Molecular and functional differences in voltage-activated sodium currents between GABA projection neurons and dopamine neurons in the substantia nigra

Abstract

GABA projection neurons (GABA neurons) in the substantia nigra pars reticulata (SNr) and dopamine projection neurons (DA neurons) in substantia nigra pars compacta (SNc) have strikingly different firing properties. SNc DA neurons fire low-frequency, long-duration spikes, whereas SNr GABA neurons fire high-frequency, short-duration spikes. Since voltage-activated sodium (Na_V) channels are critical to spike generation, the different firing properties raise the possibility that, compared with DA neurons, Na_V channels in SNr GABA neurons have higher density, faster kinetics, and less cumulative inactivation. Our quantitative RT-PCR analysis on immunohistochemically identified nigral neurons indicated that mRNAs for pore-forming Na_V1.1 and Na_V1.6 subunits and regulatory Na_Vβ1 and Na_vβ4 subunits are more abundant in SNr GABA neurons than SNc DA neurons. These α-subunits and β-subunits are key subunits for forming Na_V channels conducting the transient Na_V current (I_NaT), persistent Na current (I_NaP), and resurgent Na current (I_NaR). Nucleated patch-clamp recordings showed that I_NaT had a higher density, a steeper voltage-dependent activation, and a faster deactivation in SNr GABA neurons than in SNc DA neurons. I_NaT also recovered more quickly from inactivation and had less cumulative inactivation in SNr GABA neurons than in SNc DA neurons. Furthermore, compared with nigral DA neurons, SNr GABA neurons had a larger I_NaR and I_NaP. Blockade of I_NaP induced a larger hyperpolarization in SNr GABA neurons than in SNc DA neurons. Taken together, these results indicate that Na_V channels expressed in fast-spiking SNr GABA neurons and slow-spiking SNc DA neurons are tailored to support their different spiking capabilities.

as a key component of the basal ganglia motor circuitry, substantia nigra (SN) is populated largely by two types of projection neurons: GABA neurons in SN pars reticulata (SNr) and dopamine (DA) neurons in SN pars compacta (SNc) and also in SNr (Bolam et al. 2000; Deniau et al. 2007; Gonzalez-Hernandez and Rodriguez, 2000; Nelson et al. 1996; Parent et al. 2000). SNr GABA neurons and SNc DA neurons have strikingly different firing properties. SNr GABA neurons fire high-frequency, brief action potentials (Atherton et al. 2005; Zhou et al. 2006), whereas SNc DA neurons fire low-frequency, long-duration spikes (Hyland et al. 2002; Zhou et al. 2006), indicating potential differences in voltage-gated potassium (K_v) and sodium (Na_V) channels in these two cell types (Ding et al. 2011; Seutin and Engel 2010). Indeed, our laboratory's recent study showed that a K_v3-like current is essential to the sustained high-frequency firing in SNr GABA neurons (Ding et al. 2011).

Na_V channels are composed of one α-subunit and two β-subunits (Catterall 2000). The α-subunit forms the channel pore, whereas the two auxiliary β-subunits regulate the kinetics and function of the α-subunit. Four of the nine identified α-subunits, Na_V1.1, Na_V1.2, Na_V1.3, and Na_V1.6, and all four β-subunits, Na_Vβ1, Na_Vβ2, Na_Vβ3, and Na_Vβ4, are expressed in the central nervous system (Catterall et al. 2005; Goldin 2001; Yu et al. 2003). Na_V channels formed by different α- and β-subunits conduct the classical fast transient Na current (I_NaT) with different kinetics (Goldin 2001; Hille 2001; Smith and Goldin 1998). Consequently, different neuron types expressing Na_V channels comprised of different subunits may have different action potential waveforms and patterns (Bean 2007). For example, I_NaT in hippocampal fast-spiking interneurons recovers rapidly from inactivation and thus promotes fast spiking (Martina & Jonas 1997). Na_V channels containing Na_V1.6 and/or Na_Vβ4 may have a resurgent property that generates a resurgent Na current (I_NaR) after the inactivation of the I_NaT and upon repolarization that may facilitate fast spiking (Bant and Raman 2010; Khaliq et al. 2003; Mercer et al. 2007; Raman and Bean 1997). Additionally, Na_V channels also conduct a small but long-lasting or persistent Na current (I_NaP) that increases neuronal excitability and promotes pacemaking firing (Bean 2007; Crill 1996).

mRNAs for Na_V1.1, Na_V1.2, Na_V1.3, and Na_V1.6 α-subunits, Na_Vβ3 and Na_Vβ4 β-subunits, and also Na_V1.1 and N_V1.2 α-subunit proteins have been detected in the nigral region (Burbidge et al. 2002; Chen et al. 2000; Furuyama et al. 1993; Gong et al. 1999; Morgan et al. 2000; Yu et al. 2003). Furthermore, mRNA for Na_Vβ4, a key β-subunit for Na_V channels with resurgent kinetics, is expressed at a high level in the SNr but not in SNc (Yu et al. 2003). Since the major neuron types in SNr and SNc are the GABA projection neurons and DA neurons, respectively, SNr GABA neurons may express more Na_Vβ4 than SNc DA neuron. It also raises the possibility that gene expression levels for other Na_V α- and β-subunits are different in SNr GABA and SNc DA neurons, leading to functional differences in Na_V channels in SNr GABA and SNc DA neurons. More specifically, we reasoned that, compared with SNc DA neuron, Na_V α- and β-subunits expressed in SNr GABA neurons may form Na_V channels that conduct a larger I_NaT with faster kinetics and less cumulative inactivation, and also a larger I_NaP and I_NaR.

To test these ideas, we first used single-cell RT-PCR (scRT-PCR) on electrophysiologically characterized SNr GABA and nigral DA neurons and quantitative RT-PCR (qRT-PCR) on laser capture-microdissected immunohistochemically identified SNr GABA and SN DA neurons to profile the mRNA expression of Na_v channels. Then we used electrophysiological techniques to compare Na_V currents in SNr GABA neurons and SNc DA neurons. Consistent with our hypothesis, we found that 1) mRNAs for Na_V1.1 and Na_V1.6 and Na_Vβ1 and Na_Vβ4 are more abundant in SNr GABA neurons than in SNc DA neurons; 2) I_NaT was larger and activated and deactivated more quickly in SNr GABA neurons than in SNc DA neurons; the recovery of I_NaT from inactivation was faster and thus the cumulative inactivation was smaller in SNr GABA neurons than in nigral DA neurons; and 3) SNr GABA neurons had a larger I_NaP and also a larger I_NaR than SNc DA neurons. These differences in Na_V currents support the different spiking behaviors in fast-spiking SNr GABA neurons and slow-spiking nigral DA neurons. A preliminary analysis of these results has appeared in an abstract (Ding and Zhou 2010).

METHODS

Preparation of brain slices.

Sixteen- to 24-day old male and female Sprague-Dawley rats were used. All procedures were carried out in accordance with the National Institutes of Health guidelines and were approved by the Institutional Animal Care Committee of The University of Tennessee Health Science Center. Midbrain slices were prepared as previously described (Atherton & Bevan 2005; Ding et al. 2011; Zhou et al. 2006, 2008, 2009). In brief, rats were deeply anesthetized with urethane. After transcardiac perfusion with an oxygenated ice-cold high-sucrose cutting solution (see below), their brains were quickly dissected out and immersed in the ice-cold oxygenated cutting solution for 2 min. Coronal midbrain slices (300-μm thickness) containing the midrostral part of the SN were prepared in an ice-cold, oxygenated high-sucrose cutting solution using a Vibratome 1000 Plus (Vibratome, St. Louis MO) or Leica Zero Z VT1200S vibratome (Leica Microsystems, Wetzlar, Germany). Slices were transferred to a holding chamber containing the normal extracellular solution (see below) at 30°C for 45 min and then kept at room temperature.

Electrophysiological cell identification and recording of Na_v currents in nucleated membrane patches.

Recordings were made under visual guidance of a video microscope (Olympus BX51WI and Zeiss Axiocam MRm digital camera) equipped with Nomarski optics and a ×60 water immersion lens. Patch pipettes with resistances of 1–3 MΩ were pulled from borosilicate (KG-33) glass capillary tubing (1.10 mm ID, 1.65 mm OD; King Precision Glass, Claremont, CA) using a PC-10 puller (Narishige, Tokyo, Japan). A Multiclamp 700B amplifier, pClamp 9.2 software, and Digidata 1322A interface (Molecular Devices, Sunnyvale, CA) were used to acquire and analyze data. After electrophysiologically fingerprinting nigral GABA and DA neurons with conventional whole cell patch clamp within the first 10 s after obtaining access to the cell interior, gentle negative pressure was applied, and the patch pipette was withdrawn slowly to isolate nucleated membrane patches (Ding et al. 2011; Martina and Jonas 1997).

Voltage-clamp waveforms were generated by pClamp 9.2 software. Signals were digitized at 50 kHz and filtered at 10 kHz using the built-in low-pass Bessel filter in the patch clamp amplifier. This 10-kHz filtering, often necessitated by reducing noise, plus stray capacitance-induced filtering, was likely to have affected the measurements of the activation and deactivation time courses, since these two kinetic parameters were very fast. Cell membrane capacitive transients were compensated with the autocompensation function of the Multiclamp 700B. The capacitance readings were taken as the capacitance estimates of the nucleated membrane patch to calculate current and conductance density. Series resistance was not corrected, since it was estimated to be between 4 and 8 MΩ, and the current amplitude was under 500 pA such that the voltage error caused by series resistance was minimal. Since both electrode capacitance and cell capacitance were compensated, the filtering from residual equivalent electrode-cell resistor-capacitor circuit should also be minimal. However, unknown stray capacitance may still exist in the recording system, leading to underestimation of the rapid activation and deactivation kinetics. Leak currents were subtracted online with a P/4 protocol. K⁺ channels were blocked by internal Cs⁺ and/or external tetraethylammonium chloride (TEA, 20 mM) and 4-aminopyridine (5 mM). Ca²⁺ currents were eliminated by replacing extracellular Ca²⁺ with Mg²⁺. GABA_A receptors and ionotropic glutamate receptors were routinely blocked by 100 μM picrotoxin and 20 μM D-(−)-2-amino-5-phosphonopentanoic acid and 10 μM 6-cyano-7-nitroquinoxaline-2,3-dione disodium, respectively. Na_v currents were isolated by offline digital subtraction following application of 1 μM tetrodotoxin (TTX). Test pulses were applied every 5 s. Most recordings were made at 30°C. Additional recordings (see Figs. 12 and 13) were obtained at room temperature (25°C), filtered at 30 kHz, and sampled at 200 kHz.

Fig. 12.

Voltage-dependent activation of I_NaT in SNr GABA and nigral DA neurons at room temperature (25°C). A and B: representative traces of I_NaT evoked in nucleated membrane patches isolated from a SNr GABA neuron (A) and a nigral DA neuron (B). The same voltage protocol used in Fig. 6A was used. C: current-voltage plots for I_NaT in SNr GABA neurons (n = 9) and nigral DA neurons (n = 11). D: I_NaT activation curves in SNr GABA neurons (n = 9) and nigral DA neurons (n = 11). Continuous lines are Boltzmann fits.

Fig. 13.

I_NaT deactivation is faster in SNr GABA neurons than in SNc DA neurons at room temperature (25°C). A and B: representative traces of I_NaT deactivation in nucleated membrane patches from a SNr GABA neurons (A) and a SNc DA neuron (B). The traces at −40 mV are displayed in A′ and B′ to show more clearly I_NaT deactivation at −40 mV. The smooth gray curves are single-exponential fits. C: pooled data for deactivation time constants plotted against deactivation potentials. Filled circles, SNr GABA neurons (n = 4); open circles, SNc DA neurons (n = 5). The time constants were shorter at −80 to −40 mV, P < 0.05. D: voltage waveform used to evoke I_NaT deactivation. Holding potential, −90 mV; 50-ms pulse to −120 mV; 200-μs pulse to 0 mV; 50-ms test pulse from −100 mV to −40 mV at 10-mV increase.

Composition of solutions.

The high-sucrose cutting solution contained the following (in mM): 220 sucrose, 2.5 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 0.5 CaCl₂, 7 MgCl₂, 20 d-glucose. The normal extracellular solution contained the following (in mM): 125 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 2.5 CaCl₂, 1.3 MgCl₂, 10 d-glucose, maintaining pH at 7.4 by continuously bubbling with 95% O₂ and 5% CO₂. During recording of Na⁺ currents, 2.5 mM CaCl₂ were substituted by 2.5 mM MgCl₂, and extracellular NaCl was reduced on an equal molar basis when 20 mM TEA and 5 mM 4-aminopyridine were used. The Cs-based intracellular solution contained the following (in mM): 135 CsCl, 0.5 EGTA, 10 HEPES, 2 Mg-ATP, 0.2 Na-GTP, 4 Na₂-phosphocreatine. pH was adjusted to 7.25 with CsOH. To record normal membrane potential and action potentials, a KCl-based intracellular solution was used, containing the following (in mM): 135 KCl, 0.5 EGTA, 10 HEPES, 2 Mg-ATP, 0.2 Na-GTP, and 4 Na₂-phosphocreatine. pH was adjusted to 7.25 with KOH. The osmolarity for the intracellular solutions was 280–290 mosM.

Data analysis.

The liquid junction potential between the 135 mM KCl- or CsCl-based intracellular solution and 100 mM NaCl- and 20 mM TEA-Cl-based extracellular bathing solution, estimated by the liquid junction potential calculator in Clampex, was 4.6 and 4.7 mV, respectively, and not corrected in the data presented below. To obtain activation curves, Na⁺ conductance was calculated with the equation g_Na = I/(V − E_rev), where E_rev is the reverse potential and was estimated to be +50 mV under our recording conditions, following the method of Safronov and Vogel (1995) and Raman et al. (2000) (see Fig. 5), g_Na is Na⁺ conductance, I is current, and V is voltage. Activation and inactivation curves were fitted with Boltzmann equation: f = 1/{1 + exp[±(V − V_1/2)/k]}, where V is the membrane potential, V_1/2 is the potential at which the value of the Boltzmann function is 0.5, and k is the slope factor. Data are reported as means ± SE. Two-sample independent t-tests were used to make comparisons, with P value < 0.05 being considered statistically significant.

Fig. 5.

Higher transient Na_V current (I_NaT) density in SNr GABA neurons than in nigral DA neurons. I_NaT was evoked from a holding potential of −100 mV by a test potential ranging from −80 mV to +30 mV in 5-mV steps. I_NaT started to appear at −55 mV. For example current traces, see Fig. 6A, because Figs. 5 and 6A were derived from the same data set. The reversal potential was obtained by linear-fitting the data points between 0 mV and 30 mV (dotted lines), according to the method of Safronov and Vogel (1995) and Raman et al. (2000). The estimated reversal potential was 49.8 mV for GABA neurons and 49.9 mV for DA neurons.

scRT-PCR.

scRT-PCR procedures generally followed well-established principles and methods (Ding et al. 2011; Liss et al. 2001; Surmeier et al. 1996; Zhou et al. 2008, 2009;). Patch pipettes were autoclaved to eliminate RNase. The intracellular solution for scRT-PCR contained the following (in mM): 135 KCl, 0.5 EGTA, 10 HEPES, 2 Mg-ATP, 0.2 Na-GTP, and 4 Na₂-phosphocreatine, and was prepared using DNase-RNase-free water. After electrophysiologically fingerprinting SNr GABA neurons and nigral DA neurons, gentle suction was applied to aspirate the cytoplasm without disrupting the seal. The aspirated cell content was expelled into a 0.2-ml PCR tube and treated with DNase I (5 min at 25°C) to remove genomic DNA contamination. cDNA was synthesized using SuperScript III reverse transcriptase-based Cells-Direct cDNA Synthesis kit (Invitrogen). The synthesized cDNA was amplified using a hot-start Platinum PCR SuperMix (Invitrogen). RT-minus controls, in which the reverse transcriptase was omitted, while all other reaction components were exactly the same, were performed to verify complete removal of genomic DNA. Negative controls were performed by lowering patch pipettes into the tissue and taking them out without seal formation and suction to exclude nonspecific harvesting of surrounding tissue components.

Two-stage PCR amplification was used as previously described (Ding et al. 2011; Zhou et al. 2008, 2009). Briefly, 5 μl of 30-μl cDNA were amplified for 45 cycles in the presence of primers for the first stage (see Table 1 for primer pair sequences). The thermal cycling protocol was 2 min at 94°C for the initial denaturation, than 45 cycles of 15 s at 94°C to denature, 30 s at 48°C to anneal, and 50 s at 72°C to extend, followed by a 10-min final extension. In the second-stage PCR, the product of 1 μl from the first-stage PCR amplification was used as template, and the same primer pair that was used in the first stage was used, and 40 cycles were run.

Table 1.

Single-cell RT-PCR primer pairs for rat neuronal Na_v channels, TH, and GAD1 mRNAs

mRNA (Accession No.)		Start Position	Primers (5′ to 3′)
Nav1.1 (NM_030875)	F	4659	ggctcgttcttcactctaaa
	R	5106	ttcctacaatggagaggatg
Nav1.2 (NM_012647)	F	1641	aggcgggataggtgttttct
	R	1991	gcaaagtcattttcggaacc
Nav1.3 (NM_013119)	F	3291	ggaaggatgcaaaagggaat
	R	3586	acagcaattggcacagtcac
Nav1.6 (NM_019266)	F	2808	ggccatcatcgtcttcatct
	R	3058	ggttgccaatgaccataacc
Navβ1 (NM_017288)	F	820	ctggccattacttccgagag
	R	1118	gccatattgcttcacccatc
Navβ2 (NM_012877)	F	403	gttcctccagttccgaatga
	R	751	gtcatccgtgctcagcttct
Navβ3 (NM_139097)	F	1850	ctgggccagagatgactttc
	R	2149	acacgcccatctttgtctct
Navβ4 (AF544988.1)	F	380	cccaaactgttctttctgag
	R	729	caatcccagagttctagtgc
TH (NM_012740)	F	1130	cactgtggaattcgggctat
	R	1329	cattgaagctctcggacaca
GAD (NM_017007)	F	1567	caagttctggctgatgtgga
	R	1826	actccatcatcagggctttg

Na_v, voltage-activated sodium; TH, tyrosine hydroxylase; GAD, glutamate decarboxylase; F, forward; R, reverse.

We first detected glutamate decarboxylase 1 and TH (tyrosine hydroxylase) mRNA to identify GABA and DA neurons and to confirm the success of cytoplasm aspiration and cDNA synthesis. Then we used 5 μl of the remaining cDNA from the original 30-μl cDNAs to detect the target genes. The products from the second-stage amplification were separated by 1.5% agrose gel electrophoresis, visualized by ethidium bromide (0.05 mg/100 ml gel) or Gelgreen under UV light and photographed. The positive bands were then cut out and extracted using a Qiagen extraction kit. The extracted products were sequenced at the Molecular Resource Center of University of Tennessee Health Science Center in Memphis, Tennessee, and positively identified.

The web-based Primer3 software (http://fokker.wi.mit.edu/primer3/input.htm) (MIT, Cambridge, MA) was used to design PCR primers, according to sequences published in GenBank. The sequences of these primers are listed in Table 1. All primers were synthesized by Integrated DNA Technologies (Coralville, IA). The effectiveness of the primers was positively confirmed by using whole brain total RNA. All primers yielded amplicons of expected sizes with correct sequences.

qRT-PCR assay on laser-captured nigral neurons.

Rats were deeply anesthetized with urethane (1.5 g/kg). Brains were rapidly removed and frozen with Freeze Spray. Cryostat coronal midbrain sections (10 μm) were collected and stored at −80°C. TH and parvalbumin (PV) were used as markers for DA and GABA neurons, respectively (Gonzalez-Hernandez and Rodriguez 2000; Zhou et al. 2009). To quickly immunostain DA neurons or GABA neurons, slide-mounted cryostat sections of unfixed rat midbrain sections were removed from −80°C storage and allowed to thaw before fixation in ice-cold 100% methanol for 3 min. The slide-mounted sections were briefly dipped in cold 0.02 M PBS and incubated for 3 min with 1:25 rabbit anti-TH polyclonal antibody or 1:25 mouse anti-PV monoclonal antibody diluted in PBS. This was followed by four brief rinses in cold PBS. The tissue sections were incubated for 3 min in red fluorescent secondary donkey anti-rabbit or donkey anti-mouse antibody diluted at 1:50 in PBS. The slide-mounted sections were washed four times in cold PBS and dehydrated (1 min each in 75%, 95%, and 100% EtOH), followed by 5 min in xylene twice. After air-drying for 10 min, brain sections were visualized on the ArcturursXT fluorescent Laser Capture Microdissection (LCM) System (Applied Biosystems). TH-positive or PV-positive cells were picked by LCM and collected on the Arcturus plastic sample caps. Approximately 300–400 TH-positive or PV-positive neurons from six coronal sections of SNc or SNr were collected and pooled into a single cap. Six caps were used to collect TH-positive or PV-positive neurons for each animal. RNA was extracted and purified using the PicoPure RNA isolation kit (Applied Biosystems). RNA samples were further treated with DNase to remove any potential genomic DNA contamination and then used as template to generate cDNA using SuperScript III CellsDirect cDNA Synthesis kit (InVitrogen). Levels of mRNA for Na_v1.1, Na_v1.2, Na_v1.3, Na_v1.6, Na_vβ1–4, and β-actin (internal control) were measured by using a Roche LightCycler 480 (LC 480) qRT-PCR system and the Universal ProbeLibrary probes and primers (Roche Applied Science, Indianapolis, IN). The sequences for these primers are listed in Table 2. Using β-actin mRNA as internal control, Na_v mRNA quantification was performed employing the comparative crossing point (C_p) method in the form of 2^−ΔCp (Luu-The et al. 2005). The C_p values of the real-time fluorescence intensity curve were calculated using the second derivative method. The calculation was performed by the built-in software on LC 480. For Na_v mRNA levels in DA neurons, ΔC_{p,Nav,DA neuron} = C_{p,Nav,DA neuron} − C_{p,β-actin,DA neuron}. For Na_v mRNA levels in GABA neurons, ΔC_{p,Nav,GABA neuron} = C_{p,Nav,GABA neuron} − C_{p,β-actin,GABA neuron}. Finally, 2^−ΔCp values for Na_v mRNAs were normalized to those in SNr GABA neurons.

Table 2.

Quantitative RT-PCR primer pairs and UPL probes for rat neuronal Na_v channel and β-actin mRNAs

mRNA (Accession No.)		Start Position	Primer (5′ to 3′)	UPL Probe No.
Nav1.1 (NM_030875)	F	2748	aatatctttgatggtttcattgtgac	64
	R	2858	caacttgaagactcggagca
Nav1.2 (NM_012647)	F	2168	tggtgtccctggttggag	74
	R	2256	ccttatttctgtctcagtagttgtgc
Nav1.3 (NM_013119)	F	295	gcaccgtccattctaaccat	56
	R	387	tttagcttcttgcataagaattgc
Nav1.6 (NM_019266)	F	3951	ctggtgctggttggacttc	5
	R	4017	gcccagggcattagctataa
Navβ1 (NM_017288)	F	629	gcgtcgtcaagaagatccac	10
	R	691	cgatggatgccatatctctg
Navβ2 (NM_012877)	F	365	tggacttaccaggagtgtagca	82
	R	424	cttcattcggaactggagga
Navβ3 (NM_139097)	F	576	gagggcggtaaagatttcct	49
	R	643	tggaaggggctctccact
Navβ4 (AF544988.1)	F	154	agaagctcatcactttcatcctg	126
	R	224	cccagaggaactcacgagac
β-Actin (NM_031144.2)	F	416	cccgcgagtacaaccttct	115
	R	479	cgtcatccatggcgaact

UPL, Universal Probe Library.

RESULTS

SNr GABA neurons fire faster spikes with larger amplitude than nigral DA neurons.

As shown in Fig. 1, A–C, putative SNr GABA neurons exhibited high-frequency spontaneous spiking with brief action potential duration and weak or no “sag” in response to hyperpolarizing current injections (Fig. 1, A and C). In contrast, the presumed DA neurons in SNc and SNr (nigral DA neurons hereafter) spiked spontaneously at low frequency with long action potential duration and displayed a pronounced hyperpolarization-activated current (I_h)-induced sag in response to hyperpolarizing current injection (Fig. 1, B and C, Table 3). These characteristics are consistent with published studies for these two neuron types (Atherton and Bevan 2005; Blythe et al. 2007; Lee and Tepper 2007; Richards et al. 1997; Tepper et al. 1995; Zhou et al. 2006, 2008). Our laboratory has previously shown that a K_v3-like channel-mediated potassium current is critical to the fast-repolarization, short-spike duration and high-frequency firing in SNr GABA neurons (Ding et al. 2011). We also noticed that, compared with nigral DA neurons, the action potentials of SNr GABA neurons have more negative threshold (−40.3 ± 0.5 vs. −36.2 ± 0.4 mV), a larger amplitude (69.5 ± 1.6 vs. 59.3 ± 1.4 mV), shorter duration (1.1 ± 0.1 vs. 2.9 ± 0.2 ms), shorter 10–90% rise time (0.27 ± 0.02 vs. 0.69 ± 0.05 ms), and faster rise rate (183.8 ± 17.3 vs. 71.8 ± 8.8 mV/ms) (Fig. 1, P values < 0.05, see Table 3). Also, SNr GABA neurons showed no or little adaptation in both firing frequency and spike amplitude, whereas nigral DA neurons were adaptive in both firing frequency and spike amplitude (Fig. 1, D and E). These results indicate that, in addition to differences in K_v channel-mediated repolarization that we have already characterized (Ding et al. 2011), SNr GABA neurons and nigral DA neurons may also have qualitative and/or quantitative differences in Na_v channel expression.

Fig. 1.

Electrophysiological characteristics of substantia nigra pars reticulata (SNr) GABA neurons and nigral dopamine projection (DA) neurons. A: electrophysiological properties of SNr GABA neurons. Current-clamp recordings show spontaneous high-frequency activity and minimal hyperpolarization-activated current (I_h)-mediated depolarizing sag in response to hyperpolarizing current injection. B: electrophysiological properties of nigral DA neurons. Current-clamp recordings show spontaneous low-frequency pacemaker activity and prominent I_h-mediated depolarizing sag (arrow) in response to hyperpolarizing current injection. C: overlay of a GABA neuron spike and a DA neuron spike, showing the clear differences in the spike waveform between the two cell types. D: SNr GABA neurons show little firing frequency and amplitude adaptation to depolarizing current injection (150 pA). E: nigral DA neurons show prominent firing frequency and spike amplitude adaptation to depolarizing current injection (100 pA). Recordings were made with a 135 mM KCl-based intracellular solution.

Table 3.

Action potential and Na_v current parameters in substantia nigra pars reticulata GABA neurons and nigral DA neurons

Parameters	GABA Neuron	DA Neuron	P
Action potential
Spontaneous firing rate, Hz	9.4 ± 0.6	1.9 ± 0.2	<0.01
Spike threshold, mV	−40.3 ± 0.5	−36.2 ± 0.4	<0.05
Amplitude, mV	69.5 ± 1.6	59.3 ± 1.4	<0.01
Base duration, ms	1.1 ± 0.1	2.9 ± 0.2	<0.01
10–90% spike rise time, ms	0.27 ± 0.02	0.69 ± 0.05	<0.01
Average rise rate, mV/ms	183.8 ± 17.3	71.8 ± 8.8	<0.01
INaT
Peak INaT density (at −10 mV), pA/pF	148.9 ± 17.8	86.0 ± 14.1	<0.01
Activation 10–90% rise time (at 0 mV), μs	85 ± 5	84 ± 5	>0.05
Activation V1/2, mV	−30.2 ± 0.2	−26.7 ± 0.5	<0.05
Activation slope K	6.2 ± 0.2	7.3 ± 0.3	<0.05
Inactivation 10−90% decay time (at 0 mV), μs	191 ± 12	195 ± 15	>0.05
Steady-state inactivation V1/2, mV	−63.3 ± 1.3	−61.3 ± 2.5	>0.05
Steady-state inactivation slope K	8.1 ± 0.5	10.1 ± 1.6	>0.05
Recovery from inactivation
τf, ms	0.59 ± 0.07	1.15 ± 0.1	<0.05
τs, ms	35.1 ± 6.4	79.7 ± 13.1	<0.05
INaR
Peak density (at −40 mV), pA/pF	3.3 ± 0.5	1.4 ± 0.6	<0.01
INaP
Whole cell peak amplitude (at −40 mV), pA	185.4 ± 17.5	121.3 ± 12.7	<0.01

Values are means ± SE. DA, dopamine projection; I_NaT, transient voltage-activated sodium current; I_NaP, persistent Na current; I_NaR, resurgent Na current; V_1/2, the potential at which the value of the Boltzmann function is 0.5; τ_f, fast recovery; τ_s, slow recovery. Spike threshold was measured at the point that had the sharpest inflection of membrane potential before the full spike. Base duration starts at the spike threshold point and ends when the membrane potential returns and crosses the spike threshold level.

Na_V channel gene expression in identified SNr GABA and nigral DA neurons.

After electrophysiologically fingerprinting SNr GABA and nigral DA neurons with a 135 mM KCl-based intracellular solution, we performed scRT-PCR to profile mRNAs for Na_Vα and Na_Vβ subunits. As illustrated in Fig. 2, mRNAs for Na_V1.1, Na_V1.2, Na_V1.3, Na_V1.6, Na_Vβ1, Na_Vβ2, Na_Vβ3, and Na_Vβ4 were detected in both SNr GABA neuron and SNc DA neuron, although there were apparent differences in the rate of detection. Since scRT-PCR is a qualitative method, these results indicate that SNr GABA neurons and SNc DA neurons may express identical types of Na_V subunits at different levels. Testing this idea requires quantitative comparison of Na_V subunit mRNA levels in SNr GABA neurons and SN DA neurons. However, the quantity of mRNA obtained in a single neuron was not sufficient for qRT-PCR. To overcome this difficulty, we collected and pooled SNr GABA neurons and SN DA neurons, separately, by using LCM and performed qRT-PCR analysis on Na_V channel subunit mRNAs isolated from these neurons. SNr GABA neurons and nigral DA neurons were identified by PV and TH fluorescent immunoreactivity, respectively (Gonzalez-Hernandez and Rodriguez 2000; Zhou 2009). As shown in Fig. 3, we found that the mRNA expression level of Na_V1.1, Na_v1.6, Na_Vβ1, and Na_Vβ4 in SNr GABA neurons was 2.07-, 2.03-, 2.15-, and 1.96-fold of that in SNc DA neurons, respectively (P < 0.005). In contrast, the level of Na_V1.2, Na_Vβ2, and Na_Vβ3 mRNAs in SNc DA neurons was 1.65-, 1.69-, and 3.26-fold of that in SNr GABA neurons (P < 0.005). As in scRT-PCR, Na_V1.3 was detectable in qRT-PCR, but the C_p values (defined in methods) of SNr GABA neurons and SNc DA neurons were too high (more than 40), indicating low abundance and potentially unreliable quantification. Thus Na_V1.3 was excluded from quantitative comparison in this study.

Fig. 2.

Single-cell RT-PCR scRT-PCR detection of Na_v mRNAs in SNr GABA neurons and nigral DA neurons. A: voltage-activated sodium (Na_V) mRNAs detected in electrophysiologically identified SNr GABA neurons. No tyrosine hydroxylase (TH) mRNA was detected in any of these neurons. Numbers in the parentheses are the expected amplicon sizes in base pair. B: Na_V mRNAs detected in electrophysiologically identified nigral DA neurons. The expected amplicon sizes are the same as in A. C: summary of scRT-PCR detection ratio of Na_v channel mRNAs in SNr GABA neurons and nigral DA neurons. For each Na_V mRNA, 5–10 TH mRNA-positive DA neurons and 5–10 glutamate decarboxylase 1 (GAD1) mRNA-positive neurons were tested. In this figure and Fig. 3, the Na_V α-subunits are denoted as α1, α2, α3, and α6, while the standard nomenclature (Na_V1.1, Na_V1.2, Na_V1.3, and Na_V1.6) is used in the main text (Catterall et al. 2005).

Fig. 3.

Quantitative RT-PCR (qRT-PCR) analysis of Na_v channel mRNAs in laser microdissection-captured, immunofluorescence-identified SNr GABA neurons and nigral DA neurons. β-Actin mRNA was used as internal control. mRNA semiquantification was performed using the comparative crossing point (C_p) method in the form of 2^−ΔCp. 2^−ΔCp values for Na_V mRNAs were normalized to those in SNr GABA neurons. Na_V1.1, Na_V1.6, Na_Vβ1, and Na_Vβ4 mRNAs were higher in SNr GABA neurons than in nigral DA neurons, whereas Na_V1.2, Na_Vβ2, and Na_Vβ3 mRNAs were lower in SNr GABA neurons than in nigral DA neurons. The difference in the expression of these Na_V genes between SNr GABA neurons and nigral DA neurons was significant with P < 0.005.

Based on these qRT-PCR data on Na_V mRNAs (Fig. 3), the differences in action potential waveforms in SNr GABA neurons and SNc DA neurons (Fig. 1), and the different electrophysiological characteristics of Na_V subunits in heterologous expression systems (Goldin 2001), we made the following prediction: compared with Na_V channels in SNc DA neurons, Na_v channels in SNr GABA neurons 1) are expressed at a higher density; 2) activate at a lower threshold and/or with a steeper voltage dependence, and 3) recover faster from inactivation and thus have less cumulative inactivation; and 4) conduct larger I_NaP and I_NaR. These Na_V channel properties would support the sustained fast spiking in SNr GABA neurons.

Higher I_NaT density in SNr GABA neurons than in DA neuron.

To study Na_V currents in SNr GABA neurons and nigral DA neurons, we performed nucleated patch-clamp recordings in rat brain slices (Ding et al. 2011; Martina and Jonas 1997). The key advantage of nucleated patch clamp is the substantially reduced spatial and voltage-clamp problems because of the elimination of axonal and dendritic processes. To help isolate Na_V currents, a 135 mM CsCl-based intracellular solution was used to block potassium channels. Cs⁺ also blocks the I_h that is prominent in nigral DA neurons, but lacking or small in SNr GABA neurons (Fig. 1). Thus we performed electrophysiological identification of SNr GABA and nigral DA neurons within the first 10 s after obtaining access to the cell interior. A similar method was used by Martina and Jonas (1997). As shown in Fig. 4, A and B, during this short time window, SNr GABA neurons and nigral DA still displayed distinct electrophysiological properties. SNr GABA neurons exhibited spontaneous firing at frequencies ≥10 Hz and little or no “sag” in response to hyperpolarizing current injection. In contrast, nigral DA neurons exhibited spontaneous firing at frequencies ≤5 Hz and prominent “sag” in response to hyperpolarizing current injection. After electrophysiological identification, nucleated membrane patches were isolated from these neurons (Fig. 4, C–E).

Fig. 4.

Electrophysiological identification of SNr GABA neurons and nigral DA neurons with 135 mM CsCl-based intracellular solution and nucleated patch formation. A: electrophysiological properties of SNr GABA neurons. Current-clamp recordings within 10 s after obtaining access to the cell interior show spontaneous high-frequency activity and no I_h-mediated sag in response to hyperpolarizing current injection (−200 pA). Due to Cs⁺ infusion, the spikes were broadened, and the firing frequency was increased. B: electrophysiological properties of nigral DA neurons. Current-clamp recordings within 10 s after obtaining access to the cell interior show spontaneous low-frequency activity and prominent I_h-mediated sag in response to hyperpolarizing current injection. The spikes were broadened, and the firing frequency was increased due to Cs⁺ infusion. Despite Cs⁺ infusion, the differences in membrane properties are still striking within the first 10 s. C, D, and E: sequential images demonstrating the formation of nucleated somatic membrane patch. C shows conventional whole cell patch clamp to electrophysiologically fingerprint SNr GABA and nigral DA neurons. D shows that gentle negative pressure was being applied, and the patch pipette was being withdrawn slowly to form nucleated patches. Arrow indicates the remaining part of the neuron. Arrowhead points to the round nucleated membrane patch. E: spherical nucleated patches free of axon and dendritic processes was formed, as demonstrated by including 20 μM fluorescent Alexa 594 in the pipette solution.

We first compared I_NaT density in nucleated membrane patches from SNr GABA neurons and nigral DA neurons. I_NaT was evoked from a holding potential of −100 mV to test potentials ranging −80 to +30 mV in 5-mV steps. As shown in Fig. 5, I_NaT reached its peak amplitude around −10 mV in both SNr GABA neurons and nigral DA neurons. The peak I_NaT and thus conductance density were clearly higher in SNr GABA neurons than in nigral DA neurons: 148.9 ± 17.8 pA/pF and 2481.8 ± 296.2 pS/pF for SNr GABA neurons (n = 12) and 86.0 ± 14.1 pA/pF and 1433.5 ± 234.2 pS/pF for nigral DA neurons (n = 9, P < 0.05, Table 3). If we assume the specific membrane capacitance to be 1 μF/cm² (Hille 2001), then I_NaT conductance density was 24.8 ± 3.0 pS/μm² for SNr GABA neurons and 14.3 ± 2.3 pS/μm² for nigral DA neurons. The reversal potential used for calculating the conductance was obtained by linear fitting the data points between 0 mV and 30 mV (Fig. 5), following the method of Safronov and Vogel (1995) and Raman et al. (2000). The estimated reversal potential was 49.8 mV for GABA neurons and 49.9 mV for DA neurons. The higher I_NaT density may contribute to the faster rise and larger amplitude of the action potentials in SNr GABA neurons.

Steeper voltage-dependent I_NaT activation in SNr GABA neurons than in nigral DA neurons.

We next compared the voltage dependence and kinetics of I_NaT in SNr GABA neurons and nigral DA neurons. The voltage-dependent activation of I_NaT was studied by holding the nucleated membrane patch at −100 mV and then stepping to test membrane potentials between −80 mV and 30 mV (Fig. 6A1). Representative I_NaT traces at these test membrane potentials and averaged activation curves were shown in Fig. 6, A2–A4. I_NaT started to activate around −50 mV in both cell types. However, with increasing depolarization, the increase of I_NaT conductance was faster and reached the maximum earlier in SNr GABA neurons than in DA neurons, indicating that I_NaT had a steeper voltage-dependent activation in SNr GABA neurons than in SN DA neurons (Fig. 6A4). Boltzmann equation fitting revealed that the activation slope factor K was 6.2 ± 0.2 for SNr GABA neurons (n = 12) and 7.3 ± 0.3 for nigral DA neurons (n = 9, P < 0.05). The activation midpoint potential V_1/2 was also different for the two cell types: V_1/2 was −30.2 ± 0.6 mV for SNr GABA neurons (n = 12) and −26.7 ± 0.5 mV for nigral DA neurons (n = 9, P < 0.05, Table 3). The steeper activation curve with a more negative midpoint indicates that I_NaT in SNr GABA neurons can be activated more readily than those in nigral DA neurons, contributing to the more negative spike threshold and faster spike rise rate in SNr GABA neurons.

Fig. 6.

Voltage-dependent activation and steady-state inactivation of I_NaT in SNr GABA and nigral DA neurons. A, A1: activation voltage waveform to evoke I_NaT. A2 and A3: representative traces of I_NaT evoked in nucleated membrane patches isolated from a SNr GABA neuron (A2) and a nigral DA neuron (A3). A4: activation curves of I_NaT in SNr GABA neurons (filled circles) (n = 12) and nigral DA neurons (open circles) (n = 9). Continuous lines are Boltzmann fits. Data points below −60 mV and above 5 mV are not displayed in order to show the difference between the two curves more clearly. B, B1: voltage waveform to induce and determine I_NaT steady-state inactivation. B2 and B3: representative traces of I_NaT evoked by the steady-state inactivation protocol in nucleated membrane patches isolated from a SNr GABA neuron (B2) and a nigral DA neuron (B3). B4: steady-state inactivation curves of I_NaT in SNr GABA neurons (filled circles) (n = 12) and nigral DA neurons (open circles) (n = 9). Continuous lines are Boltzmann fits. G/G_max, normalized conductance; I/I_max, normalized current.

As expected for I_NaT, both the activation (measured by 10–90% rise time) and inactivation (measured by 10–90% decay time) of I_NaT were fast (Fig. 6, A2, A3, B2, B3, Table 3). However, we did not detect any significant difference in I_NaT rise time and decay time in these two cell types. For example, when stepped from −100 mV to 0 mV, the 10–90% rise time was 85 ± 5 μs for SNr GABA neurons (n = 12) and 84 ± 5 μs for nigral DA neurons (n = 9); the 10–90% decay time was 191 ± 12 μs for SNr GABA neurons and 195 ± 15 μs for nigral DA neurons.

Similar steady-state inactivation of I_NaT in SNr GABA neurons and nigral DA neurons.

Next, we studied the steady-state inactivation of I_NaT in SNr GABA and nigral DA neurons. Representative traces evoked by a steady-state inactivation protocol were shown in Fig. 6B. I_NaT in SNr GABA neurons and nigral DA neurons showed a similar steady-state inactivation. Boltzmann equation fitting revealed that the slope factor of steady-state inactivation was 8.1 ± 0.5 for SNr GABA neurons (n = 12) and 10.1 ± 1.6 for nigral DA neurons (n = 9), respectively (P > 0.05). The midpoint potential of steady-state inactivation was −63.3 ± 1.3 mV for SNr GABA neurons (n = 12) and −61.3 ± 2.5 mV for nigral DA neurons (n = 9), respectively (P > 0.05). These results show that steady-state inactivation of I_NaT did not differ between the two types of neurons.

I_NaT recovery from inactivation is faster in SNr GABA neurons than in nigral DA neurons.

Fast recovery (τ_f) of I_NaT from inactivation is essential for high-frequency repetitive firing. Based on the fact that the spike amplitudes decline less in SNr GABA neurons than in nigral DA neuron (Fig. 1, D and E), we hypothesized that I_NaT may recover more rapidly from inactivation such that more functional Na_V channels are available in SNr GABA neurons than in nigral DA neurons. We examined the recovery time course from inactivation with a double-pulse protocol (Fig. 7A): the holding potential was −90 mV, then a 50-ms prepulse at −120 mV, then a 300-ms pulse at 0 mV to completely inactivate I_Na, then stepping to −120 mV for increasing duration (Δt) to allow I_NaT recover, and finally a 20-ms test pulse at 0 mV (Fig. 7A). We found that the time course for I_NaT to recover from inactivation was biexponential in both SNr GABA neurons and SNc DA neurons (Fig. 7, B–D). The τ_f component contributed 52.6 ± 5.7 and 35.8 ± 3.4% (recovered within 5 ms) to the total current in SNr GABA neurons and nigral DA neurons, respectively. Mostly importantly, I_NaT recovered faster in SNr GABA neurons than in SNc DA neurons. The τ_f was 0.59 ± 0.07 ms for SNr GABA neurons (n = 5) and 1.15 ± 0.1 ms for SNc DA neurons (n = 5, P < 0.05). The slow recovery (τ_s) was 35.1 ± 6.4 ms for SNr GABA neurons (n = 5) and 79.7 ± 13.1 ms for SNc DA neurons (n = 5, P < 0.05, Table 3). These results indicate that I_NaT in the fast spiking SNr GABA neurons recovered more rapidly from inactivation than that in the slow spiking SNc DA neurons.

Fig. 7.

I_NaT recovered more rapidly from inactivation in SNr GABA neurons than in substantia nigra pars compacta (SNc) DA neurons. A: voltage protocol for recovery from a 300-ms-induced inactivation. B and C: representative traces of recovery from inactivation in a SNr GABA neuron nucleated patch (B) and a SNc DA neuron nucleated patch (C). D: pooled data on I_NaT recovery time course from inactivation by a 300-ms conditioning pulse (I). Curves represent sums of two exponential fits. Filled circles are nucleated patches from SNr GABA neuron (n = 5). Open circles are nucleated patches from SNc DA neuron (n = 5).

Because of the different recoveries from inactivation, we reasoned that, compared with nigral DA neurons, SNr GABA neurons may incur less cumulative I_NaT inactivation during repetitive depolarization. To test this idea, we performed the experiment shown in Fig. 8. The membrane patch was held at −70 mV, then stepped to a train of five pulses (3 ms at 0 mV with an interpulse interval of 100 ms at −70 mV, Fig. 8A). The ratio of the I_NaT evoked by a pulse over that evoked by the first pulse, I_n/I₁, was used as a measure of cumulative inactivation. The ratio becomes small when cumulative inactivation is large. As illustrated in Fig. 8, B–D, by the third pulse, the ratio was already significantly larger in SNr GABA neurons (n = 6) than in SNc DA neurons (n = 5), although without reaching statistical significance. By the fourth pulse, the difference was statistically significant (P < 0.05). These results indicate that, during repetitive depolarization such as spike firing, Na_V channels undergo less cumulative activation in SNr GABA neurons than in nigral DA neurons. This difference may contribute to the nonadaptive spiking in SNr GABA neurons and the adaptive spiking in nigral DA neurons, respectively (Fig. 1, D and E).

Fig. 8.

Less cumulative I_NaT inactivation in SNr GABA neurons than in SN DA neurons. A: voltage protocol: holding potential, −70 mV; 3-ms test pulses to 0 mV at 100-ms interval. B and C: representative traces of cumulative I_NaT inactivation in nucleated membrane patches from a SNr GABA neuron (B) and SNc DA neuron (C). D: pooled data on I_NaT cumulative inactivation. Peak I_NaT was normalized to the I_NaT evoked by the first test pulse I₁. *Significant difference at P < 0.05.

I_NaT deactivation is faster in SNr GABA neurons than in nigral DA neurons.

Na_V channels with slower deactivation kinetics can conduct longer-lasting inward current upon membrane repolarization and prolong action potential duration. We compared the deactivation properties of I_NaT in SNr GABA neurons and nigral DA neurons. As illustrated in Fig. 9D, I_NaT was evoked by a 200-μs pulse to 0 mV. Based on our data on I_NaT activation (Fig. 6, A2 and A3), at the end of a 200-μs pulse at 0 mV, I_NaT and thus Na_V channel opening were at or near their peak. To deactivate Na_V channels, the membrane was stepped to −100 mV through −20 mV in a 10-mV step. As shown in Fig. 9, A and B, I_NaT deactivated very quickly, as reflected by the rapid decay of the tail currents in both neuron types. These tail currents were well fitted with a single exponential function. At very negative membrane potentials (−100 mV and −90 mV), the deactivation was only slightly faster in SNr GABA neurons than in nigral DA neurons. For example, the deactivation time constant at −100 mV was 48.1 ± 2.6 μs for SNr GABA neurons (n = 7) and 54.9 ± 5.3 μs for SNc DA neurons (n = 6), but the difference was not statistically significant (P > 0.05). When the deactivation pulse became more positive, particularly at −50 mV to −20 mV, which is in the range of action potential repolarization, the deactivation time course became significantly faster in SNr GABA neurons than in SNc DA neurons (Fig. 9, A–C). For example, at −40 mV, the deactivation time constant was 99 ± 8.9 μs for SNr GABA neurons (n = 7) and 127.8 ± 12.2 μs for nigral DA neurons (n = 6, P < 0.01). The faster deactivation of I_NaT, together with a robust K_v3-like current (Ding et al. 2011), can thus repolarize the membrane more quickly in the fast-spiking SNr GABA neurons than in slow-spiking SNc DA neurons (see Fig. 1).

Fig. 9.

I_NaT deactivation is faster in SNr GABA neurons than in SNc DA neurons. A and B: representative traces of I_NaT deactivation in nucleated membrane patches from a SNr GABA neurons (A) and a SNc DA neuron (B). The two traces at −40 mV are displayed in A′ and B′ to show more clearly I_NaT deactivation at −40 mV. The smooth gray curves are single-exponential fits. C: pooled data for deactivation time constants plotted against deactivation potentials. Filled circle: SNr GABA neurons. Open circles: SNc DA neurons. The time constants were shorter at −80 to −20 mV, P < 0.05. D: voltage waveform used to evoke I_NaT deactivation. Holding potential, −90 mV; 50-ms pulse to −120 mV; 200-μs pulse to 0 mV; 50-ms test pulse from −100 mV to −20 mV at 10-mV increase.

I_NaR is larger in SNr GABA neurons than in nigral DA neurons.

I_NaR is generated during membrane repolarization after depolarization and inactivation of I_NaT (Bant and Raman 2010; Grieco et al. 2005; Raman and Bean 1997). I_NaR is much smaller but longer-lasting than I_NaT. Na_V1.6 (α6) and Na_Vβ4 (β4) are critical for I_NaR generation (Bant and Raman 2010; Do and Bean 2004; Grieco and Raman 2004; Grieco et al. 2005). In cerebellar Purkinje neurons and granule neurons and the GABA neurons in the globus pallidus, the Na_v channels containing Na_V1.6 and/or Na_vβ4 have a resurgent property that generates a I_NaR after the inactivation of the I_NaT and upon repolarization (Bant and Raman 2010; Khaliq et al. 2003; Mercer et al. 2007; Raman and Bean 1997). I_NaR has been suggested to facilitate fast spiking. Our qRT-PCR data (Fig. 3) indicated that Na_Vβ4 mRNA is expressed at a higher level in SNr GABA neurons than in nigral DA neurons. Using a voltage protocol illustrated in Fig. 10D, we observed I_NaR in both SNr GABA neurons and SNc DA neurons (Fig. 10, A and B). Consistent with our qRT-PCR data that compared with nigral DA neurons, SNr GABA neurons had more Na_v1.6 and Na_vβ4 subunit expression; our nucleated patch-clamp recordings showed that I_NaR was larger in SNr GABA neurons than in nigral DA neurons (Fig. 10, P < 0.01, Table 3). Na_v1.6 and Na_vβ4 subunit expression has been reported to be the main subunits responsible for large I_NaR in other types of neurons (Aman and Raman 2007; Bant and Raman 2010; Grieco et al. 2004; Mercer et al. 2007; Rush et al. 2005).

Fig. 10.

Resurgent Na current (I_NaR) is larger in SNr GABA neurons than in SNc DA neurons. A and B: representative traces of I_NaR in nucleated membrane patches from a SNr GABA neuron (A1) and a SNc DA neuron (B1). The boxed areas in A1 and B1 are expanded and displayed in A2 and B2, respectively. C: polled data from SNr GABA neurons (n = 5) and SNc DA neurons (n = 4). The difference was significant (P < 0.05). D: voltage waveform used to generate I_NaR. Holding potential between the pulses was −90 mV.

I_NaP is larger in SNr GABA neurons than in nigral DA neurons.

Separate studies have examined I_NaP in SNr GABA neurons (Atherton & Bevan 2005) and SNc DA neurons (Puopolo et al. 2007). But a direct, side-by-side comparison of I_NaP between the two types of neurons was lacking. So we set out to compare the amplitude of I_NaP and their possible roles in firing properties and membrane potentials between SNr GABA neurons and SN DA neurons. Since I_NaP was slow, conventional whole cell patch-clamp recording was suitable. We used the CsCl-based intracellular solution (see methods section) to block K⁺ channels. We also used a slow 4-s linear voltage ramp from −80 mV to 0 mV to inactivate the I_NaT (Fig. 11A) (Gorelova and Yang 2000; Koizumi and Smith 2008). Since I_NaP is known to be blocked by 1 μM TTX (Crill 1996; French et al. 1990; Gorelova and Yang 2000; Koizumi and Smith 2008), the slow TTX-sensitive current was taken as the I_NaP. After a stable baseline recording was obtained, 1 μM TTX was bath-applied. As shown in Fig. 11, a quite large TTX-sensitive, slow sodium current or I_NaP was recorded in both SNr GABA neurons and nigral DA neurons (Fig. 11, A2 and A3). I_NaP started to activate around −60 mV and peaked around −40 mV. The peak amplitude of I_NaP was 185.4 ± 17.5 pA in seven SNr GABA neurons and 121.3 ± 12.7 pA in six SNc SN DA neurons (P < 0.05, Fig. 11A4, Table 3). To explore the functional roles of I_NaP, we recorded the membrane potential and action potential firing in conventional whole cell mode with the 135 mM KCl-based intracellular solution and the normal extracellular solution (see methods). After obtaining a stable baseline, 10 μM riluzole, an established I_NaP blocker (Koizumi and Smith 2008; Urbani and Belluzzi 2000), was bath-applied. As shown in Fig. 11B, bath application of 10 μM riluzole induced a 11.9 ± 1.1 mV hyperpolarization from its baseline potential around −50 mV in SNr GABA neurons (n = 6) and a 7.3 ± 0.6 mV hyperpolarization from its baseline potential around −50 mV in nigral DA neurons (n = 5), leading to a cessation of spontaneous firing in both neuron types. Since the slow hyperpolarization is not likely due to a potential riluzole inhibition of I_NaT (Urbani and Belluzzi 2000), these results indicate that I_NaP was active at the baseline membrane potential around −50 mV in SNr GABA neurons and nigral DA neurons. The larger I_NaP in SNr GABA neurons may contribute to the faster pacemaking activity in SNr GABA neurons than in SN DA neurons.

Fig. 11.

Persistent Na current (I_NaP) is larger in SNr GABA neurons than in SNc DA neurons. A: voltage clamp data on I_NaP. A1: diagram showing the voltage ramp waveform consisting of a 4-s ramp from −80 mV to 0 mV. A2 and A3: representative current traces before (black line) and after (thin gray line) perfusion of 1 μM tetrodotoxin (TTX) evoked by the voltage ramp, in GABA neurons (A2) and DA neurons (A3). TTX-sensitive I_NaP (thick gray line) was obtained by subtraction traces before after perfusion of 1 μM TTX. A4: pooled data of peak I_NaP in SNr GABA (n = 7) and SNc DA (n = 6) neurons. B: current clamp data on I_NaP. B1 and B2: representative traces showing the effects of 10 μM riluzole in SNr GABA neurons (B1) and SNc DA neurons (B2). Bath perfusion of 10 μM riluzole induced a stronger hyperpolarization in GABA neurons than in DA neurons. Spike amplitudes were truncated due to slow resampling (1 kHz) for reducing the large data size of long recording segments. B3: pooled data on 10 μM riluzole-induced hyperpolarization in SNr GABA neurons (n = 6) and SNc DA neurons (n = 5). The difference was significant (P < 0.05).

I_NaT in SNr GABA neurons and nigral DA neurons at room temperature.

To verify that our conclusions based on recordings made at 30°C and filtered at 10 kHz were correct, additional experiments were performed at room temperature (25°C), low-pass filtered at 30 kHz, and sampled at 200 kHz. I_NaT activation, decay, and deactivation were reexamined because these parameters were fast and thus sensitive to recording temperature and signal filtering. As shown in Figs. 12 and 13 and Table 4, although slower than at 30°C, the kinetic differences seen at 30°C between these two cell types remained at room temperature, while the kinetic parameters that were similar at 30°C were still similar at room temperature. The only change in I_NaT kinetics that needs to be mentioned in particular is that the 10–90% decay time was prolonged more substantially than the 10–90% rise time. Although the mechanisms are not clear, differential temperature effects on the rise and the decay of I_NaT have been observed (Magatomo et al. 1998; Sah et al. 1988). The I_NaT density is similarly reduced in both cell types at room temperature compared with that at 30°C, consistent with literature data (Sah et al. 1988). These data indicate that the kinetic differences in I_NaT in nigral DA neurons and SNr GABA neurons observed at 30°C may be extrapolated to 37°C, the normal body temperature, and thus are physiologically important.

Table 4.

I_NaT current properties in substantia nigra pars reticulata GABA and nigral DA neurons at 25°C, filtered at 30 kHz, and sampled at 200 kHz

Parameter	GABA Neurons	DA Neurons	Significance P
Activation midpoint, mV	−34.81 ± 1.2 (n = 9)	−29.90 ± 1.30 (n = 11)	<0.05
Slope factor k	7.04 ± 0.17 (n = 9)	7.75 ± 0.13 (n = 11)	<0.05
10-90% Rise time (at 0 mV), ms	0.10 ± 0.004 (n = 9)	0.11 ± 0.004 (n = 11)	>0.05
90-10% Decay time (at 0 mV), ms	0.72 ± 0.03 (n = 9)	0.68 ± 0.04 (n = 11)	>0.05
Deactivation τ (at −40 mV), ms	0.16 ± 0.01 (n = 4)	0.25 ± 0.02 (n = 5)	<0.05

Values are means ± SE; n, no. of neurons.

DISCUSSION

The main findings of this study are that, compared with slow-spiking nigral DA neurons, fast-spiking SNr GABA neurons had higher gene expression levels for Na_Vα1.1, Na_Vα1.6, Na_Vβ1, and Na_vβ4; and larger amplitudes of I_NaT, I_NaP, and I_NaR. I_NaT also had a steeper voltage-dependent activation, a faster deactivation, and a faster recovery from inactivation in SNr GABA neurons than in nigral DA neurons. These differences in Na_V currents may contribute to the striking differences in spiking properties in these two cell types.

Different Na_V mRNA expression in SNr GABA neurons and nigral DA neurons.

The strikingly different spike waveform and spiking pattern in SNr GABA neurons and nigral DA neurons suggest possible molecular and functional differences in Na_V channels between the two neuron types. Our qualitative scRT-PCR analysis revealed similar types of Na_V mRNAs in SNr GABA neurons and nigral DA neurons, indicating the two cell types probably express qualitatively similar but quantitatively different Na_V subunits, leading to the formation of Na_V channels with different subunit composition, particularly with different regulatory β-subunits. Indeed, using qRT-PCR on immunochemically identified SNr GABA and SNc DA neurons, we found that SNr GABA neurons had higher expression levels of Na_V1.1, Na_V1.6, Na_Vβ1, and Na_Vβ4 and lower expression levels of Na_V1.2, Na_Vβ2, and Na_Vβ3 than SN DA neuron (Fig. 3). The high Na_vβ4 level in SNr GABA neurons compared with SNc DA neurons is also consistent with a histochemical study that detected high Na_vβ4 mRNA in SNr but not SNc (Yu et al. 2003). These differences in Na_V channel expression levels may affect Na_V channel expression density and also subunit composition and hence Na_V channel properties and function. Specifically, the higher Na_V1.1 and Na_V1.6 mRNA expression, together with Na_vβ1 and Na_vβ4 mRNA expression, likely contributes to the larger amplitude or higher density of I_NaT, I_NaP, and I_NaR in SNr GABA neurons than in nigral DA neurons. Na_Vβ1 may enhance Na_V channel cell surface expression and increase I_NaT (Isom et al. 1992, 1995; McEwen et al. 2004). Na_Vβ1 may also promote fast gating for Na_V α-subunits (Goldin 2001). Recent studies indicate that Na_vβ4 may interact with Na_V α-subunit and enhance I_NaP and I_NaR (Bant and Raman 2010; Grieco et al. 2005) and will be further discussed below. The significance of higher Na_V1.2, Na_Vβ2, and Na_Vβ3 mRNA expression in DA neurons than in GABA neurons is not clear. One possibility is that our semiquantitative qRT-PCR only compared the relative abundance of Na_V1.2 mRNA between the two cell types and did not compare the relative abundance of Na_V1.1 mRNA, Na_V1.2 mRNA, and Na_V1.6 mRNA within each cell type. Thus it is possible that Na_V1.1 and Na_V1.6 are the dominant Na_V subunits, such that the difference in Na_V1.2 is not important. Alternatively, we speculate that, while Na_V1.1 and Na_V1.6 are the major α-subunits with Na_Vβ1 and Na_Vβ4 as the regulatory subunits in SNr GABA neurons, Na_V1.2 and Na_V1.6 may be the major α-subunits with Na_Vβ2 and Na_Vβ3 as the regulatory subunits in nigral DA neurons. Future molecular studies are needed to determine the absolute mRNA expression levels for these different Na_V subunits and Na_V channel protein subunit composition.

Different I_NaT density and kinetics in SNr GABA neuron and nigral DA neuron.

In the present study, we found that I_NaT and conductance density were considerably higher in fast-spiking SNr GABA neurons than in slow-spiking nigral DA neurons: 2,481.8 ± 296.2 pS/pF vs. 1,433.5 ± 234.2 pS/pF (Fig. 5, Table 3). This may be due to higher expression of Na_V1.1 and Na_V1.6 and also auxiliary Na_vβ1 and Na_vβ4. The higher I_NaT density may contribute to the faster rise and larger amplitude of the action potentials in SNr GABA neurons.

In addition to the difference in I_NaT density, we also detected differences in I_NaT kinetics between SNr GABA neurons and SNc DA neurons. To compare the very fast kinetics of I_NaT, we used the nucleated patch-clamp techniques that provide outstanding space and voltage clamp (Ding et al. 2011; Martina and Jonas 1997). We found that the activation and inactivation time courses of I_NaT were similar in SNr GABA neurons and nigral DA neurons (Fig. 6, Table 3). However, I_NaT had a steeper voltage-dependent activation and a faster deactivation in SNr GABA neurons than those in nigral DA neurons (Figs. 6 and 9, Table 3), potentially contributing to the fast rise and short duration of spikes in the fast-spiking SNr GABA neurons. I_NaT also recovered more quickly from inactivation in SNr GABA neurons than in SN DA neurons, leading to less cumulative inactivation in SNr GABA neurons. These results are consistent with the combination of our qPCR data that SNr GABA neurons express more Na_V1.6 and less Na_V1.2 than DA neurons (Fig. 3) and published results that Na_V1.2 undergo more cumulative inactivation than Na_V1.6 in peripheral neurons (Rush et al. 2005). The different rate of recovery from inactivation may contribute to the fact that spike firing was not or was only slightly adaptive in amplitude in SNr GABA neurons, whereas it was strongly adaptive in SNc DA neurons (Fig. 1, D and E).

The differences in I_NaT kinetics between fast-spiking SNr GABA neurons and slow-spiking DA were relatively modest (Table 3). This is consistent with published studies that different heterologously expressed Na_v channel subunits or isoforms display only minor or subtle differences in their kinetic properties (Goldin 2001; Smith and Goldin 1998). However, subtle differences in I_NaT kinetics may be sufficient to affect spike initiation and waveform (Engel and Jonas 2005; Hu et al. 2009). Parts of our results on I_NaT (I_NaV density and inactivation time course) are also consistent with a recent study (Seutin and Engel 2010). However, some of our results on I_NaT kinetics are in contrast to Seutin and Engel (2010). For example, Seutin and Engel (2010) indicated a faster rise time for I_NaT in GABA neurons than in DA neurons, whereas our data indicated a similar rise time for I_NaT for both cell types. Seutin and Engel (2010) also indicated that there was no difference in activation, deactivation, and recovery from inactivation, whereas we found differences in these three parameters. Specifically, Seutin and Engel (2010) reported that the midpoint potentials of activation were −9.6 mV and −12.6 mV for DA and GABA neurons, respectively. These two values are over 10 mV more positive than the V_1/2 of I_NaT activation reported for multiple types of neurons (Colbert and Pan 2002; Gittis and Lac 2008; Hu et al. 2009; Martina and Jonas 1997; Maurice et al. 2004; Raman et al. 2000). Furthermore, Seutin and Engel (2010) reported that the midpoint potential of steady-state inactivation was more positive (−49 mV) in DA neurons than in GABA neurons (−56 mV), indicating that Na_V channels in DA neurons are less likely to inactivate, whereas we found there was no significant difference in this parameter: −63.3 mV for SNr GABA neurons and −61.3 mV for nigral DA neurons. I_NaT recovery from inactivation is important and needs to be discussed. Seutin and Engel (2010) also found no difference in time course of recovery from fast inactivation among the two cell types, while we found that I_NaT recovered more quickly from both fast and slow inactivation in SNr GABA neurons than in nigral DA neurons. We also found that there was less cumulative inactivation in I_NaT in SNr GABA neurons than in SNc DA neurons, indicating that more Na_V channels become available after action potential firing and the refractory period is shorter. Another discrepancy between our present study and Seutin and Engel (2010) is the deactivation time course. Our data indicate a faster deactivation in SNr GABA neurons than in nigral DA neurons (Fig. 9), whereas Seutin and Engel (2010) reported no difference. We believe that this discrepancy can be explained by the fact that we used a different deactivation protocol. We first used the same protocol with 300 μs at 0-mV activation pulse used by Seutin and Engel (2010) and saw no apparent difference in deactivation (data not shown). However, we noticed that, at the end of 300 μs at 0 mV, I_NaT had inactivated significantly. Thus 300 μs at 0-mV activation pulse was too long for studying deactivation under our recording conditions. Based on our activation data, I_NaT was at its peak at 200 μs after stepping to 0 mV. Thus we shortened the activation part to 200 μs in the deactivation protocol and detected the faster deactivation in SNr GABA neurons. In our opinion, our data on I_NaT kinetics can better explain the differences in spiking properties in fast spiking SNr GABA neurons and slow spiking DA neurons. For example, Na_V channels with a steeper activation slope, faster deactivation, and faster recovery from inactivation can support the fast spiking in SNr GABA neurons. Certainly, independent studies are required to resolve these discrepancies.

Larger I_NaR in SNr GABA neurons than in SN DA neurons.

The I_NaR is generated during membrane repolarization after depolarization and inactivation of I_NaT (Raman and Bean 1997). Particularly, Na_Vβ4 may act as an open channel blocker, binding to the open Na_V channels upon depolarization. Upon repolarization, Na_vβ4 peptide unbinds and allows Na⁺ flow briefly, thereby producing I_NaR (Bant and Raman 2010; Grieco et al. 2005). I_NaR is much smaller but longer-lasting than I_NaT. Evidence indicates that Na_V1.6 (α6) and Na_Vβ4 (β4) are critical for I_NaR generation (Bant and Raman 2010' Do and Bean 2004; Grieco and Raman 2004; Raman et al. 1997; Grieco et al. 2005). Since I_NaR is induced upon repolarization, it may promote repetitive firing (Khaliq et al. 2003).

Our qRT-PCR data indicated that Na_v1.6 and Na_Vβ4 mRNA are expressed at higher levels in SNr GABA neurons than in nigral DA neurons (Fig. 3). In addition, our nucleated patch-clamp recordings showed that I_NaR was larger in SNr GABA neurons than in nigral DA neurons (Fig. 10). These results are consistent with reports suggesting that Na_v1.6 and Na_vβ4 subunits are the key subunits responsible for generating I_NaR in other neuron types (Aman and Raman 2007; Bant and Raman 2010; Grieco et al. 2004; Raman et al. 1997; Mercer et al. 2007; Rush et al. 2005).

Larger I_NaP in SNr GABA neurons than in SN DA neurons.

Although the precise mechanisms generating I_NaP are not clear, studies indicate that I_NaP may be generated by the same Na_V channels that generate the I_NaT, but via different gating mechanisms that allow a small fraction of Na_V channels to remain open for long periods of time (Alzheimer et al. 1993; Bant and Raman 2010; Crill 1996; Patlak and Ortiz 1986; Taddese and Bean 2002). Evidence also indicates that Na_Vβ4 (β4) subunit, when coexpressed Na_V1.1 (α1) subunit, promotes the generation of I_NaP and I_NaR (Aman et al. 2009; Bant and Raman 2010). Although the amplitude of I_NaP is only a fraction of I_NaT, its duration is much longer than that of I_NaT, such that it can enhance neuronal excitability (Crill 1996). Consequently, I_NaP is involved in pacemaking activity in many types of neurons (Do and Bean 2003; Jackson et al. 2004; Khaliq and Bean 2010; Koizumi and Smith 2008; Mercer et al. 2007; Puopolo et al. 2007; Swensen and Bean 2003; Taddese and Bean 2002). Previous studies also examined I_NaP in SNr GABA neurons and SNc DA neurons (Atherton and Bevan 2005; Chan et al. 2007; Puopolo et al. 2007), although a direct side-by-side comparison of I_NaP in these two neuron types was lacking. Based on our qRT-PCR results, we predicted larger I_NaP in SNr GABA neurons than SN DA neurons, and our experiment proved this hypothesis. We also found that I_NaP took part in maintaining the membrane potential at a relatively depolarized range, in addition to pacemaking. Blocking I_NaP with 10 μM riluzole caused a larger hyperpolarization in SNr GABA neurons than SN DA neurons. Those results indicated that I_NaP played an important role in shaping the different firing patterns between SNr GABA neurons and SN DA neurons.

Our electrophysiological results are consistent with our qRT-PCR data that SNr GABA neurons had higher expression of Na_V1.1 and Na_V1.6 than SN DA neurons, while SN DA neurons had higher expression of Na_V1.2 than SNr GABA neurons. These results were also consistent with literature data that heterologously expressed Na_V1.1 and Na_V1.6 conduct larger I_NaP than Na_V1.2 (Smith et al. 1998).

Functional implications of differences in Na_V currents in fast-spiking SNr GABA neurons and slow-spiking nigral DA neurons.

According to our present study, I_NaT has a higher density and a steeper voltage-dependent activation in SNr GABA neurons than in nigral DA neurons. These two properties can contribute to the faster rise rate and the larger amplitude of action potentials in SNr GABA neurons than in nigral DA neurons (Fig. 1, C–F). Equally important, compared with nigral DA neurons, I_NaT in SNr GABA neurons recovered more quickly from inactivation and resisted cumulative inactivation, thus enabling or supporting the sustained high-frequency firing in SNr GABA neurons. The large I_NaP in SNr GABA neurons that activates at or below −60 mV may drive the membrane potential toward the I_NaT activation threshold potential around −50 mV, triggering action potentials. Our laboratory has previously indicated that SNr GABA neurons have constitutively active type 3 canonical transient receptor potential cation channels that can depolarize SNr GABA neurons, even when they are at very negative membrane potentials (Zhou et al. 2008). On the hand, DA neurons have a prominent I_h cation current that can depolarize the neurons to approximately −65 mV, the threshold for I_NaP activation (Fig. 1B; Fig. 10A). SNr GABA neurons also have a robust expression of K_v3-like channels that can quickly repolarize the membrane after each action potential (Ding et al. 2011). Combination of differential expression of these nonselective cation channels, K_v channels, and Na_V channels can support the strikingly different spike waveforms and spiking patterns in SNr GABA neurons and nigral DA neurons.

GRANTS

This work was supported by National Institutes of Health grants R01NS058850 and R01DA021194 and an American Parkinson Disease Association grant to F.-M. Zhou.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: S.D. and F.-M.Z. conception and design of research; S.D. and W.W. performed experiments; S.D. and F.-M.Z. analyzed data; S.D. and F.-M.Z. interpreted results of experiments; S.D. and F.-M.Z. prepared figures; S.D. and F.-M.Z. drafted manuscript; S.D. and F.-M.Z. edited and revised manuscript; S.D., W.W., and F.-M.Z. approved final version of manuscript.