Dopamine-Induced Changes in Gαolf Protein Levels in Striatonigral and Striatopallidal Medium Spiny Neurons Underlie the Genesis of l-DOPA-Induced Dyskinesia in Parkinsonian Mice

The dopamine precursor, l-3,4-dihydroxyphenylalanine (l-DOPA), exerts powerful therapeutic effects but eventually generates l-DOPA-induced dyskinesia (LID) in patients with Parkinson’s disease (PD). LID has a close link with deregulation of striatal dopamine/cAMP signaling, which is integrated by medium spiny neurons (MSNs). Olfactory type G-protein α subunit (Gαolf), a stimulatory GTP-binding protein encoded by the GNAL gene, is highly concentrated in the striatum, where it positively couples with dopamine D1 (D1R) receptor and adenosine A2A receptor (A2AR) to increase intracellular cAMP levels in MSNs. In the striatum, D1Rs are mainly expressed in the MSNs that form the striatonigral pathway, while D2Rs and A2ARs are expressed in the MSNs that form the striatopallidal pathway. Here, we examined the association between striatal Gαolf protein levels and the development of LID. We used a hemi-parkinsonian mouse model with nigrostriatal lesions induced by 6-hydroxydopamine (6-OHDA). Using quantitative immunohistochemistry (IHC) and a dual-antigen recognition in situ proximity ligation assay (PLA), we here found that in the dopamine-depleted striatum, there appeared increased and decreased levels of Gαolf protein in striatonigral and striatopallidal MSNs, respectively, after a daily pulsatile administration of l-DOPA. This leads to increased responsiveness to dopamine stimulation in both striatonigral and striatopallidal MSNs. Because Gαolf protein levels serve as a determinant of cAMP signal-dependent activity in striatal MSNs, we suggest that l-DOPA-induced changes in striatal Gαolf levels in the dopamine-depleted striatum could be a key event in generating LID.


INTRODUCTION
Human pathology has shown that Parkinson's disease (PD) results from dopamine deficiency in the neostriatum, particularly in the putamen, due to degenerative loss of nigrostriatal dopaminergic cells (Kish et al., 1988;Goto et al., 1989). Treatments with the dopamine precursor, L-3,4-dihydroxyphenylalanine (L-DOPA), remain the gold standard of drug therapy for PD. However, after prolonged and pulsatile exposure to L-DOPA, PD patients eventually develop L-DOPA-induced dyskinesia (LID; Jenner, 2008;Calabresi et al., 2010;Huot et al., 2013). LID is an adverse event that occurs in more than 50% of patients after 5-10 years (Ahlskog and Muenter, 2001;Rascol et al., 2006). Importantly, once LID has been established, its severity increases unless dopaminergic drug dosage is reduced (Brotchie, 2005). It is known that the severity of loss of nigral dopaminergic cells represents the most important factor that determines the severity of LID (Guridi et al., 2012;Bastide et al., 2015). However, the nature of the cellular and molecular key events that lead to a progressive increase in responsiveness to dopaminergic stimulation in LID remains unclear.
Olfactory type G-protein α subunit (Gα olf ), the stimulatory G-protein encoded by the GNAL gene, is highly concentrated in the striatum, where it positively couples with D 1 R and A 2A R to activate adenylyl cyclase and, thereby, increase intracellular cAMP levels in MSNs (Hervé, 2011). As Gα olf represents the rate-limiting factor for the D 1 R-and A 2A R-dependent cAMP production (Kull et al., 2000;Corvol et al., 2001), Gα olf protein level serves as a determinant of cAMP signaldependent activity in both D 1 R-expressing striatonigral MSNs (D1-cells) and D 2 R-expressing striatopallidal MSNs (D2-cells). D 1 R/Gα olf -mediated increases in intracellular cAMP levels facilitate D1-cell activity (Hervé, 2011), while the elevation of intracellular cAMP levels via A 2A R/Gα olf activation functionally opposes the actions of D 2 Rs on D2-cells (Schwarzschild et al., 2006;Fuxe et al., 2007). It is also known that Gα olf protein levels in striatal MSNs are regulated by posttranslational usage-dependent mechanism through the activation of D 1 Rs Corvol et al., 2004Corvol et al., , 2007Alcacer et al., 2012;Ruiz-DeDiego et al., 2015) and A 2A Rs .
The aim of this study was to clarify the association of striatal Gα olf protein levels with LID development. For this purpose, we used a hemi-parkinsonian mouse model with nigrostriatal lesion induced by 6-hydroxydopamine (6-OHDA). Using quantitative immunohistochemistry (IHC) and a highly-sensitive in situ proximity ligation assay (PLA), we show that in the 6-OHDAlesioned striatum, daily pulsatile injections of L-DOPA might cause changes in Gα olf levels in not only D1-cells but also D2-cells, and lead to elevated responsiveness to dopamine stimulation in both D1-cells and D2-cells. This novel finding suggests that L-DOPA-induced changes in striatal Gα olf levels in the dopamine-denervated striatum may serve as a principal cause for generating LID.

Experimental Animals
All experimental procedures involving the use of animals and the analysis of brain anatomy were approved by the Institutional Care and Use Committees of Tokushima University, Japan. Adult male C57BL/6 mice aged 8-9 weeks were purchased from Nihon SLC Co. (Shizuoka, Japan). Mice were housed in a controlled environment (23 ± 1 • C, 50 ± 5% of humidity) with 12 h light/dark cycle. Mice were allowed to take food and tap water ad libitum.

Stereotaxic Injection of 6-OHDA
Mice were anesthetized with isoflurane (Sigma-Aldrich, St. Louis, MO, USA) and were mounted on a stereotaxic frame (Narishige, Tokyo, Japan). Each mouse received a stereotaxic injection of 6-OHDA-HCl (8.2 µg) dissolved in 4 µl of saline containing 0.02% ascorbic acid. Two 2-µl injections were administered into the striatum at a rate of 1 µl/min. The needle was left in place for 5 min to allow diffusion away from the injection site. The stereotaxic coordinates according to the mouse brain atlas (Paxinos and Franklin, 2001) were anterior-posterior, +0.5; medial-lateral, +2.4; and dorsal-ventral, −4.0 and −3.0. Mice were allowed to recover for 3 weeks and then apomorphine (Sigma-Aldrich; 0.5 mg/kg)-induced rotation behavior was studied over the course of 60 min. Mice with contralateral rotations (>7 times/min) were chosen and used for further studies.

L-DOPA Treatments
Three weeks after the 6-OHDA-lesioning, mice received intraperitoneal injections of L-DOPA (Sigma-Aldrich; 20 mg/kg of free base) dissolved in 0.9% saline and intraperitoneal injections of benserazide-HCl (Sigma-Aldrich; 12 mg/kg) dissolved in 0.9% saline 20 min before daily administration of L-DOPA over 10 days. On day 11, the mice underwent behavioral studies and were then sacrificed for histological studies.

Assessment of Abnormal Involuntary Movements (AIMs)
AIM scoring was performed according to previous reports (Cenci et al., 1998;Pavón et al., 2006;Santini et al., 2007). AIM scores were obtained after the last injection of L-DOPA for 1 min every 10 min over a period of 140 min. For the evaluation, each mouse was placed in a glass cylinder (diameter of 12 cm). Purposeless movements were classified on the basis of their topographic distribution. The following four subtypes of AIMs were present: locomotive (tight contralateral turns), axial (twisted posturing of the neck and upper body toward the contralateral side), forelimb (jerky movements of the contralateral forelimb, and/or grabbing movement of the contralateral paw), and orolingual (jaw movements and tongue protrusion toward the contralateral side). Each subtype was scored as follows; 0, absent; 1, occasional; 2, frequent; 3, continuous; 4, continuous, not interrupted by sensory stimuli.

Tissue Preparations
Immediately after the last AIM scoring, the mice were intraperitoneally administered a lethal dose of pentobarbital (Sigma-Aldrich). They were then transcardially perfused with 0.01 M phosphate-buffered saline (PBS) at pH 7.2, followed by cold 4% paraformaldehyde in 0.1 M phosphate buffer at pH 7.2. The brains were removed, post-fixed overnight in the same fixative at 4 • C, and stored in a 10-30% sucrose gradient in 0.1 M phosphate buffer at 4 • C for cryoprotection. Sixteen-micrometrethick sections were cut on a cryostat and stored in PBS containing 0.05% NaN 3 until use.

Dual-Antigen Recognition In Situ PLA
Dual-antigen recognition PLA experiments were conducted using the Brightfield Duolink PLA kit reagents (Sigma-Aldrich) according to the manufacturer's recommendations with some modifications. Briefly, after blocking endogenous peroxidases in PBS containing 0.1% H 2 O 2 for 30 min, the free-floating sections were incubated in PBS containing 3% normal goat serum for 60 min. They were then incubated in PBS containing 3% normal goat serum and a rabbit polyclonal antibody against Gα olf (1:500; Santa Cruz Biotechnology) in combination with a mouse monoclonal antibody against D 1 R (1:500; Novus Biologicals) or a mouse monoclonal antibody against A 2A R (1:500; Santa Cruz Biotechnology) for 18 h at room temperature. After subsequent secondary labeling with rabbit PLA minus and mouse PLA plus probes, we used the Brightfield Duolink Detection reagents for ligation and amplification and label probe binding according to the manufacturer's instructions. For final signal visualization, we used the TSA-system with Cyanine3 (Perkin Elmer). After mounting on slides, the stained sections were counterstained with hematoxylin and were cover-slipped using 10% glycerol in PBS.

Digital Imaging and Morphometry
Digital microscopy images were captured using an Olympus BX51 microscope (Olympus, Tokyo, Japan) equipped with a DP40 digital camera (Olympus). They were imported into Adobe Photoshop CS4 and processed digitally. We adjusted contrast, brightness, and color balance. Using an image analyzer (MetaMorph, Molecular Device, Tokyo, Japan), we measured the optical densities of immunoreactive products and PLA signals in the striatum, which were represented by gray levels on non-colored digital images (Sato et al., 2008;Goto et al., 2013;Morigaki and Goto, 2015). Using the same protocol described above, we also measured optical densities of Gα olfimmunoreactive products in the striosome and matrix subfields in the striatal sections double-stained for Gα olf and MOR. We also counted the numbers of neuronal nuclei positive for c-Fos in a 0.5 mm × 0.5 mm field in the striatum and globus pallidus, as in our previous report (Tanabe et al., 2014). These morphometric analyses were carried out in a blind manner.

Statistical Analysis
All experimental values are expressed as means ± SEM. For two-group comparisons, we used a paired two-tailed t-test. Multiple comparisons were analyzed using one-way or two-way analysis of variance (ANOVA), followed by Bonferroni's post hoc tests for pair wise comparisons. Statistical analyses were performed using Stat View 5.0 (SAS Institute, Cary, NC, USA) software. P-values of less than 0.05 were considered statistically significant.

Generation of a Mouse Model with LID
To model the generation of AIMs in PD following repeated L-DOPA treatments, we employed a well-established PD mouse model in which mice first received unilateral injection of 6-OHDA into the striatum (Santini et al., 2007). After 3 weeks of recovery and an apomorphine test, the mice were subjected to L-DOPA treatment for 10 days according to standardized protocols ( Figure 1A). In this study, 6-OHDA-lesioned mice administered daily injections of benserazide-HCl (12 mg/kg) alone for 10 days were designated as ''PD'' models. Six-OHDA-lesioned mice that received daily injections of L-DOPA (20 mg/kg) and benserazide-HCl (12 mg/kg) for 10 days and finally exhibited LIDs with total AIM scores of more than 20 were designated as ''PD with Dyskinesia (PD-D)'' models. Among 6-OHDA-lesioned mice that received daily injections of L-DOPA (n = 28), 25 mice (∼90%) were grouped into the PD-D model. Mice that received no drug treatment, except for anesthetic drugs, were used as ''naïve controls''.
In both PD and PD-D mice, IHC with anti-TH antibody revealed a severe loss of nigral dopaminergic cells ( Figure 1B) and striatal dopaminergic afferents ( Figure 1C) on the side of the 6-OHDA injection. Quantitative measurements ( Figure 1D) revealed a greater-than-90% reduction in TH labeling in the dorsolateral (DL) striatum on the lesioned side when compared to the non-lesioned side (lesion side, 3.9 ± 2.1; non-lesion side, 56.5 ± 8.9; means ± SEM; n = 10; two-tailed t-test, P < 0.01). Figure 1E shows the time course of changes in LIDs as determined by AIM scoring in PD-D mice. AIMs were maximal 40 min after L-DOPA administration, declined after 70 min, and almost disappeared after 120 min.
Regional and Cellular Localization of Gα olf in the Normal Mouse Striatum Figure 2A depicts the known distributional patterns of Gα olf , D 1 R, and A 2A R in a simplified basal ganglia circuit diagram. Note that Gα olf is mainly localized with D 1 R in the D1-cells that form the striatonigral pathway, while it is localized with A 2A R in the D2-cells that form the striatopallidal pathway. Using IHC, we reappraized the localization profile of Gα olf immunoreactivity (IR) in the mouse striatum. Low-magnification microscopic images show strong Gα olf labeling in the striatum ( Figure 2B), particularly in the DL region ( Figure 2B', arrows). As in our previous report (Sako et al., 2010), Gα olf IR was differentially concentrated in the different striatal compartments, with heightened Gα olf labeling in the striosomes relative to the matrix (Figures 2C,D). Optical density measurements ( Figure 2E) also revealed that Gα olf IR in the striosomes was significantly higher than that in the matrix (striosomes, 39.8 ± 5.0; matrix,  23.5 ± 6.0; means ± SEM; n = 10; two-tailed t-test, P < 0.01).
Microscopic images with high magnification show numerous tiny dots of Gα olf IR densely distributed in the DL striatum (Figures 2F-H). In the double-labeling study, Gα olf -positive dots were frequently localized in MSNs labeled for D 1 R ( Figure 2G) or A 2A R (Figure 2H). Using dual-antigen recognition in situ PLA, which indicates that two proteins are in close proximity (Söderberg et al., 2006), we also found that dot signals indicating the presence of Gα olf protein in close proximity to D 1 R protein (D 1 R-Gα olf ; Figure 2I) or A 2A R protein (A 2A R-Gα olf ; Figure 2J) were abundantly distributed in the DL striatum.

Dopaminergic Regulation of Striatal Gα olf Protein Levels
To examine the dopaminergic regulation of Gα olf protein levels in the striatum, we performed a quantitative IHC using an anti-Gα olf antibody on striatal sections prepared from naïve control, PD, and PD-D mice. In low-magnification microscopic images, PD mice ( Figure 3A) showed dramatic increases in Gα olf IR in the dorsal striatum on the 6-OHDAlesioned side when compared to non-lesioned side. In contrast, PD-D mice (Figure 3B) only had a modest increase in striatal Gα olf IR on the 6-OHDA-lesioned side relative to non-lesioned side. Higher-magnification microscopic images of the DL striatum also show that, when compared to naïve controls (Figure 3C), there is a marked, but, slight increase in Gα olf IR in the 6-OHDA-lesioned striatal areas of PD ( Figure 3D) and PD-D ( Figure 3E) mice. As indicated in a previous report (Ruiz-DeDiego et al., 2015), it is likely that dopamine depletion increases Gα olf IR mainly in the matrix of the 6-OHDA-lesioned striatum, leading to a loss of the striosome-predominant pattern of Gα olf IR expression in PD mice. However, daily treatment with L-DOPA reverses the lesion-induced increase in Gα olf IR primarily in the matrix, leading to reappearance of the striosome-predominant pattern of Gα olf IR expression in PD-D mice. These visual impressions were confirmed by quantitative densitometry analyses of the DL striatum (Figures 3F-H), as follows. We found a significant and marked increase of 101% (P < 0.001, two-way ANOVA) in Gα olf IR levels in the 6-OHDA-lesioned striatum of PD, when compared to naïve controls. There was a 67% decrease (P < 0.01, two-way ANOVA) in Gα olf IR levels in the 6-OHDA-lesioned striatum of PD-D mice when compared to that of PD mice ( Figure 3F; PD mice: non-lesion, 102 ± 17% and 6-OHDA-lesion, 201 ± 28%; PD-D mice: non-lesion, 103 ± 19% and 6-OHDA-lesion, 134 ± 16%; % of naïve control mice ± SEM; n = 15).

Dopaminergic Regulation of Striatal Expression of D 1 R, A 2A R, and D 2 R
To examine dopaminergic regulation of D 1 R, A 2A R and D 2 R expression in the DL striatum, we performed quantitative IHC on sections prepared from 6-OHDA-lesioned striata of PD and PD-D mice (Figure 4). We observed no significant changes (P > 0.05, one-way ANOVA) in the expression levels of D 1 R IR in PD or PD-D mice when compared to naïve controls (Figures 4A,B; PD mice, 101 ± 21%; PD-D mice, 98 ± 17%; % of naïve control mice ± SEM; n = 15). The expression levels of A 2A R IR in PD and PD-D mice were not significantly different (P > 0.05, one-way ANOVA) from those in naïve controls (Figures 4C,D; PD mice, 103 ± 15%; PD-D mice, 118 ± 13%; % of naïve control mice ± SEM; n = 15). We found a significant increase in the expression of D 2 R in PD (P < 0.01, one-way ANOVA), but not PD-D (P > 0.05, one-way ANOVA), mice when compared to naïve controls (Figures 4E,F; PD mice, 123 ± 15%; PD-D mice, 112 ± 22%; % of naïve control mice ± SEM; n = 15). These findings indicate that dopamine depletion causes a significant increase in striatal D 2 R expression, which is reversed by daily treatment with L-DOPA. In addition, dopamine depletion and L-DOPA replacement cause no significant changes in striatal expression of D 1 R and A 2A R in the dopamine-denervated striatum.

Dopaminergic Regulation of Striatal Levels of PLA Signals for D 1 R-Gα olf
To examine the dopaminergic regulation of striatal levels of Gα olf protein in close proximity to D 1 R protein, we used a sensitive in situ PLA in sections prepared from 6-OHDA-lesioned striata from PD and PD-D mice (Figure 5). In low-magnification microscopic images, a marked and moderate increase in D 1 R-Gα olf PLA signals was observed in the dorsal striatum in PD and PD-D mice when compared to naïve controls (Figures 5A,B). Higher-magnification microscopic images of the DL striatum also show that compared to naïve controls (Figure 5C), there is a marked and moderate increase in the D 1 R-Gα olf PLA signals in 6-OHDA-lesioned striatal areas in PD ( Figure 5D) and PD-D ( Figure 5E) mice. Quantitative densitometry analyses of the DL striatum revealed increases of 92% (P < 0.001, one-way ANOVA) and 50% (P < 0.001, one-way ANOVA) in the D 1 R-Gα olf PLA signal in PD and PD-D mice, respectively, when compared to naïve controls. There was a decrease of 42% (P < 0.001, one-way ANOVA) in the D 1 R-Gα olf PLA signal in PD-D mice when compared to PD mice ( Figure 5F; PD mice, 192 ± 25%; PD-D mice, 150 ± 21%; % of naïve control mice ± SEM; n = 10). These findings indicate that dopamine depletion causes a marked increase in striatal D 1 R-Gα olf PLA signal, which is downregulated by daily treatment with L-DOPA. However, there is a significant increase of striatal D 1 R-Gα olf PLA signal in PD-D mice compared to naïve controls.

Dopaminergic Regulation of Striatal Levels of PLA Signals for A 2A R-Gα olf
To examine the dopaminergic regulation of striatal levels of Gα olf protein in close proximity to A 2A R protein, we used a sensitive in situ PLA in sections prepared from 6-OHDA-lesioned striata from PD and PD-D mice (Figure 6). Notably, low-magnification microscopic images show an apparent decrease in the A 2A R-FIGURE 5 | Dopaminergic regulation of striatal levels of Gα olf proteins in close proximity to D 1 R proteins. Dual-antigen recognition in situ PLA used to detect Gα olf proteins in proximity to D 1 R proteins (D 1 R-Gα olf ) was carried out on normal hemispheres of naïve controls and on lesioned hemispheres from 6-OHDA-lesioned mice treated with daily injections of benserazide alone for 10 days (PD model) and from 6-OHDA-lesioned mice that received daily injections of benserazide and L-DOPA for 10 days and exhibited dyskinesia (PD-D model). Frontiers in Cellular Neuroscience | www.frontiersin.org Gα olf PLA signal in the DL striatum of PD-D mice when compared to both naïve control and PD mice (Figures 6A,B). Higher-magnification images also show the localization patterns of A 2A R-Gα olf PLA signals in the DL striatum of naïve control (Figure 6C), PD (Figure 6D), and PD-D (Figure 6E) mice. Quantitative densitometry analyses of the DL striatum revealed decreases of 41% (P < 0.001, one-way ANOVA) and 45% (P < 0.001, one-way ANOVA) in A 2A R-Gα olf PLA signal levels in PD-D mice, when compared to naïve controls and PD mice, respectively ( Figure 6F; PD mice, 104 ± 24%; PD-D mice, 59 ± 21%; % of naïve control mice ± SEM; n = 10). These findings indicate that L-DOPA replacement, but not dopamine depletion, causes a significant decrease in the striatal A 2A R-Gα olf PLA signal in the dopamine-denervated striatum.

Differences in Striatal Responsiveness to Dopamine Stimulation between PD and PD-D Mice
To assess changes in striatal responsiveness to dopamine stimulation in PD and PD-D mice, we performed IHC using an antibody against c-Fos, which is known to be induced in the striatum and globus pallidus following the stimulation of D 1 Rs and D 2 Rs (Marshall et al., 1993;LaHoste and Marshall, 1994). We prepared striatal sections from PD and PD-D mice that received injections of L-DOPA (20 mg/kg) and benserazide-HCl (12 mg/kg) 2 h before sacrifice on day 11 (see Figure 1A). Microscopic images of the DL striatum stained for c-Fos from naïve control, PD and PD-D mice are shown in Figure 7A. Compared to naïve controls that also received the injections of L-DOPA (20 mg/kg) and benserazide-HCl (12 mg/kg) 2 h before sacrifice, we found a marked increase in the densities of c-Fos-positive (c-Fos + ) nuclei in the 6-OHDA-lesioned striatum in both PD and PD-D mice. Quantitative densitometry analyses also showed a marked increase (P < 0.001, two-way ANOVA) in the density of c-Fos + nuclei in the 6-OHDA-lesioned striatum in both PD and PD-D mice when compared to naïve controls. However, there was a decrease of ∼40% (P < 0.001, two-way ANOVA) in the density of c-Fos + nuclei in the 6-OHDAlesioned striatum of PD-D mice when compared to PD mice ( Figure 7B; naïve controls: 20 ± 12; PD mice: non-lesion, 31 ± 10 and 6-OHDA-lesion, 475 ± 55; PD-D mice: non-lesion,  An immunohistochemical study with anti-c-Fos antibody was carried out on the striatal sections from PD and PD-D mice, which received injections of L-DOPA (20 mg/kg) and benserazide-HCl (12 mg/kg) 2 h before sacrifice on day 11 (see Figure 1A), and those from naïve controls that also received the injections of L-DOPA (20 mg/kg) and benserazide-HCl (12 mg/kg) 2 h before sacrifice. The density measurements were performed by counting the numbers of c-Fos-positive (c-Fos + ) nuclei in a 0.5 mm × 0.5 mm field in the striatum and globus pallidus from each animal. (A) Representative photomicrographs of the DL striatum stained for c-Fos in naïve control, PD and PD-D mice. The inset (red open box) in the naïve control is the corresponding figure from the atlas of Paxinos and Franklin (2001) to show the striatal area that were analyzed in all naïve control, PD and PD-D mice. (B) Density quantification of c-Fos + nuclei in the DL striatum from naïve controls (n = 10), and of those on non-lesion and lesion sides in PD (n = 10) and PD-D (n = 10) mice. Data are expressed as means ± SEM. * * * P < 0.001 vs. naïve controls; ### P < 0.001 vs. PD; two-way ANOVA (F (1,36) = 140.9) followed by Bonferroni's test. (C) Representative photomicrographs of the globus pallidus stained for c-Fos in naïve control, PD and PD-D mice. The inset (red open box) in the naïve control is the corresponding figure from the atlas of Paxinos and Franklin (2001) to show the pallidal area that were analyzed in all naïve control, PD and PD-D mice. (D) Density quantification of c-Fos + nuclei in the globus pallidus from naïve controls (n = 10), and of those on non-lesion and lesion sides in PD (n = 10) and PD-D (n = 10) mice. Data are expressed as means ± SEM. * * * P < 0.001 vs. naïve controls; ## P < 0.01 vs. PD; two-way ANOVA (F (1,36) = 80.8) followed by Bonferroni's test. Scale bars: (A) = 200 µm; (C) = 100 µm.
29 ± 12 and 6-OHDA-lesion, 292 ± 49; means ± SEM; n = 10). Microscopic images of the globus pallidus stained for c-Fos obtained from naïve control, PD, and PD-D mice are shown in Figure 7C. Compared to naïve controls, we found increased densities of c-Fos + nuclei in the globus pallidus on the lesioned sides in both PD and PD-D mice. Quantitative densitometry analyses also indicated a significant increase (P < 0.001, two-way ANOVA) in the density of c-Fos + nuclei in the globus pallidus on the lesioned sides in both PD and PD-D mice when compared to naïve controls. Importantly, we found that there was an increase of ∼140% (P < 0.01, two-way ANOVA) in the density of c-Fos + nuclei in the globus pallidus on the lesioned side in PD-D mice when compared to PD mice ( Figure 7D; naïve controls: 4 ± 2; PD mice: non-lesion, 5 ± 3 and 6-OHDA-lesion, 38 ± 8; PD-D mice: non-lesion, 8 ± 7 and 6-OHDA-lesion, 92 ± 11; means ± SEM; n = 10). These findings indicate that dopamine depletion causes a marked increase in the responsiveness of striatal D1-cells to dopamine stimulation, which is downregulated by daily treatments with L-DOPA. Given the changes in striatal D 2 R expression in PD and PD-D mice (see above), it is likely that in the dopaminedenervated striatum, dopamine depletion may cause increased striatal D 2 R expression, which then enhances the responsiveness of D2-cells to dopamine stimulation (Cai et al., 2002). Notably, L-DOPA replacement could induce a further increase in the responsiveness of D2-cells to dopamine stimulation despite no obvious increase in striatal D 2 R expression in the dopaminedenervated striatum.

DISCUSSION
Here we used IHC and in situ PLA to determine the regionand cell-type-specific distributions of Gα olf proteins in the mouse striatum. Using a mouse model of hemiparkinsonism induced by 6-OHDA, we also found that daily pulsatile administration of L-DOPA might induce usage-dependent changes in Gα olf expression not only in D1-cells, but also in D2-cells in the dopamine-depleted striatum (see Figure 8). This raises the possibility that LID might result from reduced A 2A R/Gα olf /cAMP signal levels in D2-cells, which may be caused by intermittent and pulsatile activation of postsynaptic D 1 Rs in the striatum. Our results support and provide new insights into the hypothesis that LID is associated with a decrease in activity of ''indirect'' striatopallidal pathway (Crossman, 1990;DeLong, 1990;Brotchie, 2005;Guridi et al., 2012).

Strategic Localization of Gα olf Proteins in the Striatum
We used IHC to demonstrate that Gα olf IR is highly concentrated in the DL striatum, which corresponds to the motor-sensory territory in rodents and is analogous to the putamen in primates (Graybiel, 2008). This implies that Gα olf may have a unique position in regulating the activities of the cortico-thalamo-basal ganglia circuit involved in motor functions, i.e., the motor loop (Alexander and Crutcher, 1990), at the striatal level. Although a previous study revealed no obvious compartmental difference in Gα olf mRNA expression throughout striatal development in rats (Sakagami et al., 1995), we observed differential concentrations of Gα olf IR in the striosome and matrix compartments, with higher densities of Gα olf IR in the striosomes relative to the matrix. This finding suggests that Gα olf may be a key molecule for controlling differential responses of striosome-matrix systems to D 1 R activation in adult mice. There is evidence that in experimental animal models with 6-OHDA-lesions (Hervé et al., 1993;Corvol et al., 2004;Alcacer et al., 2012;Ruiz-DeDiego et al., 2015) or in those with a total absence of D 1 Rs due to D 1 R gene targeting , the upregulation of Gα olf levels in the striatum is not accompanied by a parallel increase in Gα olf mRNA expression. Thus, homeostatic regulation of striatal Gα olf protein levels is thought to occur via post-translational mechanisms, wherein the altered expression of Gα olf protein depends directly on its rate of usage (Hervé, 2011). We suggest that when compared to the matrix, the striosomes might have the lower levels of D 1 R/Gα olf stimulation, which may then lower the Gα olf degradation rate and lead to accumulation of the protein. Our assumption is supported by the present finding that changes in striatal Gα olf IR expression were primarily found in the matrix in both PD and PD-D mice.
In this study, we first used highly sensitive dual-antigen recognition in situ PLA using a combination of the Brightfield Duolink PLA kit reagents and the TSA system (see ''Materials and Methods'' Section). This dual-antigen recognition PLA technique allowed us to obtain specific and efficient fluorescent signals showing Gα olf protein in close proximity to D 1 R or A 2A R protein in the striatum. However, we cannot say that all the PLA signals detected here resulted from the direct interaction (or actual coupling) of Gα olf protein with D 1 R or A 2A R protein. Borroto-Escuela et al. (2013) have shown that PLA can indicate a close proximity between two proteins, which is not always a reflection of direct interaction. This is because in situ PLA signals can be detected when two protein epitopes are in close proximity with ranges of 10-30 nm or more. In addition, although the precise mechanisms by which Gα olf protein interact with D 1 R or A 2A R protein remains unclear, it was also noted that in striatal membrane, the content in Gα olf protein would be almost one to two orders of magnitude higher than that in D 1 R or A 2A R (Hervé, 2011).

Striatal Gα olf as a Determinant of the Increased Responsiveness of D1-cells to Dopamine Stimulation in LID
As shown in previous studies (Alcacer et al., 2012;Ruiz-DeDiego et al., 2015), we found a marked increase in striatal Gα olf protein levels in PD mice with 6-OHDA lesions. This is in line with evidence that dopamine depletion may lead to up-regulation of Gα olf protein expression in the rat striatum (Hervé et al., 1993;Marcotte et al., 1994;Penit-Soria et al., 1997;Corvol et al., 2004;Rangel-Barajas et al., 2011) and in the putamen in patients with PD (Corvol et al., 2004). Given the evidence that striatal levels of D 1 R (Shinotoh et al., 1993;Turjanski et al., 1997: Hurley et al., 2001 and other major mediators of D 1 R signaling (Girault et al., 1989;Nishino et al., 1993) are unchanged in PD patients, the dramatic increase in striatal Gα olf protein level may be a key event in the D 1 R hypersensitivity that develops in PD (Alcacer et al., 2012). In support of this notion, we detected no obvious changes in striatal D 1 R expression in PD mice.
Previous data have suggested that the up-regulation of Gα olf protein levels in the dopamine-depleted striatum is post-translational (Hervé et al., 1993;Ruiz-DeDiego et al., 2015) and results from the disuse of the D 1 Rs . Indeed, daily administration of L-DOPA for 10 days resulted in a down-regulation of the increased Gα olf protein levels in the 6-OHDA-lesioned striatum in PD-D mice. However, we also found a significant increase in striatal Gα olf levels in PD-D mice when compared to naïve controls. In agreement with the changes in striatal Gα olf levels in PD and PD-D mice, in situ PLA also revealed that striatal D 1 R-Gα olf PLA signals were dramatically increased in PD mice and moderately increased in PD-D mice. These findings imply an increased responsiveness of D1-cells to D 1 R activation in PD-D mice, although this responsiveness is lower than that found in PD mice. Our assumption is also supported by the fact that, compared to naïve controls, a significant increase in the number of striatal c-Fos + nuclei consequent to L-DOPA administration was evident in PD-D mice, although this increase was more pronounced in PD mice. Since Gα olf represents the rate-limiting factor in D 1 Rmediated cAMP production in D1-cells, these findings suggest that striatal Gα olf level acts as a determinant for the increased responsiveness of D1-cells to dopamine stimulation in LID (see Figure 8).

Striatal Gα olf as a Determinant of the Increased Responsiveness of D2-cells to Dopamine Stimulation in LID
It has been postulated that repeated exposure to dopaminergic agents leads to increased sensitivity of D2-cells to D 2 R activation in the dopamine-depleted striatum in experimental animals (Engber et al., 1989;Asin et al., 1995;Kashihara et al., 2000). However, no obvious increase in striatal D 2 R expression has been observed in PD patients treated with dopaminergic drugs (Rinne et al., 1981;Guttman and Seeman, 1985;Antonini et al., 1997;Thobois et al., 2004). In agreement with this notion, we found that pallidal c-Fos induction consequent to L-DOPA administration was more marked in PD-D mice compared to PD mice. On the other hand, there was increased expression of striatal D 2 Rs in PD mice, but not in PD-D mice. This indicates that the repeated administration of L-DOPA results in an increased responsiveness of D2-cells to striatal D 2 R activation in the dopamine-denervated striatum, and suggests that this phenomenon might underlie LID. Using an in situ PLA, we found that A 2A R-Gα olf PLA signals were markedly reduced along with Gα olf protein levels in the 6-OHDA-lesioned striatum of PD-D mice. This novel finding indicates that as in D1-cells, repeated exposure to L-DOPA causes down-regulation of Gα olf protein levels in D2-cells in the dopamine-depleted striatum. This then leads to the facilitation of the effects of dopamine on D2-cells by reducing A 2A R/Gα olf signaling-mediated cAMP production (see Figure 8). This may be the reason that PD-D mice display an increased responsiveness of D2-cells to dopamine stimulation. However, the mechanism by which repeated and pulsatile injections of L-DOPA causes a decrease in A 2A R/Gα olf PLA signals in PD-D remains a matter of speculation, as follows.
A 2A R usage by endogenous adenosine results in a basal rate of Gα olf degradation . It has been shown that in experimental animals with 6-OHDA-lesions, chronic (or persistent) dopamine depletion caused no significant changes (Ballarin et al., 1987;Herrera-Marschitz et al., 1994;Nomoto et al., 2000) or slight decrease (Pinna et al., 2002) in the extracellular adenosine levels in the striatum. In accordance with these findings, our present results also showed no significant changes in striatal levels of A 2A R-Gα olf PLA signals in PD mice. Thus, we suggest that chronic dopamine depletion per se might cause no obvious changes in A 2A R/Gα olf signaling activities that depend on the endogenous adenosine levels in striatal D2-cells. However, it is known that endogenous levels of adenosine are increased in response to the activation of Nmethyl-D-aspartate (NMDA) receptors , which can be facilitated by D 1 R stimulation (Cepeda and Levine, 2012;Morigaki and Goto, 2015), in the striatum. A landmark report has shown that in the rat striatum, transient (pulsatile) stimulation of D 1 Rs facilitates the NMDA receptor-dependent increase in extracellular adenosine levels (Harvey and Lacey, 1997). These findings suggest that in 6-OHDA-lesioned mice with D 1 R hypersensitivity, repeated exposure to L-DOPA may lead to a transient activation of D 1 Rs, which then enhances the NMDA receptor-dependent increase in adenosine release in the dopamine-denervated striatum. Moreover, Nash and Brotchie (2000) have shown that in striatal slices prepared from rats with 6-OHDA lesions, NMDA receptor activation could cause a marked increase in adenosine release and, thereby, indirectly stimulate A 2A Rs. Taken together, we speculate that in the 6-OHDA-lesioned striatum of PD-D mice, decreased Gα olf levels in D2-cells might be due to increased extracellular adenosine levels caused by the daily pulsatile activation of striatal D 1 Rs. If our assumption is correct, striatal D 1 R signals might contribute to regulation of the Gα olf protein levels in not only D1-cells but also D2-cells in the dopamine-depleted striatum.
Because adenosine/A 2A R signaling functionally opposes the actions of D 2 Rs on D2-cells by its ability to increase the A 2A R/Gα olf -dependent cAMP production, it has so far been suggested that A 2A R antagonism may boost the anti-parkinsonian action of D 2 R agonists in treating PD symptoms (Jenner, 2003;Schwarzschild et al., 2006;Fuxe et al., 2007;Huot et al., 2013). In addition, based on the evidence that striatal A 2A R expression might be increased in PD patients with dyskinesia (Calon et al., 2004;Ramlackhansingh et al., 2011) and in dyskinetic animal models of PD (Jenner et al., 2009), it has also been suggested that adenosine A 2A sites might be a potential pharmacologic target for reducing LIDs (Jenner et al., 2009;Ramlackhansingh et al., 2011;Huot et al., 2013;Kanda and Uchida, 2014). In Japan, istradefylline, an A 2A R antagonist, is currently used in clinics for treating PD patients (Kondo and Mizuno, 2015). The drug has shown to improve ''off'' time in patients with advanced PD, but has not shown anti-LID effects in the absence of a reduction in dopaminergic drug dosage. Adjunct use of istradefylline often causes dyskinetic symptoms as a major adverse effect (Kondo and Mizuno, 2015). Considering usage-dependent Gα olf degradation through adenosine/A 2A R, we assume that in PD patients treated with L-DOPA, adenosine/A 2A R antagonism might be effective in reducing the ''priming'' of LID. However, once LID is established, adenosine/A 2A R antagonism might exacerbate dyskinetic symptoms. Our assumption may corroborate the notion that A 2A R activation might be required for dyskinesia ''priming'' mechanism (Brotchie, 2005).

CONCLUSION
Because Gα olf protein level serves as a determinant of cAMP signal-dependent activity in both D1-cells and D2-cells in the striatum, Gα olf may represent an ideal target for the modulation of striatal functions under physiological and pathological conditions. Dysregulation of Gα olf expression has been associated with the pathophysiology of several brain disorders (Hervé, 2011). Of our particular interest is that the GNAL gene, which encodes Gα olf , is a causative gene in primary (torsion) dystonia (Fuchs et al., 2013). This is direct evidence that Gα olf plays a pivotal role in the ''motor loop'' of the cortico-basal ganglia circuits. Under parkinsonian conditions, dopamine depletion results in a crucial D 1 R hypersensitivity in the striatum, which leads to the beneficial effects of L-DOPA in PD patients, but also generates LID. In this study, we found that in the 6-OHDA-lesioned striatum of PD mice, daily pulsatile administrations of L-DOPA may cause usage-induced changes in striatal Gα olf levels, leading to increased responsiveness to dopamine stimulation in both D1-cells and D2-cells. Thus we suggest that L-DOPA-induced changes in Gα olf levels in the dopamine-depleted striatum may be a key event in LID development.

AUTHOR CONTRIBUTIONS
SG conceived and designed the experiments. RM, SO and SG performed the experiments; analyzed the data; contributed reagents/materials/analysis tools. SG wrote the manuscript.