The Dynamics of Dopamine D₂ Receptor-Expressing Striatal Neurons and the Downstream Circuit Underlying L-Dopa-Induced Dyskinesia in Rats

Abstract

L-dopa (l-3,4-dihydroxyphenylalanine)-induced dyskinesia (LID) is a debilitating complication of dopamine replacement therapy for Parkinson’s disease. The potential contribution of striatal D₂ receptor (D₂R)-positive neurons and downstream circuits in the pathophysiology of LID remains unclear. In this study, we investigated the role of striatal D₂R⁺ neurons and downstream globus pallidus externa (GPe) neurons in a rat model of LID. Intrastriatal administration of raclopride, a D₂R antagonist, significantly inhibited dyskinetic behavior, while intrastriatal administration of pramipexole, a D₂-like receptor agonist, yielded aggravation of dyskinesia in LID rats. Fiber photometry revealed the overinhibition of striatal D₂R⁺ neurons and hyperactivity of downstream GPe neurons during the dyskinetic phase of LID rats. In contrast, the striatal D₂R⁺ neurons showed intermittent synchronized overactivity in the decay phase of dyskinesia. Consistent with the above findings, optogenetic activation of striatal D₂R⁺ neurons or their projections in the GPe was adequate to suppress most of the dyskinetic behaviors of LID rats. Our data demonstrate that the aberrant activity of striatal D₂R⁺ neurons and downstream GPe neurons is a decisive mechanism mediating dyskinetic symptoms in LID rats.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12264-023-01054-5.

Introduction

Striatal dopamine replacement therapy with the dopamine precursor L-dopa (l-3,4-dihydroxyphenylalanine) is the most common and effective treatment for the motor deficits of Parkinson’s disease (PD). However, it induces long-term side-effects [1, 2]; L-dopa-induced dyskinesia (LID) is a major debilitating complication of chronic L-dopa therapy for PD with unclear pathogenesis [3–5]. In the past decades, significant advances have been made in the decoding of the neural mechanisms that underlie LID [6–8]. Preclinical and clinical findings have indicated that intermittent stimulation of striatal postsynaptic dopamine receptors and the subsequent aberrant activity of striatal neurons is a key mechanism underlying LID [2, 8, 9]. However, more work is needed to illuminate the role of the striatum in the pathophysiology of LID.

The striatum is central to the basal ganglia circuit of movement control [10]. The major neural inputs to the striatum are glutaminergic afferents from the cerebral cortex and dopaminergic projections from the substantia nigra pars compacta (SNc) [11, 12]. The GABAergic medium spiny projection neurons (SPNs), the principal neuron type of the striatum, form two distinct neural pathways. The direct pathway arises from SPNs expressing dopamine D₁ receptors (D₁Rs) and projects directly to the substantia nigra pars reticulata (SNr) and globus pallidus internal (GPi). The indirect pathway arises from SPNs expressing dopamine D₂ receptors (D₂Rs) and projects indirectly to the SNr/GPi through the globus pallidus externa (GPe) and the subthalamic nucleus (STN) [11, 12]. Numerous experimental studies lend support to the view that D₁Rs and the striatal direct pathway are involved in the pathophysiology of LID [2, 13]. A subset of striatal neurons in the direct pathway is activated in LID mice and in vivo fiber photometry displayed synchronized hyperactivity of striatal D₁R⁺ SPNs during dyskinesia in dyskinetic rats [8, 14]. In addition, optogenetic activation of striatal D₁R⁺ neurons induces dyskinesia-like behaviors in parkinsonian mice, and LID is inhibited by the deactivation of these neurons [7, 14]. While the role of D₁R-expressing striatal neurons in LID has been identified, the potential role of striatal D₂Rs and D₂R-expressing neurons in LID is still unclear. It has been reported that L-dopa inhibits the activity of putative striatal D₂R⁺ SPNs below rates in healthy and L-dopa-naïve PD mice [15, 16]. Early studies suggested that systemic stimulation of D₂-like receptors delays or induces dyskinesia [9, 17–19]. One recent study showed that the dysfunction of striatal D₂R-expressing neurons contributes to the etiology of paroxysmal dyskinesia [20]. Chemogenetic stimulation of putative striatal D₂R-expressing neurons by clozapine N-oxide (CNO) treatment inhibits general motor activity and LID [21]. However, another study indicated that the genetic inactivation of D₂Rs has no significant effect on the behavior of LID mice [22]. Few studies have established causal links between the aberrant activity of striatal D₂R⁺ neurons and LID.

To further investigate this issue, the LID rats were challenged with intrastriatal administration of D₂R agents, and in vivo fiber photometry was used to record the real-time GCaMP fluorescence signals of striatal D₂R⁺ neuronal populations and downstream GPe neurons following L-dopa injection in freely-moving LID and non-dyskinetic (non-LID) rats. Subsequently, we activated striatal D₂R⁺ neurons or their projections in the GPe using the optogenetic method to evaluate the effects of striatal D₂R⁺ neuronal activation on dyskinetic behaviors in dyskinetic rats. The results suggested that over-inhibition of the striatal D₂R⁺ neuronal population was causally linked to the emergence of dyskinesia upon L-dopa administration.

Materials and Methods

Laboratory Animals

Heterozygous Drd2-Cre transgenic and wild-type rats (Sprague-Dawley genetic background) were obtained from Beijing Biocytogen Co., Ltd. (Beijing, China) [14, 23]. Ninety-eight Drd2-Cre male rats and 56 wild-type male rats were used in this study. Animals weighing 225 g–250 g were group-housed under a 12-h light/dark cycle with ad libitum access to water and rodent chow. Experimental manipulations were performed following protocols approved by the Ethics Committee for Animal Experimentation of the University and the National Research Council’s Guide for the Care and Use of Laboratory Animals. All animal experiments complied with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines.

Stereotaxic Surgery and L-Dopa Treatment

Animals were treated with standard procedures (Fig. 1A). All neurosurgeries were performed with a rodent stereotaxic apparatus. Rats were anesthetized with 1%–2% isoflurane. Then, 6-hydroxydopamine (Sigma-Aldrich, St. Louis, USA; 6-OHDA, 12 µg dissolved in 4 µL ice-cold normal saline) was unilaterally infused into the medial forebrain bundle [Mfb; anteroposterior (AP), − 4.20 mm; lateral (L) − 1.25 mm; dorsal (D) − 7.8 mm, relative to bregma] through a pulled glass micropipette connected to a 10-µL Hamilton microsyringe at a rate of 4 µL/min [14, 24–26]. The effects of the 6-OHDA lesion were verified using the apomorphine-induced circling test (0.05 mg/kg, s.c.) 2 weeks after the surgery, and by post-mortem histology [24]. Only rats exhibiting >30 contralateral turns per 5 min were chosen for further investigation (Fig. S1).

Fig. 1.

Intra-striatal administration of D₂R agents affects the dyskinetic symptoms of LID rats. A Experimental scheme of the study. The dopaminergic lesion is achieved by 6-OHDA injection into the left Mfb. B, C Photomicrographs of TH staining of the SNc (B) and striatum (C) on the injected side (left; scale bar, 500 μm) compared to the contralateral side (right). D Experimental setup for the assay after intra-striatal microinjection of D₂R agents and L-dopa administration (scale bar, 500 μm). E Effect of intrastriatal perfusion of raclopride, a selective D₂R antagonist, on the abnormal involuntary movement scale (AIMS) scores per time point of dyskinetic rats following L-dopa administration (n = 6 for both the raclopride and saline groups; two-way repeated measures analysis of variance (ANOVA) with the post hoc Holm-Sidak method: F_{1, 119} = 66.857, P < 0.001; *P < 0.05, **P < 0.01). F–I Comparison of different dyskinetic parameters between the raclopride and saline groups (n = 6 for both groups). Unpaired t-test: t = − 1.944 (F); t = 3.156 (G); t = 6.325 (H) and t = 8.177 (I). J Effect of intrastriatal perfusion of pramipexole, a D₂-like receptor agonist, on the AIMS scores per time point of dyskinetic rats following L-dopa administration (n = 6 for both the pramipexole and saline groups; two-way repeated measures ANOVA with the post hoc Holm-Sidak method: F_{1, 119} = 20.810, P = 0.001; *P < 0.05, **P < 0.01). K–M Comparison of different parameters of dyskinetic behavior between the pramipexole and saline groups (n = 6 for both groups). Unpaired t-test: t = 4.008 (K); t = − 3.101 (L), and t = − 4.562 (M). N, O Representative histology (N, scale bar, 50 µm) and quantification (O, red: percentage of c-Fos-positive neurons expressing D₂-Cre-mCherry, green: percentage of D₂-Cre-mCherry⁺ neurons expressing c-Fos) of c-Fos staining in striatal brain sections of D₂-Cre LID rats after L-dopa administration. 6-OHDA, 6-hydroxydopamine; APO, apomorphine; AAV, adeno-associated virus; AIMS, abnormal involuntary movement scale; Cre, cyclization recombinase; Mfb, medial forebrain bundle; SNc, substantia nigra pars compacta; VTA, ventral tegmental area; TH, tyrosine hydroxylase; Str, striatum; LID, L-dopa-induced dyskinesia. SC, subcutaneous.

The virus was infused into the dorsolateral striatum (AP, +0.6 mm; L, − 3.6 mm; D, − 3.5 mm; Figs. 2A, 3A, 4A, 4C) or GPe (AP, − 1.1 mm; L, − 3.2 mm; D, − 5.8 mm; Fig. 5A) on the side ipsilateral to the 6-OHDA administration 1–2 weeks after the circling test. AAV5-EF1α-DIO-eYFP-WPRE-pA, AAV5-EF1α-DIO-hChR2-eYFP-WPRE-pA, AAV5-EF1α-DIO-hChR2-mCherry-WPRE-pA, AAV5-EF1α-DIO-GCaMP6m-WPRE-hGH-pA, and AAV5-hSyn-DIO-mcherry were generated by BrainVTA Co., Ltd. in China. All the adeno-associated viruses (AAVs) had titers of 1 × 10¹²–5 × 10¹² genome copies/mL. Viruses were microinfused through a pulled-glass micropipette connected to a nanoliter microinjection pump (World Precision Instruments, Sarasota, USA) at a rate of 100 nL/min. The injection volumes into the striatum and GPe were 300 nL. Optical ferrule fibers (220 μm OD, 0.37 NA) were implanted 0.3 mm–0.5 mm above the target coordinates of the virus infusion and secured to the skull with dental cement. Postoperatively, the rats were allowed 3 weeks for recovery and viral expression before the behavioral assays [27]. For the administration of intrastriatal dopamine agents, the rats were implanted with a plastic cannula targeting the ipsilateral striatum (AP, +0.6 mm; L, − 3.6 mm; D, − 3.0 mm). One week after the surgery, L-dopa plus benserazide were injected once daily for 3 weeks as described previously [14, 25]. After 21 days of chronic L-dopa administration, all rats were observed and rated based on the general procedures of abnormal involuntary movement scale (AIMS) of dyskinesia as previously described (Fig. 3B) [14, 28]. The rats with obvious AIMS scores were classified as the LID group, and rats with no apparent AIMS scores were classified as the non-LID group [25].

Fig. 2.

Striatal D₂R⁺ neuronal population activity during the dyskinetic phase of LID rats. A Experimental scheme for GCaMP_6m-based fiber photometry assay in the striatum (scale bar, 200 µm). B–E Representative striatal D₂R⁺ GCaMP fluorescence signal (dF/F, 180 s) of normal control (B), non-LID (C), and LID rats (D) at the peak phase of dyskinesia after L-dopa injection. E Representative striatal D₂R⁺ GCaMP signal of LID rats during the decay phase of dyskinesia showing a robust increase in the GCaMP signal (dF/F) correlated with the suspension of dyskinetic behavior (blue vertical lines) and a decrease in GCaMP signal (dF/F) correlated with the onset of each dyskinetic symptom (pink vertical lines). F, G Example heatmaps (upper panel) and peri-event plots (lower panel) aligned to the start of each dyskinetic movement (F) or suspension of dyskinetic movement (G) for GCaMP traces. H Power spectral density (PSD) graphs of the GCaMP signal (180 s, dF/F) from normal control, non-LID, and LID rats during the peak phase of dyskinesia (data are the mean ± SEM; n = 6 for normal and LID groups, n = 7 for the non-LID group). I Power in the 0.1 Hz–9 Hz frequency band of GCaMP fluorescence traces (dF/F) >180 s from normal, non-LID, and LID rats (n = 6 for normal and LID groups, n = 7 for the non-LID group; Kruskal–Wallis one-way ANOVA on Ranks with the post hoc Student–Newman–Keuls Method, H₂ = 13.108, P = 0.001; LID versus non-LID, q = 2.487; LID versus normal, q = 3.540; Normal versus non-LID, q = 1.186; *P < 0.05, **P < 0.01). SC, subcutaneous; PETH, peri-event time histogram; LID, L-dopa-induced dyskinesia.

Fig. 3.

Optogenetic activation of striatal D₂R⁺ neurons alleviates the dyskinetic movements of LID rats. A Experimental setup of the optogenetic assay. Stereotaxic administration of Cre-dependent AAV to express ChR2 or eYFP in striatal D₂R⁺ neurons of D₂-Cre rats (scale bar, 200 μm). B Representative photographs from dyskinetic rats showing locomotor, axial, forelimb, and orolingual abnormal involuntary movements. C Comparison of the total AIMS scores before, during, and after short (3 min) striatal blue light (473 nm) in ChR2 or eYFP rats (n = 8 for both groups; two-way repeated measures ANOVA with the post hoc Holm-Sidak method, F_{1, 143} = 33.997, P < 0.001; *P < 0.05, **P < 0.01). D–G Effects of intrastriatal optical stimulation of D₂R⁺ neurons on the locomotor (D), axial (E), forelimb (F), and orolingual (G) AIMS scores. Two-way repeated measures ANOVA with the post hoc Holm-Sidak method; F_{1, 143} = 10.389, P = 0.006 (D); F_{1, 143} = 14.299, P = 0.002 (E); F_{1, 143} = 22.968, P < 0.001 (F); F_{1, 143} = 2.588, P = 0.130 (G); *P < 0.05, **P < 0.01. H Effect of continuous optical activation on the AIMS score per time point in dyskinetic rats following levodopa administration (n = 6 for both groups; two-way repeated measures ANOVA with the post hoc Holm-Sidak method: F_{1, 119} = 107.411, P < 0.001; *P < 0.05, **P < 0.01). I–L Comparison of different dyskinetic parameters between ChR2 and eYFP groups (n = 6 for both groups). Unpaired t-test: t = − 1.449 (I); t = 3.761 (J); t = 9.160 (K) and t = 10.052 (L). LID, L-dopa-induced dyskinesia.

Fig. 4.

Striatal D₂-Cre⁺ neurons receive monosynaptic projections from several upstream brain regions and send fibers directly to the GPe. A The virus injection procedure for monosynaptic rabies virus retrograde tracing (left). Striatal D₂-Cre⁺ neurons labeled with eGFP and dsRed are starter cells for the retrograde tracing assay (right; scale bar, 100 μm). B The somatosensory cortex, ventral striatum, parafascicular nucleus of the thalamus, and substantia nigra pars compacta send monosynaptic innervation to striatal D₂-Cre⁺ neurons. TH: tyrosine hydroxylase, scale bar, 200 µm. C Fluorescent projection from striatal D₂-Cre⁺ neurons. Striatal D₂-Cre⁺ neurons are labeled by intrastriatal injection of AAV5-EF1a-DIO-EYFP-WPRE-pA in D₂-Cre rats. Str, striatum; GPi, globus pallidus internus; GPe, globus pallidus externus; Ic, internal capsule; SNr, substantia nigra pars reticulata (scale bar, 1 mm).

Fig. 5.

Overactivity of GPe neurons is involved in the dyskinetic symptoms of LID rats. A Experimental scheme for GCaMP_6m-based fiber photometry assay in the GPe (scale bar, 200 μm). B–D Representative GPe GCaMP signals (dF/F, 180 s) from normal control (B), non-LID (C), and LID rats (D) during the peak phase of dyskinesia after L-dopa injection. Pink vertical lines show distinct increases in the GCaMP signal (dF/F) correlated with the initiation of each dyskinetic behavior. E Representative GPe GCaMP signals of LID rats during the decay phase of dyskinesia. F, G Example heatmaps (F) and peri-event plots (G) aligned to the start of each dyskinetic movement. H Power spectral density (PSD) graphs of the GCaMP signal (180 s, dF/F) from normal, non-LID, and LID rats at 60 min–100 min after levodopa injection (data are the mean ± SEM; n = 6 for all groups). I Power at the 0.1 Hz–9 Hz frequency band >180 s in GCaMP traces (dF/F) from normal, non-LID, and LID rats (n = 6 for all groups; one-way ANOVA with the post hoc Holm-Sidak method, F_{2, 17} = 15.909, P < 0.001; LID versus non-LID, t = 4.557; LID versus normal, t = 5.160; normal versus non-LID, t = 0.603; *P < 0.05, **P < 0.01). The activation of GPe-projecting striatal neurons inhibits the dyskinetic behavior of LID rats. SC, subcutaneous; Ic, internal capsule; Str, striatum; GPe, globus pallidus externus; LID, L-dopa-induced dyskinesia.

Intra-Striatal Administration of D₂R Agents and Behavioral Test

The effects of intra-striatal microinjection of raclopride [S(−)-raclopride (+)-tartrate salt, Sigma-Aldrich, St. Louis, USA; 20 µg/2 µL], a selective D₂R antagonist, and vehicle on dyskinetic behavior of LID rats were investigated (Fig. 1D). To investigate the effect of pramipexole (pramipexole dihydrochloride, Sigma-Aldrich; 12 µg/2 µL in saline), an agonist of the D₂-like receptor, on the dyskinetic movement of LID rats, pramipexole was injected into the striatum simultaneously with L-dopa administration. The drug or vehicle was administrated for >1 min using a Hamilton microsyringe (10 µL) that protruded beyond the tip of the implanted cannula [24]. The AIMS of each rat were scored up to 180 min after the administration (Fig. 1E, J).

Simultaneous GCaMP6-Based Ca²⁺ Fiber Photometry and Behavioral Recordings

The multi-channel fiber photometry system (RWD Life Science Co., Ltd, Shenzhen, China) was used in the assay to record the fluorescence signals and real-time behavior video as previously described (Fig. 2A) [14, 29, 30]. Streamlined GCaMP signal data and real-time animal behavioral videos were synchronized and saved for off-line analysis. The time coordinate of each dyskinetic event was determined by instant replay review [14].

The pMAT software (Photometry Modular Analysis Tool, Piscataway, USA) and a homemade MatLab (R2022a, MathWorks, Natick, USA) scripts were used to calculate the normalized change in the motion-corrected GCaMP signal (dF/F) [14, 31]. To quantify the correlation between dyskinetic events and the striatal D₂R⁺ GCaMP signal, peri-event time heatmaps/histograms (PETHs) were generated. The event time coordinate was set at the initiation of each dyskinetic event or suspension of dyskinesia in the decay phase of dyskinesia (Fig. 2E–G) [32, 33]. The power spectral density (PSD) of the fiber photometry signal (180 s) was then estimated using Welch’s method [pwelch() in MatLab] [14]. The average band power within the dominant frequency band (0.1 Hz–9 Hz) was then computed by integrating the PSD estimate bandpower () in MatLab [32].

In Vivo Optogenetic Stimulation

An optogenetic system (Aurora120, 473 nm; Newdoon Inc., Hangzhou, China) was used to deliver and control the frequency and pulse width of the laser light [14]. The calibrated light power density (0.5 mm below the fiber tip, NA 0.37) used in the optogenetic stimulation experiment was 5 mW/mm² [27, 34–36]. For the short optogenetic stimulation assay, an LID rat was placed in the center of the chamber 60 min–90 min after the administration of L-dopa, and its dyskinetic behavior was tracked for 9 min in the assay with 3 min of light stimulation applied 3 min after starting the assay (Figs. 3C, 6B) [14]. For the continuous optogenetic stimulation assay, an LID rat was placed in the chamber immediately after the administration of L-dopa, and its dyskinetic behavior was tracked for 180 min with continuous light stimulation applied right after the start of the assay (Fig. 3H). Based on our preliminary experimental data, a 20 Hz, 5 ms pulse width of light was used in all the optogenetic activation experiments.

Fig. 6.

Optogenetic activation of striatal D₂R⁺ neuronal projections to the GPe suppresses abnormal involuntary movements of LID rats. A Diagram showing that AAV5 viruses are injected into the striatum while ferrule fibers are implanted above the GPe to activate the striatal-GPe pathway. B Changes in total AIMS scores before, during, and after short (3 min) GPe blue light (473 nm) in ChR2 or eYFP rats (n = 6 for both groups; two-way repeated measures ANOVA with the post hoc Holm-Sidak method, F_{1, 143} = 33.997, P < 0.001; *P < 0.05, **P < 0.01). SC, subcutaneous; Ic, internal capsule; Str, striatum; GPe, globus pallidus externus; LID, L-dopa-induced dyskinesia.

Neuronal Tracing and Histology

For rabies virus-based monosynaptic retrograde tracing, AAV2-EF1α-DIO-oRVG and AAV2-EF1α-DIO-H2B-eGFP-T2A-TVA (BrainVTA, Wuhan, China) were mixed in a ratio of 3:1 and infused into the dorsolateral striatum (AP, +0.6 mm; L, − 3.6 mm; D, − 3.5 mm) of D2-Cre rats with the Nanoliter microinjection system [14]. Approximately 2 weeks later, EnvA G-deleted Rabies-dsRed (BrainVTA) was injected at the same place and the rats were perfused 5 days–7 days later (Fig. 4A). For anterograde tracing, D2-Cre rats were perfused 3 weeks after the intrastriatal injection of AAV5-EF1α-DIO-eYFP-WPRE-pA (Fig. 4C).

All the rats were perfused after the behavioral test and checked for virus expression and implanted fiber or cannula location as previously described [14]. To determine the extent of the dopaminergic lesion, tyrosine hydroxylase (TH; rabbit anti-TH antibody-containing solution, 1:1200; Abcam, Cambridge, UK; ab112) immunofluorescence histochemistry of the SNc and striatum was applied on coronal sections of the brain (25 µm; Fig. 1B, C) [14]. The localization of c-Fos staining (mouse anti-c-Fos antibody-containing solution, 1:1200; Abcam, ab208942) was observed in the striatum of D2-Cre LID rats [14]. The secondary antibodies used were donkey anti-mouse Alexa Fluor 488 (Abcam, ab150073) and donkey anti-rabbit Alexa Fluor 594 (Abcam, ab150108). The sections were washed in phosphate-buffered saline mounted, and coverslipped using Fluoromount-G or DAPI-Fluoromount-G gel (SouthernBiotech, Birmingham, UK). Images of brain sections were captured under an Olympus BX51 fluorescent microscope with UPlanFL N objective lenses and a DP73 digital camera (Olympus, Tokyo, Japan). Fluorescence images for co-localization and quantification were calculated with the cell counter plug-in in Fiji (ImageJ, Dresden, UK) [37].

Quantification and Statistical Analysis

Data are represented as the mean ± SEM or median as indicated. The specific statistical tests used are detailed in each figure legend. A P-value < 0.05 was considered significant. Data were analyzed with GraphPad Prism 9.0 (GraphPad Software Inc., La Jolla, USA) and SigmaStat 3.5 (Systat Software Inc., Palo Alto, USA).

Results

Effects of Intrastriatal Administration of D₂R Agents on Dyskinetic Behavior

First, we tested whether the intrastriatal administration of D₂R agents affects the dyskinesia of LID rats. Raclopride, a selective D₂R antagonist, was administered in the striatum simultaneously with L-dopa (Fig. 1D). Compared with saline administration, raclopride caused a significant decrease in L-dopa-related AIMS scores in dyskinetic rats. The AIMS score per time point was significantly lower in the raclopride-treated group at 8 out of 10 time points (Fig. 1E; n = 6 for both the raclopride and saline groups). Intrastriatal administration of raclopride increased the latency period of dyskinesia and decreased its duration, the highest AIMS score, and the total AIMS score (Fig. 1F–I). In contrast, intrastriatal administration of pramipexole, an agonist of D₂-like receptors, resulted in a significant aggravation of dyskinetic movements in dyskinetic rats. The AIMS score per time point of the pramipexole-treated group was significantly higher at 5 out of 10 time points (Fig. 1J; n = 6 for both the pramipexole and control groups). In addition, intrastriatal administration of pramipexole following L-dopa administration decreased the latency period of dyskinesia and increased the highest AIMS score and the total AIMS score (Fig. 1K–M).

The D₂R is highly expressed in the striatum. To investigate the involvement of striatal D₂R⁺ neurons in the regulation of LID, we monitored c-Fos expression at 90 min–150 min after L-dopa injection. The D₂-Cre LID rats received an intra-striatal injection of AAV5-hSyn-DIO-mCherry 3 weeks before the assay. The results showed that L-dopa injection induced marked c-Fos expression in the striatum of LID rats (Fig. 1N). Notably, the majority (92.15% ± 1.01%) of c-Fos⁺ striatal neurons activated by L-dopa were negative for D₂-Cre-mCherry and only 7.13% ± 0.95% of the D₂-Cre-mCherry⁺ striatal neurons showed c-Fos staining (Fig. 1O, 12 brain sections from 4 rats). The results imply that most striatal D₂R⁺ neurons are unaffected or inhibited during the dyskinetic phase in LID rats.

Striatal D₂R⁺ Neuronal Population Activity in LID Rats

To further investigate how striatal D₂R⁺ neurons respond in the dyskinetic phase, GCaMP6-based fiber photometry was used to record the striatal D₂R⁺ neuronal activity of D₂-Cre LID and non-LID rats after L-dopa administration, along with D₂-Cre normal controls (n = 6 for both the normal and LID groups, n = 7 for the non-LID group). The genetically encoded green-fluorescent Ca²⁺ indicator GCaMP_6m was delivered to the D₂R⁺ neurons of the striatum by using Cre-dependent AAV5 vectors. GCaMP_6m expression was confined to the region of the dorsolateral striatum (Fig. 2A). A stable GCaMP signal of the striatal D₂R⁺ neuronal population was recorded from all groups of rats (Fig. 2B–E). The striatal D₂R⁺ GCaMP signal was mainly concentrated in the low-frequency band (<2 Hz; Fig. 2H). However, the magnitude of the striatal D₂R⁺ GCaMP signal was different among the three animal groups. Notably, during the peak phase of dyskinesia, the striatal D₂R⁺ neurons of LID rats were characterized by decreased neuronal population activity (Fig. 2B–D). Quantitative power spectrum analysis showed that the average relative power (0.1 Hz–9 Hz) of the striatal D₂R⁺ GCaMP fluorescence signal was significantly lower in LID rats than that in non-LID rats and normal controls (Fig. 2I). These results implied that the striatal D₂R⁺ neurons of LID rats are over-inhibited during peak dyskinesia. Interestingly, the activity of the striatal D₂R⁺ neurons of LID rats began to recover when the dyskinetic symptoms started to decrease in the decay phase of dyskinesia (Fig. 2E). However, it remained unclear whether the fluctuation of striatal D₂R⁺ neuronal population activity was related to specific dyskinetic movements of LID rats. To address this issue, we quantified the correlation between dyskinetic events and the striatal D₂R⁺ GCaMP signal during the decay phase of dyskinesia by plotting PETHs (Fig. 2F, G). Sorting the striatal D₂R⁺ GCaMP signal at the time points of dyskinetic behavior initiation revealed that the striatal D₂R⁺ GCaMP activity troughed at the beginning of each dyskinetic movement (Fig. 2F, Movie S1). In contrast, the striatal D₂R⁺ neuronal population activity peaked at each time point of dyskinesia suspension (Fig. 2G, Movie S1).

Optogenetic Activation of Striatal D₂R⁺ Neurons Relieves the Dyskinetic Symptoms in LID Rats

Hypoactivity of the striatal D₂R⁺ neurons was observed during the dyskinetic phase of LID rats, but the functional involvement of the striatal D₂R⁺ neuronal population in LID had not been fully established. Therefore, we virally expressed channelrhodopsin-2 (ChR2) unilaterally in the striatal D₂R⁺ neurons of LID rats (Fig. 3A), and then delivered blue light (473 nm) to these neurons during the peak phase of dyskinesia, to test the influence of striatal D₂R⁺ neuronal activation on dyskinetic symptoms in dyskinetic rats (Fig. 3A–C). When compared to the controls, acute stimulation of the striatal D₂R⁺ neuronal population inhibited locomotor, axial, limb, and total AIMS scores of ChR2⁺ LID rats (Fig. 3C–G, Movie S2; n = 8 for both groups). In contrast, optogenetic activation of striatal D₂R⁺ neurons with the same parameters did not affect the general movement of rats in the open field test (Fig. S2). Furthermore, we tested whether long-term activation of striatal D₂R⁺ neurons could affect the entire process of dyskinesia, by applying blue light continuously for 180 min immediately after the administration of L-dopa. In comparison to eYFP controls, activation of striatal D₂R⁺ neurons resulted in a significant decrease in total AIMS scores in the ChR2 group (Fig. 3H; n = 6 for both groups). In addition, we found that continuous optical stimulation significantly affected other parameters of dyskinesia in ChR2-expressing LID rats. Blue light stimulation significantly reduced the duration of dyskinesia, the highest AIMS score, and the total AIMS score in ChR2-expressing LID rats (Fig. 3I–L). Accordingly, these results suggested that hypoactivity of the striatal D₂R⁺ neuronal population is necessary for the expression of dyskinetic symptoms in LID rats.

Striatal D₂R⁺ Neurons Project to the GPe and Receive Inputs from Multiple Brain Regions

The finding that striatal D₂R⁺ neurons regulate the dyskinesia of LID rats raised the question as to which upstream neurons regulate the striatal D₂R⁺ neurons. We therefore applied Cre-dependent rabies virus-based monosynaptic retrograde tracing to screen for the upstream neurons that send projections to D₂-Cre⁺ neurons in the dorsolateral striatum of D₂-Cre rats (Fig. 4A; n = 4). The somatosensory cortex, ventral striatum, parafascicular nucleus of the thalamus, subthalamic nucleus, and dopaminergic neurons in the SNc were labeled by dsRed, the marker for monosynaptic upstream neurons of striatal D₂-Cre⁺ neurons (Fig. 4B).

In addition, we mapped the downstream projections of D₂R⁺ neurons in the dorsolateral striatum of D₂-Cre rats using Cre-dependent fluorescent tracers (n = 4). The anterograde tracing study revealed long-range projections to the GPe. In contrast, D₂-Cre-eYFP⁺ fibers were sparse in the GPi, and SNr (Fig. 4C). These results indicated that the D₂R⁺ neurons in the striatum are mainly involved in the indirect pathway of the basal ganglia.

The Striatal D₂R⁺–GPe Pathway Is Involved in the Dyskinesia of LID Rats

As noted above, the GPe is the main downstream target of striatal D₂R⁺ neurons; therefore, we investigated the neuronal dynamics of the GPe in LID rats. GCaMP_6m was delivered to the GPe neurons ipsilateral to the 6-OHDA lesion using an AAV5 vector (Fig. 5A). As shown in Fig. 5B–E, the magnitude of the GPe neuronal population GCaMP signal differed among the three rat groups (n = 6 for all groups). In contrast to the striatal D₂R⁺ neurons, the GPe neurons of LID rats were characterized by increased neuronal population activity during the peak phase of dyskinesia. Sorting the GPe GCaMP signal by the time coordinate of each dyskinetic behavior revealed that the GPe GCaMP signal peaked at the beginning of each dyskinetic movement (Fig. 5F, G). The average relative power (0.1 Hz–9 Hz) of GPe neuronal population activity in LID rats was significantly higher than that in non-LID rats and normal controls (Fig. 5H, I). These results indicated that the GPe neurons of LID rats are overactivated during dyskinesia.

The activity of GPe neurons is regulated by the GPe-projecting striatal D₂R⁺ neurons [11, 12]. To test whether the GPe-projecting striatal D₂R⁺ neurons regulate dyskinesia in LID rats, we expressed ChR2 in striatal D₂R⁺ neurons using the same Cre-dependent strategy and implanted ferrule fibers ipsilaterally, above the GPe (Fig. 6A). Optogenetic activation of the striatal D₂R⁺ neuronal projections in GPe significantly inhibited the dyskinetic behavior of LID rats after L-dopa administration (Fig. 6B; n = 6 for both groups).

Discussion

Our data showed that selective blocking of striatal D₂Rs inhibited dyskinesia in a rat model of LID, while stimulation of striatal D₂Rs induced more severe dyskinetic behaviors. We found that striatal D₂R⁺ neurons were hypoactive and downstream GPe neurons were hyperactive during the dyskinetic phase of LID rats. Importantly, selective optical activation of striatal D₂R⁺ neurons and axon terminals in the GPe resulted in distinct alleviation of the dyskinetic symptoms in LID rats.

The execution of normal movement relies on the functional coordination of the direct and indirect striatal pathways [15]. The D₁R-expressing direct pathway facilitates movements, whereas the D₂R-expressing indirect pathway is thought to stop movements [38]. The motor impairments in PD and LID are associated with dysfunction in striatal dopaminergic signaling and imbalance of the direct and indirect pathways [11, 13]. The effects of the intervention of D₂Rs on LID have been examined in many previous studies. Pioneering clinical trials have shown that early oral treatment with cabergoline, a D₂R agonist, delays the development of LID [19]. In contrast, intraperitoneal injection of D₂R agonists induced dyskinesia in a rodent model of PD [39]. Chronic subcutaneous injection of a D₂R agonist establishes dyskinesia in the L-dopa-naïve monkey model of PD [40]. However, systemic D₂R gene knockout has no significant effect on LID [22]. D₂Rs are widely expressed in many brain regions and other organs and systemic intervention of D₂Rs induces complicated and inconsistent results [41, 42]. In our study, the D₂R agents were injected locally into the striatum to avoid the side-effects induced by systemic administration. The results clearly displayed that selective blocking of striatal D₂Rs inhibits LID, while the activation of these receptors aggravates the dyskinesia of LID rats (Fig. 1D–M). Under normal conditions, D₂Rs are found on both the dopamine axon terminals from the SNc and postsynaptically on striatal neurons [43]. In parkinsonism and 6-OHDA lesioned rats, most striatal dopamine terminals have degenerated, but L-dopa can be catalyzed into dopamine in other monoaminergic terminals. Dopamine and dopaminergic agents primarily combine with dopamine receptors on postsynaptic striatal neurons [11, 14, 44, 45].

The non-physiological dopamine boost from L-dopa affects the neuronal activity and plasticity of dopamine receptor-expressing striatal neurons [15, 46]. The influence of striatal D₁R⁺ neuronal dynamics in LID has been analyzed in many previous studies [14, 15]. It has also been reported that most striatal c-Fos⁺ neurons in LID rats are positive for D₁Rs [14]. In contrast, our study showed no obvious c-Fos expression in the D₂R⁺ neurons of LID rats (Fig. 1N, O). Previous studies have reported that the firing activity of striatal D₂R⁺ neurons in LID mice is significantly lower than that of healthy and L-dopa-naïve PD mice [15, 16]. It is commonly assumed that L-dopa induces dyskinesia partially by excessive inhibition of striatal D₂R⁺ indirect pathway neurons [13]. Unfortunately, there was a lack of systematic research about striatal D₂R⁺ neuronal dynamics in LID and non-LID animal models. Our GCaMP-based fiber photometry data showed that the striatal D₂R⁺ neuronal population activity of LID rats during the peak phase of dyskinesia was significantly lower than that of non-LID rats after the same L-dopa treatment (Fig. 2B–D). Importantly, we found that the striatal D₂R⁺ neurons of LID rats showed intermittent overactivity in the decay phase of dyskinesia, which was synchronized with the discontinuous suspension of dyskinetic behaviors (Fig. 2E–G). It has been reported that the ablation of striatal D₂R⁺ neurons induces motor hyperactivity in normal mice and the striatal D₂R⁺ SPNs are the first to degenerate in Huntington’s disease [38, 47]. Therefore, we have reasons to hypothesize that the abnormal hypoactive striatal D₂R⁺ neuronal population dynamics after L-dopa injection might mediate the motor hyperactivity in LID.

To explore the potential causal link between abnormal striatal D₂R⁺ neuronal activity and LID, we selectively manipulated the neuronal activity of striatal D₂R⁺ hChR2-expressing neurons in LID rats and observed the behavioral changes. It has been reported that chemogenetic activation of striatal Adora2a-expressing neurons by systematic CNO treatment inhibits general locomotion and dyskinesia in PD mice [21]. It should be noted that CNO is not an ideal designer drug for chemogenetics. Studies suggested that CNO itself can inhibit locomotion in non-DREADD (designer receptors exclusively activated by designer drugs)-expressing rodents [48, 49]. In contrast to chemogenetics, optogenetics can selectively manipulate the activity of ChR2-expressing neurons [50]. In our present study, the acute optogenetic activation of striatal D₂R⁺ neurons effectively inhibited most of the dyskinetic symptoms in LID rats after L-dopa injection (Fig. 3C–G). Consistent with this, continuous light activation immediately after the administration of L-dopa significantly inhibited the total AIMS and the parameters of LID (Fig. 3H–L). In addition, optogenetic activation with the same parameters did not significantly affect the general locomotion of rats (Fig. S1). Our findings indicate that hypoactive striatal D₂R⁺ neurons are involved in the pathophysiology of LID and optogenetic activation of these hypoactive neurons may represent an effective therapeutic strategy for ameliorating LID. It should be noted that striatal D₂R⁺ neuronal dysfunction has been implicated in the pathophysiology of paroxysmal dyskinesia [20]. This suggests that the hypoactivity in the indirect striatal pathway may be one of the key mechanisms for the generation of dyskinesia in many disorders. Meanwhile, there is plenty of evidence that the hyperactivity of striatal D₁R⁺ neurons contributes to dyskinesia [14, 15]. We propose that dyskinesia is driven by an imbalance in the activity of striatal D₁R⁺ and D₂R⁺ neurons.

All the above findings demonstrate that aberrant hypoactivity of striatal D₂R⁺ neurons encodes dyskinetic signals and is necessary for the expression of LID. The striatal D₂R⁺ neurons are embedded in complex circuitry and their activity is regulated by afferents from other brain regions. Our findings that D₂R⁺ neurons in the dorsolateral striatum receive inputs from the cerebral cortex, ventral striatum, parafascicular nucleus of the thalamus, subthalamic nucleus, and dopaminergic neurons in the SNc suggest that these pathways might also be involved in LID (Fig. 4B). This is corroborated by several studies that show that the neuronal activity of the somatosensory cortex, subthalamic nucleus, and parafascicular nucleus is altered in PD and LID [51–53]. Further experiments are needed to clarify the role of these upstream regions in the pathogenesis of LID.

As shown in our anterograde tracing assay (Fig. 4C), many of the striatal D₂-Cre⁺ neurons in the study were GPe-projecting SPNs. Because we used D₂-Cre rats [23], the striatal D₂R⁺ neurons in our study possibly contained the majority of D₂R⁺ SPNs and a few D₂R⁺ striatal interneurons [54]. However, the potential role of D₂R⁺ striatal interneurons cannot be completely ignored. To demonstrate the key role of GPe-projecting D₂R⁺ SPNs in the pathophysiology of LID, we tested whether the stimulation of GPe-projecting D₂R⁺ SPNs could inhibit the dyskinetic behavior of LID rats. The application of optogenetics showed that activation of the axon terminals in the GPe projecting from the striatal D₂R⁺ neurons significantly decreased the total AIMS score of LID rats (Fig. 6B). In addition, the fiber photometry assay indicated that the GPe neurons of LID rats were hyperactive during the peak phase of dyskinesia (Fig. 5B–I). These results suggested that the striatal D₂R⁺ SPNs and their downstream GPe neurons are involved in the pathophysiology of LID. The GPe neurons project to many basal ganglia nuclei including the STN and the striatum [11, 55]. The specific functional impact of these GPe projections in LID remains unknown. Further experiments are necessary to determine the role of these downstream projections in the pathogenesis of LID.

Overall, we showed that aberrant hypoactivity of the GPe-projecting striatal D₂R⁺ neuronal population and hyperactivity of downstream GPe neurons are decisive mechanisms mediating the dyskinetic symptoms upon L-dopa administration in LID rats, suggesting that these neurons could serve as an alternative target for managing LID in the future.

Supplementary Information

Below is the link to the electronic supplementary material.

Data Availability

The raw data supporting the conclusions of this article will be made available by the corresponding author, without undue reservation.

Conflict of interest

The authors declare no conflict of interest.