We used a preparation of acutely dissociated neurons to quantify the ionic currents driving the spontaneous firing of substantia nigra pars compacta neurons, isolated from transgenic mice in which the tyrosine hydroxylase promoter drives expression of human placental alkaline phosphatase (PLAP) on the outer surface of the cell membrane. Dissociated neurons identified by fluorescent antibodies to PLAP showed firing properties similar to those of dopaminergic neurons in brain slice, including rhythmic spontaneous firing of broad action potentials and, in some cells, rhythmic oscillatory activity in the presence of tetrodotoxin (TTX). Spontaneous activity in TTX had broader, smaller spikes than normal pacemaking and was stopped by removal of external calcium. Normal pacemaking was also consistently silenced by replacement of external calcium by cobalt and was slowed by more specific calcium channel blockers. Nimodipine produced a slowing of pacemaking frequency. Pacemaking was also slowed by the P/Q-channel blocker ω-Aga-IVA, but the N-type channel blocker ω-conotoxin GVIA had no effect. In voltage-clamp experiments, using records of pacemaking as command voltage, cobalt-sensitive current and TTX-sensitive current were both sizeable at subthreshold voltages between spikes. Cobalt-sensitive current was consistently larger than TTX-sensitive current at interspike voltages from −70 to −50 mV, with TTX-sensitive current larger at voltages positive to −45 mV. These results support previous evidence for a major role of voltage-dependent calcium channels in driving pacemaking of midbrain dopamine neurons and suggest that multiple calcium channel types contribute to this function. The results also show a significant contribution of subthreshold TTX-sensitive sodium current.
Midbrain dopaminergic neurons play an important role in many brain functions, including motor control, motivation, reward, and cognition (Nieoullon, 2002; Schultz, 2002; Wise, 2004). Dysfunction of the dopaminergic system is involved in many neurological disorders, including Parkinson's disease, schizophrenia, and drug addiction (Koob et al., 1998; Lewis and Lieberman, 2000; Dauer and Przedborski, 2003). Midbrain dopamine neurons are spontaneously active (Grace and Bunney, 1984) and tonic firing is likely important for maintaining background dopamine levels in target areas.
Spontaneous activity of midbrain dopaminergic neurons apparently relies on a different ionic mechanism than most other pacemaking cells in the CNS. Blocking the hyperpolarization-activated cation current Ih has no effect on pacemaking in most midbrain dopamine neurons (Mercuri et al., 1995), although in a subpopulation of neurons, pacemaking is slowed but not stopped (Seutin et al., 2001; Neuhoff et al., 2002). In contrast to most other pacemaking neurons in the CNS where blocking calcium entry has little effect on pacemaking or even speeds it (Llinás and Alonso, 1992; Takakusaki and Kitai, 1997; Bevan and Wilson, 1999; Raman and Bean, 1999; Beurrier et al., 2000; Do and Bean, 2003), midbrain dopamine neurons are unusual in that blocking calcium current by cadmium or cobalt completely silences pacemaking (Fujimura and Matsuda, 1989; Grace and Onn, 1989; Harris et al., 1989). In a powerful experimental approach combining calcium imaging with voltage clamp, Wilson and Callaway (2000) showed that a subthreshold calcium current in midbrain dopamine neurons activates steeply over the voltage range from −60 to −30 mV and shows little inactivation (in contrast to expectations from low-threshold T-type current). Voltage clamp in dissociated neurons shows the existence of multiple components of calcium current, including L-type, N-type, and P/Q-type currents (Cardozo and Bean, 1995; Durante et al., 2004). The calcium current involved in pacemaking is probably at least partly L-type current, because pacemaking is slowed or stopped by L-type channel blockers but not N-type channel blockers (Nedergaard et al., 1993; Mercuri et al., 1994) (but see Fujimura and Matsuda, 1989; Kang and Kitai, 1993b). P/Q-type channel blockers have not yet been tested on pacemaking. Also, the possible role of subthreshold TTX-sensitive sodium current in pacemaking of midbrain dopamine neurons has received little attention, although it drives pacemaking of many other neuronal pacemakers (Pennartz et al., 1997; Bevan and Wilson, 1999; Raman and Bean, 1999; Taddese and Bean, 2002).
We have examined the currents driving pacemaking of midbrain dopamine neurons using the action potential clamp method, in which records of pacemaking activity are used as the voltage command in voltage-clamp experiments and currents flowing during the interspike interval can be directly quantified. Consistent with a dominant role of calcium current in pacemaking, we found a sizeable interspike current sensitive to replacement of extracellular calcium by cobalt. TTX-sensitive sodium current also contributes significant charge entry during the interspike interval. Surprisingly, the selective P/Q-channel blocker ω-Aga-IVA had a dramatic effect in slowing or stopping pacemaking.
Materials and Methods
Preparation of acutely dissociated neurons.
Acutely dissociated neurons were obtained from transgenic mice (14–18 d old) in which catecholaminergic neurons in the CNS express human placental alkaline phosphatase (PLAP) on the outer surface of the cell membrane (Gustincich et al., 1997). After anesthesia and decapitation, the brain was removed and placed in ice-cold dissociation medium consisting of the following (in mm): 82 Na2SO4, 30 K2SO4, 5 MgCl2, 10 glucose, 10 HEPES, and 0.001% phenol red indicator; pH was adjusted to 7.4 with NaOH and the solution was continuously bubbled with 100% O2. The brain was cut into 400 μm coronal slices with a vibrating tissue slicer. The region containing the substantia nigra pars compacta (SNc) was dissected out from the slices and incubated in dissociation medium containing 3 mg/ml protease type XXIII (Sigma, St. Louis, MO) for 30–40 min at 35°C. After enzymatic digestion, the tissue was washed with dissociation medium containing 1 mg/ml bovine serum albumin (Sigma) and 1 mg/ml trypsin inhibitor (Sigma) and triturated using fire-polished Pasteur pipettes of varying bores. The resulting cell suspensions were centrifuged at 1000 rpm for 3 min and the pellet was resuspended in Tyrode's solution. The dissociated neurons were directly plated on the glass bottom of concanavalin A-coated (1 mg/ml) recording chambers and kept at room temperature at least 1 h before commencement of recordings. SNc neurons were identified by labeling of their membrane with a monoclonal antibody to PLAP (De Groote et al., 1983) directly conjugated to the fluorochrome Cy3 (E6-Cy3). E6-Cy3-stained SNc neurons were identified by scanning the coverslip with an inverted microscope (Diaphot 300; Nikon, Tokyo, Japan) equipped with epifluorescence. After identification of a PLAP-positive cell, the remaining procedures were performed using visible light with Nomarski optics.
Preparation of brain slices.
Black Swiss mice (14–42 d old) were anesthetized with isofluorane. After decapitation, the brain was removed and placed in ice-cold artificial CSF (ACSF) enriched with sucrose. This solution, referred as slicing solution, consisted of the following (in mm): 87 NaCl, 2.5 KCl, 1.25 NaH2PO4, 26 NaHCO3, 1 MgCl2, 20 glucose, and 75 sucrose, continuously bubbled with 95%O2/5%CO2. Coronal slices (200–250 μm) containing the substantia nigra were obtained using a vibrating tissue slicer (DTK-1000; DSK, Dosaka, Japan). Slices were then held at room temperature in the slicing solution supplemented with 0.5 mm CaCl2. For recordings, slices were transferred to a recording chamber and mounted on the stage of an upright microscope (Nikon E600FN). Neurons were visualized using infrared differential interference contrast videomicroscopy (DAGE-MTI, Michigan City, IN). We targeted neurons with large cell bodies located near the ventrolateral margin of the SNc (i.e., close to the margin between the SNc and substantia nigra pars reticulata and far from the ventral tegmental area). Slices were continuously perfused at a rate of 2 ml/min with ACSF consisting of the following (in mm): 125 NaCl, 2.5 KCl, 1.25 NaH2PO4, 26 NaHCO3, 2 CaCl2, 1 MgCl2, and 20 glucose, bubbled with 95%O2/5%CO2.
Recordings in dissociated cells were made using an Axopatch 200B amplifier (Molecular Devices, Foster City, CA). The resistance of the patch pipettes filled with internal solution was 2.5–3.5 MΩ. The patch pipettes were wrapped to near the tip with stretched Parafilm to reduce capacitance. In voltage-clamp experiments, the series resistance of the pipettes was 5–10 MΩ and was compensated up to ∼80%. Current-clamp recordings were performed in the fast current-clamp mode to reduce distortion of action potentials (Magistretti et al., 1996, 1998). Current and voltage traces were low-pass-filtered at 2–5 kHz using the internal Bessel filter of the amplifier and sampled at 20–100 μs.
Recordings in brain slice were made using a MultiClamp 700B amplifier (Molecular Devices) using pipettes of resistance 1.3–1.8 MΩ, wrapped with Parafilm to reduce capacitance. In voltage-clamp experiments, the series resistance of the pipettes was compensated by 30–40%.
Solutions and pharmacology.
For recordings in isolated cells, unless noted otherwise, the extracellular solution for current-clamp recordings contained the following (in mm):137 NaCl, 5.4 KCl, 1.8 CaCl2, 1 MgCl2, 5 HEPES, and 10 glucose, pH adjusted to 7.4 with NaOH. The intracellular solution contained the following (in mm):120 K-gluconate, 6 NaCl, 4 MgCl2, 2 Mg-ATP, 0.3 Na-GTP, 14 phosphocreatine, 5 EGTA, 0.5 CaCl2, 10 glucose, and 10 HEPES, pH adjusted to 7.2 with KOH. In some of the experiments with nimodipine, KMeSO4 (ICN Pharmaceuticals, Costa Mesa, CA) was substituted for K-gluconate. Sodium current was obtained as the current sensitive to 1 μm TTX with most voltage-dependent potassium and calcium currents blocked with an extracellular solution containing the following (in mm): 137 NaCl, 1.8 CoCl2, 1 MgCl2, 5.4 tetraethylammonium (TEA)-Cl, 10 4-aminopyridine, 5 HEPES, and 10 glucose, pH adjusted to 7.4 with NaOH. To determine calcium currents, equimolar cobalt was substituted for calcium in an extracellular solution containing the following (in mm): 137 NaCl, 1.8 CaCl2, 1 MgCl2, 5.4 TEA-Cl, 10 4-aminopyridine, 5 HEPES, 10 glucose, and 1 μm TTX, pH adjusted to 7.4 with NaOH.
A stock solution of TTX (Research Biochemicals, Natick, MA) was made in 2 mm citric acid and stored at −20°C. Nimodipine (Research Biochemicals) was prepared as a 50 mm stock solution in polyethylene glycol and kept at 4°C. ZD 7288 was obtained from Tocris Bioscience (Bristol, UK) and prepared as a 10 mm stock solution in water. For cells in which ZD 7288 had no effect on firing, there were always other cells studied the same day in which the same ZD 7288 solution did affect firing, providing a positive control. ω-Conotoxin GVIA and ω-Aga-IVA (Peptides International, Louisville, KY) were prepared as a 1000× stock solution in water and kept at 4°C. Drugs were diluted to a final concentration in the extracellular solution before the experiment and applied onto the cell by gravity flow through an array of microcapillary tubes controlled by a rapid solution changer (RSC-200; Biologic, Claix, France). Solution exchange was complete in <1 s. ω-Aga-IVA was applied at a concentration of 200 nm. ω-Aga-IVA was applied without the addition of cytochrome C, often included to reduce adsorption of toxin to perfusion tubing (Mintz et al., 1992). A control experiment recording voltage-clamped calcium current in a cerebellar Purkinje neuron showed that application of 200 nm ω-Aga-IVA applied without cytochrome C (using the same perfusion apparatus used in studying the dopaminergic neurons) produced a saturating block of P-type current, with no additional effect after switching to 200 nm ω-Aga-IVA plus 1 mg/ml cytochrome C. Thus, although we cannot rule out some reduction in effective toxin concentration caused by adsorption to tubing, the toxin delivered was sufficient to completely block P-type calcium channels.
For slice recordings, slices were continuously perfused with ACSF. Seal and whole-cell configuration were obtained in ACSF. Afterward, the experiment was performed using HEPES-based extracellular solutions applied locally in the vicinity of the cell. Synaptic transmission was blocked by supplementing the extracellular solution with kynurenic acid (2 mm) and picrotoxin (50 μm). The extracellular solution for current-clamp recordings consisted of the following (in mm): 151 NaCl, 2.5 KCl, 2 CaCl2, 1 MgCl2, 10 HEPES, and 13 glucose, with pH adjusted to 7.4 with NaOH. The intracellular solution contained the following (in mm): 135 K-Methanesulfonate, 10 NaCl, 2 MgCl2, 4 Mg-ATP, 0.3 Na-GTP, 14 phosphocreatine, 0.1 EGTA, and 10 HEPES, with pH adjusted to 7.2 with KOH. In voltage clamp, sodium current was isolated by determination of the current sensitive to 1 μm TTX in a solution consisting of the following (in mm): 151 NaCl, 2.5 KCl, 2 CoCl2, 1 MgCl2, 15 TEA-Cl, 10 HEPES, and 13 glucose, pH adjusted to 7.4 with NaOH. Calcium current was determined by substitution of equimolar cobalt for calcium in an extracellular solution containing the following (in mm): 151 NaCl, 2.5 KCl, 15 TEA-Cl, 2 CaCl2, 1 MgCl2, 10 HEPES, 13 glucose, and 1 μm TTX, pH adjusted to 7.4 with NaOH. In some experiments on neurons in brain slice, cobalt substitution for calcium was done on a background of a solution in which TEA fully replaced sodium [(in mm): 153.5 TEA-Cl, 2 CaCl2, 1 MgCl2, 10 HEPES, 13 glucose, and 1 μm TTX, pH adjusted to 7.4 with NaOH], which might in principle better isolate calcium currents by blocking calcium-activated potassium currents more fully (as well as any currents carried by sodium that might be affected by cobalt for calcium substitution). However, recordings were less stable under these conditions, and there was little difference in the voltage dependence or magnitude of the interspike cobalt-sensitive current between experiments done with a background of 151 mm NaCl and 15 mm TEA-Cl [17 ± 2 pC, integrated from the afterhyperpolarization to −50 mV; postnatal day 35 (P35)–P42; n = 8] and those done with a background of 153.5 TEA-Cl (14 ± 2 pC; P35–P42; n = 8). Therefore, the values for brain slice experiments given in the text are confined to experiments done with a background of 151 mm NaCl and 15 mm TEA-Cl.
All of the experiments were performed at room temperature (22 ± 2°C).
Acquisition and analysis.
Data acquisition was made by using pClamp 8 or 9 (Molecular Devices) and data analysis was performed using Clampfit 8 (Molecular Devices) and IgorPro version 3.14 (WaveMetrics, Lake Oswego, OR). Reported voltages were corrected for junction potential between the internal solution and the extracellular solution measured using a flowing 3 m KCl electrode as described by Neher (1992). Data are presented as mean ± SEM.
Mouse brains were perfused with 2% formaldehyde in 0.15 m Sörensen phosphate buffer, pH 7.4, cryoprotected in 20% sucrose, and frozen in partially solidified dichlorodifluoromethane kept in liquid nitrogen. Cryostat sections were blocked in 10% normal goat serum (Vector Laboratories, Burlingame, CA), 2% fish gelatin (Goldmark Biologicals, Phillipsburg, NJ), and 0.2% bovine serum albumin (Sigma) in PBS for 1 h. They were incubated overnight in mouse monoclonal to tyrosine hydroxylase (TH) (Immunostar, Hudson, WI) 1:1000 in block solution, followed by Alexa Fluor goat anti-mouse IgG (Invitrogen, Eugene, OR) 1:500 in block solution.
For immunocytochemistry on acutely dissociated neurons, neurons were plated on a coverslip, fixed with 0.5% formaldehyde in 0.15 m Sörensen phosphate buffer, pH 7.4, washed in block solution containing 0.3% Triton X-100, and treated with a mixture of 1:100 E6 anti-human monoclonal antibody to PLAP (Gustincich et al., 1997) directly conjugated to Cy3 and 1:500 polyclonal sheep antibody to TH (Novus Biologicals, Littleton, CO) in block solution. Secondary antibody was Alexa Fluor 488 donkey anti-sheep IgG (Invitrogen) 1:500.
For demonstration of PLAP activity, vibratome sections of the fixed midbrain were warmed up to 65°C for 30 min in PBS to inactivate other phosphatases. They were subsequently incubated in the dark for 3 h in 0.1% 5-bromo-4-chloro-3-indolyl phosphate (Invitrogen) and 1% nitroblue tetrazolium in 0.1 m Tris-HCl buffer, pH 9.5, containing 0.1 m NaCl and 0.05 m MgCl2. After rinsing with 0.02 m EDTA in PBS, the sections were mounted in a polyvinyl alcohol resin (Vinol Grade 205; Air Products and Chemicals, Allentown, PA).
Images were captured by using a Nikon Eclipse E600 microscope equipped with a SPOT RT Slider camera (Diagnostic Instruments, Sterling Heights, MI).
Properties of spontaneous firing in dissociated neurons
Previous work has shown strong involvement of calcium current in generating the spontaneous firing of dopaminergic neurons in the SNc (Grace and Onn, 1989; Harris et al., 1989; Yung et al., 1991; Nedergaard and Greenfield, 1992; Kang and Kitai, 1993a,b; Nedergaard et al., 1993; Mercuri et al., 1994; Ping and Shepard, 1996; Wilson and Callaway, 2000). Also, dopaminergic SNc neurons express prominent Ih, which appears to make a significant contribution to pacemaking in a subpopulation of midbrain dopamine neurons (Neuhoff et al., 2002). The calcium current driving pacemaking in midbrain dopamine neurons appears to be especially prominent in the dendrites of the cells (Wilson and Callaway, 2000), and in some types of neurons, including hippocampal and cortical pyramidal neurons, Ih is also present at higher density in dendrites than in cell bodies (Magee, 1998; Williams and Stuart, 2000; Lorincz et al., 2002; Kole et al., 2006). This makes it uncertain whether the ionic mechanism of pacemaking can fruitfully be studied in dissociated neurons, which have the advantage of allowing higher-fidelity voltage clamp and faster solution exchange but the potential disadvantage of lacking most of the dendritic tree. Therefore, we first tested how well the basic properties of pacemaking are preserved in dissociated neurons by examining how spontaneous firing of dissociated cells is affected by block of Ih, calcium current, and sodium current, all of which have been studied previously using dopaminergic neurons in brain slice.
To identify acutely dissociated SNc neurons, we used a transgenic mouse line in which catecholaminergic neurons in the CNS express PLAP on the outer surface of the cell membrane (Gustincich et al., 1997). The line was obtained by introducing into the mouse genome PLAP cDNA linked to a promoter sequence of the gene for tyrosine hydroxylase (Banerjee et al., 1992). Figure 1 illustrates the location and morphology of midbrain dopaminergic neurons expressing PLAP and compares expression of the transgene with tyrosine hydroxylase expression. Cells staining for PLAP (Fig. 1A,C) were also labeled by antibodies to tyrosine hydroxylase (Fig. 1B) and vice versa. Similarly, after enzymatic digestion and mechanical trituration of the SNc, a population of isolated neurons were stained by both a monoclonal antibody to PLAP directly conjugated to the Cy3 fluorophore (Fig. 1D, left) and by antibodies to tyrosine hydroxylase (Fig. 1D, middle). Because PLAP resides on the external surface of the membrane, living unfixed dopaminergic neurons could be fluorescently labeled by the monoclonal Cy3-conjugated antibody after the neurons were dissociated from the SNc. This allowed electrophysiological recording from identified dopaminergic neurons.
Labeled dissociated neurons were spontaneously active in the cell-attached configuration, as evidenced by regular action currents when the pipette was sealed onto the cell. After breaking the cell membrane and switching to whole-cell current-clamp configuration, the vast majority of neurons continued to fire spontaneously (Fig. 2). Spontaneous firing in acutely dissociated neurons occurred with an average rate of 3.4 ± 0.1 Hz (n = 62) and was usually quite rhythmic, with an average coefficient of variation of interspike intervals of 45 ± 4% (n = 62). Spikes reached an average peak of +31 ± 1 mV and had an afterhyperpolarization averaging −64 ± 1 mV (n = 62). Action potentials were broad (width at half-maximum amplitude, 3.1 ± 0.1 ms; n = 62) with a prominent shoulder.
These results show that dissociated neurons retain the ionic machinery necessary to produce spontaneous pacemaking and are consistent with previous studies showing that pacemaking activity can be retained in acutely dissociated dopaminergic neurons (Hainsworth et al., 1991; Cardozo and Bean, 1995). In fact, pacemaking in the dissociated neurons is somewhat faster than is typical in brain slice recordings at room temperature (1–2 Hz). To see whether the basic ionic properties of pacemaking are similar in dissociated cells as in brain slice, we examined the effects on pacemaking of blocking Ih, calcium current, and sodium current. First, we tested the effect of blocking Ih by ZD 7288 (Fig. 2). Two different populations of neurons could be distinguished based on the effects of ZD 7288. In 17 of 22 neurons tested, 30 μm ZD 7288 produced a clear slowing of firing (Fig. 2A), by an average of 52 ± 6%. The slowing of firing was accompanied by a small hyperpolarization (measured at the trough, by −1.1 ± 0.2 mV), as expected from block of an inward current carried by Ih between spikes. In the other five of the 22 neurons tested (Fig. 2C), ZD 7288 had no effect on firing rate (change of + 2 ± 5%) and either had no effect on interspike voltage or produced a small depolarization (average change in trough by +1.2 ± 0.4 mV in these five cells).
These results are consistent with previous reports of two distinct populations of SNc dopaminergic neurons studied in brain slice, one in which ZD 7288 slows firing by ∼40% and one in which ZD 7288 has no effect (Seutin et al., 2001; Neuhoff et al., 2002; see also Mercuri et al., 1995). Neuhoff et al. (2002) found that the two populations of SNc neurons showing differential responses to ZD 7288 had different current densities of Ih and could also be distinguished by the timing of rebound firing after a hyperpolarizing current pulse, with those responding to ZD 7288 showing a shorter latency to firing. We saw the same correlation in dissociated neurons. The dissociated neurons in which ZD 7288 slowed spontaneous firing (Fig. 2A,B) had an average latency to firing of 229 ± 18 ms (n = 17), considerably faster than the group of neurons (Fig. 2C,D) in which ZD 7288 did not affect firing (854 ± 55 ms; n = 5).
These results show that spontaneous firing in dissociated neurons is affected by ZD 7288 in a very similar manner as for intact cells studied in brain slice. The dissociated neurons clearly retain substantial Ih, because all neurons (even those in which firing was not affected by ZD 7288) showed the characteristic sag in voltage during hyperpolarizing current pulses (Fig. 2D). This sag was always blocked by ZD 7288 (30 μm, seven of seven cells tested.) The results show that Ih channels survive dissociation and are expressed in the parts of the cell (soma and proximal dendrites) retained during dissociation. However, it is notable that in no cell did blocking Ih completely stop firing, suggesting that even when Ih contributes to spontaneous firing, it never plays an essential role. This again is identical with the results in brain slice (Mercuri et al., 1995; Seutin et al., 2001; Neuhoff et al., 2002).
We next tested the effects of blocking voltage-dependent calcium current on pacemaking. To rapidly block currents from all of the multiple calcium channel types known to be present in SNc dopaminergic neurons (Cardozo and Bean, 1995; Durante et al., 2004), we replaced external calcium by equimolar cobalt. In many other pacemaking central neurons, including cerebellar Purkinje neurons (Raman and Bean, 1999), suprachiasmatic nucleus neurons (Pennartz et al., 1997; Jackson et al., 2004), and subthalamic nucleus neurons (Bevan and Wilson, 1999; Beurrier et al., 2000; Do and Bean, 2003), this manipulation has little or no effect on firing rate or even speeds firing, apparently by blocking calcium-activated potassium channels. In SNc dopaminergic neurons, however, substituting cobalt for calcium consistently stopped spontaneous firing (Fig. 3). Total block of spontaneous firing was seen in 17 of 17 cells tested. The effect was immediate with the solution change (<1 s) and the cessation of firing was accompanied by a hyperpolarization, consistent with direct electrical effect attributable to inhibition of inward calcium current flowing during the interspike interval rather than a second messenger-mediated response secondary to a change in intracellular calcium levels. Cobalt substitution for calcium hyperpolarized the membrane potential of neurons (measured as the average potential between spikes) by −7 ± 2 mV (n = 17), from −59 ± 1 mV in control to −66 ± 3 mV after the replacement of calcium. In some cells, like that in Figure 3, the hyperpolarization was so pronounced that after cessation of firing the resting potential was negative even to the maximum afterhyperpolarization during control pacemaking. Thus, although SNc dopaminergic neurons have small-conductance calcium-activated potassium (SK) channels, producing substantial SK current, expected to be present between spikes (Nedergaard et al., 1993), the net interspike current sensitive to removal of calcium is inward. This is in striking contrast to other central neurons, including Purkinje neurons, subthalamic nucleus neurons, and suprachiasmatic nucleus neurons.
We next examined the effects of blocking voltage-dependent sodium channels. Previously, it has been found that, although TTX blocks the fast overshooting spikes during pacemaking of SNc neurons, in some neurons the cell membrane potential continues to oscillate in the presence of TTX (Fujimura and Matsuda, 1989; Yung et al., 1991; Nedergaard and Greenfield, 1992; Nedergaard et al. 1993; Wilson and Callaway, 2000). These oscillations are calcium-dependent and, at least in some cases, occur at a frequency similar to the frequency of spontaneous firing of full-blown spikes under control conditions, as if normal pacemaking is driven by an underlying oscillation whose depolarizing phase is entirely attributable to calcium current. In a variety of other types of pacemaking neurons, including dopaminergic neurons in the retina and in the olfactory bulb (Feigenspan et al., 1998; Steffen et al., 2003; Puopolo et al., 2005), there appears to be a strong contribution from subthreshold “persistent” sodium current to the spontaneous depolarization between spikes. It has been argued that subthreshold persistent sodium current might necessarily be present in all neurons that have large transient sodium currents (Taddese and Bean, 2002) (but see Astman et al., 2006). If so, it might be expected to contribute significantly to the spontaneous depolarization underlying pacemaking in SNc dopaminergic neurons.
As for other manipulations, we saw very similar effects of TTX on pacemaking in dissociated neurons as reported previously for neurons in slice. In some cells, TTX produced complete quiescence. An example is shown in Figure 4A, where firing stopped abruptly (<1 s) on application of TTX. In this and the other cells in which spontaneous activity was stopped by TTX, the membrane potential in TTX stabilized at a value (−64 ± 2 mV) considerably hyperpolarized relative to the average interspike voltage of the cells before TTX (−58 ± 1 mV; n = 10). Hyperpolarization to a relatively steady membrane potential was seen in 14 of 21 cells exposed to TTX. In the other seven cells, spontaneous activity continued after application of TTX, with smaller, broader action potentials arising from more depolarized interspike voltages. An example is shown in Figure 4B. In this cell, the firing frequency changed only slightly in the presence of TTX, but spikes were smaller and broader and the average interspike voltage shifted in the depolarizing direction, from −57 to −53 mV. In cells in which spontaneous oscillatory activity continued in the presence of TTX, it was consistent that spikes were smaller (average peak of +31 ± 3 mV in control, −4 ± 5 mV in TTX), much broader (width at half-maximum of 3.1 ± 0.6 ms in control, 8.2 ± 0.7 ms in TTX), and had depolarized interspike voltages (−58 ± 2 mV in control, −50 ± 1 mV in TTX). However, the direction of change of the frequency of spontaneous activity was inconsistent, ranging from slowing by 27% to speeding by 41%. This suggests complex effects in different directions that can sometimes result in very little change in the frequency, as in Figure 4B. When present, the spontaneous activity in TTX was always (seven of seven cells) stopped by cobalt substitution for calcium (Fig. 4C).
Like the effects of blocking Ih, the ability of spontaneous oscillatory activity to continue in the presence of TTX is consistent with previous experiments with SNc neurons in brain slice (Nedergaard et al., 1993, Wilson and Callaway, 2000). These results suggest that pacemaking in dissociated SNc neurons has very similar characteristics as seen for intact neurons in brain slice. In particular, the dissociated neurons apparently retain contributions from calcium channels and from Ih (in a subpopulation of neurons) similar to those seen in slice experiments. The similarity of pacemaking in dissociated neurons and brain slice neurons encouraged us to perform voltage-clamp experiments using dissociated neurons aimed at quantifying the relative roles of interspike calcium current and TTX-sensitive sodium current. To do this, we used the action potential clamp method (Llinás et al., 1982), using signal-averaging to resolve the relatively small currents flowing in between spikes (Zaza et al., 1997; Taddese and Bean, 2002). In each cell, we first recorded a 5 s segment of spontaneous activity under current clamp and then used the recorded waveform as the command voltage after switching to voltage-clamp mode in the same cell. Current carried through sodium channels was then isolated as the current sensitive to TTX, and current through calcium channels as the current sensitive to substitution of cobalt for calcium. Figure 5 shows results from such an experiment. As expected, sodium current was maximal during the upstroke of the action potential. However, there was also substantial sodium current during the spontaneous depolarization between spikes: it was near zero immediately after a spike (at the time of the afterhyperpolarization), but then gradually increased. The amplitude of sodium current during the interspike interval was measured at four different voltages in eight neurons and average values were −1.3 ± 0.4 pA at −60 mV, −2.6 ± 0.5 pA at −55 mV, −6.0 ± 1.4 pA at −50 mV, and −13.6 ± 2.5 pA at −45 mV (Fig. 6A). The steep voltage dependence of TTX-sensitive sodium current over this voltage range (more than doubling for each change of 5 mV) is typical of subthreshold, “persistent” sodium current.
The current sensitive to replacement of extracellular calcium by cobalt was also determined. This current probably reflects predominantly current through voltage-dependent calcium channels, although in principle there could also be contributions from sources such as the sodium-calcium exchanger, voltage-independent calcium-permeant channels, or calcium-activated nonselective cation channels. As expected for current through voltage-activated calcium channels, cobalt-sensitive current was maximal during the falling phase of the action potential, including the shoulder of the action potential, consistent with previous arguments that calcium current plays an important role in the wide spike of the neurons. However, cobalt-sensitive current was also present throughout the interspike interval. Cobalt-sensitive current was larger than TTX-sensitive current at voltages negative to −45 mV, and it was less steeply voltage-dependent over the range of interspike voltages. Also, unlike TTX-sensitive current, the cobalt-sensitive current was quite substantial even immediately after a spike. The average amplitude of the cobalt-sensitive current (for the same eight neurons in which sodium current was quantified) was −4.5 ± 1.0 pA at −60 mV, −5.1 ± 0.8 pA at −55 mV, −7.0 ± 1.1 pA at −50 mV, and −8.4 ± 1.9 pA at −45 mV (Fig. 6A). The relative contribution of TTX-sensitive and cobalt-sensitive currents to the spontaneous depolarization in between spikes was quantified (Fig. 6B) by integrating each current from the time of the afterhyperpolarization (−65.0 ± 0.7 mV in the eight neurons tested) to the time at which voltage reached −45 mV, before spike threshold (−41 ± 1 mV in the eight cells tested). In six of eight cells, the charge carried by cobalt-sensitive current during the interspike interval was larger than the charge carried by TTX-sensitive current, with no obvious correlation between the fractional contribution of the two currents and the firing rate. In collected results, the charge carried during the interspike interval by cobalt-sensitive current was 764 ± 177 fC, and the charge carried by TTX-sensitive current was 345 ± 60 fC, corresponding to an average of 57 ± 13% of the charge carried by cobalt-sensitive current. These results suggest that both calcium and sodium currents contribute substantially to the spontaneous interspike depolarization, with calcium current carrying about twice as much charge as sodium current.
Sodium and calcium currents during spikes
Although our main goal was to examine the relative roles of calcium and sodium currents in driving spontaneous pacemaking, the experiments shown in Figure 5 are also useful for visualizing sodium and calcium currents during the various phases of the spike itself and for quantifying the total amount of calcium or sodium entering during a firing cycle. One surprise is the relative sizes of peak sodium current and calcium current during the spike (Fig. 5B): peak calcium current is not very much smaller than peak sodium current despite the much smaller concentration of external calcium (1.8 mm) compared with sodium (137 mm), another indication of the unusually high density of calcium channels that appear to be present in SNc neurons. On average, peak calcium current (reached during the falling phase of the spike) was 61 ± 8% (n = 8) of the peak sodium current (reached during the upstroke). The large current carried by calcium during normal spikes is obviously consistent with the ability of calcium by itself to support fairly large spikes in neurons treated with TTX like that shown in Figure 4B. However, during normal spikes, the more rapid activation of the sodium current means that the calcium current makes only a minor contribution to the total inward current during the rising phase of the spike, amounting to only 7 ± 2% of the sodium current at the same time. Thus, the rising phase of spikes is almost entirely attributable to sodium current during normal firing. During the shoulder of the action potential, however, calcium current was 12 ± 4-fold greater than sodium current, presumably because of nearly complete inactivation of sodium channels. Thus, sodium current clearly plays little role in supporting the shoulder of the spike in SNc neurons. This is in contrast to action potentials in nociceptors, which have similarly wide action potentials with shoulders, but where sodium currents play a greater role than calcium currents during the shoulder (Blair and Bean, 2002).
As another index of the large amount of calcium current entering SNc cells during spontaneous firing, we integrated the total cobalt-sensitive current entering during the spike (integrated from −45 mV just before the spike to the afterhyperpolarization). This amounted to 3534 ± 744 fC (n = 8). This is both larger than the total TTX-sensitive sodium flux during the spike (1671 ± 151 fC) and also much larger than the calcium flux during the interspike interval (764 ± 177 fC). The total calcium flux during the entire firing cycle, as estimated by integrating cobalt-sensitive current, amounted to 4298 ± 809 fC, much larger than the total TTX-sensitive sodium flux of 2017 ± 165 fC.
Interspike currents in slices
The experiments so far were performed with acutely isolated neurons from relatively young animals (P14 to P18). To examine whether the results are likely to hold for more intact neurons from older animals, we also performed experiments using the action potential clamp method using cells studied in brain slice preparations from older (P35 to P42) animals. For these experiments, it was unnecessary to use transgenic animals, because dopaminergic neurons in the substantia nigra region of brain slices can be identified by their large cell bodies and regular, slow pacemaking with broad action potentials. Voltage-clamp experiments with intact cells must be interpreted cautiously, because both the uniformity and speed of voltage control are less good than in dissociated cells, but the limitations of voltage clamp should be least severe for the small, slowly changing currents between spikes that are of greatest interest. We were encouraged by previous evidence suggesting that the dendritic tree of dopaminergic neurons is reasonably electrotonically compact, at least between the cell body and the dendrite to which the initial segment of the axon is attached (Häusser et al., 1995). The results with the action potential clamp method applied to adult neurons in brain slice were similar to those obtained with dissociated neurons from younger animals with regard to the relative roles of calcium and sodium currents in driving spontaneous depolarization. Figure 7, A and B, shows an example, and Figure 7C shows average values for cobalt-sensitive and TTX-sensitive currents at various voltages during the interspike interval. Just as in dissociated neurons, interspike sodium current was very small immediately after spikes and increased with steep voltage-dependence during the interspike interval; average values for interspike sodium current were as follows: −2.0 ± 0.9 pA at −65 mV, −7.0 ± 3.0 pA at −60 mV, −23.4 ± 8.8 pA at −55 mV, and −162.4 ± 112.7 pA at −50 mV (n = 8). Also similar to dissociated neurons, cobalt-sensitive current was sizeable immediately after spikes and was larger than sodium current for most of the interspike interval (negative to −50 mV). Average values for interspike cobalt-sensitive current were as follows: −11.8 ± 2.8 pA at −65 mV, −27.8 ± 4.7 pA at −60 mV, −57.7 ± 8.6 pA at −55 mV, and −112.9 ± 14.2 pA at −50 mV. As for the experiments in dissociated neurons, we quantified the relative contribution of TTX-sensitive and cobalt-sensitive currents to the spontaneous depolarization in between spikes by integrating each current from the time of the afterhyperpolarization (−74 ± 2 mV in the eight neurons tested) to the time at which voltage reached −50 mV, before spike threshold (which was −45 ± 1 mV in the eight cells tested). Integrated cobalt-sensitive current was larger than integrated interspike sodium current in each of the eight neurons tested. Thus, interspike calcium current appears to make a larger contribution than interspike sodium current in driving the spontaneous depolarization to threshold in adult neurons studied in brain slice, as it does in acutely dissociated neurons from young mice. In another series of experiments, we quantified interspike sodium and calcium currents in neurons studied in brain slices from younger animals (P14–P18, the same age as the dissociated neurons). Although integrated interspike sodium current and calcium current both increased between P14–P18 and P35–P42, their relative magnitudes showed little change (Fig. 7E), suggesting that there is little difference in relative roles of sodium current and calcium current in driving pacemaking as animals mature from P14 to P42. The relative contribution of charge from interspike sodium current was somewhat less in neurons studied in brain slice (29 ± 8% of cobalt-sensitive charge in P35–P42 neurons, n = 8; 23 ± 5% of cobalt-sensitive charge in P14–P18 neurons, n = 11) compared with acutely dissociated neurons (57 ± 13% of cobalt-sensitive charge; n = 8). The difference is consistent with a relatively high density of calcium current in dendrites (Wilson and Callaway, 2000).
Effects of selective calcium channel blockers
Voltage-clamp experiments have shown that SNc neurons express multiple types of voltage-dependent calcium channels, including T-type, L-type, N-type, and P-type channels (Cardozo and Bean, 1995; Durante et al., 2004). We therefore tested the effect of subtype-specific calcium channel blockers. Because of the relatively slow time course of equilibration of the blockers compared with the instantaneous effect of cobalt substitution, it was not practical to perform action potential clamp experiments with subtype-specific blockers. Instead, we examined the effects of the blockers on spontaneous firing in current clamp in dissociated cells in which there was stable firing in control conditions. Blocking L-type calcium channels with 1 μm nimodipine slowed spontaneous firing in nine of 17 neurons tested (Fig. 8A,B), with complete quiescence in four neurons. In those nine neurons, nimodipine caused a modest hyperpolarization (average interspike voltage from −55.8 ± 1.6 to −58.3 ± 1.7 mV), a reduction of the firing rate from 3.3 ± 0.3 to 0.6 ± 0.1 Hz, an increase of the coefficient of variation of the interspike intervals from 45 ± 9% to 124 ± 16%, and a reduction of the speed of the spontaneous depolarization in between spikes from 51 ± 6 mV/s to 14 ± 3 mV/s. In the four neurons in which there was a complete cessation of firing in the presence of 1 μm nimodipine, the drug caused a stronger hyperpolarization of the average interspike membrane potential, from −54.5 ± 3.0 mV to −66.7 ± 2.9 mV, suggesting a major contribution of L-type calcium channels to the spontaneous interspike depolarization during the pacemaking cycle. The effects of nimodipine on firing in dissociated neurons are very similar to previous results with dihydropyridines applied to midbrain dopaminergic neurons in brain slice (Nedergaard et al., 1993; Mercuri et al., 1994).
In contrast to the effect of nimodipine, the N-type calcium channel blocker ω-conotoxin GVIA (3 μm) had no significant effect on spontaneous firing (n = 8) (data not shown). The drug had little or no effect on average interspike voltage (−54.7 ± 2.2 mV in control, −53.6 ± 2.2 mV with 3 μm ω-conotoxin GVIA) or on firing rate (2.8 ± 0.3 Hz in control, 3.0 ± 0.5 Hz with conotoxin), suggesting that N-type calcium channels do not contribute inward current to the spontaneous interspike depolarization during the pacemaking cycle in SNc neurons. However, the P/Q-type calcium channel blocker ω-Aga-IVA applied at 200 nm slowed spontaneous firing in 10 of 14 neurons tested (Fig. 9) with cessation of firing in six neurons. In those 10 neurons, the drug caused a hyperpolarization of mean interspike voltage, from −56.3 ± 0.5 mV to −59.8 ± 1.1 mV, a dramatic reduction of the firing rate from 4.5 ± 0.5 to 0.6 ± 0.2 Hz, an increase in the coefficient of variation of the interspike intervals from 39 ± 12% to 138 ± 19%, and a decrease in the speed of the spontaneous depolarization in between spikes from 53 ± 7 mV/s to 13 ± 3 mV/s. In the six neurons in which there was cessation of firing, the mean interspike membrane potential hyperpolarized more dramatically than in the other neurons, from −56.0 ± 1.0 mV to −62.1 ± 1.5 mV.
It is increasingly clear that multiple ionic mechanisms can produce spontaneous firing of different central neurons. The action potential clamp method has been used to quantify the interspike currents carried by calcium currents and TTX-sensitive sodium currents in a variety of pacemaking neurons. Of those neurons examined so far, SNc dopaminergic neurons are unique in having a larger calcium current than TTX-sensitive sodium current during the interspike interval. In contrast, interspike calcium current is <10–25% of the interspike TTX-sensitive sodium current in cerebellar Purkinje neurons (Raman and Bean 1999), subthalamic nucleus neurons (Do and Bean, 2003), tuberomammillary nucleus neurons (Taddese and Bean, 2002) (A. Taddese and B. Bean, unpublished observation), and suprachiasmatic nucleus neurons (Jackson et al., 2004). Accordingly, inhibition of all calcium current in these neurons (e.g., by substituting cobalt or magnesium for calcium) generally does not stop spontaneous firing although it sometimes decreases regularity of firing, probably by inhibiting calcium-activated potassium currents (Hallworth et al., 2004; Walter et al., 2006). Interestingly, the cells with the largest contribution of calcium entry studied previously are dopaminergic periglomerular neurons in the olfactory bulb, where interspike calcium current is on average ∼70% of interspike sodium current (Puopolo et al., 2005), and where replacing calcium by cobalt results in cessation of firing accompanied by hyperpolarization (Pignatelli et al., 2005; Puopolo et al., 2005), as we saw in the SNc dopaminergic neurons. Substitution of cobalt for calcium also hyperpolarizes and silences spontaneously active cholinergic interneurons (Bennett et al., 2000); here, however, the effect can be attributed to reduction of Ih current and other evidence suggests a minor direct role of subthreshold calcium entry in driving pacemaking. Changes in Ih cannot account for the effects of cobalt substitution in our experiments, because total block of Ih at most slowed firing only moderately (and often did nothing), whereas cobalt substitution silenced all cells.
The simplest interpretation of the interspike current sensitive to cobalt substitution for calcium is that it represents calcium current flowing directly through calcium-permeable channels. In this case, the effect of nimodipine in slowing or blocking spontaneous firing can most simply be attributed to block of L-type calcium current flowing during the interspike interval. Such a direct effect is consistent with the net hyperpolarization during the interspike interval produced by nimodipine, which contrasts with the net interspike depolarization and reduction of afterhyperpolarization seen in suprachiasmatic nucleus neurons, apparently reflecting indirect block of calcium-activated potassium current (Cloues and Sather, 2003; Jackson et al., 2004). Interspike L-type current in SNc dopaminergic neurons is consistent with previous evidence that the calcium channels contributing to pacemaking activate at potentials as negative as −60 mV and show little inactivation (Kang and Kitai, 1993b; Wilson and Callaway, 2000). Midbrain dopaminergic neurons express CaV1.3 calcium channels (Takada et al., 2001) (D. J. Surmeier, personal communication), which activate at more negative voltages than L-type current carried by CaV1.2 calcium channels (Koschak et al., 2001; Scholze et al., 2001; Xu and Lipscombe, 2001). Durante et al. (2004) found that the L-type current in dissociated dopaminergic neurons activates at more negative potentials than N-type and P/Q-type currents, consistent with a contribution by CaV1.3 channels.
Although calcium current appears to play a dominant role in driving spontaneous depolarization, subthreshold sodium current is also significant during the interspike interval, amounting on average to ∼60% of the cobalt-sensitive current. The participation of subthreshold, persistent sodium current in helping to drive pacemaking has been suggested before and is embodied in models of pacemaking of dopaminergic neurons as contributing to the spontaneous depolarization without being essential (Kuznetsov et al., 2006), as our results support.
Because SNc neurons have subthreshold persistent sodium current, which by itself appears sufficient to drive pacemaking of many types of neurons that lack significant subthreshold calcium current and fire much faster than SNc dopaminergic neurons (Taddese and Bean, 2002; Atherton and Bevan, 2005), why is it that blocking calcium current silences SNc neurons? The key factor is most likely the resting potential (i.e., the voltage reached when spontaneous activity is silenced), which in our experiments was about −65 mV when neurons were silenced by TTX or by cobalt substitution for calcium. This is negative to the voltage at which persistent sodium current typically becomes significant, about −60 to −55 mV (Bennett et al., 2000; Taddese and Bean, 2002). In contrast, many neurons whose pacemaking depends mainly on persistent sodium current have resting potentials in the range from −55 to −45 mV when silenced by TTX (Raman et al., 2000; Taddese and Bean, 2002; Atherton and Bevan, 2005), so that no other subthreshold current is necessary to produce a spontaneous depolarization. Several such neurons have been shown to possess a non-TTX-sensitive resting sodium conductance that contributes to a depolarized resting potential (Raman et al., 2000; Eggermann et al., 2003; Jackson et al., 2004; Atherton and Bevan, 2005). We can speculate that the expression of the channels underlying this conductance, which are not known, is lower in SNc neurons than in the pacemaking neurons that can fire spontaneously using only persistent sodium current.
Overall, our results fit well with a current model of spontaneous firing of midbrain dopaminergic neurons in which both soma and dendritic branches have intrinsic oscillatory activity based on subthreshold calcium current (Wilson and Callaway, 2000; Medvedev et al., 2003; Kuznetsov et al., 2006). However, it is unexpected that the isolated cell bodies show more rapid pacemaking than intact cells, because the model, supported by experimental evidence from calcium imaging (Wilson and Callaway, 2000), envisions a system of coupled oscillators (coupled by voltage) in which the dendrites have faster natural frequencies of oscillation than the cell body. The firing of dissociated cells clearly occurs mainly by subthreshold calcium current, as proposed in the model. One possible explanation of its unexpected speed would be if either calcium-activated or voltage-activated potassium currents had a lower density in the cell body compared with dendrites, which could allow faster recovery from the afterhyperpolarization. Block of calcium-activated SK channels in dopaminergic neurons both speeds firing and also decreases the regularity of spontaneous firing (Ping and Shepard, 1996; Wolfart et al., 2001). Consistent with reduced influence of SK current in cell bodies, the firing in our experiments was considerably less regular (coefficient of variation of interspike intervals of 45%) than that seen in intact neurons (∼10%) (Wolfart et al., 2001). This suggests that the system of coupled oscillators with intact dendrites serves to enhance the precision of firing (Wilson and Callaway, 2000). Regional differences in channel density between soma and dendrites are common in other neuronal types but have not been systematically examined in dopaminergic neurons.
Another unexpected result is the large magnitude of calcium entry during the action potential compared with the subthreshold current entering during the phase of spontaneous depolarization: 80% of the total calcium entry during the pacemaking cycle occurs during the spike. This is surprising because in calcium-imaging experiments, the signal from calcium entry decreases only slightly when fast spikes are blocked by TTX (Wilson and Callaway, 2000). However, the two sets of observations are not necessarily incompatible, because the calcium spikes occurring in TTX, although much smaller than normal fast spikes, are also much broader, and it is possible that equivalent calcium entry occurs with slower kinetics (and possibly through a different mixture of channels).
Perhaps the most surprising result in our experiments was the dramatic slowing of firing produced by ω-Aga-IVA. This is unexpected because ω-Aga-IVA appears to be a highly selective blocker of CaV2.1 (P/Q-type) channels, which require relatively large depolarizations to be activated. Indeed, previous voltage-clamp experiments in dissociated dopaminergic neurons have shown that ω-Aga-IVA blocks a distinct fraction of high-threshold activated calcium current (Cardozo and Bean, 1995; Durante et al., 2004), and there is no hint that the ω-Aga-IVA-sensitive current includes a subthreshold component. In particular, it seems clear that ω-Aga-IVA does not also inhibit low-threshold L-type current through a nonselective action because the fraction of total current blocked by ω-Aga-IVA is similar whether or not cells were previously exposed to nimodipine and vice versa (Durante et al., 2004). An interesting possibility is that the effect of ω-Aga-IVA on pacemaking is indirectly mediated as a consequence of block of spike-evoked calcium entry through ω-Aga-IVA-sensitive channels. Recent experiments on pacemaking cells in the cardiac sinoatrial node have suggested a major role for depolarizing current produced by the Na/Ca exchanger as it pumps calcium that enters during the pacemaking cycle (Sanders et al., 2006). It is becoming clear that, like cardiac sinoatrial cells, pacemaking of midbrain dopaminergic neurons involves massive entry of calcium during both the slow spontaneous depolarization and also during the spike. The consequences and mechanisms for removal of this unusually large calcium entry remain to be characterized in more detail and could well be involved in the unique pathophysiology of these neurons.
This work was supported by National Institutes of Health Grants NS36855 and EY01344.
- Correspondence should be addressed to Bruce P. Bean, Department of Neurobiology, Harvard Medical School, 220 Longwood Avenue, Boston, MA 02115.