L-type calcium channel blockers nimodipine (10 μm) and nifedipine (5–30 μm) on the plateau were assessed (Fig. 4). The plateau observed under control conditions, or in the presence of TTX, was completely abolished by nimodipine (286.3 ± 23.4 to 137.9 ± 16.8 ms; t = −5.44; p < 0.001; n = 11), as well as by two concentrations of nifedipine (30 μm; 231.7 ± 32.7 to 80.9 ± 17.5 ms; t = −5.97; p = 0.001; n = 7; 5 μm; 217.5 ± 9.4 to 116.7 ± 15.1 ms; t = −8.89; p < 0.001; n = 5). Identical results were obtained regardless of whether TTX was first used to block action potentials before application of nimodipine along with TTX (Fig. 4B) (n = 5). The plateau continued to be abolished by nimodipine even if the neurons were strongly depolarized, allowing the maximum voltage achieved during the depolarizing current pulse to reach that achieved during depolarization in the presence of TTX but without nimodipine (Fig. 4B, blue trace). Additionally, the prolonged increase in firing rate outlasting depolarizing current pulses delivered from rest was blocked by nimodipine (Fig. 4C,D). In contrast, neither nimodipine nor nifedipine was effective at blocking the plateau unmasked by TEA in the presence of TTX (218.5 ± 28.6 to 236.5 ± 30.9 ms; t = 1.18; p = 0.28; n = 8) (Fig. 4E). These results suggest that although the plateau potential is dependent on an L-type calcium conductance control conditions, the calcium conductance does not underlie the plateau itself.
In addition to being attenuated when calcium conductances were blocked, the plateau was enhanced when calcium conductances were increased (Fig. 5). Both the L-type calcium channel agonist Bay K 8644 (5 μm) and the NMDA receptor agonist NMDA (20–30 μm) enhanced plateaus, which were observed under control conditions and elicited a plateau after depolarization in neurons not previously exhibiting it under control conditions or if action potentials were first blocked with TTX (Bay K 8644: 105.4 ± 7.2 to 279.5 ± 13.6 ms, t = 15.75, p < 0.0001, n = 27; NMDA: 97.3 ± 8.2 to 228.7 ± 18.6 ms, t = 9.34, p < 0.0001, n = 20) (Fig. 7B,D).
The role of sodium
The previous experiments established that the plateau is dependent on calcium conductances but implied that the conductance underlying the plateau does not use calcium as its main charge carrier. Rather, calcium appeared to be activating a conductance that uses another ion as its main charge carrier. The contribution of sodium to the plateau potential was assessed by replacing the majority of the extracellular sodium ions with an equimolar concentration of choline (see Materials and Methods). Because this also abolished action potential discharge, in some cases, TTX (2 μm) was applied before switching to low-sodium buffer to minimize the differences between the two conditions. Elimination of most sodium ions from the extracellular solution completely abolished the plateau potential regardless of whether action potentials were first blocked by TTX (262.6 ± 18.0 to 107.4 ± 7.3 ms; t = −8.76; p < 0.0001; n = 13) (Fig. 6A,B). The plateau potential continued to be abolished in low-sodium buffer with TTX when the neuron was strongly depolarized, allowing the voltage achieved during the depolarizing current pulse to reach that achieved during TTX in standard buffer, thus arguing against a threshold effect (Fig. 6C). Additionally, the plateau unmasked by TEA in the presence of TTX was also abolished under conditions of low sodium (266.4 ± 25.7 to 108.8 ± 14.9 ms; t = −7.32; p < 0.0001; n = 10) (Fig. 6D). Thus, the plateau potential is caused by a TTX-resistant, calcium-activated conductance that uses sodium as its main charge carrier.
As shown previously, increasing calcium conductances through L-type calcium channels with Bay K 8644 as well as increasing NMDA conductances could elicit the plateau. This effect was also observed when action potentials were first blocked with TTX. The plateau elicited by depolarizing current pulses under these treatments was similarly completely abolished in low-sodium buffer, again regardless of whether or not sodium spiking was first blocked with TTX (Bay K 8644: 300.6 ± 27.2 to 122.3 ± 15.8 ms, t = −6.49, p < 0.001, n = 9; NMDA: 223.7 ± 20.8 to 99.1 ± 5.3 ms, t = −5.42, p < 0.001, n = 10) (Fig. 7). When these neurons were strongly depolarized in low-sodium buffer, allowing the membrane potential achieved during the depolarizing current pulse to reach that achieved with TTX and either Bay K 8644 or NMDA, the plateau continued to be abolished (data not shown).
A calcium-activated nonselective cation conductance underlies the plateau potential in nigral GABAergic neurons
The aforementioned results revealed that the plateau potential is dependent on calcium, but the conductance underlying the plateau uses sodium as a charge carrier. These are characteristics of a calcium-activated nonselective cation conductance (ICAN), which can be blocked by flufenamic acid (Shaw et al., 1995; Partridge and Valenzuela, 2000; Hill et al., 2004).
The ICAN blocker flufenamic acid (10–200 μm) was applied to neurons exhibiting a plateau under various conditions (Fig. 8). Flufenamic acid completely abolished the plateau observed after depolarization under control conditions or in the presence of TTX (246.3 ± 21.3 to 67.4 ± 13.4 ms; t = −9.26; p < 0.0001; n = 9). In addition, when the plateau was abolished by flufenamic acid in the presence of TTX, depolarizing the neuron to the level achieved in TTX alone did not reveal the plateau (Fig. 8B, blue trace). The prolonged increase in firing rate observed when neurons exhibiting the plateau were depolarized from rest (Fig. 8C) was blocked by flufenamic acid (10 μm) (Fig. 8D). The plateau unmasked by TEA in the presence of TTX was also abolished (284.1 ± 26.2 to 112.9 ± 9.8 ms; t = −7.12; p < 0.0001; n = 11).
The plateau observed after increasing calcium conductances with either Bay K 8644 or NMDA was similarly abolished by flufenamic acid (Bay K 8644: 256.6 ± 17.9 to 84.6 ± 12.0 ms, t = −9.33, p < 0.0001, n = 8; NMDA: 235.1 ± 42.9 to 62.5 ± 11.8 ms, t = −5.09, p < 0.01, n = 7). These results suggest that ICAN underlies the plateau potential observed in nigral GABAergic neurons after depolarization under control conditions as well as that observed after depolarization while increasing calcium conductances or blocking potassium conductances.
Increasing calcium conductances may cause burst firing in nigral GABAergic neurons
Application of Bay K 8644 (5 μm) caused some neurons initially exhibiting regular spiking spontaneous activity to exhibit bursting both during spontaneous activity with a small hyperpolarizing bias current (Fig. 9B) and in response to depolarization from a hyperpolarized holding potential (Fig. 9C,D). In some cases in which burst firing was observed during depolarization, a plateau potential was also present (Fig. 9C). In others, burst firing outlasted the depolarizing current pulse and appeared to be supported by the plateau (Fig. 9D). The burst firing exhibited by these neurons in the presence of Bay K 8644 was similar to the rhythmic activity exhibited by nigral output neurons in response to dopaminergic denervation, as in Parkinson's disease (Bergman et al., 1994, 1998; Nini et al., 1995; Murer et al., 1997; Hurtado et al., 1999; Raz et al., 2000; Degos et al., 2005; Heimer et al., 2006). These results suggest that activation of L-type calcium conductances leads to activation of ICAN, which likely contributes to burst firing in these neurons.
Nigral GABAergic neurons exhibit a plateau potential when depolarized from a hyperpolarized membrane potential that is caused by a nonselective cation conductance activated by calcium entry through L-type calcium conductances. In most neurons, the plateau is masked by TEA-sensitive potassium conductances, leading to its appearance after a depolarizing current pulse in only approximately one-third of neurons in control conditions. Blocking these potassium conductances or increasing calcium conductances through L-type calcium channels or NMDA receptors leads to the appearance of the plateau after depolarizing current pulses in the remaining neurons. Nigral GABAergic neurons are not unique in their ability to exhibit plateau potentials, and much can be inferred about the function of the plateau potential in these neurons by examining the effects and molecular substrates of plateau potentials in other neurons of the CNS.
Although the exact type of channel underlying the plateau potential is unknown, given its activation by calcium and permeability to sodium, it is likely to be a melastatin-related transient receptor potential cation channel (TRPM) and most likely a TRPM2 channel. TRPM2 channels are found primarily in the brain, are regulated by calcium, and are nonselective to monovalent cations as well as to calcium (Fleig and Penner, 2004). In addition, human recombinant TRPM2 channels have been shown to be blocked by flufenamic acid (Hill et al., 2004).
Plateau potentials in other neurons
In the basal ganglia, the presence of a plateau potential and calcium-activated nonselective cation conductance is not unique to GABAergic neurons in the substantia nigra. Nigral dopaminergic neurons exhibit a plateau potential in response to depolarizing current pulses after potassium conductances are blocked (Mercuri et al., 1994; Ping and Shepard, 1999). Plateau potentials are also observed in nigral dopaminergic neurons after depolarization by glutamate in the presence of nicotinic acetylcholine receptor activation and can be abolished with flufenamic acid (Yamashita and Isa, 2003a). The functional significance of ICAN in the dopaminergic neurons may be to augment excitatory inputs and contribute to burst firing (Yamashita and Isa 2003a, 2004). In these neurons, activation of ICAN appears to involve the calcium/calmodulin-CaMKIII (calcium/calmodulin-dependent protein kinase II) pathway (Yamashita and Isa, 2003b).
Neostriatal medium spiny neurons respond to metabolic stress with depolarization, which is blocked by flufenamic acid (Bao et al., 2005), and calcium-activated nonselective cation conductances may contribute to a plateau potential observed in those neurons (Hernandez-Lopez et al., 1997; Vergara et al., 2003).
Closely related to the substantia nigra pars reticulata, GABAergic output neurons of the entopeduncular nucleus also exhibit plateau depolarizations in response to depolarizing current pulses as well as stimulation of the subthalamic nucleus (Nakanishi et al., 1990, 1991).
The best studied example of ICAN in the basal ganglia is in the subthalamic nucleus, in which a subset of neurons exhibits a plateau potential in response to depolarization from a hyperpolarized holding potential either by depolarizing current pulses or in response to EPSPs (Nakanishi et al., 1987a; Overton and Greenfield, 1995; Beurrier et al., 1999; Otsuka et al., 2001, 2004; Garcia et al., 2003). Like the conductance underlying the plateau potential in nigral GABAergic neurons described here, an inward current can be induced by NMDA or activation of L-type calcium conductances in subthalamic nucleus neurons and is blocked by flufenamic acid (Baufreton et al., 2003; Zhu et al., 2005).
The function of ICAN in nigral GABAergic neurons is likely similar to that in subthalamic nucleus neurons. ICAN and its associated plateau potential is important in controlling the firing pattern of subthalamic nucleus neurons which exhibit tonic, single spike firing, burst firing, as well as a quiescent plateau depolarization under some conditions (Beurrier et al., 1999; Zhu et al., 2004; Kass and Mintz, 2006). Burst firing can be observed in these neurons after release of hyperpolarizing current pulses or application of NMDA, both of which also elicit the plateau potential in subthalamic neurons (Beurrier et al., 1999; Zhu et al., 2004, 2005). Cells that lack the plateau have been reported to similarly lack the capacity to exhibit burst firing (Beurrier et al., 1999). Flufenamic acid abolishes NMDA-induced bursting, suggesting that the presence of the plateau potential leads to the emergence of more bursty firing patterns in these neurons (Zhu et al., 2004). Thus, the plateau potential can control the discharge pattern of neurons by increasing their excitability and allowing them to fire in bursts.
Functional significance of the plateau potential in nigral GABAergic neurons
Nigral GABAergic neurons receive a glutamatergic input from the subthalamic nucleus (Hammond et al., 1978; Van Der Kooy and Hattori, 1980; Robledo and Feger, 1990; Bevan et al., 1994; Iribe et al., 1999; Ibanez-Sandoval et al., 2006) and can respond to stimulation of the subthalamic nucleus with plateau potentials (Nakanishi et al., 1987b). We have shown here that one mechanism that can elicit the plateau potential is NMDA receptor activation. It is possible that input from the subthalamic nucleus could contribute to the emergence of plateau potentials in nigral GABAergic neurons, which could lead to increased excitability and burst firing in the nigral output neurons. A recent study has demonstrated that nigral GABAergic neurons can exhibit a burst of action potentials in response to electrical stimulation of the subthalamic nucleus in vitro (Shen and Johnson, 2006). As shown here, activation of L-type calcium conductances caused bursting in some neurons. Activation of either NMDA receptors or L-type calcium channels leads to the activation of ICAN, which likely underlies the burst firing. Because the plateau potential can be enhanced by attenuating potassium conductances, it is likely that by increasing calcium conductances while at the same time attenuating potassium conductances, the effect of ICAN on the firing pattern could be heightened and become more evident.
The functional significance of burst firing in both the subthalamic nucleus and substantia nigra pars reticulata may be best recognized in terms of pathological conditions such as Parkinson's disease. Although some of the motor symptoms of Parkinson's disease are likely caused by increased output from the substantia nigra pars reticulata and internal globus pallidus (Albin et al., 1989; DeLong, 1990), the firing pattern of basal ganglia output neurons has been increasingly recognized as an important variable in basal ganglia and motor function.
The firing of nigral and internal globus pallidus GABAergic neurons is largely tonic and regular under most conditions but acquires a bursty, correlated firing pattern in Parkinson's disease or after dopaminergic denervation (MacLeod et al., 1990; Hutchison et al., 1994; Nini et al., 1995; Murer et al., 1997; Bergman et al., 1998; Hurtado et al., 1999; Wichmann et al., 1999; Raz et al., 2000; Ruskin et al., 2002; Boraud et al., 2005). Additionally, glutamatergic inputs to nigral output neurons from the subthalamic nucleus also exhibit increased output and burst firing under conditions of dopaminergic denervation (Albin et al., 1989; DeLong, 1990; Bergman et al., 1994; Rodriguez-Oroz et al., 2001; Levy et al., 2002; Sharott et al., 2005). This would presumably lead to increased, rhythmic activation of postsynaptic NMDA receptors on nigral output neurons. As shown here, activation of NMDA receptors increases activation of the ICAN, which likely contributes to the rhythmic firing of nigral output neurons in Parkinson's disease.
Whether the loss of dopamine in Parkinson's disease would result in upregulation of ICAN, or calcium conductances activating ICAN, is unknown. Subthalamic nucleus neurons recorded in vitro from dopaminergic denervated mice show increases in irregular firing, which are resistant to blockade of ionotropic glutamatergic and GABAergic receptors, suggesting that intrinsic mechanisms such as the plateau potential may be upregulated in those neurons and contribute to the shift away from more regular firing patterns (Wilson et al., 2006). However, it has also been reported that dopamine receptor activation potentiates burst firing in subthalamic nucleus neurons in vitro (Baufreton et al., 2003). It is nevertheless highly likely that the plateau potential contributes significantly to the burst firing observed in nigral output neurons in Parkinson's disease regardless of whether it is upregulated by the loss of dopamine or more strongly activated through increased activation of NMDA or other voltage-gated calcium channels.
Present treatments for Parkinson's disease such as high-frequency stimulation of the subthalamic nucleus and dopaminergic replacement have been shown to exert their effects, at least in part, by regularizing abnormal oscillations in the basal ganglia (Bergman et al., 1990; Tseng et al., 2001; Levy et al., 2002; Brown et al., 2004; Degos et al., 2005; Meissner et al., 2005; Heimer et al., 2006; Kuhn et al., 2006). It is conceivable that by blocking ICAN, the excitability of nigral output neurons could be decreased and their firing returned to more regular patterns thereby helping to ameliorate the symptoms of Parkinson's disease and other dyskinesias.
This work was supported by National Institute of Neurological Disorders and Stroke Grant NS34865. We thank Fulva Shah for excellent technical assistance and Prof. Denis Paré for helpful comments on this manuscript.
- Correspondence should be addressed to James M. Tepper, Center for Molecular and Behavioral Neuroscience, Rutgers, The State University of New Jersey, 197 University Avenue, Newark, NJ 07102.