Differential control of dopaminergic excitability and locomotion by cholinergic inputs in mouse substatia nigra

Summary

Understanding how dopaminergic (DA) neurons of the substantia nigra pars compacta (SNc) govern movements requires a detailed knowledge of how different neurotransmitter systems modulate DA neuronal excitability. We report a heterogeneity of electrophysiological properties between medial and lateral SNc neurons modulated by cholinergic neurotransmission. Lateral DA neurons received mainly excitatory (nicotinic or glutamatergic) mediated cholinergic neurotransmission. Medial DA neurons received predominantly GABAergic currents mediated by presynaptic nicotinic receptors or biphasic GABAergic and nicotinic neurotransmission conveyed by GABA and ACh corelease, which inhibited DA neurons. To examine whether cholinergic signaling in the SNc controls mouse behavior, we used optogenetics in awake behaving mice and found that activation of cholinergic terminals in the medial SNc decreased locomotion, while activation in the lateral SNc increased locomotion. Our findings provide novel insights on how cholinergic inputs to subregions of the SNc regulate the excitability of DA neurons differentially, resulting in different patterns of motor behavior.

Introduction

Midbrain dopaminergic (DA) neurons play a key role in a wide range of behaviors, from motor control, motivation, reward and reinforcement learning [1–3]. Disorders of midbrain DA signaling leads to a variety of nervous system disorders including Parkinson’s disease (PD), schizophrenia and drug addiction. As an integral component of the basal ganglia, DA neurons of the substantia nigra pars compacta (SNc) are key neural substrates for initiating voluntary movement. The activity of SNc neurons are modulated by several neurotransmitter systems including γ-aminobutyric acid (GABA), glutamate and acetylcholine (ACh) [4–6]. The major cholinergic input into the SNc are from two brainstem nuclei: the pedunculopontine tegmental nucleus (PPT) and the laterodorsal tegmental nucleus (LDT) [7,8]. ACh released from these nuclei can activate two major classes of ACh receptors on DA neurons, muscarinic and nicotinic acetylcholine receptors (nAChRs) [9–11]. We have previously shown that SNc DA neurons contain one of the highest expression of nAChRs in the brain and are powerfully excited by nicotine [4]. However, the detailed modulation of DA neuronal excitability and neurotransmitter release from endogenous ACh release from the PPT and LDT onto midbrain DA neurons have not been investigated in detail. Evidence from others has indicated that there may be a heterogeneity of DA neurons expressed in the SNc [12]. If this is the case, do these heterogeneous subpopulations of DA SNc neurons mediate different cholinergic mediated responses?

Cholinergic neurons of the PPT and LDT both innervate the substantia nigra (SN) in addition to the ventral tegmental area (VTA) [7,8,13]. Both of these nuclei are actually quite heterogeneous since they are composed of not only cholinergic but also glutamatergic and GABAergic neurons [14]. Recordings from the PPT showed that PPT neuronal activities positively correlated to movements but interestingly there was a subpopulation of PPT neurons with neuronal activity negatively correlated to movement [15,16]. How can a cholinergic nucleus such as the PPT mediate a seemingly paradoxical effect of both stimulation or inhibition of locomotor activity?

Our recordings from DA SNc neurons have revealed a profound heterogeneity between medially and laterally located SNc neurons. We investigated whether stimulation of cholinergic terminals in these two distinct regions of the SNc may mediate different cholinergic neurotransmission modalities that would differ in modulating DA neuronal excitability. We also investigated how cholinergic neurotransmission in the medial and lateral SNc modifies locomotor behavior.

Results

DA neurons in the medial and lateral SNc display different biophysical properties

To examine the heterogeneity of DA neurons in spatially distinct regions of the SNc, we performed whole-cell patch-clamp recordings of DA neurons from two different regions of SNc – medial and lateral SNc. The medial lemniscus (ml) was used as the landmark separating VTA from SNc and the oculomotor nerve was the landmark that separated medial from lateral SNc (Fig. S1). DA neurons were characterized by slow tonic firing frequency (< 5 Hz), a broad action potential (> 2 ms) and a hyperpolarization-activated inward current (Ih) (Fig. S1A). With biocytin in the recording pipette, it was evident from the recordings that medial and lateral DA neurons were distinct based on morphological and electrophysiolocal properties. In the lateral SNc, DA neuronal somata were fusiform with resting membrane potentials of −54.1 ± 1.4 mV, large Ih currents (−1071 ± 190 pA), and average firing frequency of 3.2 ± 0.7 Hz (Table S1). Medial DA neurons were, multipolar with more hyperpolarized resting membrane potentials (−57.2 ± 0.5 mV) (p = 0.01, t(16) = 2.54), smaller Ih currents (−662 ± 137 pA) (p = 0.04, t(20) = 1.79), and lower firing frequency of 1.4 ± 0.3 Hz (p = 0.03, Wilcoxon rank sum test) (Table S1).

Stimulation of cholinergic terminals in the lateral SNc mediates mainly excitatory currents on DA neurons

We recorded from SNc DA neurons from mouse brain sections from ChATcre-ChR2 mice to examine how cholinergic neurotransmission modifies dopaminergic activity. In this mouse line ChR2 is expressed only in cholinergic neurons as verified with immunohistochemistry for ChAT (Fig. S2). Whole cell voltage-clamp recordings showed that stimulation of cholinergic terminals in the lateral SNc by brief pulses (5 ms) of blue light elicited robust excitatory postsynaptic currents (EPSCs) held at −70 mV. About 93% of neurons in the lateral SNc received excitatory nAChR mediated cholinergic neurotransmission (Table S2). These EPSCs included both monosynaptic (direct) nAChR or disynaptic (indirect) glutamatergic currents, which were both blocked by a cocktail of nAChR antagonists (DHβE, MLA, and MEC), while only the latter were blocked by a glutamatergic antagonist cocktail of CNQX and AP5 (Fig. 1).

Figure 1. Lateral SNc expressed mainly excitatory glutamatergic and nicotinic mediated cholinergic neurotransmission.

(A1, A2) Blue light activation (5 ms pulse duration, blue bars) of cholinergic terminals in the lateral SNc elicited nAChR currents that were insensitive to DNQX and AP5 but completely inhibited by the nAChR inhibitor cocktail of MEC, MLA and DHβE. (A3) Current traces of nAChR responses at different holding potentials (−100 to +40 mV) and the corresponding I–V relationship (A4). (A5) Supportive evidence that nAChR EPSCs were monosynaptic since 4-AP (100 M) was able to restore the blue light evoked response after abolishment with TTX (1 μM). (B1, B2) Disynaptic glutamatergic EPSCs mediated by presynaptic nAChRs as evidenced by inhibition with CNQX and AP5 and the cocktail of nAChR antagonists. (B3) Blue light evoked disynaptic glutamatergic responses at different holding potentials and I–V plot (B4) at the earlier (black circles) and latter (red triangles) time points displayed AMPA and NMDA currents, respectively. (B5) Supportive evidence that glutamatergic EPSCs were disynaptic since 4-AP was unable to restore the blue light evoked response after inhibition with TTX. (C) Injection of a floxed ChR2-YFP AAV into the PPT of a ChAT-cre mouse showing expression of ChR2-YFP in the PPT and fibers in the SNc. Blue light evoked nAChR EPSCs that were inhibited by a cocktail of nAChR antagonists. (D) Injection of AAV floxed ChR2-YFP into the LDT of ChAT-cre mice, showing expression of ChR2-YFP in the LDT and in the fibers in the SNc. Also shown are blue light evoked disynaptic glutamatergic EPSCs that were reversibly inhibited by DNQX and AP5. See also Figures S1–S3, and Tables S1 and S2.

Moreover, we confirmed the monosynaptic nicotinic and disynaptic glutamatergic EPSCs by recording blue light evoked currents at different holding potentials from −100 mV to +40 mV in order to analyze the I–V relations for these currents. For the glutamatergic currents, the I–V curves had two components separated in time. The earlier component had a linear I–V relationship that reversed near 0 mV, which was characteristic for AMPA currents (Fig. 1). The delayed component had a non-linear I–V relationship, which also reversed around 0 mV but rectified at hyperpolarized potentials, characteristic of an NMDA current (Fig. 1).

To confirm that the evoked nicotinic EPSCs were action potential dependent and that activation of ChR2 alone was insufficient for transmitter release, we applied 1 M tetrodotoxin (TTX) and completely inhibited the EPSC response. In the presence of TTX, when we applied the K+ channel blocker 4-aminopyridine (4-AP, 100 μM) to increase further the membrane depolarization from ChR2 activation, we observed a rescue of the response (Fig. 1A5), supporting that the evoked nicotinic EPSCs were monosynaptic [17]. However, for blue light evoked glutamatergic EPSCs 4-AP was unsuccessful in rescuing the complete inhibition of response with TTX (Fig. 1B5); this provides evidence that glutamatergic EPSCs are disynaptic.

In a select experiments, we injected adeno-associated virus (AAV) encoding a cre-dependent floxed ChR2-YFP into the LDT of ChAT-cre mice and three weeks later we recorded blue light evoked disynaptic glutamatergic EPSCs from a lateral SNc DA neuron (Fig. 1D). In one ChAT-cre mouse injected with AAV floxed ChR2-YFP into the PPT we recorded blue light evoked nAChR EPSCs from a lateral SNc DA neuron (Fig. 1C).

In addition to cholinergic mediated EPSCs, 26% of the cholinergic stimulated responses also resulted in nAChR mediated disynaptic inhibitory postsynaptic currents (IPSCs) when DA neurons were held at −20 mV (Table S2). These GABAergic currents were completely abolished by either the GABAA inhibitor gabazine or the cocktail of nAChR inhibitors.

The medial SNc mediates mainly disynaptic inhibitory or monosynaptic biphasic currents produced by ACh and GABA coreleased onto DA neurons

Our recordings in medial SNc DA neurons expressed largely indirect outward currents or biphasic inward and outward currents. In 47% of all medial DA SNc neurons recorded, we observed disynaptic inhibitory postsynaptic currents (IPSCs) mediated by presynaptic nAChRs (indirect IPSCs) in response to 5 ms of blue light stimulation of cholinergic terminals (Fig. 2 and Table S2). These indirect IPSCs were sensitive to both nAChR antagonist cocktail and gabazine (Fig. 2 and Table S2). Plotting current against voltage at different holding potentials from −100 mv to +40 mV for a direct GABA_A mediated current in the presence of nAChR antagonists indicated that the currents reversed at −82 mV, near the reversal potential for chloride (E_Cl = −83 mV) (Fig. 2C, D). While 38% of recorded DA neurons showed biphasic current responses, in which in addition to blue light evoked nicotinic EPSCs held at −70 mV, we also recorded blue light evoked monosynaptic (direct) IPSCs from the same cells held at −20 mV (Table S2). Although for the biphasic currents, the outward IPSCs were resistant to nAChR antagonists but sensitive to bicuculline (GABAA receptor antagonist) inhibition, the inward currents were blocked by nAChR antagonists (Fig. 2E, Table S2). We suspected that we discovered a population of cholinergic terminals in the medial SNc that in addition to releasing ACh also coreleased GABA.

Figure 2. Medial SNc expressed mainly cholinergic mediated GABAergic neurotransmission and biphasic GABAergic and nAChR currents including coreleased GABA and ACh responses.

(A, B) Blue light activation of cholinergic fibers in the medial SNc elicited a fast IPSC mediated by presynaptic nicotinic receptors (Vh = −20 mV). The disynaptic (indirect) response was blocked by both GABAA receptor and nAChR antagonists. (C) Monosynaptic (direct) GABA currents held at different potentials in the presence of nAChR antagonists and its corresponding I–V plot (D). (E) Bicuculline sensitive monosynaptic (direct) GABAA current (Vh= −20 mV) and nAChR antagonist cocktail (MEC, MLA and DHβE) sensitive nicotinic current (Vh= −70 mV) recorded in the same DA neuron, indicating corelease of ACh and GABA. The lack of inhibition of GABAA currents with nAChR inhibitors suggests release of GABA from cholinergic neurons. (F) Direct monosynaptic GABAA currents were significantly reduced in ChAT-ChR2-VGAT KO mice as compared to ChAT-ChR2-VGAT WT. No differences were found in blue light evoked disynaptic (indirect) GABAA currents between ChAT-ChR2-VGAT WT and ChAT-ChR2-VGAT KO mice. (G) Traces of biphasic direct and indirect GABAA mediated IPSCs with their corresponding nAChR currents. (H) Onset latencies for direct and indirect IPSCs over different blue light stimulation intensities. (I) The mean latency of current onset for nAChR currents was significantly greater than that of direct GABAergic currents, while the mean latency of current onset for nAChR currents iwas significantly less than that of indirect GABAergic currents. (J) Further evidence that coreleased nAChR EPSCs and GABAergic IPSCs are monosynaptic since 4-AP was able to restore the EPSC and IPSC following abolishment with TTX. (K) Meanwhile, indirect GABA IPSCs were disynaptic since 4-AP was unable to rescue the elimination of the response with TTX. (L) Injection of a AAV floxed ChR2-YFP into the PPT of ChAT-cre mice resulted in blue light evoked indirect GABAergic IPSCs that were reversibly inhibited by nAChR inhibitors. The PPT injected mouse also expressed coreleased direct GABAergic IPSCs and nAChR EPSCs that were blocked by GABAzine and nAChR inhibitors. See also Figures S1–S3, and Tables S1 and S2.

To address whether a population of medial DA SNc neurons received coreleased ACh and GABA neurotransmission, we performed whole-cell recordings from brain slices obtained from ChATcre-ChR2-VGAT KO mice, in which the VGAT GABA transporter was floxed only in cholinergic neurons and compared to ChATcre-ChR2-VGAT WT mice. Our results were supportive of coreleased ACh and GABA neurotransmitters, as recordings from ChAT-ChR2-VGAT KO showed fully intact nAChR currents but significantly diminished GABAA current responses (42 ± 8 pA, n= 5 neurons) as compared to WT (263 ± 16 pA, n = 5 neurons) (Fig. 2F) (t(8) = 12.4, p < 0.0001, t-test). While, we did not see any difference in the amplitude of indirect GABA_A currents between VGAT KO (63 ± 22 pA) and VGAT WT mice (60 ± 10 pA)(Fig. 2F) (t(6) = 0.16, p = 0.44).

A second evidence of corelease was the difference in latencies of the direct and indirect GABAergic currents. By plotting the current latencies against blue LED power stimulation, we found that for direct GABAergic responses latencies were significantly less (4.6 ± 0.3 ms) than that of for indirect GABAergic currents (8.4 ± 0.4 ms) (t(6) = 8.3, p < 0.0001) and the blue light stimulation threshold of direct GABA was less than indirect GABA, suggesting that direct responses were monosynaptic, while, indirect responses were disynaptic (Fig. 2H). Furthermore, for biphasic GABA-nAChR current responses showed that the direct monosynaptic GABAergic current preceded the nicotinic current, while the onset of the indirect disynaptic GABAergic current occurred following the onset of the nAChR current (Fig. 2G). Another piece of evidence that both coreleased GABA IPSCs and nAChR EPSCs were monosynaptic was that both GABA IPSCs and nAChR EPSCs were rescued by 4-AP in the presence of TTX (Fig. 2J). However, the indirect GABA IPSCs were shown to be disynaptic since 4-AP was unable to rescue the abolishment of the response with TTX (Fig. 2K).

In a select experiment we injected AAV ChR2-YFP in the PPT of a ChAT-cre mouse and later recorded blue light evoked indirect GABAergic currents from a voltage clamped lateral SNc DA neuron (Fig. 2L). We were also able to record coreleased ACh mediated nAChR EPSC with a monosynaptic GABAergic IPSC in a medial SNc DA neuron from a PPT injected ChAT-cre mouse (Fig. 2L).

Although 84% of medial DA neurons expressed cholinergic mediated GABA currents, some of which were biphasic, there were a minority of cells in the medial SNc (16%, 14 of 90 cells) that expressed purely an excitatory current, which were mostly nicotinic responses (4 of 90 cells) except for one cell containing glutamatergic responses (Table S2). We further characterized the receptor subtypes conveying the evoked nicotinic responses from either medial or lateral SNc using pharmacology, and found some responses sensitive to inhibition with DHβE, suggesting α4β2* nAChRs (* indicates that there may be additional subunits in the pentameric channel). We also found α7 nAChRs, as these evoked nicotinic currents were inhibited by MLA, and some inward current responses that were blocked by α-conotoxin PeIA[A7V,S9H,V10A,N11R,E14A] (specific for α6β2* nAChRs [18]) (Fig. S3). Interestingly, when analyzing the kinetic properties of the AMPA, α4β2* and α7 nAChR responses we observed distinct activation (F(2,10) = 7.38, p = 0.01) and deactivation kinetics (F(2,10) = 23.4, p = 0.0002) (Fig. S3). The disynaptic glutamatergic responses had the fastest 10–90% rise time of 0.93 ± 0.08 ms and decay tau of 2.98 ± 0.43 ms but were insignificantly different than that for α7 mediated responses, which had a rise time of 1.7 ± 0.4 ms and decay tau of 4.7 ± 1.0 ms (p = 0.74 and p = 0.70, respectively, Tukey HSD). α4β2 mediated responses had the slowest kinetics with a mean rise time of 4.7 ± 1.3 ms and decay tau of 16.4 ± 2.5 ms that were significantly slower than AMPA glutamatergic (p = 0.01 and p = 0.0002, respectively, Tukey HSD) and α7 nAChR responses (p = 0.04 and p = 0.0009, respectively, Tukey HSD) (Fig. S3).

ACh and GABA colocalization in brainstem cholinergic nuclei and cholinergic terminals in SNc

As further evidence of ACh and GABA corelease, we performed immunohistochemistry using antibody against GABA on brain sections from knock-in mice (n = 3) that cre-dependently expressed tdTomato in ChAT positive cells (ChATcre-tdTomato). Our immunohistochemistry results showed that in the PPT 57% of ChAT positive neuronal soma also expressed GABA, while in the LDT 44% of neurons positive for ChAT also expressed GABA immunostaining (Fig. 3). To ensure that in the medial SNc there is colocalization of ACh and GABA containing terminals, we immunolabeled for vesicular acetylcholine transporter (VAChT) and vesicular GABA transporter (VGAT) and imaged z-stacks of images with confocal microscopy. Results showed that some terminals labeled as small (~1 μm diameter) puncta for both VAChT and VGAT, indicating colocalization of ACh and GABA in individual terminals (Fig. 3). To ensure that this was true colocalization and not alignment of the two different labeled puncta due to chance, we performed quantitative colocalization analysis by calculating the Manders coefficients of the degree of overlap of VAChT label to VGAT label (M1) and vice versa (M2). Then we rotated the image of one of the images 90° relative to the other and recalculated the Manders coefficients so that if there was nonspecific colocalized labeling due to chance then the Manders coefficients should not change with image rotation. Both M1 and M2 of unrotated images were significantly greater than those in which one of the images was rotated 90°, indicating that VAChT and GABA were truly colocalized in the terminals (M1 vs rotated M1: p < 0.0001, Wilcoxon rank sum test; M2 vs rotated M2; p < 0.0001, Wilcoxon rank sum test) (Fig. 3D).

Figure 3. Colocalization of ACh and GABA in cholinergic terminals in the medial SNc and colocalization of ACh and GABA in PPT and LDT neurons.

(A–C) Immunostaining of cholinergic (VAChT) and GABAergic (VGAT) markers in presynaptic terminals in the medial SNc. DA neurons were immunolabeled against tyrosine hydroxylase (TH). (D) Quantitative analysis of colocalization by calculating the Manders coefficients (M1 and M2). M1 colocalization index was significantly greater than rotated M1 when one of the two color image channels was rotated 90 degrees relative to the other (p < 0.0001, Wilcoxon rank sum test); ditto for M2 (p < 0.0001, Wilcoxon rank sum test). (E–H) Immunohistochemistry and quantification of PPT brain slices from ChAT-tdTomato mice showing some neuronal soma with colocalization of GABA and ChAT (arrow heads). (I–L) Immunohistochemistry and quantification of LDT brain slices from ChAT-tdTomato mice also showed neurons with colocalized GABA and ChAT (arrow heads). (M–P) Immunostaining of PPT from a brain section of a ChAT-ChR2-YFP-VGAT WT mouse showing VGAT colocalization in the cell bodies positive and negative for ChAT. (Q–T) However, in the PPT of a brain section from a ChAT-ChR2-YFP-VGAT KO mouse there was a lack of colocalization of VGAT and ChAT; though there was colocalization of VGAT and GABA in ChAT negative neurons. (P, T) White filled arrow heads: colocalized GABA, VGAT and ChAT; empty white arrow heads: ChAT only; empty magenta arrow heads: colocalized GABA and VGAT; yellow filled arrow heads: colocalized GABA and ChAT but not VGAT.

To complement and support our electrophysiological experiments of ACh and GABA corelease with the ChAT-ChR2-VGAT KO mice, we performed immunohistochemistry and imaged GABA, VGAT and ChR2-YFP in the PPT of brain sections from ChAT-ChR2-VGAT KO vs ChAT-ChR2-VGAT WT mice (Fig. 3M–T). In VGAT WT brain sections, we saw evidence of colocalization of VGAT in GABA positive neurons that were either ChAT positive or ChAT negative (Fig. 3M–P), while in VGAT KO brain sections, we witnessed VGAT immunstaining only in GABAergic neurons that were ChAT negative but not in ChAT positive neurons (Fig. 3Q–T).

Release probabilities of ACh and GABA during cotransmission differ and depend on stimulation frequency

We next examined short term synaptic plasticity of ACh and GABA synaptic cotransmission when stimulating with a 1.5 sec train of blue light stimuli at two different frequencies (5 Hz and 15 Hz). Under voltage-clamp mode, we recorded nicotinic EPSCs at −70 mV and GABAergic IPSCs at −20 mV. We found that the cholinergic synapses underwent facilitation of ACh release at 5 Hz optical stimulation, with the first response starting at −22 ± 5 pA and eventually increasing to −73 ± 19 pA (n=6), while for GABA release there was a slight facilitation in GABAergic currents at 5 Hz stimulation (from 162 ± 30 pA to 209 ± 29 pA)(Fig. 4A). However, at a higher stimulation frequency of 15 Hz there was a steady depression of GABAergic responses during the train (from 287 ± 62 pA to 177 ± 54 pA, n=7) (Fig. 4B), while the nAChR currents displayed a similar facilitation of current at 15 Hz as it did at 5 Hz (−30 ± 9 pA to −75 ± 23 pA)(Fig. 4B). These results suggest that with a decline of GABAergic current at 15 Hz stimulation that there would be a tipping of the balance from inhibition to excitation during higher frequency stimulation. Furthermore, the dynamic patterns of GABA_A mediated currents and nAChR mediated responses during the train stimulation suggests that in the medial SNc GABA has a high probability of release, while ACh has a low probability of release. For biphasic indirect GABA and nAChR current responses in the medial SNc DA neurons there was facilitation for both GABAergic and nAChR currents during both 5 Hz and 15 Hz stimulation (Fig. 4C, D).

Figure 4. Frequency dependent changes in GABAergic and nAChR currents mediated by cholinergic neurotransmission in the medial SNc.

(A) Voltage-clamp recordings of medial SNc DA neurons receiving coreleased GABA and ACh showed that with repeated blue light stimulation at 5 Hz with 5 ms pulse duration there was a robust repeatable outward GABAergic current (held at −20 mV) and a weak nAChR current (held at −70 mV), which facilitated. (B) At 15 Hz stimulation there was a depression of GABAergic currents, while nAChR currents facilitated. (C, D) For medial SNc neurons displaying indirect GABAergic and nAChR currents, both 5 Hz and 15 Hz stimulation resulted in facilitation of GABAA and nAChR currents. See also Figure S4.

The pattern of ACh release in the lateral SNc was very different than in the medial SNc. In the lateral SNc there was a high probability of release for ACh at both 5 and 15 Hz (Fig. 6A–D). At 5 Hz stimulation, the first nAChR response in the stimulation train was robust at −43 ± 6 pA but plateaued by the second response to 50% of the first nicotinic EPSC and remained steady thereafter in the train (Fig. 6A, B). At 15 Hz stimulation, the first nAChR response in the train was at −65 ± 16 pA and subsequently there was an exponential decay in the nAChR EPSCs that settled at zero current by the eleventh response in the train (Fig. 6C, D).

Figure 6. Frequency dependent changes in nAChR current and neuronal excitability of lateral SNc DA neurons.

(A) A voltage-clamp trace of a lateral SNc DA neuron at 5 Hz blue light stimulation. (B) Plot of mean current amplitude showing that nAChR currents undergo a moderate depression at 5 Hz stimulation. (C) A voltage-clamp trace of a lateral SNc DA neuron at 15 Hz blue light stimulation. (D) Evoked nAChR EPSCs at 15 Hz optical stimulation showed a more severe depression of the nAChR EPSCs. (E–J) Current-clamp recordings of the lateral SNc DA neurons. (E) A current-clamp trace showing that 5 Hz blue light stimulation increased firing of the DA neuron. (F) Histogram showing the average firing frequency of six DA neurons and (G) Raster plot. (H) A current-clamp trace at 15 Hz stimulation showing increased firing of DA neuron. (I) Histogram time course of the mean firing frequency of eight DA neurons and (J) raster plot of action potential events.

ACh and GABA corelease in the medial SNc have different sensitivities to extracellular Ca²⁺

We further investigated the differential probability of release of GABA and ACh during cotransmission onto medial SNc DA neurons by recording both GABA and ACh mediated currents at different concentrations of extracellular Ca²⁺. The hypothesis is that since ACh has a low probability of release while GABA has a high probability of release, then ACh release should be more vulnerable to being inhibited when extracellular Ca²⁺ concentration is lowered than GABA release. Both cholinergic evoked GABAergic currents (F(3,12) = 15.03, p = 0.0002) and nAChR currents (χ²(3) = 11.9, p = 0.008) were sensitive to lowering of extracellular Ca²⁺ (Fig. S4A–D). GABA release was less sensitive to changes in extracellular Ca²⁺ than ACh release, since there was no significant decrease in GABAergic current when extracellular Ca²⁺ was dropped from 2 mM to 1.2 mM (t(6) = 2.02, p = 0.18 with Bonferoni’s correction) (Fig. S4A, B). However, for ACh release, the nicotinic responses were completely abolished when extracellular Ca²⁺ was dropped from 2 mM to 1.2 mM (p = 0.036, Wilcoxon signed rank test). A very low Ca²⁺ (0.2 mM) caused a marked depression but not entire abolishment of the GABA current (t(6)=5.79, p = 0.002 with Bonferoni’s correction), while the nAChR current was completely eliminated (p = 0.036, Wilcoxon signed rank test) (Fig. S4A–D). Furthermore, nAChR currents had greater sensitivity to blockage of voltage-gated Ca²⁺ channels than GABA currents since 150 μM Ni²⁺ attenuated the nAChR current but had no effect on the GABAergic current (Fig. S4E, F).

DA neuronal excitability depends on frequency of stimulation and the subregions of SNc

We wanted to determine how cholinergic neurotransmission would impact DA neuronal excitability in the medial vs lateral SNc. We performed current-clamp recordings on DA cells when cholinergic terminals were optically stimulated at 5 and 15 Hz. For recordings in the medial SNc of DA neurons receiving coreleased ACh and GABA, 5 Hz stimulation of cholinergic terminals of ChAT-ChR2 brain slices inhibited action potential firing (Fig. 5A). In contrast, 15 Hz stimulation caused an initial inhibition followed by a sharp increase in the firing frequency of DA neurons during blue light stimulation (Fig. 5B). We then examined over a wider range of blue light stimulation frequencies from 5 Hz to 60 Hz and found an upside-down U-shaped relationship of peak firing frequency that peaked in excitability at 15 Hz stimulation but went back close to baseline at 60 Hz stimulation (F(5,27) = 14.47, p < 0.0001, n=5 to 6 cells) (Fig. 5C2, C3). The peak firing frequency followed an initial inhibition of spike activity for all stimulation frequencies. To examine whether coreleased GABA was responsible for the initial inhibition of the biphasic pattern of DA excitability, we performed experiments on brain slices from ChAT-ChR2-VGAT KO mice and when cholinergic terminals were excited by blue light at 15 Hz there was only an increase in action potential firing and no inhibition as with the ChAT-ChR2-VGAT WT mice (Fig. 5D).

Figure 5. Frequency dependent changes in DA neuronal excitability in the medial SNc.

(A–D) Current-clamp recordings of medial SNc DA neurons with coreleased ACh and GABA. (A1) Recordings from a ChAT-ChR2-VGAT WT mouse brain showing a current-clamp trace at 5 Hz blue light stimulation train (5 ms pulse duration), which inhibited action potential firing. (A2) Raster plot showing the time course of nine DA neurons with 5 Hz blue light stimulation, which produced inhibition of neuronal excitability. (A3) Histogram of the time course of mean firing frequency across all nine DA neurons. (B1) Recordings from a ChAT-ChR2-VGAT WT mouse brain section showing a current-clamp trace in which DA neuronal excitability initially showed an inhibition and then an enhancement in action potential firing when the frequency of blue light stimulation was increased to 15 Hz. (B2) Raster plot of six DA neurons and (B3) histogram of the mean firing frequency. (C1) Histograms of the mean firing frequency of 5–6 neurons over a range of different blue light stimulation frequencies from 5–60 Hz. Graphs summarizing the mean peak and baseline firing (C2) and change in peak firing (C3) over different frequencies of blue light stimulation. (D1) Current-clamp trace from a ChAT-ChR2-VGAT KO mouse showing an increase in neuronal excitability during blue light stimulation, the raster plot (D2) and histogram showing the time course of the mean firing frequency of the five DA neurons (D3). (E) Although DA neurons having nAChR and indirect currents are rare, stimulation at 5 Hz results in a very sharp and brief increase in neuronal excitability, which was also found at 15 Hz stimulation (F).

Unlike the coreleased GABA and nAChR responses, neurons that received biphasic evoked nAChR responses followed by indirect GABAergic responses showed only a sharp increase in frequency of action potential firing that lasted very briefly (<3 sec) for both 5 Hz and 15 Hz blue light stimulation (Fig. 5E, F)

Analysis of DA neurons firing in the lateral SNc showed that cholinergic terminal stimulation by blue light at 5 and 15 Hz augmented action potential firing during the period of stimulation (Fig. 6E–J). Surprisingly, we did not find any significant difference in the maximal peak of action potential firing frequency at 5 Hz (3.4 ± 1.5, n=6) and 15 Hz (3.9 ± 1.4, n=8) blue light stimulations (p = 0.16, Wilcoxon rank sum test), this is likely due to the greater depression of nAChR currents at 15 Hz (Fig. 6C, D) than at 5 Hz (Fig. 6A, B).

Cholinergic neurotransmission in lateral and medial SNc differentially modulates locomotion behavior

To examine how cholinergic activity in these two subregions of the SNc affect movement, we stereotactically implanted fiber optics bilaterally into either the medial or lateral SNc (Fig. S5A, B) of ChATcre-ChR2 mice and optically stimulated cholinergic terminals during the open field locomotor task. The protocol used for monitoring locomotion comprised of a 5 min baseline (no optical stimulation), 5 min of discontinuous photostimulation at 5 Hz, 5 min recovery (no stimulation), 5 min of discontinuous photostimulation at 15 Hz, and 5 min recovery (no stimulation) (Fig. 7A). The 5 min periods of discontinuous photostimulation consisted of a 20 sec baseline followed by 1 min of LED train stimulation at either 5 Hz or 15 Hz with 5 ms pulse durations, 1 min no stimulation recovery, then 1 min LED stimulation at the chosen frequency with 5 ms pulse durations, then 1 min no stimulation recovery, and then ending with 40 sec photostimulation train (Fig. 7A). Analysis of recorded videos showed that stimulation of the lateral SNc increased motor activity at both 5 and 15 Hz (F(4,16) = 12.8, p < 0.0001) (5 Hz vs baseline: t(4) = 4.3, p = 0.012; 15 Hz vs recovery 1: t(4) = 5.5, p = 0.005, Bonferroni’s correction, n= 5 mice)(Fig. 7B). Interestingly, 5 Hz photostimulation of cholinergic terminals in the medial SNc significantly decreased locomotor activity (F(4,16) = 4.08, p = 0.018, t(4) = 3.05, p = 0.038, with Bonferonni’s correction), whereas 15 Hz photoexcitation significantly increased locomotor activity as compared to 5 Hz stimulation (t(4) = 3.7, p = 0.02, with Bonferonni’s correction) (Fig. 7E). We confirmed that the changes in locomotor activity were mediated by activation of nAChRs as 5 mg/kg mecamylamine (MEC) abolished any change in locomotor activity whether light stimulation was performed in the lateral SNc (p = 0.98, Tukey HSD, n = 4 mice) (drug effect: F(1,22) = 17, p = 0.0004; blue light effect: F(1,22) = 7.7, p = 0.011; drug × blue light effect: F(1,22) = 5.4, p = 0.029) (Fig. 7C) or the medial SNc (p = 0.93, Tukey HSD, n = 2 mice) (drug effect: F(1,22) = 20, p = 0.0003; blue light effect: F(1,22) = 6.5, p = 0.02; drug × blue light effect: F(1,22) = 3.8, p = 0.069) (Fig. 7F).

Figure 7. Optogenetic stimulation of the lateral SNc increased locomotion while stimulation of the medial SNc depressed locomotion.

(A) Experimental design performed on mice implanted with fiber optics showing the 5 min periods of baseline, blue light stimulation at 5 Hz (blue segments), recovery no stimulation, blue light stimulation at 15 Hz, and then followed by recovery no stimulation. (B) In ChAT-ChR2 mice with fiber optics implanted into the lateral SNc the top trajectory plots show baseline activity (black trace) and activity during blue light stimulation at 5 Hz and 15 Hz (blue traces). Graph shows that photostimulation of lateral SNc in ChAT-ChR2 mice induced significant enhancement at 5 Hz (p = 0.012, paired t-test, Bonferonni’s correction, n=5 mice) and 15 Hz stimulation (p = 0.005, paired t-test, Bonferonni’s correction, n=5 mice). (C) Open field locomotion during 5 Hz blue light stimulation of the lateral SNc was compared to baseline in mice injected (+MEC) or not injected (−MEC) with mecamylamine (5 mg/kg). There was significant increase in locomotion with blue light stimulation with uninjected mice (p = 0.008, Tukey HSD, n = 9 mice). Injected mice showed no significant change in locomotor activity (p = 0.98, Tukey HSD, n = 4 mice). (D) Blue light photostimulation of α4YFP mice with fiber optic implants into the lateral SNc did not significantly alter locomotion. (E) With fiber optics implanted into the medial SNc of ChAT-ChR2 mice photostimulation at 5 Hz decreased locomotor activity (p = 0.038, paired t-test, Bonferonni’s correction, n=5 mice) but not at 15 Hz stimulation, which showed significantly greater locomotion than with 5 Hz (p = 0.02, paired t-test, Bonferonni’s correction, n=5 mice). (F) Mecamylamine injected ChAT-ChR2 mice did not show any significant change in open field locomotion (p = 0.93, Tukey HSD, n = 2 mice) while the uninjected mice decreased their locomotor activity (p = 0.028, Tukey HSD, n = 8 mice) with blue light. (G) Blue light stimulation of the medial SNc of ChAT-ChR2-VGAT KO mice resulted in a significant increase in locomotion (p = 0.003, paired t-test, n = 4 mice). (H) A schematic diagram of the cholinergic circuitry in the SNc summarizing cholinergic mediated postsynaptic currents, their effect on DA neuronal excitability and locomotion. See also Figure S5.

To ensure that the changes in locomotion with blue light stimulation were due to activation of ChR2 expressing cholinergic neurons and not due to potential phototoxicity, we performed negative control experiments in which blue light stimulation of fiber optics implanted in the lateral SNc were performed in α4YFP knock-in mice. These mice have the α4 nAChR subunits fused to YFP, which are highly expressed on the membranes of DA neurons [4] but are not activated by blue light. Blue light stimulation at 5 or 15 Hz of the lateral SNc of α4YFP knock-in mice did not produce any changes in locomotion (F(4,12) = 0.97, p = 0.46, n=4 mice)(Fig. 7D). We also plotted the instantaneous velocity of movement for each frame analyzed and averaged over all the mice in relation to the blue light flashes used to stimulate the SNc (Fig. S5C–H).

To ascertain the role of GABA that is coreleased with ACh on motor function, we performed open field locomotor tests of ChATcre-ChR2-VGAT WT vs ChATcre-ChR2-VGAT KO mice. We found that in a 5 min session the ChATcre-ChR2-VGAT KO mice traveled a significantly longer distance than the ChATcre-ChR2-VGAT WT mice (Fig. S5I) (t(11) = 2.24, p = 0.024). Furthermore, we noticed that the ChATcre-ChR2-VGAT KO stopped less frequently to explore their surroundings and had significantly lower number of rearings (15.6 ± 2.5, n=7 mice) than ChATcre-ChR2-VGAT WT mice (25.3 ± 4.2, n=6 mice) (t(11) = 2.06, p = 0.03). Consistent with these results, in ChATcre-ChR2-VGAT KO mice that had fiber optics implanted into the medial SNc, blue light stimulation at 5 Hz resulted in a significant increase in locomotion as compared to baseline locomotion without stimulation (t(3) = −8.5, p = 0.003; F(4,12) = 3.5, p = 0.04).

Discussion

We found that light evoked ACh release onto the lateral SNc DA neurons resulted in mainly excitatory currents (direct nicotinic or indirect glutamatergic) and a gain in neuronal excitability, while medial DA neurons mainly expressed a biphasic current (direct nicotinic and GABA) or solely an indirect GABAergic response, which mainly inhibited neuronal excitability. In a subpopulation of the medial DA neurons there was evidence of coreleased ACh and GABA. To our knowledge, this is the first study to discover coreleased neurotransmission onto DA neurons in the SNc. Interestingly, ChR2 stimulation in awake behaving mice showed increased locomotion upon stimulation of cholinergic terminals in the lateral SNc, while there was decreased locomotion with cholinergic stimulation in the medial SNc at only 5 Hz.

The PPT and LDT are heterogeneous nuclei containing cholinergic, glutamatergic and GABAergic neurons [14,19]. Stimulation of the PPT largely excites SNc DA neurons [7,20]. We found an increase in neuronal excitability when we stimulated cholinergic terminals in the lateral SNc. The cholinergic neurotransmission that these neurons received were mainly direct monosynaptic nicotinic EPSCs or indirect disynaptic AMPA and NMDA glutamatergic EPSCs conveyed by presynaptic nAChRs (Fig. 1). Furthermore, our study establishes that postsynaptic nAChRs do mediate fast direct nAChR mediated neurotransmission in SN DA neurons due to endogenous release of ACh. With the advent of optogenetics, fast direct nAChR neurotransmission has been more commonly found in the CNS [21,22]. Very likely, two modes of cholinergic neurotransmission could be involved that includes both fast direct nAChR neurotransmission in addition to diffuse volume transmission [20,23,24]. Interestingly, ultrastructural studies have shown that in the CNS there appears to be two populations of cholinergic synapses. One is a population of ACh containing presynaptic terminals that do not form synaptic junctional complexes but are involved in diffuse volume transmission. The other population form synaptic structures with postsynaptic densities [25]. The cholinergic input is likely from PPT and LDT as witnessed from other studies using anterograde tracing and optogenetics [7,19,25]. The glutamatergic input into the SNc is unknown but may be from the subthalamic nucleus [6], local glutamatergic SNc neurons [27], PPT glutamatergic neurons [14] or the somatosensory/motor cortices [28]. Cholinergic neurotransmission in the medial SNc largely involved indirect disynaptic GABAergic neurotransmission, mediated by GABAA receptors and presynaptic nAChRs (Fig. 2), and biphasic responses consisting of GABA and postsynaptic nicotinic responses. The source of the GABAergic neurotransmission is likely the SN pars reticulata (SNr) [4].

There were two kinds of biphasic responses, one that comprise monosynaptic nAChR and disynaptic GABAergic responses, which were very rare (3 of 90 medial SNc cells), and another population of biphasic responses resulting from ACh and GABA corelease resulting in GABAergic responses that slightly precede the nAChRs responses (31 of 90 medial SNc cells). As a result, the ACh-GABA corelease leads to net inhibition of DA neuronal firing with 5 Hz stimulation. However, 15 Hz stimulation resulted in a switch from neuronal inhibition to excitation during the stimulation train. This was due to the fact that at higher frequency, GABAergic currents show depression, which was not the case at 5 Hz. While nAChR currents show facilitation at both 5 and 15 Hz (Fig. 4); therefore, tipping the balance toward excitation at 15 Hz. These data also suggest that for cotransmission, GABA containing vesicles have a high probability of release, while ACh containing vesicles have a low probability of release. Thus, during cotransmission, ACh and GABA are released from separate vesicles. Many reasons could explain the difference in probability of release between ACh and GABA including potentially that ACh containing vesicles are farther from the Ca²⁺ channels in the presynaptic active zone than the GABA vesicles or that the ACh vesicles contain an isoform of synaptotagmin with lower Ca²⁺ sensitivity than the isoform of synaptotagmin tethered to GABA vesicles. In the present study we showed that ACh release had greater Ca²⁺ sensitivity than GABA release. ACh release was also completely inhibited when applying the voltage-gated Ca²⁺ channel blocker Ni²⁺ at 150 M, which had no effect on GABA release (Fig. S4). Interestingly, this is in contrast to the lateral SNc, which showed a high probability of ACh release (Fig. 6). Hence, probability of neurotransmitter release was not necessarily specified by the identity of the neurotransmitter but perhaps the expression of molecular components involved in vesicular release in specific brain circuits. As far as the origin of ACh and GABA corelease, we did find that 44% LDT neurons and 57% of PPT neurons contained both GABA and ChAT, making these nuclei likely candidates. This is in agreement with other studies showing GABA and ACh colocalization in PPT and LDT neurons [14,19]. We did show from one recording of an SNc DA neuron from a ChAT-cre mouse injected with AAV floxed ChR2-YFP into the PPT evidence of coreleased nAChR EPSCs with direct GABAergic IPSCs (Fig. 1L).

The pattern of activity of cholinergic neurons or DA neurons related to movement is quite complex. Dodson and colleagues reported that most DA neurons in the SNc had a pause in activity during onset of movement [29]. Others have shown that increased SNc activity is critical for initiation and acceleration of movement [2,30–32]. Some PPT neurons can increase in activity with movement, while others a decrease in firing with movement [15,16]. We report that activation of cholinergic terminals in the lateral SNc resulted in increased locomotion regardless of frequency of cholinergic firing (5 or 15 Hz)(Fig. 7). However, in the medial SNc we found that low frequency stimulation at 5 Hz led to decreased locomotion (Fig. 7E), consistent with the largely cholinergic mediated disynaptic IPSCs and biphasic coreleased ACh-GABA, which attenuated SNc DA firing (Fig. 4A, 5A). Interestingly, higher frequency (15 Hz) stimulation of coreleased ACh and GABA in the medial SNc significantly enhanced locomotor activity relative to 5 Hz (Fig. 7E), consistent with the attenuation followed by enhanced action potential firing of SNc neurons during blue light stimulation (Fig. 5B, C). We have summarized the major cholinergic synapses in the medial and lateral SNc, and their effects on modulating neurotransmitter release, neuronal excitability and locomotor behavior (Fig. 7H).

Our results have shed new light on how cholinergic neurotransmission in different subregions of the SNc modulate DA neuronal excitability and locomotor activity. Future work needs to be performed to determine to what extent the medial and lateral SNc are innervated by separate cholinergic nuclei, the PPT or LDT, or whether they are innervated by different subpopulations of cholinergic neurons within the same nucleus. We have only begun to address this issue in this study. Furthermore, much work needs to be performed to determine the source of ACh and GABA corelease onto DA neurons in the medial SNc.

STAR*METHODS

Contact for Reagent and Resource Sharing

Further information and requests for reagents may be directed to and will be fulfilled by the Lead Contact, Dr. Raad Nashmi (raad@uvic.ca)

Experimental Model and Subject Details

All experimental procedures were conducted in accordance with the Canadian Council for Animal Care and a protocol approved by the Animal Care Committee of the University of Victoria. Experiments were performed using ChATcre-ChR2 mice which are produced from the cross of ChAT-cre mice (JAX stock# 006410), knock-in mice which express cre-recombinase driven by the endogenous choline acetyltransferase (ChAT) promoter, with a knock-in cre-dependent channelrhodopsin-yellow fluorescent protein chimera (ChR2) mouse line (JAX stock# 012569). The other mouse line used in this study was ChAT cre-ChR2-VGAT KO, which was produced from the cross of Vgat^flox (JAX stock# 012897) and ChAT cre-ChR2 mice. We also used ChAT-tdTomato mice, a cross between ChAT-cre and Ai9 mice (JAX stock# 007909). α4YFP knock-in mice [4], a knock-in mouse with α4 nAChR subunit fused to yellow fluoresent protein (YFP), and C57BL/6J mice were also used. All the mice were housed under a 12 h light/dark cycle and give ad libitum access to both food and water.

Method Details

Brain slice preparation for electrophysiology

Acute brain slices were acquired from mice aged 20–25 days old of either sex. Mice were anesthetized by isofluorane inhalation and rapidly decapitated. Brains were removed and held for 30 sec in cold (2–4°C) cutting solution containing: 92 mM NMDG, 2.5 mM KCl, 1.25 mM NaH2PO4, 30 mM NaHCO3, 20 mM HEPES, 4.5 mM D-glucose, 5 mM Na-ascorbate, 3 mM Na-pyruvate, 0.5 mM CaCl2, and 10 mM MgCl2 (pH between 7.3–7.4). The brain was blocked in melted 3% agar-A (CAS#9002-18-0, Bio Basic Canada Inc), then placed on the slicing platform and sectioned coronally at 320 μm thickness with a vibratome (Leica VT 1000S) containing cold (2–4°C) bubbled (95% O2/5% CO2) cutting solution. Sections that included the SNc were transfered into continuously carbogenated pre-warmed (32–34°C) cutting solution for a period of 12 min time for initial recovery. Then the slices were transferred into a room temperature carbogenated holding solution containing: 119 mM NaCl, 2.5 mM KCl, 1.2 mM NaH2PO4, 24 mM NaHCO3, 12.5 mM D-glucose, 5 mM Na-ascorbate, 3 mM Na-pyruvate, 2 mM CaCl2, and 2 mM MgCl2 for period of 30 min as second recovery before recording. For recordings of AAV injected mice, mice were approximately three months old. These mice were deeply anaethetized by isofluorane inhalation and intracardially perfused with cold (2–4°C) bubbled cutting solution before rapid removal of the brain for brain sectioning.

Electrophysiological recordings

A brain slice was transferred onto a recording chamber on an upright Nikon FN1 microscope equipped with a CFI APO 40× W NIR objective (0.80 numerical aperture, 3.5 mm working distance). The chamber was continuously perfused with 32°C carbogenated recording solution containing: 122 mM NaCl, 2.5 mM KCl, 1.2 mM NaH2PO4, 24 mM NaHCO3, 12.5 mM D-glucose, 5 mM Na-ascorbate, 3 mM Na-pyruvate, 2 mM CaCl2, 0.1 mM MgCl2. In select experiments muscarinic acetylcholine receptors were inhibited with 100 nM atropine to confirm nAChR currents (SKU# A0132, Sigma-Aldrich). Dopaminergic neurons were visualized in the SNc via video monitored infra-red differential interference contrast (IR DIC) illumination microscopy using the 40× objective. Whole cell patch-clamp recordings were performed using patch pipettes with resistances between 5–8 M. Recording pipettes were prepared from borosilicate glass capillaries (1B150F-4, WPI, USA) and for current-clamp recordings they were filled with pipette solution (280–290 mOsm/L, pH 7.4) containing: 130 mM K gluconate, 5 mM EGTA, 10 mM HEPES, 2 mM MgCl2, 0.5 mM CaCl2 2H2O, 5 mM phosphate Tris, 3 mM Mg-ATP, and 0.2 mM GTP Tris. For voltage-clamp recordings the pipettes were filled with a modified pipette solution (280–290 mOsm/L, pH 7.4) containing: 135 mM CsMeSO4, 5 mM QX314 chloride, 0.6 mM EGTA, 10 mM HEPES, 2.5 mM MgCl2, 5 mM phosphate Tris, 3 mM Mg-ATP, and 0.2 mM GTP Tris. All the recordings were amplified using a MultiClamp 700B amplifier (Molecular Devices), low-pass filtered at 4 kHz, sampled at 10 kHz with a Digidata 1440A data acquisition system (Molecular Devices) and recorded using pCLAMP 10.2 acquisition software (Molecular Devices). For recording evoked excitatory postsynaptic currents (eEPSCs) and evoked inhibitory postsynaptic currents (eIPSCs) cells were held at −70 mV and −20 mV, respectively, after correction for liquid junction potential and the series resistance was corrected 40%. In the current-clamp mode bridge balance and capacitance neutralization were applied. In some recordings, biocytin (0.5%) (Cat. # 3349, Tocris Bioscience) was in the recording pipette.

Optogenetic stimulation of brain slices

After establishing whole-cell recording and identifying DA neurons, in order to stimulate ChR2-containing cholinergic axons, we used 5 ms wide-field illumination through the 40× objective with a 470 nm blue LED (Thorlabs, part # M470L3-C5). Square pulses of blue light, at half-maximum intensity, were controlled by a controller box which was driven by pCLAMP 10.2 (Molecular Devices) through Digidata 1440 (Molecular Devices). To evaluate the effects of endogenous release of ACh on DA neurons, we stimulated cholinergic axons with the 5 ms pulse at a 5 Hz or 15 Hz stimulation train for a train duration of 1.5 sec and repeated every 30 sec.

Optical fiber implant construction for in vivo optogenetics

To construct implantable optical fibers, we used step-index multimode fiber (200 μm core, 0.5 NA, Thorlabs, Item# FP200URT). Using a microstriper (Thorlabs, Part# T12S21), we stripped ~30 mm of fiber and then cut it with a fiber cutter, while leaving 10 mm of unstripped fiber. One drop of heat-curable epoxy (Thorlabs, Part# F112) was placed at the flat side of a ceramic ferrule (Precision Fiber Products, Inc, SKU: MM-FER2007C-2300) and the Stripped end of the fiber was inserted through the ferrule. Twelve mm of the stripped fiber was left exposed and the epoxy was cured with a heat gun. The ferrule with fully cured epoxy was secured from the unstripped end to a flat surface by a piece of scotch tape and then the fiber at the convex end of the ferrule was scored with a fiber cutter as well as the stripped end (8 mm or 5 mm in length). To polish the convex end of the implantable optic fiber, we used a hemostat to hold the ferrule perpendicular to polishing paper and made 20 eight-shaped rotations. We used four grades of polishing paper in the order of 5, 3, 1, and 0.3 μm (Thorlabs, Part# LF5P, LF3P, LF1P, and LF0.3P). The optic fibers with concentric light transmission and > 60% light output of the input light from the patch cable were considered acceptable for implantation.

Surgery for optical fiber implantation

The ChATcre-ChR2 and α4YFP mice, aged between 70 to 120 days old, were anesthetized with inhaled 2% isofluorane using an anesthetic machine. To make sure the animal was deeply anesthetized we examined for the lack of a toe pinch reflex. The anesthetized mouse was placed and stabilized on a stereotaxic frame (SKU# 51615, Stoelting Co.) and sufficient amount of tear gel was applied to mouse eyes to minimize drying. 50 μl of lidocaine was injected right underneath the scalp and then a 1 cm midline incision was made through the scalp. The mouse head was leveled by measuring bregma and lambda dorsal-ventral coordinates. A bone anchor screw (SKU# 51462, Stoelting Co.) was inserted into the skull 3.5 mm caudal from the optic fiber insertion site to ensure a properly secured headcap. Two holes were drilled bilaterally into the skull for either the medial SNc (bregma 3.25 mm, lateral ±0.70 mm, ventral 4.15 mm) or lateral SNc (bregma 3.25 mm, lateral ±1.35 mm, ventral 3.9 mm). The implantable optic fiber was attached to a stereotaxic cannula holder (SCH_1.25, Doric Lenses) and lowered into the brain (ventral: 4.15 mm for medial SNc and 3.90 mm for lateral SNc). For the medial SNc, in order to fit these two closely spaced optic fibers, first we implanted the shorter (5 mm long) optic fiber and then the longer (8 mm) one. Optic fibers, were then glued with cyanoacrylic glue and then a sufficient amount of dental cement was applied with a spatula. After ~10 min the incision was sutured and the mouse was put under red light in a clean cage and monitored for full recovery. Approximately seven days after surgery, we commenced the optogenetic behavioral experiments.

Open-field locomotor behavior test

Experiments were performed in a dark room lit by a red lamp. Video recordings were performed with a video camera (Sony Digital HD video camera recorder, Handycam, HDR-SR1) mounted 70 cm above a 42 cm × 20 cm cage. Individual mice were placed in the cage at least 10 min before testing to acclimatize to the new environment. To deliver blue light for stimulating ChR2-containig axons in the SNc, a 470 nm blue LED (cat# M470F3, Thorlabs) was attached to a monofiber optic patch cord (cat# MFP_200/220/900-0.53_1m_FCM-SMA, Doric Lenses), then to a fiber optic rotary joint (cat# FRJ_1×1_FC-FC, Doric Lenses) and then to a branched fiber optic pach cord (BFP(2)_200/220/900-0.53_1m-FCM-2xZF1.25(F), Doric Lenses), which was connected to the implanted optic fibers with zirconia ceramic sleeves (cat# SM-CS125S, Precision Fiber Products Inc). LED blue light was delivered at 5 ms square pulse durations set at maximal intensity using a custom made LED driver box, which was triggered with a Grass S48 stimulator (Grass Instruments). Each open field test took 25 min and comprised of 5 min baseline locomotor activity, 5 min locomotor activity during discontinious photostimulation at 5 Hz, 5 min recovery, 5 min locomotor activity during discontinious photostimulation at 15 Hz, and 5 min recovery (Fig 7A). The discontinuous photostimulation consisted of two 1 min of photostimulation and a 40 sec photostimulation period, which were preceded by 20 sec and interspaced by 1 min of no photostimulation (Fig. 7A). The recorded videos were analysed using ImageJ software (version 1.50i, https://imagej.nih.gov/ij/). The videos were converted to avi files using FFmpeg software. The stack of images for each video was converted to 8 bit grey scale Gaussian blur filtered at 5 and thresholded to detect the mouse body. The center of mass X and Y coordinates of the thresholded mouse body in each frame was calculated using the “Analyze particles” function. From this the total distance travelled for each mouse was calculated.

Immunohistochemistry

Mice were anaesthetized with ketamine (100 mg/ml) and dexmedetomidine hydrochloride (0.5 mg/ml) and intracardially perfused with 20 ml PBS with heparin (pH= 7.6), followed by 25 ml of 4% paraformaldehyde (PFA) in PBS (pH= 7.6) and 20 ml of 5% sucrose in PBS (pH= 7.6). The brain was extracted and kept in 30% sucrose (pH= 7.6) for 24 hrs, frozen in O.C.T. mounting compound and then sectioned coronally 30–40 m thick with a cryostat and mounted on coated slides (cat# 15-188-48, Superfrost Plus Gold, Fisher Scientific). For the fiber optic implanted mice, they were intracardially perfused with 20 ml PBS with heparin (pH= 7.6), followed by 25 ml of 4% PFA. Then the extracted brains were post-fixed overnight in 4% PFA, then cut 50–60 m thick coronal slices in PBS with a vibratome (VT1000 S, Leica). The slices, contaning optic fiber tracts, were used for immunohistochemistry.

The slides with brain sections were thawed for 7 min, washed twice with PBS 10 min each, and then incubated with 0.25% Triton-X for 7 min. The slides were washed twice for 10 min with PBS and then blocked with 10% donkey serum for 30 min. All primary antibodies were diluted in 3% donkey serum in PBS at 1:250 dilution. For detecting DA neurons in the SNc, the slides were incubated with a primary antibody against tyrosine hydroxylase (Pel-Freez, cat# P40101-150, host: rabbit; abcam, cat# AB76442, host: chicken; Millipore, cat# AB1542(CH), host: sheep), GABA antibody (abcam, cat# AB62669, host: chicken), GAD67 monoclonal antibody (Millipore, cat# MAB5406, host: mouse) and ChAT (Millipore, cat# AB144P, host: goat) for 24 hrs at 4°C. To evaluate the extent of ACh and GABA colocalization in the cholinergic terminals in the SNc of C57BL/6J mice, we used primary antibodies against vesicular acetylcholine transporter (VAChT) (Millipore, cat# ABN100, host: goat) and vesicular GABA transporter (VGAT) (Millipore, cat# AB5062P, host: rabbit; or for VGAT KO: Synaptic Systems, cat# 131002). After incubation with primary antibodies, slides were washed with PBS three times 10 min each and incubated with a secondary antibody (Alexa Fluor 405 IgG secondary antibody, Invitrogen, cat# A-31556; Alexa Fluor 488 IgG secondary antibody, Cy5 IgG secondary antibody, Jackson ImmunoResearch Labs, cat# 715-175-150) diluted in 3% donkey serum at a 1:300 concentration and incubated for 24 hrs at 4 °C. Brain slices were then washed with PBS three times 15 min each and then dried out for 2 min in 37 °C before mounting with 30 l Immu-Mount, pH= 8.2, (cat# 9990402, Thermo Scientific Shandon) and then coverslipped.

To visualized recorded DA neurons filled with biocytin, brain slices were fixed in 4% paraformaldehyde (pH= 7.6) in PBS for 24 hrs at 4°C. The slices were washed three times with PBS 10 min each, and then incubated with 0.25% Triton-X for 10 min. After twice washing with PBS 5 min each, the slices weretransfered into 10% donkey serum for 30 min and then they were incubated with Alexa Fluor 555 conjugated streptavidin, (cat# S32355, Thermo Fisher Scientific) for 24 hrs at 4°C. The slices were then washed three times with PBS, placed on slides, dried and mounted with 30 μl Immu-Mount.

Confocal microscopy

Coverslipped slides were imaged using a Nikon C1si spectral confocal microscope. A 20× CFI Plan Apochromat (0.75 NA, 1.0 mm working distance) and 60× oil CFI Plan Apo VC objectives (1.40 NA, 0.13mm working distance) were used to acquired images. Lambda-stack images were collected simultaneously with one laser sweep onto an array of 32 photomultiplier tubes, each sampling a 5 nm wavelength band that spanned in total over 150 nm band width. Images were obtained at 512 pixels × 512 pixels with the pixel dwell time of 5.52 μsec and a spectral detector gain at 220. Pinhole was set to medium (60 m diameter) and 405, 488 and 560 nm laser lines were used at intensities which did not saturate the signal.

Quantification and Statistical Analysis

Statistical analyses

Values are expressed as mean ± standard error. Statistics were analyzed using R statistical analysis software (www.r-project.org). Parametric statistical comparison tests were performed provided the data were normally distributed as determined by the Shapiro-Wilk test and the variances of the groups of data did not significantly differ from each other as determined with the Fligner-Killeen test of homogeneity of variances. Otherwise, non-parametric statistical tests were performed. To compare between two groups of data a one tailed t-test or paired t-test were used assuming parametric criteria, while a Wilcoxon rank-sum test or repeated Wilcoxon rank sum test were used for non-parametric data. To analyze the means of more than two groups that were repeated over different conditions, a one-way repeated measures ANOVA was used for parametric data, while a Friedman rank sum test was used for non-parametric data. Post-hoc analyses were either a Tukey’s HSD or t-tests with Bonferroni correction for parametric data, while Wilcoxon rank sum tests were performed on nonparametric data. Data were considered significantly different at p < 0.05. For electrophysiological recordings “n” represents the number of recorded neurons, while for behavioral experiments “n” represents the number of mice.

Colocalization analysis of confocal microscopy images

Analysis of colocalization of cholinergic and GABAergic terminals using confocal microscopy images was performed using ImageJ software, version 1.51a. Images were converted to 8-bit and a Gaussian blur filter set to 1.0, was applied to all images of the stacks. Automated thresholding was applied that successfully selected all puncta while maximally excluding background noise for the majority of slices. Corresponding thresholded slices from both labeled stacks were analyzed using JACoP colocalization plugin, which calculated two Manders coefficients (M1: proportion of pixels from image A overlapping with pixels from image B; M2: proportion of pixels from image B overlapping with pixels from image A). As a control experiment, the slices were re-analyzed using JACoP after rotating one of the images 90 degrees relative to the other image. This was done to rule out the possibility that the overlap found in the correct orientation was due to chance rather than true colocalization. Truly colocalized pixels should be moved out of alignment after rotation resulting in a decrease in Manders coefficients, while the amount of random overlap should remain very similar and therefore have similar Manders coefficients before and after rotation.

Highlights.

• ACh release evokes IPSCs and EPSCs in the medial and lateral SNc, respectively.

• Medial SNc receives coreleased ACh and GABA.

• Cholinergic activation in the medial SNc inhibits locomotion.

• Cholinergic activation in the lateral SNc stimulates locomotion.