Deep brain stimulation alleviates tics in Tourette syndrome via striatal dopamine transmission

Abstract

Tourette syndrome is a childhood-onset neuropsychiatric disorder characterized by intrusive motor and vocal tics that can lead to self-injury and deleterious mental health complications. While dysfunction in striatal dopamine neurotransmission has been proposed to underlie tic behaviour, evidence is scarce and inconclusive. Deep brain stimulation (DBS) of the thalamic centromedian parafascicular complex (CMPf), an approved surgical interventive treatment for medical refractory Tourette syndrome, may reduce tics by affecting striatal dopamine release. Here, we use electrophysiology, electrochemistry, optogenetics, pharmacological treatments and behavioural measurements to mechanistically examine how thalamic DBS modulates synaptic and tonic dopamine activity in the dorsomedial striatum.

Previous studies demonstrated focal disruption of GABAergic transmission in the dorsolateral striatum of rats led to repetitive motor tics recapitulating the major symptom of Tourette syndrome. We employed this model under light anaesthesia and found CMPf DBS evoked synaptic dopamine release and elevated tonic dopamine levels via striatal cholinergic interneurons while concomitantly reducing motor tic behaviour.

The improvement in tic behaviour was found to be mediated by D₂ receptor activation as blocking this receptor prevented the therapeutic response. Our results demonstrate that release of striatal dopamine mediates the therapeutic effects of CMPf DBS and points to striatal dopamine dysfunction as a driver for motor tics in the pathoneurophysiology of Tourette syndrome.

Parafascicular nucleus deep brain stimulation is an effective treatment for medically refractory Tourette syndrome. Using an animal model of the disease, Rusheen et al. show that deep brain stimulation reduces motor tics via enhanced striatal dopamine activity at D2 receptors.

Introduction

Tourette syndrome is a debilitating neurodevelopmental disorder characterized by the chronic presence of both motor and vocal tics. The Centers for Disease Control and Prevention (CDC) estimates 350 000–450 000 people are affected by this syndrome in the USA alone.¹ While primarily a disease of childhood that improves with age, a subset (∼5%) of patients are refractory to conventional therapy and continue to suffer severe disability into adulthood.² This can lead to a multi-faceted decrease in quality of life, and in children, can interfere with scholastic performance and social development.^3,4

Addressing Tourette syndrome requires an improved understanding of its underlying pathoneurophysiology. It has been hypothesized that striatal dysregulation mediates tic behaviour.^5-7 Consistent with this hypothesis, post-mortem studies of the brains of patients with Tourette syndrome have revealed a reduction of striatal GABAergic and cholinergic interneurons.^8,9 These neurons are important for regulating afferent dopamine neurons responsible for high dopaminergic tone in the striatum^10,11 and for gating efferent medium spiny neuronal activity that regulate movement in the direct and indirect pathways of the basal ganglia.^12-19 Based on neuroimaging studies in patients, it has been postulated that high dopaminergic tone mediates tic generation.²⁰ Here, we have instead hypothesized that decreased striatal dopamine transmission may account for the susceptibility to anomalous tic behaviour, and increased transmission is therapeutic.

Deep brain stimulation (DBS) of the thalamic centromedian parafascicular complex (CMPf) has emerged as a treatment strategy for medically refractory Tourette syndrome patients.²¹ The CMPf is located in the intralaminar thalamus and studies suggest it primarily functions to process salient stimuli and shift behavioural patterns.²² Importantly, the CMPf sends dense topographically organized glutamatergic projections to the dorsal striatum (caudate nucleus and putamen in humans).²³ Previous studies in normal animals have shown that optogenetic activation of intralaminar thalamic nuclei generates cholinergic interneuron-mediated phasic dopamine release.^24,25 Therefore, characterization of the effects of CMPf DBS may help determine if increased dopamine transmission improves tic behaviour.^9,26

Striatal dopamine is critical for regulating movement, reward-related behaviour and habit formation.^27,28 Dopamine signalling occurs by two primary mechanisms.²⁹ The first is phasic firing, defined as fast direct synaptic transmission between neurons. The second is tonic firing, defined as slower extra-synaptic volume transmission that affects distal dopamine receptors to modulate the gain of phasic signalling. Measurement of tonic dopamine levels with high spatiotemporal resolution has recently become possible with our development of multiple cyclic square wave voltammetry (M-CSWV).³⁰In vivo characterization of both signalling processes provides crucial insights to understand the pathoneurophysiology for improved therapy.

Previous studies have recapitulated human anatomic findings by focal disruption of GABAergic transmission in rats, and demonstrated these animals express repetitive motor tics with morphology and neural activity analogous to humans.³¹ In this study, we employed this model under light anaesthesia to investigate striatal neuronal activity and dopamine dynamics in response to parafascicular nucleus (Pf, the rat analogue of CMPf) DBS.³² We found that Pf DBS elevated alpha and beta oscillations in the dorsomedial striatum (dmSTR), thought to reflect cholinergic interneuron and dopaminergic neuronal activity. Further investigation showed Pf DBS activates thalamostriatal glutamatergic neurons to evoke synaptic dmSTR dopamine release and elevate tonic dopamine tone, while also reducing tic behaviour. DBS-induced changes in both local field potentials (LFPs) and dopamine release were found to be dependent on cholinergic interneuron activity at nicotinic acetylcholine receptors (nACh-R). Finally, the therapeutic effect on tic behaviour is mediated by dopamine activity at striatal D₂ receptors.

Materials and methods

Subjects

Adult male and female Sprague-Dawley rats (Envigo) weighing 200–300 g (2–4 months old) were used for in vivo recordings. Animals were housed in an AAALAC accredited vivarium in plastic cages under standard 12-h light/dark cycles and conditions (21°C, 45% humidity, water and food provided ad libitum). The rats were allowed to acclimate for at least 2 weeks prior to any experimentation. Guidelines published by the NIH Guide for the Care and Use of Laboratory Animals (Department of Health and Human Services, NIH publication No. 86–23, revised 1985) were followed for all facets of animal care, and the Mayo Clinic Institutional Animal Care and Use Committee (IACUC) approved the experimental procedures.

Electrodes

Carbon fibre microelectrodes (CFMs) were constructed using T300 carbon fibre (Toray) with exposed fibre lengths of 100 µm for in vivo phasic dopamine recordings and AS4 carbon fibre (Hexel) with fibre lengths of 50 µm for in vivo tonic dopamine recordings. AS4 electrodes were coated via electropolymerization in a low-density PEDOT:Nafion solution.³³

Tourette syndrome rat model

To establish the Tourette syndrome rat model, a Hamilton syringe (Hamilton Co.) was connected to a micropump for continuous injection of the GABA_A antagonist bicuculline. A CFM was mounted medial to the needle using a custom-designed adaptor that ensured 1 mm x-y and 1 mm z distance (Supplementary Fig. 1A). This distance of separation allows diffusion of bicuculline to act on neurons in the recording location to recapitulate the disease state,³⁴ as well as prevent tissue damage from (i) the infusion needle; and (ii) the volume of infusate from confounding our recordings. Both the CFM and needle were simultaneously implanted in the dorsal striatum, and the CFM target was used for positioning [Supplementary Fig. 1B; anterior-posterior (AP): +1.2 mm, medial-lateral (ML): ±2 mm, dorsal-ventral (DV): 4 mm]. An initial bolus of bicuculline was infused (1 µl/min, 1 µg/µl, 1 min infusion) to establish the model.³² After tic emergence in the lightly anaesthetized rat and 1 h from the initial injection, bicuculline was chronically infused at 15 nl/min (equivalent to 0.9 µl/h) throughout the experiment as per Vinner et al.³⁵

Optogenetic viral vectors

The pAAV-CamkIIa-Chronos-eGFP plasmid was obtained from Addgene and packaged in AAV5 (10¹²–10¹³ infectious units per ml) by Vector Biolabs with permission of the Boyden Lab at MIT (Cambridge, MA, USA). pAAV-CaMKIIa-eNpHR3.0-EYFP from the Deisseroth Laboratory was prepared by and obtained from the University of North Carolina Vector Core at >10¹³ infectious units per ml.

Optogenetic viral injections

Approximately 1000 nl of AAV was injected into the Pf nucleus (AP: −4.3, ML: ±1.3, DV: −5.6) at a rate of 100 nl/min using a Hamilton syringe. Experiments were performed at a minimum of 4 weeks after stereotaxic injection in a randomized fashion. Injection sites and optical fibre placement locations were confirmed in all animals through coronal sectioning and light microscopy.

General acute surgical procedure

Rats were initially anaesthetized with a single intraperitoneal dose of ketamine (60 mg/kg) and xylazine (10 mg/kg). A catheter was placed intraperitoneally to deliver a constant infusion of ketamine/xylazine (60 mg ketamine/10 mg xylazine/kg/h). Small rate adjustments were sometimes needed to maintain rats under light anaesthesia, as increased anaesthesia prevented emergence of motor tics. For experiments with urethane, a single dose of 1.5 g/kg was used. EMG needle electrodes were placed in the arm extensor muscles and dorsorostral snout muscles of the rat contralateral to infusion/stimulation/recording site and secured with Vicryl sutures. The animals were then placed in a stereotaxic head frame (Kopf Instruments). Body temperature was maintained with a heating pad. Fur on the scalp was removed using an electric shaver and the scalp cleaned with betadine. The skull was then exposed by creating a small (10 mm) scalp incision and placing a hemostat. Small trephine holes (0.5 mm) were drilled to place electrodes into the brain. A neurochemical recording CFM was implanted into the dorsal striatum (AP: +1.2 mm, ML: ±2 mm, DV: 4 mm), and an Ag/AgCl reference electrode was implanted into the contralateral cortex. Coordinate targeting was derived from the Paxinos and Watson rat brain atlas and calculated from bregma.³⁶

Deep brain stimulation experiments

A Rhodes bipolar concentric simulating electrode (SNE-100, MicroProbes) was implanted (AP: −4.2 mm, ML: ±1.3 mm, DV: −5.5–5.7 mm). Stimulation was conducted with a Mayo Clinic WINCS Harmoni system for phasic dopamine recording experiments,³⁷ and an A-M Systems Model 2100 stimulator for tonic dopamine recording experiments. The waveform for stimulation was biphasic and charge balanced. For initial phasic recordings, a series of stimulation parameters in the range 50–130 Hz (20 Hz increments) with a 0.4 ms pulse width and 200–400 µA (100 µA increments) current amplitude was randomized and given over a period of 2 s, with a 5-min interval between each stimulation. Subsequent experiments were then performed at 130 Hz, 0.4 ms pulse width and 0.4 mA. Tonic recording experiments were also performed with these parameters for 30 min epochs. ‘No stimulation’ control experiments were performed with all equipment connected and powered on, but no stimulation applied.

In vivo optogenetic experiments

Rats were implanted with a 200 nm optical fibre (ThorLabs) at AP: −4.3, ML: ± 1.3, DV: −5.5. The optical fibre was connected to a Plexon PlexBright^® system. For optogenetic activation, 15 mW of 465 nm light (130 Hz, 4 ms) was used. For optogenetic silencing, 2 mW of 590 nm light (130 Hz, 4 ms) was used. For tonic dopamine recording experiments, optogenetic light was applied for 30 min periods. Experiments were performed in a randomized order.

Electrophysiologic recording experiments

Electrophysiology recordings were performed with a CFM connected to an FHC head-stage and amplifier (FHC), a CED Power1401 data acquisition interface (Cambridge Electronic Design Ltd.) and Spike2 data acquisition software (Cambridge Electronic Design Ltd.). Stimulation was conducted with a Mayo Clinic WINCS Harmoni system; 130 or 50 Hz, 0.4 ms pulse width and 0.4 mA stimulations were applied for 2–4 s. Post-processing of exported time series data was performed in MATLAB (MathWorks Inc., Natick, MA, USA) as per Goyal et al.³⁸

Electrochemical recordings

For phasic recordings of dmSTR extracellular dopamine release following electrical stimulation, fast-scan cyclic voltammetry (FSCV) was performed using the WINCS Harmoni device and software.³⁷ FSCV was conducted at a 1.3 V peak potential using a scan rate of 400 V/s and 10 Hz repetition rate. Brief stimulations were then applied with the parameters described above. Tonic recordings were performed with M-CSWV using a commercial electronic interface (NI USB-6363, National Instruments), base-station PC and custom MATLAB software. The recording electrode was stabilized in the brain for 120 min prior to any stimulation or pharmacology. After experiments, post-calibration for phasic experiments was performed using a flow cell with a 200 nM concentration of dopamine, and post-calibration for tonic experiments was performed in a beaker containing a 200 nM concentration of dopamine.

Electromyography

Needle electrodes in the face and paw were connected to a g.tec biosignal amplifier (g.tec) and PowerLab 16/35 data acquisition system (ADInstruments). Data were acquired using LabChart software (ADInstruments). EMG data were analysed using custom code written in MATLAB. A fast Fourier transform was used to determine the tic frequency in each experiment and data were then bandpass filtered. Tics were identified in the waveform as spikes lasting >50 ms and >3 SD from the baseline when analysed in 3 s bins. Tic frequency was calculated as tics/min. Identified tics were cross-validated against video recordings. Face tics were identified as brief downward jerks of the lower jaw, and paw tics identified as brief contractions of the contralateral forelimb. It is important to note that prior to administration of bicuculline, rats did not demonstrate movement in these areas (Supplementary Video 1).

Pharmacology

Pharmacologic confirmation was conducted after a dopamine response with the following drugs: alpha-methyl-p-tyrosine (AMPT; 250 mg/kg, i.p.), a tyrosine hydroxylase inhibitor; and nomifensine (20 mg/kg, i.p.), a dopamine reuptake inhibitor. Mechanistic studies were performed using the following drugs: mecamylamine (3 mg/kg, i.p.), a nACh-R antagonist; sulpiride (20 mg/kg, i.p.), a selective D₂ receptor antagonist; and SCH23,390 (0.5 mg/kg, i.p.), a selective D₁ receptor antagonist. Experiments were performed in a randomized order. All chemicals were purchased from Sigma-Aldrich.

Cresyl violet staining and immunohistochemistry

Upon completion of recordings, tissue at the stimulation electrode tip was lesioned by passing a direct current (1 mA for 10 s). Anaesthetized rats underwent intracardiac perfusion using normal saline and 4% paraformaldehyde (PFA) with or without potassium ferricyanide (Sigma-Aldrich). After fixation, the brain was extracted and placed in PFA solution overnight, and then placed in a 25% glycerol sinking solution for an additional day. Coronal sections were sliced to 40 µm using a microtome (Leitz).

For cresyl violet staining, slices were mounted on glass slides, stained and imaged with light microscopy for electrode trajectory and targeting.

For immunohistochemistry, slices were washed three times in 1× PBS. Slices were then incubated in blocking solution (2% bovine serum albumin, 3% goat serum, 0.5% Triton X-100) for 2 h at room temperature in the dark. Slices were subsequently incubated with 1:1000 anti-glutamate primary antibody (AB133, EMG Millipore) in blocking solution overnight at 4°C in the dark. The following day, slices were washed three times with 1× PBS. Next, slices were incubated at room temperature with 1:500 goat anti-rabbit 647 (Alexa Fluor, ab150083, Abcam) in blocking solution for 2 h. Slices were washed three times with 1× PBS and mounted on glass slides using VectaShield Antifade mounting media (Vector Laboratories). Slices were imaged on a Zeiss LSM 780 confocal light microscope (Zeiss) under identical conditions, including laser line attenuation and detector gain. Images were obtained with the following bandpass filter setting: eGFP (bandpass filter: 490–560), eYFP (bandpass filter: 515–570) and Alexa Fluor 647 (band-pass filter: 570–690).

Statistical analysis

A detailed report of the statistics used, including experimental measures, sample sizes and statistical tests can be found in Supplementary Table 1. Sample sizes were according to comparable publications.^39-41 For all data, the normality and homogeneity of variance was checked to determine if parametric or non-parametric analyses would be performed. For parametric analyses, paired or unpaired t-tests were performed, as well as one-way ANOVA. For data collected in sequence, repeated measures one-way ANOVA was performed. The Bonferroni post hoc test was performed for all ANOVAs. For non-parametric data, Mann–Whitney test or Wilcoxon signed-rank matched pair tests were performed, as well as the Kruskal–Wallis test followed by Dunn’s multiple comparison test. For line plots, outliers were found and replaced using the shape-preserving piecewise cubic spline function in MATLAB, and a low pass IIR filter (filter order: 5, half power frequency: 0.05, Butterworth filter design) was applied to individual datasets. For tonic dopamine tracings, data were normalized to the average dopamine release from all parameters for each animal.⁴² The Δdopamine for each experiment was determined as the mean difference between the non-normalized 10 data-points prior to and the non-normalized 10 data-points after stimulation for each animal. Linear regression and multiple linear regression was performed for categorical data. Data are presented as mean ± SEM. Statistical significance was set as P < 0.05. All statistical tests were performed in Prism v8 (GraphPad Software, San Diego, CA, USA).

Data availability

The data that support the findings of this study are openly available in Figshare at https://doi.org/10.6084/m9.figshare.c.6423353.

Results

Pf DBS increases alpha- and beta- but decreases delta- and theta-oscillation power in the dmSTR

To investigate the mechanistic therapeutic effect of Pf DBS on tic behaviour, we employed an established rat model of the disease. This model replicates post-mortem human findings of decreased striatal GABAergic interneuron density through local infusion of the GABA_A receptor antagonist bicuculline. Validation of this model found that tic behaviour was generated in the basal ganglia,⁴³ and it elicits similar brain electrophysiological activity patterns found in humans.³¹ A recent tracing study found the central Pf projects to the dmSTR, and these form a parallel associative circuit that involves somatosensory and motor cortices.⁴⁴ Thus, we recorded neural oscillations with a CFM implanted in the dmSTR, while electrically stimulating the central Pf with a concentric bipolar electrode in light ketamine anaesthetized rats (Fig. 1A). To locally administer bicuculline, we implanted a cannula adjacent to the recording electrode and chronically infused bicuculline to maintain a stable tic model throughout the experiments (Supplementary Fig. 1). The location of injection within the striatum generated tics in the contralateral face and forelimb. Neurophysiologic analysis of EMG data revealed a mean burst duration of 356.67 ± 38 ms (SupplementaryFig. 2; n = 6 rats).

Figure 1.

Parafascicular nucleus deep brain stimulation (Pf DBS) increases alpha and beta power and decreases delta and theta power in the dorsomedial striatum. (A) Illustration of the surgical approach. Electrical stimulation of the pPf using an implanted concentric bipolar stimulation electrode (SE). Voltametric and electrophysiology recordings were performed at a carbon fibre microelectrode (CFM) implanted in the dorsomedial striatum (Str). An infusion cannula (Can) was implanted 1 mm lateral to the CFM for chronic bicuculline infusion to establish and maintain motor tics in the Tourette’s model. (B) Histologic verification of correct SE and CFM placement. (C) Illustration of the experimental paradigm for electrophysiological (Ephys) recordings. Recordings were initially performed in the ‘normal state’ prior to bicuculline infusion, and after infusion of bicuculline and visualization of tics. Short bursts of stimulation (2–4 s) were conducted in triplicate at 130 and 50 Hz. EMG was performed concurrently in the contralateral face and paw to evaluate tics. (D) Scalogram of a representative electrophysiology recording experiment. Beneath the scalogram are high-pass filtered (250 Hz) EMG recordings demonstrating a pause in tics (stars) in response to electrical stimulation. The arrow indicates non-specific activity in the immediate post-stimulation state. (E) Representative pre- and post-stimulation power spectra for both the normal and Tourette states. The dotted line is the 1/f baseline fitted with non-linear regression. Shaded areas of grey represent frequencies 1 SD above baseline. (F) A comparison of power by frequency band in the pre- and post-stimulation normal state. Delta power was significantly lower after stimulation, while beta power was significantly higher. n = 6 rats. ***P < 0.0001. Comparisons were made with paired t-tests. (G) A comparison of power by frequency band in the pre- and post-stimulation Tourette state. Delta and theta were significantly lower after stimulation, while alpha and beta powers were significantly higher. n = 6 rats. **P < 0.001, ***P < 0.0001. Comparisons were made with paired t-tests. (H) A comparison of the local field potentials (LFPs) pre-stimulation between the normal and Tourette states. Delta and alpha powers were significantly lower in the Tourette model state, while theta and beta were significantly higher. n = 6 rats. *P < 0.01, **P < 0.001, ***P < 0.0001. Comparisons were made with paired t-tests. (I) A comparison of the LFPs post-stimulation between the normal and Tourette states. Beta was found to be significantly higher in the Tourette model state. Delta and theta were found to be close to significantly lower. n = 6 rats. *P < 0.01, ^†P < 0.05. Comparisons were made with paired t-tests. (J–M) Comparisons of LFPs in response to 50 Hz stimulation. Analyses were performed identical to F–I. All data are presented as mean ± SEM.

Owing to the small structures being targeted, we performed histologic confirmation that showed accurate electrode placement (Fig. 1B). Our experimental paradigm for electrophysiological recordings included two identical stimulation periods, one prior to bicuculline infusion (referred to hereafter as ‘normal state’) and one afterwards (referred to hereafter as ‘Tourette state’). Stimulation consisted of 2–4 s charge-balanced electrical pulses at both 130 and 50 Hz. These frequencies were chosen to replicate high frequency 130 Hz stimulation used clinically to treat Tourette syndrome and low frequency non-therapeutic 50 Hz stimulation for comparison. In these experiments, LFPs were recorded to assess neural response to stimulation, and simultaneous face and forelimb EMG measurements were taken to evaluate behavioural sequelae. Figure 1C summarizes this experimental protocol.

Brief stimulation of the Pf elicited marked changes in neural oscillations in the Tourette state. Figure 1D shows a representative electrophysiology scalogram and EMG data before, during and after stimulation. Stimulation resulted in a clear pause in tic activity followed by a reduction in tic frequency. Interestingly, in most experiments, stimulation elicited a fine tremor in the face and forelimbs co-occurring with striatal non-specific activity peaking at 13 Hz (Supplementary Fig. 3). For further analysis, we divided the power spectrum for each group into characteristic frequency bands (delta: 1–4 Hz, theta: 4–8 Hz, alpha: 8–12 Hz, beta: 13–30 Hz and gamma: 30–70 Hz). For both the normal and Tourette model states we analysed power spectra before and after stimulation (Fig. 1E). In the normal state stimulation elicited a significant decrease in delta power and a significant increase in beta power (Fig. 1F; n = 6 rats, paired t-tests: P-values < 0.001). In the Tourette model state, stimulation resulted in a significant decrease in delta and theta power, while increasing alpha and beta power (Fig. 1G; n = 6 rats, paired t-tests: P-values < 0.01).

A comparison between the normal and Tourette states prior to stimulation found that infusion of bicuculline resulted in significantly less delta and alpha power and significantly elevated theta and beta power (Fig. 1H; n = 6 rats; paired t-tests: P-values ≤ 0.01). Comparison of these groups after stimulation found that bicuculline treatment resulted in significantly more stimulation-induced increase in beta power (Fig. 1I; n = 6 rats; paired t-test: P ≤ 0.01). Similar patterns of activity were found with 50 Hz stimulation (Fig. 1J–M; n = 6 rats).

Pf DBS evokes phasic dopamine release in the dmSTR

Previous studies have demonstrated phasic dopamine release both by activation of cholinergic interneurons and thalamostriatal afferents.²⁴ Together with our findings above, we hypothesized that Pf stimulation induces synaptic dopamine release (termed phasic). To explore this, we employed the rat model of motor tics in Tourette syndrome and recorded phasic dopamine release with FSCV using the same experimental paradigm described above (Fig. 2A). First, due to organized heterotopy in Pf projections to the striatum, we sought to determine if dopamine release at our fixed dmSTR recording site was affected by stimulation location in the Pf. We performed stimulation repeatedly while adjusting electrode depth from a dorsal to ventral position and found the magnitude of evoked dopamine directly scaled with depth, showing that heterotopy in glutamatergic neurons may affect the magnitude of dopamine release [Supplementary Fig. 4A and B; n = 13 rats; simple linear regression: R² = 0.58, F(1,7) = 9.76, P = 0.02]. Control stimulations in the cortex did not evoke dopamine release in the dmSTR, and the difference between the two stimulation sites was significant (Supplementary Fig. 5A and B; n = 7 rats; Wilcoxon matched-pairs signed rank test: P = 0.02).

Figure 2.

Parafascicular nucleus deep brain stimulation (Pf DBS) evokes phasic dopamine release in the dorsomedial striatum. (A) Illustration of the experimental paradigm for conducting phasic dopamine (DA) recordings. Recordings were initially performed in the ‘normal state’ prior to bicuculline infusion, and after infusion of bicuculline and visualization of tics (Tourette state). Short bursts of stimulation (2–4 s) were conducted in triplicate at 130 and 50 Hz. EMG was performed concurrently in the face and paw to evaluate tic production. (B) Representative fast-scan cyclic voltammetry (−0.4 V, +1.3 V, 400 V/s scan rate, 10 Hz) colour plot demonstrating characteristic oxidation and reduction potentials for DA in response to Pf electrical stimulation in the normal state. The white inset is the cyclic voltammogram recorded at the dashed white line. The red bar indicates when stimulation was applied. (C) Representative fast-scan cyclic voltammetry (−0.4 V, +1.3 V, 400 V/s scan rate, 10 Hz) colour plot demonstrating the characteristic oxidation and reduction potentials for DA in response to Pf electrical stimulation in the Tourette state. The white inset is the cyclic voltammogram recorded at the vertical dashed line. Beneath the colour plot is synced EMG data demonstrating reduction in tics (stars) in response to stimulation. The horizontal dashed line in the colour plot indicates the line plot (right). (D) Quantification of the +0.6 V oxidation potential and conversion to dopamine concentration over time in response to stimulation from C. (E) Change in stimulation-evoked DA release (ΔDA) is compared between the normal state and Tourette state. No significant difference was found (100.4 ± 12.2 nM versus 106.7 ± 9.3 nM; n = 6; comparison made with the Wilcoxon matched-pairs signed-rank test: P > 0.05). (F) Change in stimulation-evoked dopamine release is compared between high frequency (130 Hz) and low frequency (50 Hz) stimulation. Fifty hertz evoked significantly less dopamine release (106.7 ± 9.3 nM versus 75.2 ± 10.8 nM; n = 6, *P < 0.01). Comparison made with the Wilcoxon matched-pairs signed-rank test. Bar plots represent the mean. (G) A brief period of 130 Hz electrical stimulation significantly decreased tics in the immediate post-stimulation period by 50 ± 14.9% (n = 5, *P < 0.05), but (H) 50 Hz did not (n = 5, P > 0.05). Comparisons made with the Wilcoxon matched-pairs signed-rank test.

We then explored phasic dopamine release in the normal state and found that stimulation resulted in voltammogram peak potentials corresponding to oxidation and reduction of dopamine (Fig. 2B).⁴⁵ Stimulation in the Tourette state evoked dopamine release at the same characteristic oxidation and reduction potentials (Fig. 2C). In addition, stimulation arrested tic behaviour as demonstrated in the representative EMG data below the colour plot. Extraction and conversion of current at dopamine’s oxidation potential (+0.6 V) illustrates stimulation-evoked release and subsequent return to baseline (Fig. 2D). To determine if the Tourette state of motor tics (GABA_A antagonism) affected dopamine release, we compared it to normal animals. We found that evoked dopamine release was not significantly different between states (100.4 ± 12.2 nM versus 106.7 ± 9.3 nM; Fig. 2E; n = 6 rats; Wilcoxon matched-pairs signed rank test: P > 0.05). In addition, we sought to determine if our light ketamine sedation preparation, an antagonist at NMDA receptors, affected stimulation-induced dopamine release. As anaesthesia may have a major effect on tic activity, we performed experiments with urethane anaesthesia. We found no significant difference in tic frequency between urethane and ketamine/xylazine (106.7 ± 9.3 nM versus 88.4 ± 8.7 nM; Supplementary Fig. 6; n = 6 ketamine/xylazine and n = 16 urethane; Mann–Whitney test: P > 0.05). We next sought to determine if dopamine release differed between a clinically therapeutic level of stimulation (130 Hz) and non-clinical low frequency stimulation (50 Hz). We found that decreasing the frequency from 130 to 50 Hz significantly reduced phasic dopamine release (106.7 ± 9.3 nM versus 75.2 ± 10.8 nM; Fig. 2F; n = 6 rats; Wilcoxon matched-pairs signed rank test: P = 0.03). We further parameterized stimulation by scaling the stimulation amplitude from 10 to 400 μA and the frequency from 50 to 150 Hz (Supplementary Fig. 7). The overall regression was statistically significant [n = 4 rats; multiple linear regression: R² = 0.58, F(2,114) = 77.95, P < 0.01] and both the effects of amplitude (β₁ = 0.2, P < 0.01) and frequency (β₂ = 0.18, P = 0.01) alone significantly predicted the magnitude of dopamine release to stimulation.

We next examined tic frequency in response to both 130 and 50 Hz stimulation. We found that a brief period of 130 Hz electrical stimulation significantly decreased tics in the immediate post-stimulation period by 50 ± 14.9% (Fig. 2G, n = 5, Wilcoxon matched-pairs signed rank test: P < 0.05), while 50 Hz had no effect (Fig. 2H, n = 5, Wilcoxon matched-pairs signed rank test: P > 0.05).

To confirm the measured species was dopamine and not a known interferent, such as serotonin, we employed the dopamine reuptake inhibitor nomifensine and the dopamine synthesis blocker AMPT. Experiments with nomifensine found significantly increased stimulation-evoked dopamine release compared to control group (Δdopamine +45.74 ± 10.7 nM; Supplementary Fig. 8A; n = 5 rats; Mann–Whitney test: P = 0.03). Experiments with AMPT resulted in a significant reduction in the measured species (−44 ± 10.9 nM; Supplementary Fig. 8B; n = 5 rats; Mann–Whitney test: P = 0.016).

Pf DBS elevates tonic dopamine in the dmSTR and reduces tic behaviour

We next sought to examine the response of tonic dopamine levels to DBS. In the Tourette state, we modelled human clinical DBS by applying 30 min of Pf stimulation while simultaneously performing M-CSWV and EMG (Fig. 3A) in the dmSTR. Figure 3B displays M-CSWV colour plots from a representative experiment that show a stimulation-induced increase in oxidation potentials that are characteristic of tonic dopamine levels. Figure 3C and Supplementary Video 2 show representative experiments where tonic dopamine levels increased in response to simulation. Figure 3D summarizes tracings showing a reproducible increase in tonic dopamine in response to Pf stimulation compared to a no simulation control group. On average, dopamine was found to significantly change by 10.63 ± 3.0 nM (Fig. 3E; n = 9 rats for stimulation, n = 5 rats for no stimulation; Mann–Whitney test: P < 0.01). We found that stimulation significantly decreased the rate of tics (tics/min) by 31.8% ± 6% (Fig. 3F; n = 18 rats; paired t-test: P < 0.0001) when measured in the immediate post-stimulation period. Supplementary Video 3 demonstrates this result in a representative experiment. Non-therapeutic 50 Hz stimulation had no effect on tic frequency (Fig. 3G, n = 3 rats, Wilcoxon matched-pairs signed rank test: P > 0.05).

Figure 3.

Parafascicular nucleus deep brain stimulation (Pf DBS) elevates tonic dopamine in the dorsomedial striatum and reduces tic behaviour. (A) Illustration of the experimental paradigm. 30-min DBS-like stimulation (130 Hz, 4 ms pulse width, 0.4 mA), modelled after clinical parameters, was applied in the Tourette model state. Tonic dopamine (DA) levels were tracked before, during, and after stimulation. EMG was applied to evaluate the potential effects on tics. (B) Representative multiple cyclic square wave voltammetry colour plots demonstrate the oxidation potential (arrowhead) before and after stimulation. (C) Representative tracing of tonic DA levels before, during and after stimulation. Dopamine increased throughout stimulation and leveled out after stimulation ceased. (D) Dopamine tracings from n = 9 stimulation experiments were compared against n = 5 sham stimulation experiments. The solid lines represent the mean, and the dotted lines represent the ± SEM. Concentrations were normalized to the average DA release from all parameters for each animal due to inter-experiment differences in baseline DA levels. (E) Comparison of tonic DA concentrations between both groups, demonstrating a significant increase of 10.63 ± 3.0 nM. ΔDA for each experiment was determined as the difference between the average of the 10 data-points prior to and the 10 data-points after stimulation (n = 9 rats for stimulation and n = 5 rats for no stimulation, **P < 0.01). Comparison made with the Mann–Whitney test. (F) DBS-like stimulation at 130 Hz significantly decreased tics by 31.8 ± 6%. n = 18, ****P < 0.0001. Comparison made with a paired t-test. (G) Non-therapeutic 50 Hz stimulation had no effect on tic frequency (n = 3, P > 0.05, Wilcoxon matched-pairs signed-rank test). All data are presented as mean ± SEM.

Chemical species confirmation was performed with nomifensine and AMPT. As expected, we found that recorded levels increased (100.7 ± 35.9 nM; Supplementary Fig. 9A; n = 6 rats; Wilcoxon matched-pairs signed rank test: P = 0.03) and decreased (70.7 ± 27.2 nM; Supplementary Fig. 9B; n = 5 rats; Wilcoxon matched-pairs signed rank test: P = 0.03), respectively.

Optogenetic activation of thalamostriatal glutamatergic neurons evokes dmSTR dopamine release

To test our hypothesis that Pf DBS evokes dopamine release via activation of thalamostriatal glutamatergic neurons, we employed optogenetics. Two viral vectors were injected into the Pf, both containing the CamkIIa promotor specific to glutamatergic neurons: the activating ultra-fast channel rhodopsin Chronos and the inactivating channel rhodopsin eNpHR3.0. The vectors were tagged with GFP and YFP, respectively, to confirm transfection and expression in glutamatergic neurons (Fig. 4A). Experiments were performed with implantation of a fibre optic cannula, high frequency stimulation and simultaneous tonic dopamine recording (Fig. 4B).

Figure 4.

Optogenetic activation evokes tonic dopamine release. (A) Infusion of AAV1:CamkIIa:Chronos:eGFP, AAV1:CamkIIa:eNpHR3.0:eYFP viruses into the Pf nucleus and histologic confirmation of viral expression. (B) Illustration of the stimulation and recording paradigm. CFM = carbon fibre microelectrode; Can = cannula; Opto = optogenetic fibre; Str = striatum. (C) Dopamine tracings from n = 5 activation experiments were compared against n = 5 inhibition experiments. The solid lines represent the mean, and the dotted lines represent ±SEM. Concentrations were normalized to the average dopamine release from all parameters for each animal due to inter-experiment differences in baseline dopamine levels. (D) Activation significantly increased tonic dopamine by 21.2 ± 7 nM compared to −2.7 ± 2.2 nM for optogenetic inhibition. Inhibition had no significant effect on basal levels compared to pre-optogenetic application (n = 5 rats/group, Mann–Whitney test, *P < 0.05).

High frequency optogenetic activation elevated tonic dopamine release over the course of the stimulation period (Fig. 4C). High frequency optogenetic inhibition had no effect on tonic dopamine release (Fig. 4D). Activation elicited a 21.2 ± 7 nM increase in dopamine over the stimulation period compared to −2.7 ± 2.2 nM after optogenetic inhibition (n = 5 rats/group; Mann–Whitney test; P = 0.016).

nACh-Rs mediate dmSTR electrophysiologic, dopaminergic and behavioural outcomes of Pf DBS

Our initial electrophysiological findings suggested stimulation-evoked cholinergic interneuron activity in the dmSTR. Furthermore, previous studies have shown synchronized cholinergic interneuron firing elicits dopamine release via activation of nACh-Rs in the dmSTR.²⁴ Therefore, we sought to determine if cholinergic nACh-R antagonism with the nACh-R antagonist mecamylamine could alter the electrophysiologic, dopaminergic and behavioural sequelae of stimulation. Figure 5A shows representative power spectra from before and after stimulation, revealing power changes from noise that significantly differed with stimulation. Delta and theta power were found to decrease, while alpha and beta power increased (n = 6 rats, paired t-tests: P-values ≤ 0.01). After mecamylamine administration, DBS significantly reduced the theta power, while alpha and beta power were significantly higher (Supplementary Fig. 10; n = 6 rats, paired t-tests: P < 0.0001 for all comparisons). Comparison of post-stimulation power between characteristic frequency bands revealed that mecamylamine significantly reversed stimulation-induced power changes at the delta, theta and beta frequency bands (Fig. 5B; n = 6 rats; multiple ANOVAs with Bonferroni correction for multiple analyses; delta: F = 3.69, P = 0.045, P = 0.0023 for bicuculline versus mecamylamine; theta: F = 5.54, P = 0.013, P = 0.0052 for bicuculline versus mecamylamine, P = 0.0013 for normal versus bicuculline; beta: F = 7.39, P = 0.0058, P = 4.41 × 10⁻⁴ for bicuculline versus normal, P = 2.49 × 10⁻⁴ for bicuculline versus mecamylamine). The results suggest nACh-R antagonism had a consequential effect on DBS therapy and dopamine release.

Figure 5.

Nicotinic cholinergic receptors mediate dorsomedial striatal electrophysiologic, dopaminergic and behavioural outcomes of parafascicular nucleus deep brain stimulation (Pf DBS). (A) Representative pre- and post-stimulation power spectra after application of mecamylamine in the Tourette model state. The dotted line is the 1/f baseline fitted with non-linear regression. Shaded areas of grey represent frequencies 1 SD above baseline. Below is a comparison of frequency bands between the two groups. Delta (δ)and theta (θ)were significantly lower after stimulation, while alpha (ɑ) and beta (β) and powers were significantly higher (n = 5 rats, ***P < 0.0001, comparisons were made with paired t-tests). (B) A comparison of frequency bands post-stimulation between the normal, Tourette state and Tourette state + mecamylamine (+Mec). Administration of mecamylamine normalized the response to stimulation at δ, θ and β frequency bands (n = 6 rats, analysis made with multiple ANOVA with Bonferroni correction for multiple analyses, *P < 0.01). γ = gamma. (C) Representative fast-scan cyclic voltammetry colour plots of stimulation-evoked dopamine in the Tourette state and after mecamylamine administration. The magnitude of stimulation-evoked phasic dopamine release is reduced with administration of mecamylamine prior to stimulation. The bar indicates stimulation. (D) Representative cyclic voltammograms of stimulation-evoked dopamine (DA) between the two conditions from C. (E) Comparison of peak phasic dopamine release with stimulation demonstrating administration of mecamylamine significantly reduces phasic dopamine release (n = 6, *P < 0.05, comparison made with the Wilcoxon matched-pairs signed-rank test). (F) Tonic dopamine tracings in Tourette state rats are compared between no stimulation (blue), stimulation (red) and stimulation plus administration of mecamylamine (green). The solid lines represent the mean, and the dotted lines represent the ± SEM. Thirty minutes of DBS-like (130 Hz, 0.4 ms, 0.4 mA) stimulation was applied (shading). Concentrations were normalized to the average dopamine release from all parameters for each animal due to inter-experiment differences in baseline dopamine levels. n = 5 no stimulation rats, n = 9 stimulation rats, n = 4 stimulation + mecamylamine. (G) Quantified changes in tonic dopamine levels before and after stimulation are compared between groups from F. Mecamylamine administration prevents stimulation-evoked tonic dopamine release as compared to the no stimulation group. ΔDA for each experiment was determined as the mean difference between the 10 data-points prior to and the 10 data-points after stimulation for each animal. (*P < 0.05, comparisons made with the Kruskal–Wallis test with Dunn’s correction for multiple analyses). (H) The stimulation-induced reduction in tics is significantly reversed with administration of mecamylamine (n = 4, **P < 0.01). Comparisons made with repeated measures one-way ANOVA and Bonferroni correction for multiple analyses. (I) Analysis of the effect of mecamylamine alone found no significant change in tic frequency (n = 4, P > 0.05, Wilcoxon matched-pairs signed-rank test). All data are presented as mean ± SEM.

Figure 5C shows representative FSCV colour plots before and after systemic administration of mecamylamine in the Tourette state which show a reduction in phasic dopamine release in dmSTR. Figure 5D shows representative cyclic voltammograms from these same studies. We found that addition of mecamylamine significantly reduced stimulation-evoked phasic dopamine release by 59 ± 13.2 nM (Fig. 5E; n = 6 rats; Wilcoxon matched-pairs signed rank test: P = 0.03). Additionally, administration of mecamylamine significantly reduced stimulation-evoked tonic dopamine concentrations to levels not significantly different from the no stimulation group (Fig. 5F and G; n = 9 stimulation rats, n = 4 stimulation + mecamylamine rats, n = 5 no stimulation rats; Kruskal–Wallis test with Dunn’s correction for multiple analyses: P = 0.01 for no stimulation versus stimulation and P = 0.68 for no stimulation versus stimulation + mecamylamine). Furthermore, we found that mecamylamine reversed the antagonistic effect of stimulation on tics [Fig. 5H; n = 4 rats; repeated measures one-way ANOVA with Bonferroni correction for multiple analyses: F(2,6) = 15.6 and P = 0.0042, P = 0.003 for no stimulation versus stimulation, P = 0.32 for no stimulation versus stimulation + mecamylamine]. We also analysed the effect of mecamylamine alone found no significant change in tic frequency (Fig. 5I; n = 4; Wilcoxon matched-pairs signed-rank test: P > 0.05). These data suggest that Pf stimulation evokes dmSTR dopamine release via cholinergic activity where nACh-Rs are necessary in the therapeutic mechanism, and that dopamine may play an important role in tic reduction.

D₂ receptor blockade reverses the therapeutic effect of stimulation

To determine if dopamine is necessary for tic reduction, we applied DBS during the Tourette state to evaluate both tonic dopamine release and tics with application of either the D₁ receptor antagonist SCH23,390 or the D₂ receptor antagonist sulpiride. We found that D₁ and D₂ receptor antagonism did not affect stimulation-induced elevations in tonic dopamine (Fig. 6A; n = 7 SCH23,390 rats, n = 5 sulpiride rats; Mann–Whitney test: P = 0.88). D₁ receptor antagonism did not affect the therapeutic effect of stimulation [Fig. 6B; n = 7 rats; repeated measures one-way ANOVA and Bonferroni correction for multiple analyses: F(2,12) = 7.72 and P = 0.007, P = 0.03 for no stimulation versus stimulation, P = 0.005 for no stimulation versus stimulation + SCH23,390]. Intriguingly, D₂ receptor antagonism reversed the therapeutic effect of stimulation [Fig. 6C; n = 7 rats; repeated measures one-way ANOVA and Bonferroni correction for multiple analyses: F(2,12) = 12.49 and P = 0.0012, P = 0.001 for no stimulation versus stimulation, P = 0.4 for no stimulation versus stimulation + sulpiride].

Figure 6.

D₂ receptor antagonism reverses the therapeutic effect of parafascicular nucleus deep brain stimulation (Pf DBS). (A) Tonic dopamine (DA) elevations in response to 30 min stimulation after application of the D₁ receptor antagonist SCH23,390 and the D₂ receptor antagonist sulpiride. No significant difference was found between groups (n = 7 for SCH23,390, n = 5 for sulpiride, comparison made with the Mann–Whitney test). (B) Stimulation-induced reduction in tics is not reversed with SCH23,390 (n = 7, **P < 0.01, comparisons made with the repeated measures one-way ANOVA and Bonferroni correction for multiple analyses). (C) Stimulation-induced reduction in tics is reversed with sulpiride. (n = 7, *P < 0.05, **P < 0.01, comparisons made with the repeated measures one-way ANOVA and Bonferroni correction for multiple analyses). (D) Analysis of the effect of SCH23,390 alone found a significant 9.5 ± 4.3% reduction in tic frequency (n = 7, *P < 0.05, Wilcoxon matched-pairs signed-rank test). (E) Sulpiride alone had no effect on tic frequency (n = 7, P > 0.05, Wilcoxon matched-pairs signed-rank test). All data are presented as mean ± SEM.

We also sought to determine if D₁ receptor and D₂ receptor antagonism alone influenced tic frequency. SCH23,390 administration conferred a significant 9.5 ± 4.3% reduction in tic frequency (Fig. 6D; n = 7 rats; Wilcoxon matched-pairs signed-rank test: P = 0.047), while sulpiride had no effect (Fig. 6E; n = 7 rats; Wilcoxon matched-pairs signed-rank test: P > 0.05).

Discussion

Tourette syndrome is a complex childhood-onset neurodevelopmental disorder that can severely reduce quality of life. At present, there is limited understanding of the underlying pathoneurophysiology. Current research suggests an important role for striatal microcircuits and dopamine in the mechanism of Tourette syndrome tic behaviour.⁴⁶ Using a suite of electrophysiological, electrochemical, optogenetic, pharmacologic and behavioural approaches in a rat model of motor tics, we present data that shows Pf DBS activates thalamostriatal glutamatergic neurons to evoke dmSTR dopamine release, both phasic and tonic, and reduces tics. Pharmacologic blockade of nACh-Rs indicates that dopamine release occurs via a disynaptic mechanism involving cholinergic interneurons (see Fig. 7 for circuit diagram). We further show that stimulation-evoked dopamine release is necessary for the therapeutic (tic reducing) response by acting specifically at D₂ receptors.

Figure 7.

Illustration of proposed circuit diagram in the dorsomedial striatum for deep brain stimulation-induced reduction in tics. We provide evidence for modulation of thalamostriatal glutamatergic pathways that results in acetylcholine release from cholinergic interneurons (ChI). Acetylcholine acts on nACh-Rs to induce dopamine release from afferent dopaminergic fibres. Dopamine acts on D₂ receptors, but not D₁ receptors, to diminish tics. These findings build on previous evidence for this pathway as referenced (numbers) in the figure. The lightning bolt indicates DBS and the yellow outlines indicate modulated neurons. SN = substantia nigra, VTA = ventral tegmental area, Glu-N = glutamatergic neuron, MSN = medium spiny neuron, DA = dopamine, ACh = acetylcholine, Glu = glutamate.

Significant research into Tourette syndrome has focused on dysfunction of the parallel interacting cortico-striato-thalamo-cortical (CSTC) circuits and the various neurotransmitters involved in its organization and maintenance. Of these neurotransmitters, neuroanatomic and brain imaging studies have suggested a major role for dopamine in Tourette syndrome.²⁰ The primary location for dopamine activity in CTSC circuits occurs in the striatum, and the dorsal component is principally involved in movement regulation. Previous studies have shown that thalamostriatal glutamatergic activation triggers phasic dopamine release.^24,25 Therefore we sought to determine if DBS uses this circuitry to reduce motor tics in a rat model of Tourette syndrome. Indeed, we found that Pf DBS triggered phasic dopamine release, elevated tonic dopamine levels and reduced tic behaviour.

Electrical high frequency stimulation of the Pf is not specific for efferent glutamatergic neurons and therapeutic action could occur via neuronal inhibition or via action on fibres of passage. However, previous studies have shown direct, preferential and topographically organized thalamostriatal innervation of the dmSTR by the Pf.^44,47 Indeed, we found a direct correlation of stimulating electrode depth with phasic dopamine release, which may reflect better targeting of glutamatergic neurons that project to the fixed recording location in the dmSTR. To ascertain if Pf DBS evokes dmSTR dopamine release via activation of thalamostriatal glutamatergic neurons, we used optogenetics to specifically activate or inactive these neuronal cells. Our data demonstrated that high-frequency activation increased tonic dopamine levels, confirming DBS is acting within the Pf nucleus. In addition, we found high-frequency optogenetic inhibition had no effect on tonic dopamine activity, suggesting that basal PF glutamatergic activity does not play a role in the maintenance of baseline dopamine levels in the striatum.

To create our model of motor tics in Tourette syndrome, we administered the GABA_A antagonist bicuculline. This led to generation of tics and concurrent marked broadband changes in LFPs with a long temporal structure. These broadband changes have previously been theorized to represent recruitment and activation of a large local population of neurons to drive tic generation.⁴³ Pf DBS induced activity in cholinergic interneuron activity is likely mediated by GABAergic transmission in the striatum. Previous studies have shown that Pf stimulation results in acetylcholine release when GABA-A receptors are antagonized.^10,48,49 This finding is supported by literature demonstrating axon collaterals from direct pathway medium spiny neurons (MSNs) provide GABAergic input to cholinergic interneurons.^50,51 In addition to GABAergic control from MSNs, striatal cholinergic interneurons have been found to be reciprocally connected via GABAergic interneurons. Thus, blockade of GABAergic transmission likely allows propagation of activity from the Pf through cholinergic interneurons to drive dopamine release.

Elevated thalamic alpha oscillation power has been recorded in patients undergoing DBS and may reflect the electrically coupled network of striatal GABAergic interneurons which have a tendency toward entrainment at 7–15 Hz.^52,53 These GABAergic interneurons, particularly the parvalbumin positive subtype, provide wide-spread feedforward inhibition that is thought to mediate action selection.^54-56 Burst firing of these neurons has been observed during high voltage spindles which occur in the 7–13 Hz frequency range.^57,58 High voltage spindles are often associated with fine tremor of the whiskers and face,^57,59-61 which we observed in the present study immediately after stimulation periods. There is controversy surrounding their functional significance, but a recent report suggests that they may reflect a resonant mode of a thalamocortical circuit during periods of inactivity, possibly secondary to dopamine blockade of action selection.⁵² Collectively these findings agree with our data which suggest dopamine action at inhibitory D₂ receptors reduce tic behaviour.

Previous studies have shown theta oscillations in basal ganglia circuitry correspond with hyperkinetic movements, such as involuntary movements in dystonia or motor tics in Tourette syndrome.^62,63 In fact, theta oscillation power has been found to directly correlate with the severity of motor tics and are a signature of tics in this disease.^64,65 When Pf DBS was applied, theta power and tic behaviour were significantly reduced. Remarkably, this finding is substantiated by data from a small clinical cohort which found a reduction in CMPf theta oscillation power in patients who responded best to CMPf DBS.⁶⁶ Blockade of nACh-Rs prevented DBS therapeutic reduction in tics and concomitantly prevented a decrease in theta oscillation power. This demonstrates the importance of nACh-Rs, and thus cholinergic interneurons, in the therapeutic circuitry of Pf DBS and the pathoneurophysiology of Tourette syndrome.

Cholinergic interneurons in the striatum are critical for modulating afferent cortical and thalamic glutamatergic neurons to shift behavioural patterns and have been shown to influence dopamine signalling. Interestingly, neuropathologic studies of Tourette syndrome patients have demonstrated a relative paucity of cholinergic interneurons.^8,9 These cholinergic interneurons receive innervation by glutamatergic inputs from the CMPf via axodendritic connections.⁶⁷ In addition, they synapse onto dopaminergic inputs from the substantia nigra pars compacta to modulate their activity.^21,68-71 To explore how Pf DBS affects cholinergic interneuron activity, we recorded dmSTR LFPs to decode changes in neuron population activity. We found that high-frequency stimulation of the Pf selectively increased alpha and beta oscillation power. Increased alpha and beta oscillation powers have been recorded during optogenetic activation of striatal cholinergic interneurons and with cholinergic receptor agonism.⁵² Activation of cholinergic interneurons has been reported to cause hypokinetic movements,⁷² which may reflect shifting of basal ganglia activity to change current behaviour. Furthermore, synchronized activity of striatal cholinergic interneurons has been shown to mediate dopamine release through activation of nACh-Rs and not muscarinic ACh-Rs on dopaminergic axons.²⁴ Hence, we ascertained if antagonism of cholinergic interneuron activity at nACh-Rs affected behaviour, electrophysiological activity and dopaminergic neurotransmission in response to DBS.

Nicotinic receptor antagonism significantly attenuated DBS-evoked beta oscillation power and dopamine signalling. Dopamine axons have been demonstrated to exhibit pre-synaptic nACh-Rs, and nACh-R antagonism has been shown to prevent thalamostriatal activation-induced dopamine release in vitro.^24,73 A recent study that attempted to correlate striatal dopamine activity with LFP oscillations reported the polarity of dopamine and beta oscillations was complex and nuanced, dependent on spatial, temporal, and task-related factors.⁷⁴ In the context of movement disorders, increased beta oscillation power is seen in the basal ganglia of patients with Parkinson’s disease, correlating both with bradykinesia and rigidity as well as hypodopaminergia.⁷⁵ In addition, subcortical beta oscillation power is a predictor of pre-operative Yale Global Tic Severity Scale score and corresponds with therapeutic improvement.⁶⁴ Considering beta oscillation power is thought to be related to the hyper-direct pathway and rapid stop responses, elevated beta oscillation power in our study may reflect dopamine’s role in stabilization of the cortical-basal ganglia network to inhibit tic generation.^64,76-80

In our study, nACh-R antagonism ameliorated the therapeutic response of DBS. Research has shown that dmSTR striatal acetylcholine efflux selectively increases during reversal learning (an effective psychologic treatment in patients) and blockade of acetylcholine release impairs reversal learning.^16,81 One possibility, therefore, is that excitatory input from the Pf induces cholinergic interneuron mediated dopamine release to facilitate motor activity changes and reduce abnormal tic behaviour.

We show that DBS-evoked striatal dopamine release ameliorates tic behaviour through a D₂ receptor mechanism. D₂ receptors are G-coupled protein receptors that inhibit neuronal firing and are found characteristically on indirect pathway MSNs and as presynaptic autoreceptors.^82-84 Research into voltage-dependent channels demonstrates that D₂ receptors reduce excitability in response to glutamatergic input by both decreasing AMPA receptor currents in MSNs and reducing pre-synaptic glutamate release.^85-88 It is therefore possible that dopamine activity at D₂ MSNs acts to reduce aberrant corticostriatal glutamatergic inputs.⁸⁹ Some research suggests that D₂ MSNs encode unselected, sometimes aberrant behaviours such as tics, and inhibition of D₂ MSNs preferentially promotes selected behaviours encoded by D₁ MSNs.^55,90 The idea of a relative balance between D₁ and D₂ MSNs to determine action selection is supported by recent literature.^91-94 Thus, DBS-evoked dopamine activity at D₂ MSNs may help shift the balance away from tic generation in specific microcircuits. It is also possible that dopamine acts at different points in basal ganglia circuitry to reduce motor tics. D1 receptors are found primarily on direct pathway MSNs,⁹⁵ and their activation in transgenic mice induces stereotypic tic behaviour.⁹⁶ Inhibition of D1 MSNs in a pathologic circuit may reduce such behaviour, which is supported by a recent clinical trial of the D1 receptor antagonist ecopipam and our own data with SCH23,390.⁹⁷ D2 receptors are found on indirect pathway MSNs, cholinergic interneurons, and as autoreceptors on afferent dopaminergic neurons.^98,99 DBS-induced dopamine agonism of D2 receptors may reduce tics via inhibition of interneuron firing or autoregulation of afferent dopaminergic input. Additional research is needed to better characterize the effect of DBS and pharmaceutics on dopamine receptor subtypes and MSN activity in Tourette syndrome.

The finding of therapeutic activity via D₂ receptor agonism appears at odds with D₂ receptor antagonism underlying mainstay pharmacologic treatment with atypical antipsychotics for Tourette syndrome.¹⁰⁰ There are a few possibilities to explain this incongruity. There may exist a more nuanced circuit-based mechanism in DBS treatment compared to pharmacologic treatment. Indeed, tracing studies have demonstrated heterotopy in thalamostriatal innervation as well as specialization of dopaminergic neuron activity depending on striatal projection location.^44,101 Another possibility is that atypical antipsychotics may act therapeutically primarily at D₂ autoreceptors. At therapeutic concentrations, atypical antipsychotics antagonize both D₂ postsynaptic receptors and presynaptic autoreceptors.¹⁰² This possibility is supported by evidence indicating partial D₂ receptor agonist aripiprazole and the full D₂ receptor agonist pergolide are effective by acting predominantly on D₂ autoreceptors.^103-106 It also may be the case that sulpiride doses given in our experimentation may primarily antagonize autoreceptors, and not D₂ postsynaptic receptors as assumed.¹⁰⁷ A third possibility is the Tourette syndrome model used in our experiments, despite its validation, may not accurately model the human disease state. In opposition to this notion, it has been reported that administration of aripiprazole effectively reduced tics in this model.¹⁰⁸ Nevertheless, model validity is complicated by a lack of understanding if Tourette syndrome is a phenotype of various etiologies, as has been suggested.⁴⁶

We found a reduction in tics with high-frequency, but not low frequency, DBS. There is a paucity of literature concerning frequency-specific effects of Pf DBS on subcortical oscillations and therapeutic benefit. Most clinicians program high frequency 130 Hz stimulation as reported in the original Lancet publication.¹⁰⁹ One study that analysed the effects of low (10 Hz) and high frequency (100 Hz) electrical stimulation found qualitatively similar activity in striatal neurons but improved recruitment with high frequency.¹⁰ Our data on phasic dopamine release demonstrates greater evoked dopamine release with higher stimulation, supporting this notion. Therefore, clinical benefit may differ due to the improvement magnitude of effect at 130 Hz and possibly better overall network recruitment.

Several limitations exist in our study. First, cholinergic interneuron activity is indirectly assumed by antagonism of nACh-Rs. Previous studies with optogenetic activation of cholinergic interneurons evoked dopamine release with nACh-R subtype specificity.²⁴ There are, however, other sources of cholinergic innervation to the striatum, including neurons from the pedunculopontine nucleus (primarily dorsolateral) and the laterodorsal tegmental nucleus (primarily medial) in the hindbrain.¹¹⁰ These neurons have direct activity on cholinergic interneurons and MSNs via nACh-Rs and can modulate dopaminergic neurons.^111,112 Nevertheless, direct stimulation of Pf is not thought to activate these hindbrain cholinergic neurons and it is more likely that the direct synaptic connection between Pf afferent glutamatergic neurons and cholinergic interneurons evokes acetylcholine release. Another limitation is the use of anaesthetics for behavioural experimentation. Use of light sedation allowed expression of tics while allowing simultaneous stimulation, recording, and infusion of drugs. Performing this in an awake animal would present an immense technical challenge. Nonetheless, while sedation with the NMDA-antagonist ketamine has not been demonstrated to affect striatal dopaminergic function,^113,114 it may present a confounding variable in our results. We compared DBS-evoked phasic dopamine release with urethane, an anaesthetic with a broad receptorial profile that is widely used in rodent research due its limited effect on postsynaptic dopamine receptor signalling in the striatum.^115,116 Our comparison revealed no statistical difference in phasic dopamine release, suggesting ketamine anesthesia may not impact our results. Nevertheless, anesthesia depth was found to affect tic emergence, and efforts are underway to develop a chronic implantation model to address this concern.

In summary, our findings reveal that high frequency DBS of Pf-striatal glutamatergic projections ameliorates tics through enhanced dopaminergic and cholinergic transmission in the dmSTR. Future work is necessary to elucidate the roles of MSN and interneuron activity in disease pathoneurophysiology and DBS mechanisms. In addition, it will be prudent to appreciate if dopamine transmission can be used as a biomarker to predict therapeutic response, enhance DBS and pharmacologic efficacy, or develop targeted therapies.

Funding

This research was supported by the National Institutes of Health, NIH R01NS112176, R01NS129549, and R42NS125895. Training grant funding for A.E.R. was supported by the NIH F31NS115202-01A1, NIH R25GM055252-23, NIH TL1TR002380-03, and NIH T32GM065841-17. This research was also supported by the Brain Convergence Research Program of the National Research Foundation (NRF) funded by the Korean government (MSIT, No. NRF-2021M3E5D2A01022391 and 2021R1A2B5B02002437).

Competing interests

The authors and Mayo Clinic have a financial competing interest in the technology used in the research and the authors and Mayo Clinic may stand to gain financially from the successful outcome of the research.

Supplementary material

Supplementary material is available at Brain online.