Electro-acupuncture alleviates motor decits and maladaptive striatal plasticity in partial-lesioned parkinsonian mice

Parkinson's disease is characterized by abnormal synaptic transmission in the corticostriatal circuit that leads to decits in motor abilities. Electro-acupuncture has shown to improve the motor behaviors in parkinsonian models. However, the potential mechanisms underlying the electro-acupuncture treatment, specically in the partial-lesioned model, remain unclear. Methods By utilizing multiple approaches, including electrophysiological, immunohistochemistrical, molecular and behavioral methods, we assessed the effect of electro-acupuncture on the motor dysfunction and striatal synaptic plasticity in a partial-lesioned mouse model induced by intrastriatal injection of 6-hydroxydopamine. NaHCO 3 26, CaCl MgSO recording. Whole-cell patch clamp striatal paired-pulse ratio Fisher Scientic, USA). RNA extracts were immediately subjected to reverse transcription using FastQuant RT Kit (TIANGEN, China). The qPCR assays for gene expression analysis were performed by adding 0.5 µl of cDNA sample, 5 µl Power SYBR Green PCR Master Mix (Thermo Fisher Scientic, USA), 0.5 µl each primer and RNase free water to a 10 µl total volume. The reaction was initiated with activation of Taq polymerase by heating at 95 °C during 2 min followed by 40 cycles of a 15 s denaturation step at 95 °C and a 15 s annealing and elongation step at 60 °C. The uorescence was measured after the extension step by the QuantStudio 5 Real-time PCR system (Thermo Fisher Scientic, USA). After the thermocycling reaction, a melting curve was performed with slow heating, starting at 55 °C and with a rate of 0.5 °C per 10 s, up to 95 °C. The assay included a non-template control (sample was substituted by RNase- Dnase-free sterile water). All reactions ran in triplicates.


Background
The key pathophysiologic characteristics of Parkinson's disease (PD) are the degeneration of dopaminergic neurons in the substantia nigra pars compacta (SNc) and the denervation of dopaminergic projection in the striatum [1]. Motor symptoms of PD appear after 50% of SNc dopaminergic neurons are degenerated [2,3] and 70 ~ 80% of nigrostriatal dopaminergic innervation is lost [1,4]. So far, neither clinical medication such as L-3,4-dihydroxyphenylalanine nor surgical therapy such as deep brain stimulation can effectively cure or stop the development of these motor symptoms in the advanced stage [5]. Therefore, the optimal therapeutic strategy aims to delay the pathological deterioration and alleviate motor symptoms in the early period of PD.
The neurotoxin 6-hydroxydopamine (6-OHDA) has been used extensively to induce animal models of PD [6]. Typically, rodents develop severe movement de ciencies and damage of the nigrostriatal dopamine (DA) system after 6-OHDA injection in the medial forebrain bundle (MFB) or SNc. In addition, unilateral injection of 6-OHDA into the striatum, especially in a low dose, would induce the incomplete depletion of striatal DA [7][8][9]. As such, the 6-OHDA lesion in the striatum can be utilized to explore the early pathogenic events of PD. Targeting the early events is essential for the development of therapeutic interventions, although the neuroanatomical, functional, and molecular mechanisms underlying the motor disorders in the early stages of PD remain largely elusive.
Loss of dopaminergic innervation in the striatum leads to the de cits in motor skills and triggers aberrant striatal synaptic plasticity [7,[10][11][12][13]. DA depletion at different extents have been shown to differentially affect striatal synaptic plasticity, subsequently leading to PD motor symptoms of various severity [7,14,15]. A complete DA denervation triggers maladaptive synaptic plasticity including loss of both long-term potentiation (LTP) and long-term depression (LTD) at corticostriatal synapses [7,16,17] and induces severe motor behaviors. However, denervation in a partial lesion rat model of PD altered LTP but not LTD, accompanied by mild motor symptoms [7]. Apparently, striatal synaptic plasticity is closely linked to motor performances [7,18]. Intervening striatal synaptic plasticity, especially in the early stage of PD, may thus be crucial for producing signi cant improvement of motor symptoms.
As an alternative and complementary therapy, acupuncture or electro-acupuncture (EA) has long been used to alleviate the symptoms of patients with PD and improve their quality of life [19][20][21]. Our studies have consistently demonstrated that high-frequency EA stimulation (100 Hz) alleviated movement disorders in multiple parkinsonian models [22][23][24][25]. However, whether EA affects motor skills or striatal plasticity in the early stage of PD has not been elucidated. In the present study, we used a partial-lesioned 6-OHDA model to analyze the changes of synaptic plasticity in the early stage of DA depletion. We rst investigated the effect of EA treatment on motor skills in the early stage of PD. We then examined whether the therapeutic effect of EA was based on the modulatory role of EA in striatal synaptic plasticity. Finally, we determined the biochemical targets sensitive to EA and their roles in processing the effect of EA on the aberrant striatal synaptic plasticity in partial-lesioned mice.

Animals
Male C57BL/6J mice, aged 6-8 weeks (body weight 20-25 g) were provided by the animal facility of Capital Medical University, Beijing, China. Mice were housed in groups of 5 mice/cage on a 12/12 h light/dark cycle and with ad libitum access to food and water throughout, except for mice during food deprivation for staircase test. All experimental procedures were approved by the Animal Ethics Committee of Capital Medical University and carried out in accordance with National Institutes of Health Guide for the Care and Use of Laboratory Animals. All efforts were made to reduce the number of animals to be used for the study and to minimize their suffering.

Experimental design
The timeline for experimental treatment is present in Fig. 1a. Brie y, mice were randomly divided into two groups: sham (saline-injection) group and 6-OHDA-lesioned group. Two weeks after 6-OHDA injection, mice were tested with apomorphine (APO) and those who performed more than 5 turns per minute were chosen as parkinsonian model mice, which were further randomly divided into two groups, named 6-OHDA and 6-OHDA + EA group, respectively. During the following four weeks, the mice of 6-OHDA + EA group received EA treatment. The effect of EA on motor performance and skills were tested at 3 different time points: before EA, 2nd and 4th weeks after EA stimulation. After 4-week EA treatment, mice were sacri ced for different biological tests.

Unilateral 6-OHDA lesion
Mice were anaesthetized by an intraperitoneal (i.p.) injection of pentobarbital sodium (30 mg/kg) and placed into a stereotaxic apparatus (Kopf, Germany). 6-OHDA (H116, Sigma) was dissolved in normal saline containing 0.02% ascorbic acid and injected in two deposits in the striatum at the following coordinates: AP = + 0.5 mm anterior to bregma; ML = -2.0 mm lateral to bregma, DV = -3.0 mm and then DV = -2.0 mm ventral to dura. At each site, 6-OHDA (4 µg) was injected in a total volume of 0.78 µl (5.14 µg/µl) at a rate of 0.26 µl/min with a 10 µl microsyringe. The microsyringe was left for additional 2 min for fully diffusion of drug before being slowly removed and the wound was cleaned and sutured. Mice in sham group underwent the same surgery protocol but received the same volume of normal saline solution containing 0.02% ascorbic acid. Animals were allowed 2 weeks for recovery before behavioral testing commenced. Animal health and behavior were monitored daily following the surgery.

EA treatment
EA stimulation was performed as described previously [24,26]. In brief, hindlimbs were cleaned with 75% alcohol and then each leg was inserted two sterilized stainless-steel needles (0.18 mm diameter × 3 mm length) at two acupoints: one is Zusanli (ST36, 2 mm lateral to the anterior tubercle of tibia) and the other is Sanyinjiao (SP6, 2 mm proximal to the upper border of medial malleolus, at the posterior border of the tibia) (Fig. 1b). Bidirectional square wave electrical pulses (0.2 ms duration, 100 Hz), generated from a Han's acupoint nerve stimulator (HANS, Neuroscience Research Institute, Peking University), were connected to the stainless-steel needles in both hindlimbs simultaneously and were administered for a total of 30 min per day, 5 days per week, during which the intensity of the stimulation was stepwise increased from 1.0 mA to 1.2 mA and then to 1.4 mA, and each step lasted for 10 min. The duration of EA treatment was limited to 4 weeks. To evaluate the neuronal activation in the striatum for the acute EA stimulation, c-fos expression was detected after sham or model mice received an acute 30-min EA treatment. Following a 90 min recovery period, mice were sacri ced for histological detection.
2.5. Behavioral tests 2.5.1. Apomorphine-induced rotation Two weeks after surgery, mice were individually placed in cylinders with the diameter of 90 mm and height of 120 mm, in a closed room to avoid any environmental disturbance, and allowed to habituate for 5 min before APO (0.05 mg/kg, Sigma) injection. APO dissolved in normal saline containing 0.02% ascorbic acid was delivered i.p.. Mice were monitored for 30 min, during which the highest 3 consecutive minutes of contralateral minus ipsilateral rotations were used for analysis. Mice with the net number of contralateral rotations > 5 turns/min were validated as successful model.

Rotarod
Before the surgery, all mice were trained on rotarod (Ugo Basil, Italy) for ve consecutive days, in order to reach a stable performance on the constant and accelerating rod. On the test day, mice were placed on the rotarod with an initial rotation speed at 4 rotations/min (rpm), which was accelerated to 60 rpm within 5 min. The time taken to fall was automatically recorded when mice landing on the base of the apparatus. Each mouse was given three trials and the averaged latency time was recorded.

Open eld test
Mice were placed individually in the center of an open-eld arena (50 × 50 cm). The behaviors of the mice were recorded using a CCD camera positioned above the cage for 30 min. The total movement distance was recorded and analyzed simultaneously (SuperMaze, Shanghai Xinruan). All experiments were performed during 9:00 am to 16:00 pm. The environment was kept dark and quiet during the entire procedure. The apparatus was carefully cleaned with 75% alcohol and water between each trail.

Staircase test
Staircase apparatus (Campden Instruments, UK) was used to measure the coordinated grasping skill in mice following unilateral lesions. As described previously [27], mice were encountered a 20 h food deprivation regime during the training and testing period to increase their motivation to retrieve the pellets. They were given food access for 4 h daily immediately after each session. Mice were rst familiarized with the food pellets by placing approximately 50 into each home cage on three consecutive days. They were then familiarized to the test box by placing food pellets along the surface of the central trough as well as on the staircase steps for a further day. On the subsequent three days, the double staircase was baited with two pellets per step, i.e. 16 on each side of 8 steps and the mouse was placed in the start compartment. Each session lasted for 15 min daily and only the 3rd day was considered for measurement. At the end of each session, the number of pellets retrieved (16 minus pellets remained) in 15 min were counted on both sides.
After the four-week EA treatment, mice were deeply anaesthetized with pentobarbital sodium (30 mg/kg) and intracardially perfused with 0.9% sodium saline followed by ice-cold 4% paraformaldehyde (PFA).
Brains were removed, post-xed in 4% PFA for 24 h and then transferred to 20% and 30% sucrose in PBS for tissue cryoprotection, consecutively. Coronal sections were cryosectioned at 40 µm thickness. Sections were collected and stored at 4 °C in an antifreeze solution.
For immunohistochemistry, free-oating sections were rinsed with 0.01M PBS before incubated with 0.3% Triton X-100. Endogenous peroxidase activity was then quenched with 3% H 2 O 2 . After washing, sections were blocked in normal horse serum obtained from ABC-peroxidase kit (PK-4002, Vector Laboratories).
They were then incubated with anti-tyrosine hydroxylase antibody (TH, 1:2000, SAB4200699, Sigma) or anti-c-Fos antibody (1:1000, ABE457, Millipore) diluted in the same blocking solution as described above, overnight at 4 °C. Sections were then incubated with biotinylated anti-mouse secondary antibody, before treated with avidin-biotin peroxidase complex reagent from the ABC-peroxidase kit. Following washing, the reaction product was developed using 3,3′-diaminobenzadine (DAB kit, Zhongshan Golden Bridge, China). Tissue sections were mounted, dehydrated in graded ethanol dilutions, cleared in xylene and cover slipped with mounting medium.
Slices collected after patch-clamp recordings were xed with 4% PFA overnight in 4 °C and stained with Alexa Fluor-488-conjugated avidin (A21370, Invitrogen). The z-stack images of individual cells (0.5 µm between successive images) and 3-4 dendrites per cell (0.3 µm between successive images) were acquired by a confocal microscope (Leica SP8, Germany) equipped with a 63 × objective. Threedimension reconstruction of the cells was performed using Imaris 8.0 (Bitplane, Switzerland) to form a continuous 3D representation of the entire cell structure.

Histological quanti cation
A total of 4-6 consecutive sections were selected from each brain to examine TH immunoreactive cells in SNc and TH positive bers in the striatum. Unbiased stereology was used to estimate the number of dopaminergic neurons of each section under a 20 × objective with Stereo Investigator software (MBF Bioscience, USA). The relative remained number of TH neurons was calculated as a ratio of the number of the lesioned side relative to the unlesioned side. Images of TH-stained striatal sections were obtained by light microscope (Olympus BX51, Japan). Immunoreactive optical densities of TH positive bers in the striatum were calculated using Image-Pro Plus 6.0 (Media Cybernetics, USA). 3-4 mice from each group were used for quanti cation. Additionally, consecutive sections stained for c-fos of each mouse were scanned with Pannoramic scan (3DHISTECH, Hungary). In the ipsilateral dorsolateral striatum, the number of c-fos positive cells were counted in a 0.5 mm × 0.5 mm region of interest using Image-Pro Plus 6.0 (Media Cybernetics, USA). The c-Fos counts of each mouse were represented by average counting per section.

High performance liquid chromatography (HPLC)
Tissue levels of DA and Glutamate were determined as previously reported [28]. The striatum was rapidly dissected, frozen in liquid nitrogen, and stored at -80 °C. Striatal tissue was weighed and homogenized in bath for 60 min. Supernatant was mixed with mobile phase solution and was kept from light for 60-minice-bath, followed by centrifugation for 20 min at 12,000 rpm at 4 °C. The resultant supernatant was ltered with a 0.22 mm membrane. The content of DA was detected by injecting an aliquot of the resulting solution into the HPLC-ECD pump (CoulArray, USA) with four potentials of -150, 100, 220 and 400 mV and ow rate at 1 ml/min. Chromatographic separation was performed using a HR-C18 reversephase column (80 × 4.6 mm I.D., 3 µm, 100A, ESA Inc., USA). The mobile phase (pH = 4.3) contains 63.5 mM citric acid monohydrate, 60.9 mM trisodium citrate dihydrate, 0.1 mM EDTA, 0.5 mM 1octanesulfonic acid sodium salt, and 8% methanol. The glutamate quanti cation was later carried out in the HPLC-electrochemiluminescence system. Each concentration was adjusted with respect to the standard and quanti ed from a standard curve. evoked by a stimulating bipolar electrode (FHC, USA) placed on the corpus callosum between the cortex and the dorsal striatum. AMPA and NMDA currents were measured at -70 mV and + 40 mV, respectively. 10 sweeps were recorded at both holding potentials separated by time intervals of 15 s. The traces were analyzed and averaged o ine using Patchmater (HEKA, Germany). The current peak at + 70 mV was extracted as the AMPA component and the decay of the AMPA current was used to establish the time window for the measurement of the NMDA component. A 10 ms time window starting 50 ms after the stimulation was used at + 40 mV to measure the NMDA current. Recordings were ampli ed and digitized (20 kHz) using HEKA EPC 10 USB. Patch pipettes were pulled with a micropipette puller (Narishige PC-10, Japan) and had initial resistances of 6-8 MΩ for current clamp and 4-6 MΩ for voltage clamp experiments. Liquid junction potential was not corrected during the experiment. Recordings were performed with pipette capacitance and access resistance compensated throughout the experiment. Data were discarded when access resistance increased beyond 30 MΩ.

Extracellular multichannel electrophysiology
The process of preparing acute brain slices were similar to those described previously [29,30]. Mice were sacri ced by decapitation after anesthesia with pentobarbital sodium (30 mg/kg). Subsequently, the whole brain was rapidly removed and immediately soaked in NMDG cutting solution containing (in mM) NMDG 92, KCl 2.5, NaH 2 PO 4 1.25, NaHCO 3 30, HEPES 20, glucose 25, thiourea 2, Na-ascorbate 5, Napyruvate 3, CaCl 2 0.5 and MgSO 4 10. With the portions containing the midbrain and cerebellum being trimmed, the remaining brain block containing the striatum was placed on the ice-cold stage of a vibrating tissue slicer (Dosaka, DTK-1000, Japan). The stage was immediately lled with oxygenated NMDG cutting solution. The thickness of each tissue slice was set at 300 µm. Each slice was gently transferred with a homemade pipet into a holding chamber containing oxygenated NMDG cutting solution, allowing the initial protective recovery to proceed for 12 min at 32-34 °C. After the initial recovery period, the slices were transferred into another holding chamber containing room-temperature HEPES holding solution consist of (in mM) NaCl 92, KCl 2.5, NaH 2 PO 4 1.25, NaHCO 3 30, HEPES 20, glucose 25, thiourea 2, Na-ascorbate 5, Na-pyruvate 3, CaCl 2 2 and MgSO 4 2 until transferred to recording dish contained ACSF consist of (in mM) NaCl 119, KCl 2.5, NaH 2 PO 4 1.25, NaHCO 3 24, glucose 12.5, CaCl 2 2 and MgSO 4 2. Mg 2+ -free ACSF was prepared for LTP recording procedures.
Recordings were carried out in a multielectrode dish (Panasonic, MED 64 planar microelectrodes). After a 15-min adaptation of the slice, one of the 64 available planar microelectrodes (located in the corpus callosum), which produced the highest amplitude was selected for stimulation. Field potentials evoked at the remaining sites (located in the dorsolateral striatum) were ampli ed by the 64-channel main ampli er and then digitized at a 20 kHz sampling rate. An input-output curve was rst determined for each slice via the measurements of fEPSP amplitude in response to a series of stimulation intensities starting at 10 µA. The intensity of the test stimulus was then adjusted to elicit 30-50% of the maximum amplitude. A baseline was then recorded for additional 15 minutes once every 1 minute. High-frequency stimulation (HFS) consisted of three bursts, each containing four pulses at 100 Hz with an inter-burst interval of 200 ms (3 × 3 HFS), both LTD and LTP were induced by HFS except Mg 2+ was omitted from the ACSF when induced LTP [31,32]. After 3 × 3 HFS, the test stimulus was repeatedly delivered for 2 h for observations of any changes in LTP or LTD magnitude. Traces were obtained and analyzed using Mobius (Alpha Med Science Inc, Osaka, Japan). 6-Cyano-7-nitroquinoxaline-2,3-dione (CNQX, 2 µM, Sigma), (s)-α-Amino-3-hydroxy-5-methylisoxazole-4-propionic acid ((s)-AMPA, 10 µM, Sigma), 5,7-Dichlorokynurenic acid (DCKA, 10 µM, Sigma), glycine (150 µM, Sigma) were bath applied to slices of separate groups to observe their roles on plasticity of the striatum.
Immunoreactive bands were visualized using Odyssey imaging system (LI-COR Biosciences). Optical densities of protein bands were normalized to the density of actin bands visualized on the same membrane.

qRT-PCR
The brains of mice were removed after decapitation. The striatum was quickly dissected and frozen in liquid nitrogen. Total RNA was isolated using TRIzol reagent (Thermo Fisher Scienti c, USA) according to the manufacturer's protocol. Concentrations and purity of isolated DNA and RNA were determined with NanoDrop 2000c spectrophotometer (Thermo Fisher Scienti c, USA). RNA extracts were immediately subjected to reverse transcription using FastQuant RT Kit (TIANGEN, China).
The qPCR assays for gene expression analysis were performed by adding 0.5 µl of cDNA sample, 5 µl Power SYBR Green PCR Master Mix (Thermo Fisher Scienti c, USA), 0.5 µl each primer and RNase free water to a 10 µl total volume. The reaction was initiated with activation of Taq polymerase by heating at 95 °C during 2 min followed by 40 cycles of a 15 s denaturation step at 95 °C and a 15 s annealing and elongation step at 60 °C. The uorescence was measured after the extension step by the QuantStudio 5 Real-time PCR system (Thermo Fisher Scienti c, USA). After the thermocycling reaction, a melting curve was performed with slow heating, starting at 55 °C and with a rate of 0.5 °C per 10 s, up to 95 °C. The assay included a non-template control (sample was substituted by RNase-Dnase-free sterile water). All reactions ran in triplicates.

EA ameliorated motor skill de cits in partial-lesioned mice
The present experiment, unilateral 6-OHDA injection in the striatum induced a partial reduction of TH positive bers in the ipsilateral striatum (Fig. 2a, b; P < 0.0001) and TH immunoreactive neurons in the ipsilateral SNc (Fig. 2a, v; P = 0.013). Western blotting revealed that TH expression was decreased in 6-OHDA mice compared with sham mice (Fig. 2d, P < 0.0001). We also examined DA concentration in the striatum using HPLC. The level of DA was decreased in 6-OHDA-treated mice (Fig. 2e, P = 0.008). These data validate that 6-OHDA-treated mice are conformed to the standards of a partial-lesioned parkinsonism model, which mimics the early PD pathology. EA treatment for 4 weeks increased TH immunoreactive terminals (Fig. 2a, b; P = 0.002) and TH protein expression (Fig. 2d, P = 0.001) in the striatum. However, EA had no signi cant in uences over the dopaminergic neuronal degeneration in the SNc (Fig. 2a, c), the release of DA or DA metabolites in the striatum (Fig. 2e, Fig. S2a-c), and the transcription levels of DA receptors (Fig. S5). Thus, it seems that EA treatment exerted region-dependent neuroprotection of the dopaminergic innervation in the local striatum, but not dopaminergic neurons in the SNc. In addition, the impact of EA stimulation on the striatal neuronal response was detected in this experiment. The expression of c-fos, a marker of neuronal activation, obviously elevated in the ipsilateral dorsolateral striatum in both sham (Fig. S1, P = 0.034) and model group (Fig. S1, P = 0.005), indicating the activation of striatal neurons by an acute EA stimulation.
After 4-week EA treatment, we found that both the total movement distance in open eld test ( Fig. 2f; week 4, P = 0.009) and the latency time in rotarod test ( Fig. 2g; week 4, P = 0.038) were signi cantly increased as compared with 6-OHDA mice that did not receive EA. Thus, EA alleviated the gross motor behaviors, including ambulatory activity and coordinated balance ability, in the partial-lesioned mice. We also performed staircase test to detect the grasping ability of ipsilateral and contralateral forepaws of mice, which has been established to measure ne motor skills [33]. Although the number of pellets retrieved by ipsilateral forepaws remained relatively stable among three groups (Fig. 2i), the number of pellets retrieved by contralateral forepaws in EA-treated 6-OHDA mice were signi cantly lower than that seen in sham mice before and 2 weeks after EA treatment (Fig. 2h). However, EA obviously improved the grasping performance in 6-OHDA mice at the end of 4th week. Thus, EA treatment alleviated both the gross and ne motor skills in the partial-lesioned 6-OHDA mice, which likely involves the modulatory effect of EA on target(s) within the local striatum.

EA did not alter structural plasticity of striatal MSNs.
Due to the fact that 4-week EA interference was a chronic treatment, we set forth to determine whether chronic EA treatment modi ed striatal neuronal morphology. We traced striatal MSNs and their dendrites using biocytin-lled pipette solution during patch-clamp recordings to detect cellular morphology and complexity of these cells (Fig. 3a-c). Although the Sholl analysis displayed that dendrites 56-89 µm from soma exhibited less intersections in 6-OHDA mice compared to sham mice (Fig. 3d), the total branch points (Fig. 3e) and terminal points (Fig. 3f) showed no alterations among three groups. We also detected the dendrites of MSNs and there was neither difference in the density (Fig. 3g) nor morphology ( Fig. S3af) of MSN spines. Taken together, in partial-lesioned mice, the complexity of dendritic arborization of MSNs and striatal synapses remain intact, while EA did not alter the structural plasticity.

EA rescued striatal LTP in partial-lesioned mice
We next set out to detect functional alterations of synaptic plasticity in the corticostriatal pathway. To this end, we used a microelectrode array system to record eld EPSPs in the striatum (Fig. S4). The inputoutput curve (Fig. 4a) and PPR (Fig. 4b) were similar among three groups surveyed, indicating that basal synaptic transmission and presynaptic release were intact in 6-OHDA mice with or without EA treatment. The intact LTD (Fig. 4c) in 6-OHDA-treated mice is consistent with the result from the previous report [7], which established the phenotype of the partial DA denervation model of PD. Notably, LTP could not be induced in the striatum of 6-OHDA mice ( Fig. 4d; P = 0.019), whereas it was readily induced in 6-OHDA + EA group ( Fig. 4d; P = 0.012). Another type of LTP known as chemical LTP is induced by glycine. Similar to the LTP induced by high frequency electrical stimulation, glycine induced a complete LTP in sham group and the blunted LTP in 6-OHDA lesioned group (Fig. 4e, P < 0.0001). Interestingly, this glycinemediated blunted LTP was effectively reversed by EA treatment (Fig. 4e, P < 0.0001). Additionally, we did not observe changes of glutamate concentrations in the striatum after either 6-OHDA lesion or EA treatment (Fig. 5f). Overall, these data imply that EA treatment selectively rescued the disrupted LTP, likely via a mechanism involving the postsynaptic rather than presynaptic elements, in our partial-lesioned model of PD.

EA rescued AMPA and NMDA receptor-mediated synaptic activities in striatal MSNs
To examine the effect of EA on glutamate receptor activity in striatal MSNs, we performed patch-clamp recordings of MSN that labeled with biocytin, targeting corticostriatal synaptic circuity in the dorsolateral striatum. AMPA receptor-mediated mEPSCs were recorded at -70 mV under whole-cell voltage clamp (Fig. 5a). The amplitude of mEPSCs in 6-OHDA mice were decreased compared to sham mice (Fig. 5b, P = 0.008). This decrease was rescued in EA-treated mice ( Fig. 5b; P = 0.001). In contrast, the mEPSC frequency was not altered among three groups (Fig. 5c), indicating that presynaptic neurotransmitter release was intact. These results together suggest that the partial dopaminergic lesion impairs synaptic transmission at the postsynaptic but not presynaptic site. In addition, we observed no signi cant changes in the AMPA/NMDA ratio in the dorsolateral striatal MSNs following 6-OHDA intervene alone or 6-OHDA + EA treatment (Fig. 5d, e), indicating that both AMPA-and NMDA-mediated EPSCs were altered by 6-OHDA or 6-OHDA + EA to a similar extent.

Antagonists of AMPA and NMDA receptors inhibited the effects of EA on striatal LTP
We further clari ed the role of AMPA and NMDA receptors in processing the effect of EA on LTP, in the presence of their antagonists. The AMPA receptor antagonist CNQX was bath-perfused during LTP recordings in sham or EA-treated mice. We observed that LTP, albeit present, was signi cantly decreased in its amplitude compared to that in the control ACSF perfusate (Fig. 4d, 6 k), showing that blockade of AMPA receptors inhibited LTP in slices from sham and EA-treated mice. On the contrary, (s)-AMPA, an agonist of AMPA receptors, signi cantly reversed the reduction of LTP in 6-OHDA mice (Fig. 6k). These evidences support that AMPA receptor activity is required for the effect of EA on LTP. Similarly, the role of NMDA receptors on LTP was observed. Activation of NMDA receptors with glycine signi cantly restored the diminished LTP in 6-OHDA mice (Fig. 6l). DCKA, an NMDA receptor antagonist, decreased the fEPSC amplitude in the slices from both sham and EA-treated mice (Fig. 6l). These results substantiate the notion that both AMPA and NMDA receptors participate in the effectiveness of EA in restoring synaptic plasticity in the striatum of partial-lesioned mice.

Discussion
In contrast to the 6-OHDA injection into the MFB which leads to motor impairments resembling the advanced phase of PD, the injection of 6-OHDA into the striatum mimics the early phase of PD as a result of partial damage of the local dopaminergic system. In the present study, 6-OHDA was stereotaxically injected into two sites of the right striatum, which resulted in degeneration of 42% nigral dopaminergic cells and loss of 64% of striatal bers. Meanwhile, a moderate DA depletion (70%) was found in the striatum of 6-OHDA mice, indicating a partial loss of DA contents in PD mice [7]. In a different partial model generated by intranigral administration of the proteasome inhibitor lactacystin, a moderate impairment of motor coordination occurred in the rotarod performance [14]. In addition, unilateral injection of 6-OHDA into the MFB at a low dose (3 µg) caused mild motor de cits, including forelimb akinesia in the stepping test [8]. Intracerebroventrical infusion of 6-OHDA induced forelimb grasping inability in mice, which was assessed as an earlier motor sign and accepted as bradykinesia in PD [34]. Consistent with these ndings, our partially lesioned mice with 6-OHDA showed impaired motor skills, including reduced movement distance and shortened latency time. Besides these gross motor defects, the ne motor impairment such as forelimb grasping activities was reduced as well. Given the good correlation between motor phenotypes and degree of DA denervation [8], behavioral and biochemical changes we observed indicate a partial lesion in the striatum, which largely mimics an early stage of PD.
Evidences revealed that acupuncture or EA could ameliorate motor symptoms and reduce dopaminergic neuronal degeneration in PD models [35][36][37][38]. In our present experiment, chronic EA treatment ameliorated gross motor signs (ambulatory activity and coordinated balance ability) and ne forelimb motor performance in a partial-lesioned model. In details, the partial lesion in the striatum was su cient to trigger de cits of motor skills, such as the reduction of movement and the slow execution of movements [34], cardinal features of PD observed during the early stage. EA mitigated the striatal lesion and behavioral de cits in a closely correlated manner. However, our results suggest that EA did alleviate motor skills but not prevent the degeneration of nigral dopaminergic neurons (Fig. 2a, c). Contrasting with application of EA immediately after toxic administration [35] [36] [37], we delivered EA two weeks after 6-OHDA lesion. Thus, it is speculated that chronic 6-OHDA lesion induced irreversible neuronal death in the SNc which could not be reversed by subsequent EA treatment [39]. Although the different acupoints were utilized in other studies [35,37], EA performed at these acupoints (ST36 and SP6) is known to effectively relieve motor symptoms in PD mice [24,26]. Thus, EA achieves its motor effect in the early stage of the disease by modulating the striatal function without restoring the loss of nigral dopaminergic neurons.
In the DA-denervated model of PD, morphological ndings showed that spine density is signi cantly reduced in striatal MSNs [40]. However, in the partial PD model induced with lactacystin, an insigni cant change in the density of postsynaptic spine was seen, while a signi cant increase in both the length and area of postsynaptic density occurred at corticostriatal synapses [14]. Similarly, we observed a slight but not signi cant reduction in the density of postsynaptic spines in the partial-lesioned 6-OHDA model.
Further studies need to determine whether morphological alterations act as a compensatory mechanism to maintain motor function under conditions of partial dopamine depletion.
The de cits in functional synaptic plasticity may precede the dendrite atrophy during the early period of DA depletion. In fact, an early study found that the complete depletion of striatal DA led to the loss of both LTP and LTD at corticostriatal synapses, while an incomplete DA denervation altered the LTP but not LTD at these synapses [7]. Other studies with transgenic mice also revealed early changes in corticostriatal plasticity [41]. In the present study, EA did not signi cantly improve morphological plasticity or LTD, while it reversed the de cits in LTP and meanwhile improved motor skills in the partiallesioned model. Therefore, the functional speci c modulatory effects of EA on striatal synaptic plasticity, especially LTP, could serve as a synaptic mechanism underlying the EA-induced improvement of motor skills. This evidence demonstrated that the new role of EA is to modulate the functional strength of corticostriatal synapses in the partial-lesioned PD model.
Synaptic plasticity in the striatum is regulated by the interaction between two major local transmitters, DA and glutamate [42,43]. Garcia-Munoz et al. have postulated that DA presynaptically inhibits the release of glutamate from corticostriatal terminals in the striatum [42]. As a result, a complete denervation of DA causes an increase in local glutamate release, which may represent an important physiological feature of PD and underlie motor de cits in the parkinsonian state. However, in the partially DA-depleted striatum, no changes were found in basal extracellular levels of glutamate in 6-OHDA treated rats [44]. Consistent with this observation, no signi cant change in glutamate levels was observed in the partial DA-depleted striatum in our experiments. Together, these ndings suggest that after partial loss of DA, a lower but signi cant level of DA retained in the striatum, which may su ciently carry out its function in inhibiting local glutamate release.
Thus, it is interesting to investigate whether any changes in glutamate receptors occur in the partial DAdepleted striatum in light of the lack of changes in glutamate release. A possible assumption is that the maladaptive response of glutamate receptors participated in the compensatory mechanisms underlying the early symptoms of PD. LTP is NMDA receptor-dependent in the striatum, while LTD is not [45]. NMDA receptors are abundant in the striatum, consisting of GluN1, GluN2A and GluN2B subunits. We found that in the partial DA-depleted striatum, expression of GluN1 subunits was decreased, while GluN2A and GluN2B subunits were unchanged. Similar reduction of GluN1 expression was previously reported in both 6-OHDA-treated rats and MPTP-treated monkeys [46,47]. Perhaps the loss of GluN1 subunits causes downregulation of NMDA receptor activity, leading to the LTP failure and the nal motor skills de cits.
However, Paille et al. reported that GluN2A and GluN2B subunits were increased and decreased, respectively, to produce mild motor symptoms in a partial lesion rat model of PD [7]. This underscores the complex of roles of different NMDA subunits in constructing synaptic plasticity and motor activity in response to different models of PD.
In addition to NMDA receptors, we monitored AMPA receptor activity by recording whole-cell mEPSCs. We found no alteration in the frequency of mEPSCs, whereas the magnitude of mEPSCs was decreased in 6-OHDA-lesioned mice. It implies that the number of AMPA receptors was downregulated in the postsynaptic MSNs. In support of this, neurochemical assays revealed that GluA1 expression was reduced in the striatum of PD mice [48]. It has been demonstrated that incorporation of GluA1-containing AMPA receptors into synaptic membranes is a crucial mechanism for LTP and the reduction of GluA1 is thus bound to weaken LTP [49]. In the present work, the reduction of mRNA and protein expression of GluN1 and GluA1 in PD mice was reversed by EA treatment. Meanwhile, the disrupted LTP in PD mice was recovered following EA treatment. Furthermore, AMPA and NMDA receptor antagonists blocked the effect of EA on LTP. These data support a notion that GluN1 and GluA1 subunits are key substrates of EA for normalizing the corticostriatal LTP in a partial-lesioned model of PD.

Conclusions
In summary, we uncovered that chronic EA administration improved the motor skills in a partial-lesioned model of PD. EA has been shown to affect the glutamate receptor-dependent striatal synaptic plasticity

Availability of data and materials
Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.

Competing interests
The authors declare that they have no competing interests.  Two-way ANOVA analysis followed by Tukey post-test for the behavior tests. *P < 0.05, **P < 0.01 vs.