All experiments on animals were carried out in accordance with the approved animal care and use guidelines of the Laboratory Animal Research Center in Sungkyunkwan University School of Medicine and all experimental protocols were approved by the Laboratory Animal Research Center in Sungkyunkwan University School of Medicine (SKKUIACUC2018-05-10-5). For midbrain slices recording we used the STOCK Tg (TH-eGFP) DJ76Gsat/Mmnc line (NIH Mutant Mouse Regional Resource Centers), which was maintained as heterozygous mice by breeding with ICR (CrljOri: CD1) inbred mice [14]. Horizontal midbrain slices which contain the substantia nigra pars compacta (SNc) area, were prepared in postnatal day 21–28 transgenic mice expressing enhanced green fluorescent protein (eGFP) driven by the tyrosine hydroxylase (TH) promoter. For acutely dissociated DA neurons, postnatal day 9–12 Sprague Dawley rats used. The mouse and rat brains were removed rapidly after decapitation and then sliced into 300–400 µm thicknesses in ice-cold, oxygenated, artificial cerebrospinal fluid (ACSF in mM: 125 NaCl, 2.5 KCl, 10 glucose, 1.25 NaH₂PO₄, 25 NaHCO₃, 1 MgCl₂, 2 CaCl₂, pH 7.4) or ice-cold high-glucose solution (in mM: 135 NaCl, 5 KCl, 10 HEPES, 1 CaCl₂, 1 MgCl₂, 25 D-glucose, pH 7.3) using a vibratome (VT 1000S; Leica Microsystems, or Series 1000; Technical Products International). The acute mice brain slices were incubated in ACSF for 30 min at 33°C in dark chamber during the recovery period. However, for dissociated DA neurons, the SNc regions were cut out from the midbrain slices without recovery time, and then the DA neuron-containing tissues were incubated in oxygenated high-glucose solution with 8 U/ml papain (Worthington Biochemical Corporation) for 30 min at 36°C–37°C. After enzymatic digestion, the tissue was washed three times with 36°C high-glucose solution. After that, the dissociated DA neurons were obtained from the tissue with gentle serial agitation with Pasture pipettes of varying pore sizes. Some dissociated DA neurons were also obtained using the same dissociation protocol from postnatal day 9–12 Sprague Dawley rats [14]. The acutely dissociated DA neurons were attached to a poly-D-lysine (0.01%)-coated cover slip for 30 min at room temperature (20°C–25°C).

Two-photon confocal imaging was performed using a Zeiss LSM 510 Meta confocal/two-photon microscope system in the Research Core Facility of the Samsung Biomedical Research Institute (SBRI). Neuron morphologies were visualized by Alexa Fluor 594 (30 µM, Invitrogen) and eGFP with a tunable (690–1,020 nm) Ti::sapphire laser (Mai Tai; Spectra Physics). Intracellular Ca²⁺ signals were measured with a Ca²⁺ indicator, Oregon Green BAPTA-1 (OGB-1, 200 µM, Invitrogen) and 800 nm wavelength before and after removing AIS from the DA neurons (800 nm for GFP or OGB and 720 nm for axon laser cutting).

To remove an AIS from the DA neuron after visualizing the thin axon, the fluorescence dyes were fully loaded into recording neurons via low resistance recording electrodes (2–3.5 MΩ) in resting state for ≥ 15 min. The axon was identified by a characteristic axon retraction ball, a non-spiny membrane, and non-tapering diameters. The 720 nm wavelength of laser was used for axon laser cutting with the line scanning mode or minimized region of interest (ROI) on target regions. The origination site of an axon close to the dendrite, where is potentially including AIS, was exposed gradually increased laser power. Laser power was increased from 5% to 50% until the axon was cut from the dendrite. Axons were usually cut at around 40% of the laser output power, and then a sudden increase in spontaneous firing frequency occurred with structural changes including bleb formation. To allow the neurons to stabilize and recover after axon removal, we injected a small amount of negative currents to keep the membrane potentials in a resting state for 15–20 min.

To stimulate glutamate receptors or GABA receptors on the soma or dendrites, 100 µM 4-methoxy-7-nitroindolinyl-caged L-glutamate (MNI-glutamate) (Tocris Bioscience) with 10 µM glycine or 100 µM γ-Aminobutyric Acid, α-Carboxy-2-Nitrobenzyl Ester, Trifluoroacetic Acid Salt (O-(CNB-Caged) GABA) (Invitrogen) was added to the bath solution. Since caged-compounds are very sensitive of lights, whole procedures were avoided light exposure and we made fresh working solutions for every experiment to keep high efficiency of caged-compounds. In the local caged Ca²⁺ photolysis, we used an internal solution with 50 µM o-nitrophenyl EGTA (NP-EGTA) (Tocris Bioscience). Photolysis was performed using a Zeiss 510 confocal microscope with a UV laser (wavelengths 351 and 364 nm) and an oil-immersion objective lens (40× with a NA of 1.3, Olympus). Detailed procedures were described previously [14].

For cytosolic Ca²⁺ measurements in tissue slices, DA neurons were loaded with Alexa-594 (red dye, 30 µM, Invitrogen) and Oregon Green Bapta-1 (green dye, OGB-1, 200 µM, Invitrogen) together. Optical signals were acquired at 800 nm of two-photon excitation beam to simultaneously excite both dyes. ROI images were acquired during frame scanning (512 × 512 pixels) with 10–20 ms time intervals and Ca²⁺ levels from the ROIs were quantified as changes in green Ca²⁺ fluorescence from OGB-1 divided by morphological red fluorescence of Alexa-594 (G/R). Ratio between OGB-1 and Alexa-594 was used to minimize interference of the fluorescence photobleaching.

Acutely dissociated DA neurons were incubated with 3–5 µM Fluo 4-AM in high-glucose solution at room temperature (20°C–25°C) for 30 min. The fluorescence intensities of the neurons were measured using a Zeiss 510 confocal microscope (40× oil immersion objective lens or 60× water immersion objective lens). Fluo 4-AM Ca²⁺ indicators were excited at 488 nm (argon laser) and cytosolic Ca²⁺ signals were collected through 550 nm long-pass filter. Ca²⁺ level changes represented delta fluorescence intensity devided by basal level of fluorescence (ΔF/F₀). To measure cytosolic Ca²⁺ concentration in some cases, we used a calibration kit (Calcium Calibration Buffer Kit #1; Invitrogen).

We used the patch clamp systems (EPC-9 with Patchmaster 2.73; HEKA Elektronik) to measure the electrical activity of DA neurons in single or two-photon confocal microscopes. Low-resistance (2.0–3.5 M_Ω) patch electrodes were prepared from borosilicate glass capillaries with a 1.5 mm outer diameter (World Precision Instruments) using a Narishige puller (MODEL PP-830). Spontaneous firing was recorded in the current-clamp mode at a sampling rate of 10 kHz filtered at 1 kHz under the cell-attached, nystatin-perforated, or whole-cell patch-clamp configurations. For nystatin-perforated patch-clamp recordings the patch pipettes were filled with the solution containing (in mM) 140 K-gluconate, 5 KCl, 10 HEPES, 5 MgCl₂ and Nystatin (200 µg/ml, MP biomedical), and pH 7.2 was adjusted with KOH. For whole-cell patch-clamp recordings, the patch pipettes were filled with (in mM) 130 K-gluconate, 1 KCl, 10 HEPES, 2 Mg-ATP, 0.3 Na-GTP, 4 Na-Vit C, 10 Na-phosphocreatine, at pH 7.2 adjusted with KOH. The normal bath solution for acutely dissociated DA neurons contained (in mM) 140 NaCl, 5 KCl, 10 HEPES, 10 D-glucose, 1 CaCl₂, and 1 MgCl₂, at pH 7.4 adjusted with NaOH.

The midbrain slices and the dissociated DA neurons were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) at 4°C for 20 min and 15 min, respectively. After that, the slices or dissociated neurons were washed in PBS four times for 15 min or three times for 10 min, respectively. Slices were incubated for 15 min in permeabilization buffer (PBS containing 2% normal goat serum [NGS] and 1% Triton X-100) at 4°C. After permeabilization of midbrain slices, we incubated slices in blocking buffer (PBS containing 2% NGS) for 20 min with gentle shaking. For permeabilization of the dissociated neurons, we used PBS containing 1% NGS and 0.2% Triton X-100. After 10 min at 4°C incubation, neurons were rinsed in PBS and blocked for 20 min in blocking buffer (PBS containing 1% NGS). For TH and AIS staining, we used mouse anti-TH antibody (1:500 or 1:300, Abcam Inc.) and rabbit anti-ankyrin G antibody (1:300 or 1:200, Abcam Inc.). For primary antibody reactions in brain slices and dissociated neurons, they were incubated overnight in the primary antibodies in permiabilization buffer at 4°C. Primary antibody-loaded brain slices and neurons were rinsed four times with blocking buffer and incubated in blocking buffer with the secondary antibodies: Alexa 488-conjugated goat anti-mouse IgG (1:100, Invitrogen) and Alexa 647-conjugated goat anti-rabbit IgG (1:100, Invitrogen). After incubation for 2 h at room temperature fluorescence images were obtained using a confocal or two photon laser microscope.

Statistical analyses were performed with Origin 8.1 (Origin Lab Corporation). All presented numeric values and graphic representations represent mean ± standard error of the mean, and statistical analyses used one-way ANOVA tests followed by Tukey’s multiple-comparison posttest. p-values < 0.05 were regarded as significant.

In midbrain DA neurons, both the soma and multiple dendrites participate in pacemaking [9,14]. But, it is not clear whether the AIS also contributes to pacemaking. Therefore, we examined what happens to the spontaneous firing of DA neurons after removal of an axon with the two-photon laser-cutting technique in mouse midbrain slices. DA neurons were identified with GFP driven by the promoters of TH [27]. A DA neuron in the SNc extended several long primary dendrites in various directions (left image, Fig. 1A) and a non-tapering thin axon was identified with a terminal retraction ball formed during slice preparation (middle upper image, Fig. 1A), as previously described [8,12]. After spontaneous firing was stabilized in the midbrain slices (right, cell-attached recording, Fig. 1A), the axon was cut using strong two-photon laser illumination (720 nm) at the origination site near the dendrite (the red line in the middle upper image of Fig. 1A). A scar was left as a bleb in this case (middle lower image, Fig. 1A), but very often the neuron ruptured and became invisible because of dye leakage. In this typical surviving DA neuron, the spontaneous firing rate temporally increased as soon as the axon was detached from the primary dendrite, but the resting spontaneous firing rate was restored after 10–20 min (Fig. 1A). After restabilization, spontaneous firing rate did not differ significantly from that before the axon-cutting (spontaneous firing rate _{before axon-cut} = 3.45 ± 0.01 Hz, spontaneous firing rate _{after axon-cut} = 3.54 ± 0.01 Hz, p = 0.356, n = 3, Fig. 1B), suggesting that the AIS does not significantly contribute to pacemaking. Nevertheless, after AIS removal, the variance in the interspike interval (ISI) did statistically increase (variance _{before axon-cut} = 0.0034 ± 0.00048, variance _{after axon-cut} = 0.0256 ± 0.0065, p < 0.05, n = 3, Fig. 1C), suggesting that the AIS lowers the irregularity of spontaneous firing.

In multipolar DA neurons, it is well known that sAP always initiates from ABD and then propagates to the soma and other dendritic trees [7,8]. So, after removing axons, we wonder if sAP still initiates from the same ABD. First, we recorded spontaneous firing at the soma with a whole-cell patch-pipette and simultaneously compared corresponding Ca²⁺ spikes between the ABD and opposite nABD to detect the sAP initiation site [9,11,28]. After sAP onset (P1, blue lines and circles), Ca²⁺ spikes always started immediately from the ABD and later on the contralateral nABD (P2, red lines and circles) (Fig. 1D–F), as previously reported [7,8]. But, interestingly, after AIS removal with a two-photon laser cut (the red line in Fig. 1D), Ca²⁺ spikes from the ABD did not always start first (time-lag _{before axon-cut} = 0.52 ± 0.10 ms, time-lag _{after axon-cut} = 5.36 ± 2.11 ms, p < 0.05, n = 15 and 39 spikes from 3 neurons). Sometimes, Ca²⁺ spikes in the opposite nABD occurred before those in the previous ABD (Fig. 1H–J). In addition, after AIS removal, some paired Ca²⁺ spikes measured in the two dendrites (P1 and P2) did not match exactly with sAP onset times, with concurrent delays in both spikes (Ca²⁺ spikes 1, 2, 4, 6, and 8 in Fig. 1I), indicating that some sAPs developed in another primary dendrite. These data suggest that, if the AIS is removed, sAPs could be generated variably from many primary dendrites, including the previous ABD.

Regarding this aspect, it is interesting to see the changes in variances of the time-lags of dendritic Ca²⁺ spikes against sAPs which were measured with patch pipettes at the soma (Fig. 1F). The variances in the time-lags between the Ca²⁺ spikes in ABDs and the sAPs were much smaller than those in the nABDs (Fig. 1G), but after AIS removal, the variance in the ABDs increased to the level of the nABDs (p < 0.005, n = 3). All these data indicate that the slowly depolarizing pacemaker potentials of multiple primary dendrites are not equal and strictly coupled to each other. The pacemaker potentials of each primary dendrite appear to develop with some variation, but one of them reaches the AP threshold more quickly than the soma (see diagram in Fig. 1A). Thus, AIS appears to lower the ISI variability in the fluctuating ABD by triggering earlier sAPs therein. Consequently, AIS removal appears to remove the dominance of the ABD over the nABDs, thereby revealing the presence of variable sAP initiation sites scattered among many primary dendrites (Fig. 1H–J).

Acutely dissociated DA neurons attached with multiple primary dendrites from the rat SNc are a good model for studying pacemaker mechanisms, due to the large cell bodies attached with long dendrites compared with those of mice, preservation of intrinsic regular firing properties, easy spatial Ca²⁺ measurements from the dendrites on cover-glasses, and clear environmental conditions [14,28,29]. In the dissociated DA neurons, we rarely found axon-like structures, unlike the DA neurons in midbrain slices (Fig. 2A). They appeared to be lost during the isolation procedures. When we stained AIS in these neurons with ankyrin G, a marker for AIS, axon was not found in more than half of the cells, and even in stained cells, significantly reduced in length (Fig. 2B–D). Consequently, as seen below, the AIS in dissociated DA neurons seems to be disabled or non-functional. Therefore, we are able to investigate the endogenous dendritic properties of pacemaking without any interference from the AIS in dissociated DA neurons.

In one dissociated DA neuron of the rat to that three primary dendrites were attached, we measured Ca²⁺ spikes along one long primary dendrite, together with sAPs using a patch-pipette at the soma (Fig. 3A). In this typical DA neuron, we found two different types of Ca²⁺ spike propagations: in one case, Ca²⁺ spikes started from the proximal dendritic region and propagated bidirectionally to the soma and the distal dendritic region (Fig. 3B). In the other case, Ca²⁺ spikes occurred unidirectionally from the soma to the proximal and then distal dendritic regions (Fig. 3C). When we analyzed the time-lags of the Ca²⁺ spikes between the soma and variable sites along a dendrite in this neuron, within the proximal dendritic regions < 80 µm from the soma, the dendritic Ca²⁺ spikes preceded the somatic Ca²⁺ spikes in 13 out of 24 sAPs (black open circles in Fig. 3D) and the reversed occurred in 11 out of 24 sAPs (red open circles in Fig. 3D). However, in the distal dendritic regions > 80 µm from the soma, no dendritic Ca²⁺ spike preceded the somatic Ca²⁺ spikes at all (gray circles in Fig. 3D). We obtained similar results from six different neurons. These data, together with the previous results, suggest that in AIS-disabled DA neurons, sAPs begin variably from one of the primary proximal dendritic regions < 80 µm from the soma and then propagate to other parts of the neuron. When the somatic Ca²⁺ spikes preceded both the proximal and distal dendritic Ca²⁺ spikes (Fig. 3C), the sAPs could probably be originated from one of the unmeasured PDCs. These results are compatible with our previous report [14] that PDCs < 80 µm from the soma are important for pacemaking and more excitable than the distal dendritic compartments. To examine this point further, in another dissociated DA neuron attached with four primary dendrites (Fig. 3E), we simultaneously measured Ca²⁺ spikes at opposite sites of two primary dendrites equally distanced from the central soma (~40.2 µm), together with the center of the soma (P1 as red dots, P2 as green dots, and the soma as black dots). In this case, there were variable combinations of Ca²⁺ spike initiation site orders among them (Fig. 3E–G). As expected, the initiation of the first Ca²⁺ spikes was observed evenly between the two contralateral primary dendrites (Fig. 3E, G), indicating that both primary dendrites could generate their own sAPs. However, most notably and importantly, in many cases (37.1% of 300 Ca²⁺ spikes in 13 cells, Fig. 3G) both the contralateral primary dendrites developed their own Ca²⁺ spikes before the soma initiated any Ca²⁺ spikes (Ca²⁺ spikes 2, 3, 4, 6, and 11 in Fig. 3E and the first trace in Fig. 3F), indicating that these two dendrites are able to generate their own sAPs independently. We also found some cases in which the somatic Ca²⁺ spikes preceded the two dendritic Ca²⁺ spikes (Fig. 3E–G), possibly due to the propagation from initiation sites of other primary dendrites elsewhere. Taken together, these data suggest that multiple primary PDCs in DA neurons drive pacemaking in DA neurons (Fig. 3H).

In spontaneously and regularly firing dissociated DA neurons, the intracellular Ca²⁺ oscillations synchronized with the pacemaking cycle reflect depolarization-induced Ca²⁺ influxes [10,28]. Also, small-conductance Ca²⁺-activated K⁺ channels (SK) are abundantly expressed in DA neurons [30-33]. Therefore, the amount of Ca²⁺ influx is not only a good indicator of the depolarizing power of pacemaker activities, but also an important regulator of the pacemaking cycle [9,10]. Thus, we analyzed Ca²⁺ oscillations or spikes in regularly firing dissociated DA neurons (Fig. 4). When we measured intracellular Ca²⁺ spikes synchronized with sAPs in the soma and PDCs (Fig. 4A), the amplitude of Ca²⁺ spikes in the PDCs was higher and more variable ([Ca²⁺]_dendrites = 0.118 ± 0.007 ∆F/F₀, SD = 0.0448, n = 36) than that of the somatic Ca²⁺ spikes ([Ca²⁺]_soma = 0.051 ± 0.005 ∆F/F₀, SD = 0.0302, n = 31). Consequently, the areas under the curves (AUCs) of Ca²⁺ spikes in the PDCs were significantly larger than those in the soma (Fig. 4B), and the variations in the AUCs of Ca²⁺ spikes in the PDCs were also higher than those in the soma (CV_soma = 0.427 ± 0.012, CV_dendrites = 0.472 ± 0.016, p < 0.05, n = 21, Fig. 4C). It is also noteworthy that during pacemaking cycles, the lowest levels of Ca²⁺ oscillations or spikes in the PDCs were significantly higher than those in somatic Ca²⁺ concentrations ([Ca²⁺]_dendrites = 93.40 ± 0.94 µM, [Ca²⁺]_soma = 88.06 ± 1.04 µM, p < 0.001, n = 6, Fig. 4D). When spontaneous firing was blocked by TTX application, the Ca²⁺ level in the PDCs dropped to the same level in the soma ([Ca²⁺]_dendrites = 76.81 ± 0.41 µM, [Ca²⁺]_soma = 77.24 ± 0.88 µM, p = 0.667, n = 6), indicating that Ca²⁺ oscillations in the PDCs during pacemaking cycles are higher and more variable than those in the soma. All these data strongly indicate the higher excitability of the PDCs compared with the soma, and they are in line with our previous conclusion that multiple PDCs dominantly drive pacemaking in multipolar DA neurons. Therefore, if many primary PDCs in multipolar DA neurons generate variable and stochastically-fluctuating Ca²⁺ influxes by their own higher excitability, they could be responsible for minor firing irregularities observed in normal pacemaking cycles (Fig. 1A, 4A). Compatible with this idea, isradipine, which inhibits the Ca_v1.3 channel, the major voltage-dependent Ca²⁺ channel in DA neurons [11], decreased the ISI variance (73.2 ± 14.3%, p < 0.001, n = 7) by reducing variable Ca²⁺ influxes (Fig. 4E).

If the soma and PDCs are basically different in intrinsic excitabilities (Fig. 4F), the firing responses of these two different compartments to external stimuli should be different. Therefore, we stimulated the soma and PDCs using the local glutamate uncaging technique by continuously increasing the uncaging area in these two compartments, respectively (Fig. 5A). The sequential increments of the uncaging area gradually increased the maximum frequencies of the evoked firings in both the soma and PDCs. But, as expected, the slope of firing frequency increase was higher than that of the soma (r2_{prox dendrite} = 0.91, p < 0.001; r2_soma = 0.83, n = 7, p < 0.01, Fig. 5B), indicating that the PDCs were more sensitive than the soma. In addition, when we performed local GABA uncaging on the soma and PDCs in spontaneously firing DA neurons using the same uncaging area (78.4 ± 0.82 µm²) (Fig. 5C, D), the firing pauses were induced immediately as soon as PDCs were exposed to GABA, whereas the pauses were induced only after some delay in the soma (latency_soma = 2.55 ± 0.28 s, latency_dendrite = 0.61 ± 0.29 s, p < 0.05, n = 6, Fig. 5E), indicating that the PDCs respond more quickly than the soma. All these data are consistent with the previous conclusions that the primary PDCs in DA neurons are a stochastically-fluctuating, highly-excitable energetic compartment, but the large soma is a slowly and stably-oscillating inertial compartment.

Next we examined how the large regularly-oscillating soma is coupled with the small, stochastically-fluctuating, and highly-excitable PDCs in DA neurons. Because cellular Ca²⁺ removal depends on the SVR of compartments, the high-SVR PDCs can remove elevated Ca²⁺ faster than the low-SVR soma [9,34]. Fig. 6A illustrates the case in which Ca²⁺ removal was dramatically faster in the PDCs than in the soma when we instantly raised free Ca²⁺ levels in the soma or PDCs using local Ca²⁺ uncaging (NP-EGTA) in dissociated DA neurons (half-maximal decay-time_dendrites = 0.59 ± 0.04 s, half-maximal decay-time_soma = 2.24 ± 0.13 s, p < 0.001, n = 7, Fig. 6B, C). Despite the significant difference in Ca²⁺ removal times between these two physically distinct compartments (Fig. 6C), Ca²⁺ spikes of the PDCs in pacemaking DA neurons were well synchronized with those in the soma (Fig. 4A, D), indicating that the stochastically-fluctuating PDCs are tightly coupled with the large regularly-oscillating soma electrically, as previously reported [9]. The tight coupling of membrane potentials appears to cause changes in Ca²⁺ influxes/effluxes in the coupled PDCs, resulting in synchronized Ca²⁺ oscillations.

Therefore, we examined how strongly the PDCs are coupled with the soma in regularly firing DA neurons by observing the correlations in Ca²⁺ oscillations between them. When we analyzed the peaks of Ca²⁺ oscillations between the soma and different sites of a dendrite (Fig. 6D), the coefficients of variance (CV) increased according to the distance from the soma (Fig. 6E), and the correlation coefficients decreased (Fig. 6F). Thus, the coupling of Ca²⁺ oscillations between the regularly oscillating soma and irregularly fluctuating PDCs is distance-dependent. When we examined Ca²⁺ removal in the soma and PDCs during normal pacemaking cycles (Fig. 6G), we found no significant difference between them in the half-maximal decay times of Ca²⁺ spikes (dendrites = 0.09 ± 0.01 s, soma = 0.11 ± 0.01 s, n = 9, p = 0.53, Fig. 6H, I), despite the different amplitudes in their Ca²⁺ spikes (amplitude_soma = 0.12 ± 0.009 ∆F/F₀, amplitude_{dendritic compartment} = 0.24 ± 0.011 ∆F/F₀, p < 0.05, n = 9). It may suggest that the tight electrical coupling of PDCs with the large soma might restrict the vulnerable and fluctuating Ca²⁺ oscillations of PDCs, thereby subjugating dynamic dendritic Ca²⁺ changes during slow pacemaking cycles. Taken together, all of these data strongly suggest that the dynamic interaction between the large, stably-oscillating, inertial soma and the multiple, stochastically-fluctuating, energetic PDCs creates the slow pacemaker activity of DA neurons.

Since the dynamic interaction between two functionally different compartments produces regular spontaneous firing, this interaction model may underlie or affect other types of firing, such as evoked firing, other than slow pacemaking. Therefore, we next examined how DA neurons respond to fast external stimuli. Midbrain DA neurons can produce burst firing in respond to sufficient excitatory glutamatergic synaptic inputs [17,35]. It is well known that glutamatergic action on DA neurons is dualistic [36,37]: glutamate can activate several inotropic glutamate receptors that depolarize the membrane potential, generating burst discharges (fast excitatory components), but glutamate can also evoke a following Ca²⁺ transient due to the Ca²⁺ influxes and Ca²⁺ release from the endoplasmic reticulum store, consequentially suppressing spontaneous firing via activation of SK channels (slow inhibitory components) (Fig. 7A) [28,37,38]. Thus, a brief strong stimulation with glutamate mostly evokes two distinct firing responses in DA neurons: the initial burst discharge and subsequent postfiring pause (Fig. 7A). To see how glutamate works in the dynamic interaction model, we stimulated a different part of a dendrite in a dissociated DA neuron using the local glutamate uncaging technique (the numbered black circles in Fig. 7B). Usually, responsibility of DA neurons to local glutamate uncaging was strong in the proximal dendritic region and disappeared in the distal dendritic regions [14,26]. Accordingly, within the PDCs the typical burst-pause firing pattern was clearly observed (Fig. 7C). Interestingly, we found that the burst firing (the excitatory component, marked as red in Fig. 7C, D) was well preserved within the PDCs, but that the burst firing abruptly declined around 80 µm from the soma, and no more responses in the distal dendritic regions, indicating the higher excitability of the PDCs than distal dendritic compartments. This result is compatible with our previous report [14,26] that PDCs < 80 µm from the soma in rat DA neurons are more excitable than the distal dendritic compartments without a significant difference in glutamate receptor expressions. However, the postfiring pause (the inhibitory component, marked as blue in Fig. 7C, D) decreased exponentially with the distance from the soma in the same PDCs. While the excitatory response was well-preserved within the PDCs, the inhibitory response in the same region was exponentially decaying like a passive electrotonic dissipation of local membrane potentials. There was a clear dissociation between the excitatory and inhibitory responses to glutamate.

To investigate why the inhibitory response decreases with the distance from the soma within the PDCs, we performed local stimulation experiments on the soma and PDCs. Because Ca²⁺ is the main cause for glutamate-induced firing pauses [36,37], local glutamate uncaging on different sites of a dendrite might evoke different levels of Ca²⁺ rise. But it was not the case. Although the same local glutamate uncaging on different sites of a dendrite evoked similar Ca²⁺ spikes (Fig. 8A), the duration of the postfiring pause decreased with the distance from the soma (Fig. 8A, C), whereas the excitatory responses of firing to glutamate (the maximum frequencies of the evoked firings) were preserved within PDCs (Fig. 8A, B). When SK channels were suppressed by a specific blocker apamin (100 µM), or intracellular Ca²⁺ was buffered with a high concentration of BAPTA (20 µM), the postfiring pause was completely abolished (control = 2.44 ± 0.72 s, n = 6; apamin = 0.57 ± 0.03 s, n = 6, p < 0.05; BAPTA = 0.57 ± 0.01 s, n = 3, p < 0.05, Fig. 8D), indicating that the firing pauses are entirely mediated by intracellular Ca²⁺ rises. However, the excitatory response was not significantly affected (control = 13.01 ± 0.97 Hz, n = 9; apamin = 12.88 ± 1.55 Hz, n = 6, p = 0.997; BAPTA = 15.03 ± 2.29 Hz, n = 3, p = 0.649, Fig. 8E). Therefore, to understand why local Ca²⁺ rises at a different part of PDCs differently suppress spontaneous firing, we directly performed local Ca²⁺ uncaging (NP-EGTA) in a dendrite (Fig. 9A-C). When local Ca²⁺ spikes were serially evoked along a dendrite (Fig. 9A), the duration of the postfiring pause decreased exponentially with the distance from the soma (Fig. 9A, C), which is very similar to the result with glutamate uncaging (Fig. 7C-D). Interestingly, according to the distance from the soma we found an exponential decrease of the hyperpolarization of the soma (Fig. 9A, B). In addition, when higher Ca²⁺ spikes were evoked with stronger Ca²⁺ uncaging on the same site of a dendrite, stronger hyperpolarization of the soma occurred, together with correspondingly longer suppression of spontaneous firing (Fig. 9D). The magnitudes of the Ca²⁺ rises correlated with the degree of hyperpolarization in the soma (r = 0.76, n = 5, Fig. 9E). All these data strongly suggest that the extent of hyperpolarization of the central large soma determines the duration of the postfiring pause. Thus, the closer the dendritic stimulation site is to the soma, the larger its impact on the soma (Fig. 9A–C). Unless the large, centrally-connected soma is affected, stimulation of a small part of the dendritic regions cannot suppress the spontaneous firing of multipolar DA neurons (Fig. 9A). Therefore, we can now conclude that the key factor determining the inhibitory responses to glutamate in multipolar DA neurons is the hyperpolarization of the large inertial soma.

It seems now clear that the functionally distinct soma and PDCs are independently responsible for the two different components of burst-pause firing to sudden glutamate stimulation in DA neurons. In actual conditions, if the summed effects of glutamatergic synaptic activities throughout all the dendrites are sufficient to affect one of the highly-excitable PDCs, high-frequency burst firing would occur in that PDC and propagate rapidly into the other somatodendritic trees, causing sequential Ca²⁺ elevations in the propagated areas and then activating regional SK channels. If the large inertial soma is sufficiently hyperpolarized, the postfiring pause would then occur. To confirm this hypothesis, we strongly stimulated one region of a dendrite in a spontaneously firing DA neuron and observed both the spontaneous firing patterns and Ca²⁺ changes in the soma and stimulated dendritic region simultaneously. Upon glutamate uncaging (the green rectangular area in Fig. 9F), the Ca²⁺ level (green line) went up abruptly in the uncaged dendritic region with an immediate burst discharge. However, somatic Ca²⁺ (blue line) increased relatively slowly, together with a mirrored change in the hyperpolarization of the soma. Consequently, after the immediate burst firing (the excitatory component), the postfiring pause (the inhibitory component) followed with a slow somatic Ca²⁺ rise. The dendritic Ca²⁺ level, which increased immediately with the glutamate uncaging, was restored to its previous level more quickly than the somatic Ca²⁺ level. Notably, the reappearance of spontaneous firing depended on the recovery of the somatic Ca²⁺ levels, but not dendritic Ca²⁺ levels. When the somatic Ca²⁺ levels dropped by 60.07 ± 7.35% (n = 7), spontaneous firing reappeared. In that condition, the increased dendritic Ca²⁺ levels had already returned to their unstimulated level (Fig. 9G). Thus, it is clear that the large inertial soma is responsible for the inhibitory components of firing to glutamate, whereas the PDCs are responsible for the excitatory firing responses to glutamate. This complementary interaction between the two compartments appears to determine the final firing output patterns in multipolar pacemaker DA neurons (Fig. 9H).