TRPC3 and NALCN channels drive pacemaking in substantia nigra dopaminergic neurons

Midbrain dopamine (DA) neurons are slow pacemakers that maintain extracellular DA levels. During the interspike intervals, subthreshold slow depolarization underlies autonomous pacemaking and determines its rate. However, the ion channels that determine slow depolarization are unknown. Here we show that TRPC3 and NALCN channels together form sustained inward currents responsible for the slow depolarization of nigral DA neurons. Specific TRPC3 channel blockade completely blocked DA neuron pacemaking, but the pacemaking activity in TRPC3 knock-out (KO) mice was perfectly normal, suggesting the presence of compensating ion channels. Blocking NALCN channels abolished pacemaking in both TRPC3 KO and wild-type mice. The NALCN current and mRNA and protein expression are increased in TRPC3 KO mice, indicating that NALCN compensates for TRPC3 currents. In normal conditions, TRPC3 and NALCN contribute equally to slow depolarization. Therefore, we conclude that TRPC3 and NALCN are two major leak channels that drive robust pacemaking in nigral DA neurons.


Introduction
Dopamine (DA) neurons in the substantia nigra pars compacta (SNc) are essential for controlling the motivational parts of brain functions such as voluntary movement, action selection, future movement, and reward-based learning (Grace and Bunney, 1984;Schultz, 2007;da Silva et al., 2018). Dysfunction of DA neurons is associated with many neuropsychiatric diseases including Parkinson's disease and schizophrenia (Bozzi and Borrelli, 2006). Midbrain DA neurons are slow pacemakers that continuously generate spontaneous action potentials, consequentially sustaining ambient DA levels in target areas including the striatum (Grace et al., 2007;Morikawa and Paladini, 2011). However, in response to unexpected reward stimuli, DA neurons also produce high-frequency burst discharges that accompany DA surges (Schultz, 1998(Schultz, , 2013. Thus, the intrinsic basal activity of pacemaking is the basis on which DA neurons operate. The rate and regularity of DA neuron pacemaking are robust against many kinds of pharmacological and molecular perturbations including most known specific blockers for classical candidate pacemaker ion channels, such as HCN and voltage-activated Ca 2+ channels (VACCs; Paladini et al., 2003;Kim et al.,2007;Guzman et al., 2009). The robustness of pacemaking suggests the necessity and biological importance of basal DA levels and DA signaling in the brain (Guzman et al., 2009;Surmeier et al., 2012). Therefore, understanding the pacemaking processes of DA neurons is of paramount importance not only for basic physiological knowledge but also for pathological mechanisms useful in therapeutic approaches.
The ion channels underlying autonomous pacemaking, therefore, have been the subject of investigation for more than three decades. The key feature of DA neuron pacemaking is the subthreshold slow depolarization during interspike intervals that In this study, we use TRPC3 KO mice, specific TRPC3 channel blockers (Kiyonaka et al.,2009;Schelifer et al., 2012), and NALCN channel blockers that we recently identified (Hahn et al., 2020) to report that TRPC3 and NALCN are two major leak channels essential for robust pacemaking in SNc DA neurons.

Blockade of either nonselective cation channels or TRPC3 channels stops pacemaking of SNc DA neurons
Endogenous firing activity of SNc DA neurons was measured by whole-cell patchclamp recording in midbrain slices of TH-eGFP transgenic mice in which DA neurons express eGFP (Figure 1A, top;Jang et al., 2015). The recorded DA neuron was visualized by Alexa-594 dye in patch pipette and/or presented as a 3-D reconstructed image ( Figure 1A and 1E). Most SNc DA neurons recorded from midbrain slices exhibited a very regular firing rhythm with an average firing rate of 3.28 ± 0.13 Hz (Figure 3-figure supplement 1, n = 17). DA neuron pacemaking can be characterized by slow depolarization during the interspike interval, which is further divided into three phases: I, the initial afterhyperpolarization due to the Ca 2+ overload and SK channel activation (Nedergaard et al., 1993); II, the middle, very long, steady, and slow depolarization; and III, the last accelerated depolarization ( Figure 1B). Phase II, occupying more than 3/4 of the interspike interval ( Figure 1C), most clearly denotes slow depolarization ( Figure 1B, green dotted line). Slow depolarization is absolutely necessary for pacemaking (Nedergaard et al., 1993;Grace and Onn, 1989;Ping and Shepard, 1996) and determines the pacemaking rate ( Figure 1D). A mild and linear 6 slope in the slow depolarization ( Figure 1B) implies the presence of small but continuous depolarizing currents during interspike intervals. HCN and VACCs have been at the center of debate in pacemaking mechanisms of DA neurons (Chan et al., 2007;Branch et al., 2014). Isradipine blocks dihydropyridine (DHP)-sensitive Ca 2+ channels including CaV1.3 channels, a dominant type of VACCs in DA neurons (Guzman et al., 2009), and ZD-7288 is an HCN channel blocker (Chan et al., 2007;Zolles et al., 2006). As previously reported (Kim et al., 2007;Guzman et al., 2009 indicating that our results from the midbrain slices are not related to either incomplete channel antagonism in brain slice experiments (Guzman et al., 2009) or perturbation of the intracellular milieu by whole-cell pipette solution (Zolles et al., 2006). Previously, using freshly dissociated SNc DA neurons we reported that nonselective cation channels (NSCCs) are essential for pacemaking (Kim et al., 2007). Consistent with this report, nonspecific TRPC channel blockers, such as SKF-96365 and 2-ABP, strongly inhibited the spontaneous firing of DA neurons in the midbrain slices (Figures S1C and S1D). TRPC channels constitute a large and functionally versatile family of nonselective cation channel proteins (Gees et al., 2010). Among many TRPC members, TRPC3 is known to be highly expressed in the brain (Sylvester et al., 2001;Clapham, 2003;Clapham et al., 2005) and a constitutively active ion channel (Dietrich et al., 2003;Zhou et al., 2008). Therefore, we tried to apply selective TRPC3 channel blocker pyr3 or pyr10 (Kyonaka et al., 2009;Schleifer et al., 2012) to DA neurons in the midbrain slices, and then observed the complete abolition of spontaneous firing (Figures 1E and 1F), together with the disappearance of dendritic Ca 2+ oscillations, which were measured by Fluo-4 in the whole-cell patch pipette ( Figures 1E and 1G).
Double immunofluorescence staining of the midbrain slices containing the SNc (Figures 1H and 1I) showed that TH-positive DA neurons were mostly overlapped with TRPC3-positive neurons (overlaps = 92%). In addition, single-cell RT-PCR from the isolated SNc DA neurons showed expression of TRPC3 mRNA in all cells examined ( Figure 1J). Taken together, these results raise the possibility that TRPC3 is a nonselective cation channel essential for the pacemaking of SNc DA neurons.

TRPC3 encodes a nonselective cation channel essential for the slow depolarization of SNc DA neurons during interspike intervals
In SNc DA neurons of midbrain slices, pyr10 eliminated slow depolarization and spontaneous firing completely (Figures 1E and 2A). Under this condition, the injection 8 of a small amount of continuous current, which resembles a leak current, revived spontaneous firing, at a rate that correlated with an injected current size (Figures 2A and 2B). When the firing rate was resuscitated to the control level, voltage traces of the regenerated slow depolarization and action potential were completely aligned with those before pyr10 treatment (Figure 2A, right bottom), indicating that the channel inhibited by pyr10 could be a leak-like channel. Consistent with these data, the firing rate was gradually increased within the pacemaking range by a slow ramp-like increase in current injection (Figure 2A,, implying that the amount of leak current determines the pacemaking rate in SNc DA neurons. Next, we examined whether the blockade of the TRPC3 channel affects subthreshold membrane potentials of SNc DA neurons. In spontaneously firing DA neurons, application of tetrodotoxin (TTX), a voltage-dependent Na + channel blocker, completely suppressed spontaneous action potentials but slow oscillatory potentials (SOPs) survived in a slightly more depolarized state ( Figures 2C and 2D), as previously reported (Nedergaard et al., 1993;Estep et al., 2016). These SOPs are not essential for pacemaking in DA neurons (Guzman et al., 2009), but are mediated by cyclic interactions between L-type Ca 2+ channels and Ca 2+ -dependent SK channels within an adequate range of membrane potentials (Ping and Shepard 1996;Wilson and Callaway, 2000). Consistent with these reports, further treatment with isradipine suppressed SOPs and lowered the membrane potential ( Figures 2C and 2E). In contrast, pyr10 not only suppressed SOPs but also hyperpolarized membrane potentials more strongly than isradipine ( Figures 2D and 2E), indicating that pyr10inhibited channels are able to depolarize the membrane potential even in the more hyperpolarized state. After pyr10 hyperpolarized the membrane potential, the total replacement of external Na + with equimolar large cation N-methyl-d-glucamine 9 (NMDG) further hyperpolarized the membrane potential ( Figure 2E), at a value similar to that predicted by the Goldman-Hodgkin-Katz (GHK) equation (VGHK = -72.57 mV; see Method details). All these data suggest that TRPC3 substantially contributes to slow depolarization of the membrane potential in DA neurons, but not alone.  Figures 3B and 3C). These data strongly indicate that the inhibitory action of pyr10 on DA neuron pacemaking in WT mice should be mediated by selective TRPC3 channel antagonism. In addition, unlike WT mice, SKF-96365, which blocks all members of TRPC channels (Nilius and Flockerzi, 2014;Hahn et al., 2020), did not affect the pacemaking rate of DA neurons in TRPC3 KO mice ( Figures   3B and 3C). The same results were obtained from acutely isolated SNc DA neurons ( Figure 3-figure supplement 2). Therefore, we next examined whether pyr10 affects the subthreshold membrane potentials of DA neurons in these mice. To measure the membrane potentials more exactly, we pretreated DA neurons with TTX and ZD-7288 to remove interferences from action potential-induced membrane potential fluctuations and hyperpolarization-activated currents by HCN channels, respectively. Under this condition, pyr10 hyperpolarized the membrane potential of DA neurons in WT mice as expected (Figrue 3d, left), but not in TRPC3 KO mice ( Figure 3D, right). When whole extracellular Na + was replaced by equimolar NMDG after treatment with pyr10, the membrane potential of DA neurons was maximally hyperpolarized ( Figures 3E and   3F). These data suggest that the sustained inward currents produced by TRPC3 channels must be compensated by other Na + -permeable ion channels in TRPC3 KO mice.

Selective
In TRPC3 KO mice, the input resistance of DA neurons did not differ from that in WT mice ( (Nilius and Flockerzi, 2014;Lievremont et al., 2005). In addition, ZD-7288 and isradipine had no significant effect on the spontaneous firing rate of DA neurons in TRPC3 KO mice (Figure 3-figure supplement 3A, B), suggesting that L-type Ca 2+ and HCN channels do not compensate for pacemaking in these mice. Therefore, it is highly likely that the leak-like inward current induced by TRPC3 channels appears to be completely compensated by SKF-96365-resistant and Na + -permeable ion channels in TRPC3 KO mice.
NALCN is another non-selective cation channel essential for the slow depolarization of SNc DA neurons.
Recently, the Na + leak channel NALCN has been reported to contribute to resting Na + permeability and basal excitability in neurons (Lu et al., 2007;Ren, 2011). However, the neonatal lethality of NALCN KO mice and lack of specific blockers (Lu et al., 2007;Shi et al., 2016;Lutas et al., 2016;Yeh et al., 2017) has hampered investigation of the potential role of NALCN in the pacemaking of DA neurons. Recently, we have found that N-benzhydryl quinuclidine (NBQN) compounds, including L-703,606, are a potent blocker for NALCN channels without affecting TRPC channels (Hahn et al., 2020). Therefore, we applied L-703,606 to SNc DA neurons in midbrain slices and observed the complete abolition of pacemaking ( Figure 4A). Under this condition, stepby-step linear current injections by whole-cell patch pipette to the neuron restored regular pacemaking activity (Figures 4B and 4D), which is similar to in TRPC3 channels (Figures 2A and B). When the firing rate was resuscitated to the control level, the voltage traces between the revived firings and those before L-703,606 treatment were completely aligned to each other ( Figure

DA neurons in TRPC3 KO mice.
NALCN is a G-protein coupled receptor channel that can be activated by substance P and neurotensin (NT; Lu et al., 2009;Hahn et al., 2020). Therefore, to examine more clearly whether NALCN currents are increased in TRPC3 KO mice, we measured NTevoked NALCN currents in freshly dissociated SNc DA neurons directly. Because of the rapid desensitization of NT-evoked NALCN currents in normal DA neurons (Hahn et al., 2020), we used a micropressure puff system (10 μM, 1 s). The NT-evoked NALCN currents in dissociated DA neurons from TRPC KO mice were significantly larger than those in WT mice (Figures 5A and 5B), demonstrating that the NALCN currents are increased in TRPC3 KO mice.
Next, we examined the expression of NALCN mRNA levels in both SNc tissues and dissociated single DA neurons from TRPC3 KO and WT mice using quantitative RT-PCR (qRT-PCR) ( Figures 5C and 5D). In both cases, the expression of NALCN mRNAs in TRPC3 KO mice was significantly higher than in WT mice. However, mRNAs of the other members of TRPC channels in TRPC3 KO mice were not changed ( Figure 5C). This is in line with our previous conclusion that NALCN channels compensate for the leak currents in TRPC3 KO mice rather than other members of TRPC. Furthermore, using western blot analysis, we confirmed that NALCN proteins were more significantly increased in heterozygous and homozygous TRPC3 KO mice than in WT littermates ( Figures 5E and 5F).

Equal contribution of TRPC3 and NALCN to subthreshold depolarization of membrane potentials in SNc DA neurons
Finally, we examined the relative contributions of the TRPC3 and NALCN channels to the subthreshold depolarization of membrane potentials in SNc DA neurons. Because Ironically, when both pyr10 and L-703,606 were co-applied to SNc DA neurons without ZD-7288, the hyperpolarization was similar to those induced by a single treatment with each blocker (Figures 6A and 6B). However, in the presence of ZD-7288, the coapplication of pyr10 and L-703,606 further hyperpolarized the membrane potential ( Figures 6A and 6C), suggesting that the blockade of these two channels together causes enough hyperpolarization to activate HCN channels in SNc DA neurons.
Therefore, HCN channels appear to activate in response to sufficiently significant hyperpolarization of the membrane potential in SNc DA neurons. These results are consistent with previous data showing that HCN channels contribute very little to normal pacemaking processes (Figure 1-figure supplement 2E, G). Analysis of the degrees of hyperpolarization induced by pyr10 and/or L-703,606 in the presence of ZD-7288 ( Figure 6D) suggests that the TRPC3 and NALCN channels contribute equally to more than two-thirds of the leak conductance responsible for slow depolarization in SNc DA neurons.

Discussion
The main finding from our combined electrical, molecular, immunohistochemical, and pharmacological experiments is that TRPC3 and NALCN channels are two major leak channels essential for the robust pacemaking of SNc DA neurons. Subthreshold slow depolarization during the interspike interval drives pacemaking of SNc DA neurons that basically generates regular spontaneous firing at 2-6 Hz (Grace and Bunney, 1984). SNc DA neurons strongly express TRPC3 and NALCN channels, which constitute a sustained leak-like inward current responsible for slow depolarization in SNc DA neurons. In addition, knocking out TRPC3 channels increases the functionally identical NALCN channels in the subthreshold range of membrane potential and consequently preserves normal pacemaking activity, demonstrating the flexible adaptive nature of the pacemaking in SNc DA neurons. This slow depolarization driven by multiple leak channels provides the reason why DA neuron pacemaking is robust and resistant to many kinds of pharmacological and molecular perturbations (Guzman et al., 2009). This is the first report to uncover molecular identities and properties of leak channels responsible for the pacemaking in SNc DA neurons.
SNc DA neurons express a variety of K + , Na + , Ca 2+ , and NSCCs, and many ion channels and various factors contribute to pacemaking processes (Gantz et al., 2018).
The voltage-dependent L-type Ca 2+ channels and HCN channels have long been the subject of controversy as the major pacemaker channels in SNc DA neurons (Guzman et al., 2009;Surmeier et al., 2012). The easiest and best way to find the ion channels responsible for pacemaking is to use specific blockers to see if autonomous spontaneous firing is blocked. However, consistent with the latest report (Guzman et al., 2009), using both the brain slices and acutely dissociated DA neurons we came to the same conclusion that L-type Ca 2+ channels and HCN channels are not necessary for the pacemaking of SNc DA neurons. The HCN channels encoded by four genes (HCN1-4) are a hyperpolarization-activated nonselective cation channel and referred to as a pacemaker channel in many cells (Zolles et al., 2006;Wahl-Schott and Biel, 2009). They appear to have a wider range of activation and inactivation kinetics depending on the expression types, binding proteins, and many other factors such as several nucleotides and phosphatidylinositol-4,5-bisphosphate (PIP2; Zolles et al., 2006;Wahl-Schott and Biel, 2009). Although HCN channels appear to have little participation in the pacemaking of the SNc DN neurons, they can be activated in response to a sudden and/or significant hyperpolarization of membrane potentials (Gantz et al., 2018). Therefore, HCN channels may play a role in stabilizing pacemaking activities upon sudden hyperpolarizing challenges. Regarding VACCs, Cav1.3 channels are known to be expressed abundantly in SNc DA neurons (Gantz et al., 2018). The Cav1.3 channel containing α1D subunits can be characterized by low-voltage activation and slowly inactivating Ca 2+ currents, so it seems more ideal for slow depolarization of DA neurons than other members of the Cav1 family (Xu and Lipscombe, 2001). However, Cav1.3 channels give rise to intracellular Ca 2+ concentration ([Ca 2+ ]c) via Ca 2+ influxes and form a microdomain with Ca 2+ -activated SK channels in DA neurons (Wolfart et al. 2001). Therefore, SK channels have an opposite effect on Ca 2+ channel-mediated depolarization of the membrane potential (Wolfart et al., 2001). In addition, the continuous spontaneous firing of DA neurons itself causes continuous Ca 2+ influxes through many kinds of VACCs, thus the global [Ca 2+ ]c of DN neurons is always fluctuating and significantly higher than that in the TTX-treated silenced state (Choi et al., 2003). SK channels are abundantly expressed in SNc DA neurons (Wolfart et al., 2001) and can be already in a state of partial activation in such a [Ca 2+ ]c level of spontaneously firing DA neurons (Choi et al., 2003;Kim et al., 2007). It is, therefore, very likely that elevated [Ca 2+ ]c generates repolarizing outward K + currents, not only at microdomain levels but even at global levels. For this reason, Na + channels seem to be more important than Ca 2+ channels for the pacemaking of SNc DA neurons. In this aspect, the highly Na + -permeable NALCN and TRPC3 channels appear to be ideal for driving pacemaking in SNc DN neurons.
In the midbrain DA neurons, insensitivity or resistance of the pacemaking to the antagonism of many ion channels have led to the multichannel hypothesis that the pacemaking activity results from the sum of activities of many ion channels (Paladini et al., 2003;Chan et al., 2007;Guzman et al., 2009). In addition, the slow depolarization in DA neurons appears to depend on a leak-like conductance (Khaliq and Bean, 2010). Therefore, many simulation models for DA neuron pacemaking have assumed a considerably large background leak conductance (Guzman et al., 2009;Kuznetsova et al., 2010). In fact, DA neurons have a very low input resistance ranging from 30 to 300 M (Grace and Bunney, 1983;Branch et al., 2014) and large multipolar DA neurons receive several thousands of synaptic inputs throughout the somatodendritic compartment from network neurons (Schultz, 1998). Therefore, the DA neurons could be not only electrically very leaky but also geometrically unstable in terms of maintaining the homogenous membrane potential throughout the entire somatodendritic compartment (Jang et al., 2014). In such conditions, to generate and maintain a stable and steady rhythm of pacemaking, voltage-dependent or Ca 2+permeable channels may not be a good choice. Therefore, unlike the classical pacemaker cells, the slow depolarization of DA neurons appears to depend on multiple voltage-independent Na + -dependent leak channels, such as TRPC3 and NALCN channels. Although TRPC3 and NALCN channels belong to NSCCs, they primarily conduct Na + ions under physiological conditions (Lu et al., 2007;Cochet-Bissuel et al., 2014). The main characteristic of these two channels is their constitutive activation at very low membrane potentials and the linearity of the IV relationship in the subthreshold range of the membrane potential (Dietrich et al., 2003;Zhou et al., 2008;Lu et al., 2007). Nevertheless, because blocking one channel stops pacemaking completely, the pacemaking of SNc DA neurons depends on multiple channels in normal conditions. To analyze the relative contributions of TRPC3 and NALCN channels to the subthreshold depolarization, we measured the membrane potential shift from the K + equilibrium potential in the presence of TTX and ZD-7288. Given that blocking each of these channels leads to similar changes in the membrane potential shifts, TRPC3 and NALCN channels appear to equally contribute to slow depolarization in DA neurons. Na + substitution experiments show that these two channels are responsible for about 66.7% of the subthreshold leak current, but do not cover the entire leak current completely. Still, some other channels are engaged in slow depolarization. Of course, HCN channels can also contribute to depolarization under excessively hyperpolarized conditions. In fact, when these two main channels are blocked completely, HCN channels can participate in the depolarization of the membrane potential in the over-hyperpolarized conditions. Although TRPC3 channel blockers completely stop autonomous firing in DA neurons, the pacemaking of DA neurons in TRPC3 KO mice is perfectly normal, indicating compensation by a functionally identical leak channel. In the neural intrinsic excitability, loss of specific ion channels can often be compensated by other channels due to homeostatic mechanisms (Marder and Goaillard, 2006). Since SNc DA neurons express a variety of ion channels including VACCs, HCN, and TRPC channels (Chan et al., 2007;Branch et al., 2014;Tozzi et al., 2003), we initially suspected that Cav1.3 channels, HCN channels, and TRPC channels may compensate the pacemaking in TRPC3 KO mice. However, all these channels do not appear to participate in the compensation of pacemaking in TRPC3 KO mice. In general, since TRPC3, TRPC6, and TRPC7 have similar primary sequences and activation mechanisms and cluster closely on the phylogenic tree, TRPC6 and TRPC7 were prime candidates for compensation of the pacemaking in TRPC3 KO mice, However, although SKF96365 can block all members of TRPC channels, it did not have any effect on spontaneous firing in TRPC3 KO mice. Therefore, other members of the TRPC family do not appear to compensate the pacemaking in TRPC3 KO mice. Consistent with this result, RT-PCR experiments show that mRNAs of all TRPC channels are not increased in TRPC3 KO mice. On the other hand, the Na + leak channel NALCN is known to be widely expressed in the brain including DA neurons (Ren, 2011) and, therefore, has been suspected as a pacemaker channel in DA neurons (Surmeier et al., 2012;Philippart and Khaliq, 2018). Very recently, NBQN compounds were found to act as a NALCN 19 channel blocker (Hahn et al., 2020). They do not affect TRPC channels but can partially inhibit voltage-dependent Na + and Ca 2+ channels in high concentrations. By taking this advantage, we find that NALCN is not only another important channel for pacemaking in DA neurons, but also compensates for TRPC3 leak currents in TRPC3 KO mice. NALCN and TRPC3 channels have many binding proteins and upstream regulators, including neurotransmitters and hormones (Cochet-Bissuel et al., 2014;Amaral and Pozzo-Miller, 2007). Therefore, the mechanism of pacemaking regulation

Acutely dissociated dopamine neuron preparation
To obtain dissociated dopamine neurons, whole brains were quickly removed from 18-26 days postnatal mice and immersed in ice-cold oxygenated (100% O 2 gas) high glucose HEPES-buffered saline, which contains (in mM: 135 NaCl, 5 KCl, 10 HEPES, 1 CaCl2, 1 MgCl2, and 25 D-glucose, pH adjusted to 7.4 with NaOH). Horizontal midbrain slices of 300 μm thickness were obtained with a TPI vibratome 1000 tissue sectioning system (TPI, USA). SNc regions from the slices were dissected out with a scalpel blade and digested with oxygenated high glucose HEPES-buffered saline containing papain (5-8 U/ml, Worthington, USA) for 20-30 min at 37°C. Next, the slice segments were rinsed with enzyme-free HEPES-saline, and then tissues were gently agitated with varying sizes of fire-polished Pasteur pipettes. The agitated cells were gently attached to a poly-D-lysine (0.01%)-coated glass coverslip for 30 min at room temperature. Recordings were performed from 40 min to 3 hr after being attached.
For the application of agonists by pressure microinjection 'puff experiments', we used a IM 300 microinjector (Narishige). The injection glass pipette filled with the HEPESbuffered saline containing 10 μM neurotensin (Sigma-Aldrich, N6383) was positioned near proximal dendrites about >20 μm of target neurons to prevent the mechanical effect. The single-pulse duration was 1 s, during which time the pressure was 200-300 kPa.

Single-cell RT-PCR and qRT-PCR
Single dissociated neurons were aspirated into a microelectrode pipette with a sampling solution containing 10 mM dithiothreitol (DTT), 50 U/ml RNasin® RNase inhibitor (Promega, USA, N2611) in diethylene pyrocarbonate-treated water. After collecting sample cells, RNA was extracted using RNEasy kits (Qiagen, Germany, 744004). For single-cell RT or qRT-PCR, RNA was purified by ethanol precipitation procedures (57). Purified RNA was heated to 50°C for 30 min. cDNA was synthesized from cellular mRNA through the addition of SuperScript III Kits (Thermo Fisher, 11752050). The reaction mixture was incubated sequentially at 25°C for 10 min, 50°C for 30 min, and then heated to 85°C for 5 min. After reverse transcription, cDNA samples were chilled at 4°C. For single-cell qPCR, cDNA was purified according to a published procedure (57). After purification, a single cell cDNA sample was used as a template for conventional PCR amplification. Cycle conditions were as follows: 95°C for 15 s, 55°C for 30 s, 72°C 45 s. RT-PCR was performed using the following primers

Quantification and Statistical Analysis
All graphical illustrations were performed using CorelDraw ® 8 and 2019 software (Corel Corporation, USA). For drug applications, 20 sec in each condition (control or drug) was used for measuring the mean of firing frequency. All data collected were analyzed using Origin 7.0 software (Origin Lab Corporation, USA) and electrophysiological data were analyzed using Igor Pro 4.01 (Wavemetrics, USA). All numeric data are presented as mean ± standard error of the mean (S.E.M). Data were 28 summarized as box plots, with the centerline showing the median, the top and bottom of the box indicating the 25-75% range, and whisker representing the 5-95% range.