6‐Hydroxydopamine lesion and levodopa treatment modify the effect of buspirone in the substantia nigra pars reticulata

Background and Purpose l‐DOPA‐induced dyskinesia (LID) is considered a major complication in the treatment of Parkinson's disease (PD). Buspirone (5‐HT1A partial agonist) have shown promising results in the treatment of PD and LID, however no 5‐HT‐based treatment has been approved in PD. The present study was aimed to investigate how the substantia nigra pars reticulata (SNr) is affected by buspirone and whether it is a good target to study 5‐HT antidyskinetic treatments. Experimental Approach Buspirone was studied using in vivo single‐unit, electrocorticogram, local field potential recordings along with microdialysis and immunohistochemistry in naïve/sham, 6‐hydroxydopamine (6‐OHDA)‐lesioned or 6‐OHDA‐lesioned and l‐DOPA‐treated (6‐OHDA/l‐DOPA) rats. Key Results Local buspirone inhibited SNr neuron activity in all groups. However, systemic buspirone reduced burst activity in 6‐OHDA‐lesioned rats (with or without l‐DOPA treatment), whereas 8‐OH‐DPAT, a full 5‐HT1A agonist induced larger inhibitory effects in sham animals. Neither buspirone nor 8‐OH‐DPAT markedly modified the low‐frequency oscillatory activity in the SNr or synchronization within the SNr with the cortex. In addition, local perfusion of buspirone increased GABA and glutamate release in the SNr of naïve and 6‐OHDA‐lesioned rats but no effect in 6‐OHDA/l‐DOPA rats. In the 6‐OHDA/l‐DOPA group, increased 5‐HT transporter and decreased 5‐HT1A receptor expression was found. Conclusions and Implications The effects of buspirone in SNr are influenced by dopamine loss and l‐DOPA treatment. The present results suggest that the regulation of burst activity of the SNr induced by DA loss may be a good target to test new drugs for the treatment of PD and LID.


| INTRODUCTION
Parkinson's disease (PD) pathophysiology involves the progressive loss of different peripheral and central neuronal populations. Degeneration of dopaminergic neurons of the substantia nigra pars compacta, leading to a decrease in striatal dopamine (DA) levels, underlies the appearance of motor deficits (Kish, Shannak, & Hornykiewicz, 1988). The current treatment of Parkinson's disease motor symptoms focuses on restoring dopaminergic neurotransmission using the DA precursor L-DOPA, dopaminergic receptor agonists and DA metabolism inhibitors. L-DOPA provides the greatest symptomatic benefit, although its prolonged use is associated with motor complications and decreased efficacy (Olanow, 2008). The appearance of abnormal involuntary movements, known as L-DOPA-induced dyskinesia (LID), complicates the management of motor symptoms and, in some cases, can greatly affect patient quality of life. Only amantadine (an N-methyl-D-aspartate (NMDA) receptor antagonist) is currently marketed as antidyskinetic, but it lacks long-term efficacy (Perez-Lloret & Rascol, 2018).
In spite of this lack of clinically efficacy of 5-HT drugs, several data support the therapeutic relevance of the 5-HT system in L-DOPAinduced dyskinesia as, for example, the fact L-DOPA is taken up, metabolized to dopamine and released from 5-HT nerve terminals (for review, see Miguelez, Benazzouz, Ugedo, & De Deurwaerdère, 2017;Politis et al., 2014). Furthermore, results from studies in patients with Parkinson's disease show a correlation between 5-HT 1A /SERT binding sites and tremors (Caretti et al., 2008;Doder, Rabiner, Turjanski, Lees, & Brooks, 2003;Loane et al., 2013;Qamhawi et al., 2015) as well as a relationship between post-mortem SERT binding and dyskinesia-free survival time (Rylander et al., 2010).
When investigating the possible effect of antidyskinetic drugs, it is important to take into account the results collected in animal models of Parkinson's disease that reveal the structural and functional changes in the 5-HT system induced by DA depletion and/or chronic administration of L-DOPA (see review Vegas-Suarez et al., 2019). In this regard, we have observed that buspirone, a 5-HT 1A receptor partial agonist, inhibits neuronal activity in the subthalamic nucleus in control rats but has no effect after DA depletion (Sagarduy et al., 2016). Another study found that it efficiently improves the behavioural score and reduces certain molecular changes associated with L-DOPA-induced dyskinesia (Azkona et al., 2014). In addition, DA depletion and/or chronic administration of L-DOPA have been associated with dysfunctional neuronal activity in several brain structures among which is the substantia nigra reticulata (SNr) (Aristieta, Ruiz-Ortega, Miguelez, Morera-Herreras, & Ugedo, 2016; Tseng, Riquelme, Belforte, Pazo, & Murer, 2000). The SNr, one of the main basal ganglia output nuclei, is densely innervated by 5-HT nerve fibres and contains 5-HT and 5-HT 1A receptors (see Di Matteo et al., 2008). Furthermore, electrical stimulation targeting the SNr alleviates some Parkinson's disease symptoms (Collomb-Clerc & Welter, 2015;Hidding et al., 2019;Weiss et al., 2013), while striatonigral optostimulation induces dyskinesia in mice (Keifman et al., 2019).
The present study aimed at defining whether the SNr is involved in the effects of buspirone and that it might be a good target to investigate antidyskinetic drugs. For this purpose, electrophysiological, immunohistochemical and neurochemical approaches were applied in 6-hydroxydopamine (6-OHDA)-lesioned rats with and without prolonged treatment with L-DOPA.

What this study adds
• Buspirone reduces SNr-burst activity and amino acid release more markedly after dopamine (DA) loss.
• Expression of 5-HT 1A receptors is reduced and 5-HT transporter is enhanced in 6-OHDA/L-DOPA rats What is the clinical significance • The findings shed light on the mechanism of action of buspirone in parkinsonism and LID.
water. Experiments involving animals were approved by the Local Eth-

| Experimental design
Experiments were performed and results were analysed under blinded conditions according to British Journal of Pharmacology (Kilkenny et al., 2010). A total of 105 rats were used in this study (six rats were excluded for technical reasons). Animals were divided into three groups which are referred to as sham/naïve, 6-OHDA and 6-OHDA/ L-DOPA groups. Animals were randomly assigned to each group and treatment. Only male animals were used, in order to avoid sex variability. Initially, all groups had equal size, but after behavioural test or immunohistochemistry verification of dopamine loss, results from six animals were excluded. The size of each group for every study design was determined by considering the accuracy and reproducibility of methods used according to our previous published work. Figure 1 shows the timeline and the number of animals (in brackets) used in each experiment. For electrophysiology recordings (experiment 1) and for 5-HT 1A and SERT immunostaining (experiment 3), rats received 6-OHDA or vehicle injections into the medial forebrain bundle and the severity of the dopaminergic lesion was screened using the cylinder test at least 2 weeks later. L-DOPA was then administered daily for 3 weeks and abnormal involuntary movements (AIMs) were evaluated at baseline and on the 21st day. Electrophysiological recordings or perfusion were performed at least 3 weeks after the lesion (sham or 6-OHDA group) or 24 h after the last dose of L-DOPA (6-OHDA/L-DOPA-treated group). At the end of the electrophysiological experiments, animals were deeply anaesthetized (1.2 gÁkg −1 urethane, i.p.), perfused with saline followed by 4% ice-cold paraformaldehyde prepared in 0.1-M phosphate buffer and brains removed for histological verification. For microdialysis experiments (experiment 2), rats received 6-OHDA injections and were screened using the bar and drag tests 3 weeks later. L-DOPA was administered for 3 weeks and abnormal involuntary movements were evaluated. At least 3 weeks after the lesion or 24 h after last L-DOPA injection, rats were implanted with microdialysis probes. Each animal underwent two microdialysis sessions (24 and 48 h after surgery) and received Ringer solution or buspirone administration in a randomized fashion. Samples were collected every 20 min for a total of 4 h. At the end of the experiment the animals were killed by isoflurane overdose and the brains removed and stored at −80 C. Coronal sections (40 μm) were cut using a cryostat LEICA microtome and the correct placement of the microdialysis probe within the SNr was verified.
F I G U R E 1 Schematic representation of the experimental design. At the beginning of the three experiments, animals received a vehicle or 6-OHDA injection into the right medial forebrain bundle. Three weeks later, dopaminergic depletion was evaluated by different asymmetry tests (cylinder or bar and drag test). After that, rats received a 21-day treatment of daily L-DOPA injections (6 plus 12 mgÁkg −1 benserazide). Abnormal involuntary movements (AIMs) were scored at baseline and at the end of the treatment. In experiments 1 and 2, electrophysiological recordings and microdialysis assays in the SNr were performed after the L-DOPA chronic treatment. Finally, animals were perfused transcardially and processed for histological examination. In experiment 3, sacrifice and tissue processing was performed after chronic treatment. * The group was divided into three subgroups with the respective sizes indicated by bold numbers. TH, tyrosine hydroxlase 2.3 | 6-OHDA lesions 6-OHDA lesions were performed according to our established protocols (Aristieta et al., 2016). Thirty minutes before surgery, rats were pretreated using desipramine (25 mgÁkg −1 , i.p.) and pargyline (50 mgÁkg −1 , i.p.) in order to preserve the noradrenergic system and the degradation of the toxin, respectively. Then rats were deeply anaesthetized using isoflurane (1.5-2%, Esteve) and placed in the stereotaxic frame (David Kopf ® Instruments). 6-OHDA (3.5 μgÁμl −1 , in 0.02% ascorbic acid) or vehicle was injected using a 10-μl Hamilton syringe at a rate of 1 μlÁmin −1 . A total of 7.5 and 6 μl, respectively, was injected at two sites in the right medial forebrain bundle-2. −8 mm with toothbar at +3.4 (Paxinos & Watson, 1997). The parkinsonian animals used in experiment 2 (see Figure 1) were lesioned following the same procedure with slight modifications according to Arcuri et al. (2018).

| Cylinder test
At least 2 weeks after the 6-OHDA lesion, forelimb asymmetry was evaluated using the cylinder test as described previously (Miguelez, Aristieta, Cenci, & Ugedo, 2011). Rats were placed in a transparent glass cylinder (20 cm of diameter) with two mirrors positioned at a 90 angle behind the cylinder to allow a complete view of the exploratory activity. Each animal was left in place, freely exploring, until it made at least 20 weight-bearing contacts on the glass walls using either forelimb (maximal exploration time of 5 min). The percentage of contralateral forelimb use with respect to the total number of contacts was calculated. Only animals with an ipsilateral contact percentage ≥80% were enrolled in the experiments 1 or 3.

| Bar test
Three weeks after the 6-OHDA lesion, the bar test was performed as previously described (Arcuri et al., 2018). The ipsilateral and contralateral forelimbs were alternatively placed on three blocks of increasing heights (3, 6 and 9 cm). The immobility time (in seconds) of each forelimb on the blocks was recorded for each step. Only animals that reached a 20-s immobility time with the contralateral forelimb were included in experiment 2.

| Drag test
Three weeks after the 6-OHDA lesion, the drag test was performed (Arcuri et al., 2018). Briefly, each rat was lifted from the abdomen, leaving each forelimb on the table and dragged backwards along a fixed distance of 1 m at a constant speed of 20 cmÁs −1 . Two different observers counted the number of contacts made by each forelimb.
Animals with a percentage of ipsilateral contacts with respect to total contacts ≥80% were included in experiment 2.
2.4.4 | Long-term L-DOPA treatment and abnormal involuntary movements rating Abnormal involuntary movements were induced in 6-OHDA-lesioned rats by daily injections of L-DOPA (6 mgÁkg −1 , i.p.) in combination with the peripheral decarboxylase inhibitor benserazide (12 mgÁkg −1 , i.p.) over 3 weeks (Aristieta et al., 2016;Aristieta, Ruiz-Ortega, Morera-Herreras, Miguelez, & Ugedo, 2019;Azkona et al., 2014;Dekundy, Lundblad, Danysz, & Cenci, 2007). Abnormal involuntary movements were scored according to the scale described by Cenci and Lundblad (2007). On the testing days, rats were placed individually in transparent empty plastic cages for at least 10 min before drug administration. Each rat was observed for one full minute every 20th minute along the 200-min testing period. The severity of the three subtypes of dyskinetic movements (axial, limb and orolingual [ALO]) were rated from 0 to 4 based on the amount of time for which the abnormal movement was observed (i.e. 0, not present to 4, continuous). In addition, the amplitude of ALO abnormal involuntary movements were rated on a scale from 0 to 4 separately from locomotive abnormal involuntary movements. Data are expressed as the abnormal involuntary movements score/session for ALO, which were calculated by multiplying the severity score by the amplitude scores during the monitoring period, with all of these products summed for each testing session (Lindgren et al., 2010). All 6-OHDA L-DOPA rats developed severe dyskinetic movements that peaked between 40 and 100 min after a single injection of L-DOPA and had a total score >100 on the 21st treatment day ( Figure S1).

| Electrophysiological procedures
The electrophysiological recordings of SNr neurons were performed at least 3 weeks after 6-OHDA or vehicle injection for the sham or 6-OHDA group or 24 h after the last L-DOPA injection in the 6-OHDA L-DOPA group (Aristieta et al., 2016). Rats were anaesthetized using urethane (1.2 gÁkg −1 , i.p.) and the right jugular vein was cannulated for systemic drug administration. Next, each rat was placed in a stereotaxic frame with its head secured in a horizontal orientation. The rats were allowed to breath spontaneously. The recording electrode was lowered into the right SNr (relative to bregma and dura, AP: −5.3 mm, ML: −2.5 mm, DV: −7.5 to 8.5 mm) (Paxinos & Watson, 1997). Neuronal spikes were digitized using the CED micro 1401 interface and Spike2 software (Cambridge Electronic Design). The basal firing rate was recorded for 300 s and at least 150 s after each systemic drug dose. For local administration, a calibrated pipette glued adjacent to a recording micropipette was filled with buspirone 0.25 M dissolved in Dulbecco's buffered saline solution (NaCl 136.9 mM; KCl 2.7 mM; NaH 2 PO 4 8.1 mM; KH 2 PO 4 1.5 mM; MgCl 2 0.5 mM; and CaCl 2 0.9 mM, pH 7.4) as previously performed by Sagarduy et al. (2016). Local buspirone injection in the SNr was applied with pressure pulses (50-150 ms) using a Picospritzer™ II (General Valve Corporation, Fairfield, NJ, USA). The injected volume was simultaneously monitored with soft movements in a calibrated pipette in which each pulse corresponded to the injection of 2 nl of solution. Several ejection pulses were applied in each application. The effect of each dose on firing rate was recorded until the neuron recovered. At the end of the experiments, a 5-μA cathodal current was passed through the recording electrode to deposit a discrete mark of Pontamine Sky on the recording site.
The electrocorticogram (ECoG) was simultaneously recorded via a 1-mm-diameter steel screw, which was juxtaposed to the dura mater above the right frontal somatic sensory-motor cortex (AP: +4.5 mm and ML: −2.5 mm to bregma) (Paxinos & Watson, 1997) as described by Aristieta et al. (2016). The signal was pre-amplified (10×), amplified (200×) and bandpass filtered (0.1-1000 Hz) in an amplifier (Cibertec S.A., model amplifier 63AC). The discriminated electrocorticogram activity (sampled at 2500 Hz) was digitized, stored and analysed using computer software. The local field potential (LFP) in the SNr was recorded through the same glass electrodes that were used for singleunit extracellular recordings. The signal was pre-amplified (10×) and then amplified (10×) in a high-input impedance amplifier (Cibertec S.A., model amplifier AE-2) where the signal was also bandpass filtered (0.1-5000 Hz). This signal was divided into two different signals in a second amplifier (Cibertec S.A., model amplifier 63AC), namely, the single-unit extracellular signal and the local field potential signal.
In this second amplifier, the local field potential signal was amplified (10×) and bandpass filtered (0.1-100 Hz). The discriminated local field potential activity (sampled at 2500 Hz) was digitized, stored and analysed using computer software.

| Analysis of electrophysiological data
All electrophysiological parameters were analysed off-line using the Spike2 software (version 7). Firing patterns were determined by analysing the interspike interval histogram:-firing rate and the coefficient of variation (CV), which represents the measure of firing regularity and is defined as the percentage ratio of the SD to the mean interval histogram, which is obtained applying a Spike2 script ("meanix.s2s").
The percentage of neurons exhibiting burst firing pattern from the total and the burst-related parameters: the number of bursts (total number of burst detected in the spike train), mean duration of burst (mean duration of the total burst detected in the spike train and expressed in ms), spikes per burst (mean of number of burst detected), recurrence of burst (number of burst per min) and intraburst frequency (mean of the interval of spike patterns within burst) were analysed during a period of 90 s applying a Spike2 script ("surprise.
s2s") based on Poisson Surprise method. For oscillatory activity and synchronization parameters analysis, local field potential and electrocorticogram signals (sampled at 2500 Hz) were smoothed to 1 ms and SNr action potentials were converted to a series of events (sampled at 2500 Hz). These series of events were then transformed to a continuous waveform (1-ms smoothing period) using a custommade Spike2 software script. The power spectrum of smoothed local field potentials, electrocorticograms and SNr spike waveforms and the coherence analyses between SNr spike waveforms and local field potentials, SNr spike waveforms and electrocorticograms, and SNrlocal field potential and electrocorticograms were also analysed using the fast Fourier transform (8,192 blocks size) in the low-frequency range (0-5 Hz) from 90 s of data applying a Spike2 script ("coherac90. s2s"). The significance of the coherence was determined by the equation described by Rosenberg, Amjad, Breeze, Brillinger, and Halliday (1989): 1 − (1 − α)1/(L − 1), where α is 0.95 and L is the number of windows used. The AUC of coherence and power spectrum curves were calculated in the 0-to 5-Hz low-frequency range.

| Analysis of endogenous GABA and Glutamate
Glutamate (Glu) and GABA were measured using HPLC coupled with fluorometric detection as previously described (Bido, Marti, & Morari, 2011 The excitation and the emission wavelengths were set at 370 and 450 nm, respectively. Under these conditions, the limits of detection for glutamate and GABA were $1 nM (i.e., $147 pgÁml −1 ) and $0.5 nM (i.e., $51 pgÁml −1 ) and maximum sensitivity was $10 μM for both amino acids. The retention times were between 2.8 and 4 min and between 16 and 19 min for glutamate and GABA, respectively.

| Histological procedures and analysis
Animals were deeply anaesthetized (1.2 gÁkg −1 urethane i.p.) and transcardially perfused with saline followed by 4% ice-cold paraformaldehyde and 0.2% picric acid prepared in 0.1-M phosphate saline buffer. Brains were removed and the following day they were transferred to a 25% sucrose solution until they sank. The brains were serially cut in coronal 40-μm sections using a freezing microtome and slices were conserved in a cryoprotectant solution at −20 C until further processing.

| Immunohistochemistry
The Immuno-related procedures used comply with the recommenda-

| Drugs
The following drugs were used in this study: urethane, 6-OHDA was dissolved in Milli-Q water containing 0.02% ascorbic acid. All drug solutions were prepared on the day of the experiment.

| Statistical analysis
The The sphericity of RM two-way ANOVA was assumed by GraphPad Prism software. The threshold of statistical significance was set at P < 0.05 with corresponding 95% confidence intervals. Data are presented as mean ± SEM. All statistical details are described in Table S1.  Recurrence of burst (n burstÁmin −1 )

| Nomenclature of targets and ligands
8.9 ± 1.7 13.2 ± 4.3 14.6 ± 5.9 Intraburst frequency (spike/s) 49.9 ± 5.7 59.0 ± 6.4 69.1 ± 9.7 Note: Data from the firing rate, coefficient of variation, and percentage of neurons exhibiting burst firing pattern (of n rats) and burst-related parameters defined as number of bursts, mean duration of the burst, mean number of spikes per burst, the recurrence of burst per minute, and intraburst frequency or the interval of spike patterns within bursts. Burst activity was detected in the spike train in a period of 90 s based following Surprise method. * P < 0.05 versus sham (one-way ANOVA followed by Bonferroni's post hoc test). did not modify either neuron activity or buspirone effect. Table S2 F I G U R E 2 Local administration of buspirone on SNr neuron activity. Representative firing rate histogram illustrating the inhibitory local effect of buspirone on neurons in sham group (a). Representative firing rate histogram illustrating the blockage of the inhibitory local effect of the previous buspirone administration induced by WAY-100635 (b). Buspirone (0.25-2 nmol) locally injected into the SNr induced a dosedependent inhibitory effect in all three experimental groups (c). The inhibitory local effect of buspirone (white circles; n = 5 neurons from two sham rats) was blocked by the systemic administration of WAY-100635 (0.1 mgÁkg −1 , i.v.) (Student's t-test, P < 0.05) (grey circle; n = 6 neurons from the same two sham rats) (d). Data are expressed as mean ± SEM. # P < 0.05 versus respective baseline (RM two-way ANOVA followed by Bonferroni's post hoc test) from sham (n = 5), 6-OHDA (n = 5), and 6-OHDA/L-DOPA (n = 5) groups

| Effect of 5-HT 1A drugs on the low oscillatory activity and synchronization in the SNr and the cortex
Simultaneously with the single-unit recordings of SNr neurons described above, we also recorded SNr-local field potential and the electrocorticogram. Later, we analysed the power spectrum and the coherence in the 0-to 5-Hz range to determine the parameters of low-frequency activity and synchronization in the SNr and between this nucleus and the cerebral cortex. In agreement with our previous results (Aristieta et al., 2016), the analysis showed low oscillatory activity in the electrocorticogram and the local field potential with a peak near 1 Hz in all three experimental groups (Figure 5a,b). The compilation of the AUC values of the electrocorticogram power spectrum revealed no significant differences among groups (Figure 5a), whereas the local field potential AUC value was higher in the 6-OHDA group (Figure 5b). The coherence analysis showed that synchronization between electrocorticogram and SNr spikes and between local field potential and SNr spikes was higher after 6-OHDA lesion and L-DOPA treatment, with the AUC value obtained from the coherence curves being significantly larger in the 6-OHDA and 6-OHDA/L-DOPA groups (P < 0.05; Figure 6a (Figure 6d). Nevertheless, buspirone produced a slight but significant effect between local field potential and SNr spikes (P < 0.05; Figure 6e). Synchronization between electrocorticogram and local field potential was not affected much by buspirone (Figure 6f).

Finally, we investigated the effect of systemic 8-OH-DPAT
administration on the oscillatory activity and synchronization. As it is shown in Figure 5e,f, 8-OH-DPAT did not cause any change in the AUC of the electrocorticogram and local field potential power spectrum. In addition, analysis of coherence revealed that 8-OH-DPAT had a slight effect on AUC of the coherence between electrocorticogram and SNr spikes, but no differences between the three experimental groups were observed (Figure 6g). 8-OH-DPAT did not cause any effect on the AUC of the coherence between local field potential and SNr spikes (Figure 6h). In the coherence between electrocorticogram and local field potential, there was a significant reduction in 6-OHDA/L-DOPA group with the first dose of 8-OH-DPAT, but no effects were observed for subsequent doses (Figure 6i).

| Effect of buspirone on amino acid release in the SNr from conscious-freely moving rats
We next investigated whether buspirone was able to modulate nigral GABA and glutamate levels. Immediately before buspirone perfusion, basal GABA and glutamate levels in the dialysate from the probe implanted in the SNr did not differ among the three experimental groups 7.28 ± 2.65 nM and 94.35 ± 11.99 nM, respectively, in naïve animals (n = 11), 8.44 ± 1.89 nM and 110.27 ± 16.01, respectively, in 6-OHDA animals (n = 11), and 8.73 ± 1.58 and 71.87 ± 7.65, respectively, in 6-OHDA L-DOPA animals (n = 8). Local perfusion of buspirone (50, 150 and 500 nM) raised dialysate GABA levels in naïve and 6-OHDA groups but had no effect in 6-OHDA/L-DOPA animals ( Figure 7a). However, the GABA enhancement was significantly lower in the 6-OHDA group when compared to that in the naïve group ( Figure 7b,c). Likewise, buspirone also elevated dialysate glutamate release in naïve and 6-OHDA groups, but it had no effect in 6-OHDA/L-DOPA animals (Figure 7d-f). Again, the effect of buspirone on 6-OHDA-lesioned rats was significantly less when compared to the effect on naïve rats (Figure 7d,e,f).

| DISCUSSION
We have used 6-OHDA-lesioned rats, a well-characterized experimental model of Parkinson's disease, to elucidate whether the substantia nigra pars reticulata (SNr) could be a good target to evaluate pharmacological treatments for the motor symptoms associated with Parkinsonism. Buspirone is one of the most effective drugs in reducing motor symptoms and it can reverse certain molecular changes induced by L-DOPA in parkinsonian rats (Azkona et al., 2014;Dekundy et al., 2007;Eskow, Gupta, Alam, Park, & Bishop, 2007).
Here, we show that in anaesthetized rats buspirone (1) reduces the firing rate of SNr neurons when administered directly to the nucleus, (2) decreases bursty activity when it is applied systemically after DA loss but not in control conditions and (3) has no effect on local field potential and synchronization between the cortex and the SNr and within the nucleus. In addition, buspirone modulates GABA and glutamate levels depending on the integrity of DA pathways. Thus, amino acid levels were increased in control conditions, were little affected after DA degeneration and were not modified after 6-OHDA lesion and L-DOPA treatment.

| Effect of buspirone on SNr neuron activity
Although numerous behavioural studies have shown the efficacy of full or partial 5-HT 1A receptor agonists in decreasing dyskinesias, no study has examined how these drugs may affect the altered output basal ganglia activity in Parkinson's disease and l-DOPA-induced dyskinesia. Here, we first show that the firing rate of SNr neurons was reduced when buspirone was administered directly into the nucleus.  (Cowen, Power, Ware, & Anderson, 1994;Eison & Temple, 1986), where its effect will depend on 5-HT levels at the synaptic cleft. Notably, following DA depletion and L-DOPA treatment, the effect of 8-OH-DPAT was abolished. In that sense, several studies have shown that dysfunction of 5-HT 1A receptors in rats after DA degeneration decreases or abolishes the effect of 8-OH-DPAT on GABAergic interneurons (Hou et al., 2012) and pyramidal neurons (Wang, Zhang, Liu, Wu, Ali, et al., 2009a).
Interestingly, systemic administration of buspirone reduced the number of bursty SNr neurons after DA loss but not in control conditions where 8-OH-DPAT has a more pronounced effect. Precisely, an increment in burst activity was the major change we observed in the SNr neurons from the 6-OHDA and 6-OHDA/L-DOPA groups, which is F I G U R E 5 Effect of buspirone on low oscillatory activity and synchronization of SNr neurons and electrocorticogram (ECoG) of the motor cortex. Power spectra (0-to 5-Hz frequency range) and AUC values obtained from the ECoG (a) and SNr-local field potential ( also in line with data from electrophysiological studies in anaesthetized rats, which have reported changes in firing patterns rather than in the firing frequency (Aristieta et al., 2016;Meissner et al., 2006;Tseng et al., 2000). This enhancement in burst activity may be due to changes in synaptic plasticity induced by DA loss (see Lobb, 2014). It seems that bursting activity is more relevant in information-carrying signals (Chergui, Nomikos, Mathé, Gonon, & Svensson, 1996) and neurotransmitter release rather than single action potentials (Gonon, 1988). Therefore, the decrement in burst activity induced by buspirone will produce less intense GABA releases in agreement with our microdialysis results discussed below.

| Effect of buspirone on SNr neuron oscillatory activity and synchronization
Cortical slow-wave activity induced by urethane anaesthesia favours low-frequency oscillations in the cortex and the basal ganglia nuclei in rats, including the SNr (Aristieta et al., 2016;Brown, 2006;Clement et al., 2008;Magill, Bolam, & Bevan, 2001). Thus, the later cited studies demonstrated that these electrophysiological properties are altered in 6-OHDA-lesioned and L-DOPA-treated rats, which show that coupling between the cortex and the SNr increased. In agreement with these findings, we also observed oscillatory activity in the low-F I G U R E 6 Synchronization between the SNr and electrocorticogram (ECoG) of the motor cortex in low-frequency range (0.5-5 Hz) was higher in 6-OHDA-lesioned groups. Coherence power between the ECoG and SNr spikes simultaneously recorded in the sham (n = 19), 6-OHDA (n = 17), and 6-OHDA/L-DOPA (n = 18) groups. AUC value obtained from coherence between the ECoG and SNr spikes was augmented in 6-OHDA L-DOPA group (a). AUC values obtained from SNr-local field potential (LFP)/SNr spikes coherence curves showed an increase in 6-OHDA L-DOPA group (b). AUC values obtained from ECoG/SNr-LFP coherence curves showed no differences (c). Effect of administration of acute doses of buspirone (0.6125-5 mgÁkg −1 , i.v.) on the synchronization between ECoG and SNr spikes (d), between SNr-LFP and SNr spikes (e), and between ECoG and SNr-LFP (f) in the sham (n = 8), 6-OHDA (n = 7), and 6-OHDA/L-DOPA (n = 8) groups. Effect of administration of acute doses of 8-OH-DPAT (20-160 μgÁkg −1 , i.v.) on the synchronization between ECoG and SNr spikes (g), between SNr-LFP and SNr spikes (h), and between ECoG and SNr-LFP (i) in the sham (n = 5), 6-OHDA (n = 5), and 6-OHDA L-DOPA (n = 5) groups. Data are expressed as mean ± SEM. @ P < 0.05 versus sham (one-way ANOVA followed by Bonferroni's post hoc test), * P < 0.05 versus sham, and # P < 0.05 versus respective baseline (RM two-way ANOVA followed by Bonferroni's post hoc test) frequency band (0-5 Hz) in the cortex and the SNr, and an increased synchronization between them after DA loss and L-DOPA treatment.
This low-frequency oscillatory activity and synchronization were not affected by systemic buspirone or 8-OH-DPAT administration at the same doses that modified bursty activity. So far, there are no studies regarding the effect of antidyskinetic drugs on this low oscillatory activity, but our previous work found that an acute L-DOPA challenge had little effect on it (Aristieta et al., 2016). It seems, therefore, that while low-frequency oscillatory activity and synchronization do not have a relevant role in l-DOPA-induced dyskinesia, they may be important in Parkinson's disease. In fact, recent published evidence obtained from 6-OHDA-lesioned rats indicates that oscillations in distinct spectral bands are differently affected at every Parkinsonism stage and may be differently modulated by pharmacological treatments. Thus, recordings in awake animals have shown that, apart from decreasing dyskinesia, 5-HT-antidyskinetic drugs suppressed 80-Hz oscillation activity but not the oscillations in other bands (5-10 Hz and 110-140 Hz) (Brys et al., 2018). Similarly, oscillations in the β band (24-36 Hz) were elevated, but those in the θ band (5-8 Hz) were unchanged after the 6-OHDA lesion. However, θ activity was elevated and correlated with abnormal involuntary movements score in 6-OHDA-lesioned rats treated with L-DOPA and was reduced by eltoprazine, a 5HT 1A/B agonist (Wang et al., 2019).

| Effect of buspirone on SNr neuron amino acid levels
In this study, using microdialysis approaches we found no significant changes in GABA and glutamate basal levels after DA loss with or without prolonged L-DOPA treatment. This finding is consistent with those of other studies performed in freely moving 6-OHDA-lesioned rats (Bianchi et al., 2003;Mela et al., 2007). However, we found that the enhancement in GABA and glutamate release induced by locally perfused buspirone in control animals was significantly reduced after 6-OHDA lesion and almost abolished after prolonged L-DOPA treatment. The sources of GABA in the SNr include the striatal and globus pallidus terminals, and collaterals of GABAergic neurons of the SNr, while the subthalamic nucleus provides the major glutamatergic input to the SNr (Kita & Kitai, 1987). In control conditions, GABA and glutamate release in the SNr may be regulated by 5-HT 1A receptors localized on 5-HT terminals coming from the raphé nuclei, striatal, globus pallidus and subthalamic terminals and nigral GABAergic neurons. The fact that the inhibitory effect of buspirone on SNr neuron activity was unchanged after 6-OHDA lesion rules out the possibility that changes in 5-HT 1A receptors on GABAergic neurons are responsible for the reduction of GABA released. However, this may reflect altered responsiveness from afferent terminals as shown for nigral GABA F I G U R E 7 Effect of local perfusion of buspirone (50-500 nM) on nigral amino acid levels of GABA (a-c) and glutamate (Glu) (d-f). Local perfusion of buspirone raised nigral GABA levels in naïve (n = 9) and 6-OHDA (n = 11) groups, but with less intensity, and it had no effect in 6-OHDA/L-DOPA (n = 7) group (a). Buspirone also raised nigral Glu levels in naïve (n = 10) and 6-OHDA (n = 9) groups, with less intensity, and it had no effect in 6-OHDA/L-DOPA (n = 7) group (d). Bar representation (mean ± SEM) of the AUC (60-240 min) (b and e) and AUC 60 min after administration of the respective buspirone concentration dose (c and f). Data are expressed as percentage of basal levels calculated as the mean of the two samples preceding buspirone perfusion. * P < 0.05 versus naïve group and x P < 0.05 versus respective dose of 6-OHDA group (RM two-way ANOVA followed by Bonferroni's post hoc test) VEGAS-SU release following striatal kainate application and for glutamate release following local application of K + in 6-OHDA-lesioned rats (Bianchi et al., 2003). Furthermore, buspirone, besides being a 5-HT 1A partial agonist also has D 2 and D 3 /D 4 receptor antagonist activity (Bergman et al., 2013). These receptors, whose activity is modulated by the 6-OHDA lesion (Avalos-Fuentes et al., 2013), are present in the pallidonigral and striatonigral terminals, modulate GABA release (Acosta-García et al., 2009;Avalos-Fuentes et al., 2015) and are involved in the antidyskinetic effect of buspirone (Azkona et al., 2014;Shin et al., 2014). Therefore, changes in GABA and glutamate release induced by buspirone may involve modulation of several receptors apart from 5-HT 1A receptor. In fact, in the present study, we only found a decrease in optical density of 5-HT 1A receptors in the SNr from 6-OHDA-lesioned rats with prolonged L-DOPA treatment where buspirone effect on amino acid release was negligible. However, no changes were observed after DA degeneration, while buspirone effect was reduced compared to that in control conditions.  (Rylander et al., 2010). The 5-HT 1A receptor density was decreased in that case in disagreement with published results showing no striatal changes (Riahi et al., 2012). However, although in both cases the change is significant, it is only around a F I G U R E 8 Coronal sections immunoassayed for SERT axon terminals (a) and 5-HT 1A receptors (b) of the SNr. Micrographs at 20× magnifications for the three experimental groups. In the top of the figure, a representative brain coronal section of the SNr (Nissl staining) with delimited sections from ipsilateral (IP) and contralateral (CL) hemispheres. Scale bar = 50 μm. While the IOD of SERT terminals left/right ratio shows a significant increase in SERT terminals (c), 5-HT 1A receptor left/right ratio shows a significant decrease in 6-OHDA/L-DOPA group (d). Sham (n = 5), 6-OHDA-lesioned (n = 5), and 6-OHDA/L-DOPA (n = 5). Each bar represents the mean ± SEM. * P < 0.05 versus sham (one-way ANOVA followed by Bonferroni's post hoc test) 10% of the control value and was only present in 6-OHDA-lesioned animals that had undergone prolonged treatment of L-DOPA. It could be speculated that changes in 5-HT 1A receptors take place on terminals rather than on SNr neurons, which could explain why local buspirone administration has no effect. On the other hand, these results could help to understand the lack of the effect of buspirone on amino acid release in 6-OHDA-lesioned and prolonged L-DOPAtreated animals but not the decreased effect of buspirone on 6-OHDA-lesioned rats. Therefore, it is hard to consider that changes in density are responsible of the altered response to buspirone observed after 6-OHDA lesion and complex mechanisms underlying the decreasing receptor sensitivity are probably involved.

| Experimental considerations
This study was performed only in male rats to avoid gender influence in the results, since electrophysiological differences between male and female have been observed in Parkinson's disease patients (see review Picillo et al., 2017). In this regard, recordings in the STN have shown sex-related differences in spike parameters and high-γ band oscillations activity (Marceglia et al., 2006;Mrakic-Sposta et al., 2008). The development of l-DOPA-induced dyskinesia and the response to deep brain stimulation also differ between female and male subjects (Accolla et al., 2007;Hassin-Baer et al., 2011). In this line, previous work from our laboratory confirmed gender differences in the STN neurons from 6-OHDA-lesioned rats with and without L-DOPA prolonged treatment (Sagarduy et al., 2016). A limitation of the present study is that recordings were performed in anaesthetized animals, and it may be argued that this does not accurately represent clinical situations. However, we have recently characterized the electrophysiological parameters in this model of DA loss and demonstrated a relationship between the observation of abnormal involuntary movements and recording parameters under anaesthesia.
Thus, we chose to use buspirone in this study because we knew from previous studies that a single dose of buspirone applied before L-DOPA was able to reduce abnormal involuntary movements and certain molecular changes observed in 6-OHDA-lesioned and L-DOPAtreated rats (Azkona et al., 2014;Sagarduy et al., 2016).

| CONCLUSION
The relevance of the SNr in Parkinson's disease and L-DOPA-induced dyskinesia is evidenced by aberrant changes of nigral neuron activity observed after DA depletion and L-DOPA treatment in animal models, as well as the symptomatic efficacy of therapeutic interventions that normalize nigral activity in both animal models and patients with Parkinson's disease. Our results show that buspirone modulates amino acid release and neuronal activity, mainly burst activity, in the SNr of the rat. However, these effects are modulated by DA loss and longterm L-DOPA treatment. These novel findings indicate that the regulation of burst activity in the SNr induced by DA loss may be a good target to test new drugs for the treatments of Parkinson's disease and dyskinetic patients and point out the importance of using experimental models of Parkinson's disease and L-DOPA-induced dyskinesia when investigating the potential therapeutic effect of new drugs.