Inhibition of striatonigral autophagy as a link between cannabinoid intoxication and impairment of motor coordination

The use of cannabis is rapidly expanding worldwide. Thus, innovative studies aimed to identify, understand and potentially reduce cannabis-evoked harms are warranted. Here, we found that Δ9-tetrahydrocannabinol, the psychoactive ingredient of cannabis, disrupts autophagy selectively in the striatum, a brain area that controls motor behavior, both in vitro and in vivo. Boosting autophagy, either pharmacologically (with temsirolimus) or by dietary intervention (with trehalose), rescued the Δ9-tetrahydrocannabinol-induced impairment of motor coordination in mice. The combination of conditional knockout mouse models and viral vector-mediated autophagy-modulating strategies in vivo showed that cannabinoid CB1 receptors located on neurons belonging to the direct (striatonigral) pathway are required for the motor-impairing activity of Δ9-tetrahydrocannabinol by inhibiting local autophagy. Taken together, these findings identify inhibition of autophagy as an unprecedented mechanistic link between cannabinoids and motor performance, and suggest that activators of autophagy might be considered as potential therapeutic tools to treat specific cannabinoid-evoked behavioral alterations.


Introduction
Cannabis is one of the most common drugs of abuse in the world (Alpár et al., 2016;Englund et al., 2017;Volkow et al., 2014). Consequently, its major intoxicating constituent, the cannabinoid D 9 -tetrahydrocannabinol (THC), is the third most popular recreational addictive chemical following ethanol and nicotine. Of note, several states in the USA, as well as a few countries in the world, have legalized the recreational use of cannabis. Cannabis preparations have also been used in medicine for millennia, and nowadays there is a vigorous renaissance in the study and application of their therapeutic effects (Pertwee, 2012). In this context, THC and other cannabinoids are already approved by various regulatory agencies, including the Food and Drug Administration (FDA), the European Medicines Agency and Health Canada, as anti-emetic, anti-cachexic, analgesic and anti-spastic compounds (Hill, 2015;Whiting et al., 2015). Moreover, medical-grade cannabis dispensation programs have been implemented in about half of the states in the USA and in a growing number of countries globally. However, cannabis use is associated to several undesired and possibly dangerous side effects, so it is crucial that innovative procedures aimed to understand and potentially reduce cannabis-evoked harms are explored (Alpár et al., 2016;Englund et al., 2017;Volkow et al., 2014).
THC exerts its biological effects mainly by activating cannabinoid CB 1 receptor, one of the most abundant metabotropic receptors in the mammalian central nervous system (Katona and Freund, 2008;Pertwee et al., 2010). This receptor is particularly expressed in discrete brain areas involved in the control of learning and memory (cortex, hippocampus), motor behavior (striatum, cerebellum), emotions (amygdala), and autonomic and endocrine functions (hypothalamus, pons, medulla), therefore participating in the control of a wide plethora of biological processes (Katona and Freund, 2008;Mechoulam and Parker, 2013). A family of retrograde lipid messengers, the endocannabinoids, biologically engages the CB 1 receptor, mediating a feedback mechanism aimed to prevent excessive neuronal activity and, thereby, tuning the functionality and plasticity of many synapses (Castillo et al., 2012;Piomelli, 2003). Recent evidence suggests that the CB 1 receptor can control autophagy, a highly conserved and pleiotropic process of cellular 'self-digestion' in which cytoplasmic materials are sequestered into double-membrane vesicles called autophagosomes, and subsequently delivered to lysosomes for degradation or recycling (Costa et al., 2016;Hiebel and Behl, 2015). Autophagy is an essential mechanism of cellular quality control, and the knowledge on its biological functions in the brain and other organs is rapidly increasing (Menzies et al., 2017;Ohsumi, 2014). Strikingly, in some cell-culture settings cannabinoids via the CB 1 receptor enhance autophagy (Koay et al., 2014;Salazar et al., 2009), while in others they inhibit autophagy (Hiebel et al., 2014;Piyanova et al., 2013). Moreover, it is not known yet whether the CB 1 receptor controls autophagy in the brain in vivo, and, eventually, what the functional consequences of this potential CB 1 receptor/autophagy connection could be. Here, we show that THC inhibits autophagy selectively in the mouse striatum, and that this process participates in the THC-induced impairment of motor coordination. Moreover, administration of clinically safe autophagy activators to mice prevents the dyscoordinating effect of THC. These findings unveil an unprecedented link between cannabinoids, autophagy and motor performance, and provide preclinical evidence for the design of potential new therapeutic strategies aimed at treating specific cannabinoid-induced behavioral alterations.

THC impairs striatal autophagy both in vivo and in vitro
To study the effect of THC on autophagy in the brain we first treated wild-type mice with a single i.p. injection of the drug at 10 mg/kg or its vehicle. After 4 hr, we evaluated the status of key autophagy protein markers. This dose and time window allows assessing persistent and pharmacologically tractable behavioral actions of THC administration, as previously reported (Metna-Laurent et al., 2017;Puighermanal et al., 2013). We analyzed the expression pattern of microtubule-associated light chain three protein (LC3), the most widely used marker of autophagic vesicles (autophagosomes) (Mizushima et al., 2011;Ohsumi, 2014), in representative brain structures. Upon induction of autophagy, LC3 is converted from a soluble, non-lipidated form (LC3-I) to an aggregated, phosphatidylethanolamine-conjugated form (LC3-II), thereby becoming recruited to autophagosomal membranes (Mizushima et al., 2011). THC increased LC3-II levels in the striatum, either when referred to LC3-I (THC vs. vehicle, t = 4.680; df = 10; p=0.0009) or to b-actin (THC vs. vehicle, t = 4.331; df = 10; p=0.0015) as control, but not in other representative brain regions as the cortex, the hippocampus and the cerebellum ( Figure 1A). An elevation of LC3-II levels, however, may indicate either that THC increases autophagosome generation (and so increases autophagic flux) or that THC decreases autophagosome clearance (and so decreases autophagic flux) (Mizushima et al., 2011;Ohsumi, 2014). To discern between these two possibilities, we measured the levels of p62 (sequestosome 1), a pivotal adaptor protein that carries cargo proteins to the autophagosome, being subsequently degraded upon fusion of the autophagosome to the lysosome (autophagolysosome or autolysosome) (Katsuragi et al., 2015;Mizushima et al., 2011). Hence, an Figure 1. THC impairs autophagy in the mouse striatum. Wild-type C57BL/6N mice were treated with THC (10 mg/kg as a single i.p. injection) or its vehicle. Four hours later, the striatum (St), cortex (Cx), hippocampus (Hc) and cerebellum (Cb) were dissected for Western blot analysis. (A) Effect of THC on autophagy markers in the different brain regions. (B) Relative levels of LC3-I and p62 in the different brain regions from vehicle-treated animals. In both panels, representative blots of each condition, together with optical density values relative to those of the respective loading controls, are shown (n = 6 animals per group). Blots were cropped for clarity. Electrophoretic migration of molecular weight markers is depicted on the right-hand side of each blot. **p<0.01 from vehicletreated group by unpaired Student t-test. Raw numerical data and further statistical details are shown in Figure 1-source data 1. The online version of this article includes the following source data for figure 1: Source data 1. Source data for THC impairs autophagy in the mouse striatum. increase of LC3-II together with a decrease of p62 usually denotes an active clearance of autophagosomes (and so an increased autophagic flux), while a simultaneous elevation of LC3-II and p62 usually defines an impaired clearance of autophagosomes (and so a decreased autophagic flux) (Katsuragi et al., 2015;Mizushima et al., 2011). THC induced an accumulation of p62 in the striatum (THC vs. vehicle, t = 5.303; df = 10; p=0.0003), but not in the other brain regions tested ( Figure 1A), thus indicating that THC impairs the execution of autophagy and that this process occurs selectively in the striatum. Of note, the levels of LC3-I and p62 were not significantly different in the striatum than in the cortex, hippocampus or cerebellum from vehicle-treated mice ( Figure 1B), suggesting that the selective impact of THC on striatal autophagy does not rely on the basal expression of those two key autophagy proteins but on additional, hitherto unknown molecular factors.
Taken together, these data support that cannabinoids inhibit autophagic flux in striatal neurons both in vivo and in vitro.
Temsirolimus prevents the THC-induced impairment of striatal autophagy and motor coordination in vivo THC and other cannabinoids modulate various intracellular signalling pathways in the brain by engaging CB 1 receptors (Castillo et al., 2012;Pertwee et al., 2010). One of the most relevant CB 1 receptor-evoked actions is the activation of the phosphatidylinositol-3-kinase/Akt/mammalian target of rapamycin complex 1 (mTORC1) pathway (Blázquez et al., 2015;Gó mez del Pulgar et al., 2000;Ozaita et al., 2007;Puighermanal et al., 2009). The serine/threonine kinase mTOR, the catalytic component of mTORC1, is critically involved in the control of neural plasticity through the regulation of protein synthesis and other basic cellular functions (Bockaert and Marin, 2015;Lipton and Sahin, 2014). Of note, mTORC1 is also the most relevant signaling platform that exerts a negative control on autophagy by phosphorylating UNC-51-like kinase 1 (ULK1), and so inhibiting autophagosome formation (Dunlop and Tee, 2014;Saxton and Sabatini, 2017). However, it is not known whether a cannabinoid-evoked activation of the mTORC1 pathway would be linked to an inhibition of autophagy, and, especially, what the biological consequences of this process could be.
Taken together, these findings provide further support to the notion that an inhibition of striatal autophagy participates in the motor-dyscoordinating activity of THC.
Cannabinoid CB 1 receptors located on the direct pathway, but not on cortical projections, are required for the THC-induced impairment of striatal autophagy and motor coordination in vivo We subsequently studied the neuroanatomical substrate of the observed THC effects. As THC exerts most of its neurobiological effects by activating CB 1 receptors, we first tested the effect of the CB 1 receptor-selective antagonist SR141716 (rimonabant) on the THC-evoked inhibition of motor coordination. Mice were treated with rimonabant (3 mg/kg, i.p.) or vehicle for 20 min, and, subsequently, with THC (10 mg/kg, i.p.) or vehicle. Four hours after injection, THC impaired RotaRod performance, and rimonabant, under conditions that did not influence behavior by itself, abrogated the effect of THC (F (3,16) = 12.86; THC vs. vehicle, p=0.0020; rimonabant + THC vs. THC, p=0.0002) (Figure 5figure supplement 1).
Taken together, all these findings indicate that CB 1 receptors located on D 1 R-MSNs, but not on corticostriatal projections, are required for the autophagy-inhibiting and motor-dyscoordinating activity of THC. Figure 5. Cannabinoid CB 1 receptors located on D 1 R-MSNs, but not on glutamatergic neurons, are required for the THC-induced impairment of striatal autophagy and motor coordination in vivo. (A-C) D 1 R-CB 1 R KO mice and CB 1 R-floxed control littermates were treated with THC (10 mg/kg as a single i. p.injection) or its vehicle for 4 hr. Panel A, Motor coordination (RotaRod test, time to fall relative to pre-treatment; n = 7 animals per group). Panel B, Western blot analysis of autophagy markers in the striatum. Representative blots of each condition, together with optical density values relative to those of loading controls, are shown (n = 5 animals per group). Blots were cropped for clarity. Electrophoretic migration of molecular weight markers is depicted on the right-hand side of each blot. Panel C, Immunofluorescence analysis of p62 (p62 fluorescence intensity per DARPP32-positive cell) in the dorsal striatum (n = 4 animals per group). Representative images with an encircled example of high-intensity cell are shown. (D-F) Glu-CB 1 R KO mice and CB 1 R-floxed control littermates were treated with THC (10 mg/kg as a single i.p. injection) or its vehicle for 4 hr. Panel D, Motor coordination (RotaRod test, time to fall relative to pre-treatment; n = 5 animals per group). Panel E, Western blot analysis of autophagy markers in the striatum. Representative blots of each condition, together with optical density values relative to those of loading controls, are shown (n = 5 animals per group). Blots were cropped for clarity. Electrophoretic migration of molecular weight markers is depicted on the right-hand side of each blot. Panel F, Immunofluorescence analysis of p62 (p62 fluorescence intensity per DARPP32-positive cell) in the dorsal striatum (n = 4 animals per group). Representative images with an encircled example of high-intensity cell are shown. *p<0.05, **p<0.01 from the corresponding vehicle-treated group, or # p<0.05, ## p<0.01 from the corresponding THC-treated CB 1 R-floxed group, by two-way ANOVA with Tukey's multiple comparisons test. Raw numerical data and further statistical details are shown in Figure 5-source data 1. The online version of this article includes the following source data and figure supplement(s) for figure 5:

Discussion
Here, we identify impairment of autophagy as an unprecedented mechanism involved in cannabinoid-induced motor alterations. On molecular grounds, our data favour a 'two-hit' model by which engagement of striatal CB 1 receptors may impair autophagy. First, CB 1 receptor activation, by coupling to the phosphatidylinositol-3-kinase/Akt/mTORC1 pathway, would lead to ULK1 phosphorylation, which, subsequently, would inhibit autophagosome formation/autophagy initiation. Second, CB 1 receptor activation, by a hitherto undefined mechanism that may conceivably involve an impact on lysosomal function (Hiebel and Behl, 2015), would inhibit autophagosome clearance/autophagy completion. We are aware, however, that our work has several shortcomings that could limit the generalization of its conclusions. Specifically, (i) the data (except for the cell-culture experiments) come from a single cannabinoid agonist (THC) given at a single dose (10 mg/kg, i.p.), and were collected at a single time point after administration (4 hr); (ii) only two (albeit well-established) motor behavior measures were examined (RotaRod and open field); and (iii) only male animals were studied.
By targeting mTORC1 with temsirolimus, we report a feasible pharmacological intervention to rescue the concerted THC-evoked impairment of autophagy and motor coordination. Temsirolimus prevents other unwanted effects of THC, such as short-term memory loss and anxiety, leaving potential therapeutically sought cannabinoid actions as analgesia and anxiolysis unaffected (Puighermanal et al., 2013). Temsirolimus has similar potency and specificity for mTOR than rapamycin, but longer stability and increased solubility, and is already approved by the FDA as first-line treatment for metastatic renal cancer patients classified as poor risk (Hudes et al., 2007). In these patients, temsirolimus is well-tolerated, increases overall survival, and improves quality of life (Zanardi et al., 2015). Taken together, these pieces of evidence suggest that administration of temsirolimus, or other FDA-approved rapalogs like everolimus (Janku et al., 2018;Lebwohl et al., 2013), might help to counteract some particular unwanted effects of cannabis.
Dietary manipulation with trehalose also prevented the THC-evoked impairment of autophagy and motor dyscoordination. Trehalose, a nontoxic disaccharide found in numerous plants, microorganisms and invertebrates, contains an a,aÀ1,1-glucoside bond between two a-glucose units, thus being an extremely stable sugar. In many countries, including USA, trehalose is added to various food products as nutritional supplement and 'natural flavor' (Richards et al., 2002). On physiological grounds, trehalose is believed to stabilize proteins and to protect them from stress-induced unfolding, aggregation and degradation (Emanuele, 2014;Hosseinpour-Moghaddam et al., 2018). Vertebrates cannot synthesize trehalose, but exogenous trehalose administration induces the clearance of toxic protein aggregates in cultured mammalian cells, and exerts therapeutic effects in a plethora of mouse models of protein-misfolding disorders (including Huntington's disease, Alzheimer's disease, Parkinson's disease and amyotrophic lateral sclerosis) concomitantly to autophagy induction (Hosseinpour- Moghaddam et al., 2018;Menzies et al., 2017). Although its mechanism of action is not completely understood (Lee et al., 2018), trehalose has been proposed to activate autophagy via competitive inhibition of GLUT glucose transporters, thus impairing cellular energy supply and stimulating AMPK (DeBosch et al., 2016). In our hands, however, the phosphorylation state of the main AMPK-dependent site in ULK1 remained unaffected upon trehalose treatment. We also observed that, in line with some reports (e.g. Sarkar et al., 2007), trehalose did not affect the basal activity of mTORC1-pathway molecular markers; and, in line with other reports (e.g. DeBosch et al., 2016), it attenuated stimulus-evoked mTORC1 overactivation. Thus, it is likely that AMPK, mTORC1, and the contextual crosstalk between these two pivotal signalling axes (Alers et al., 2012) are required for the full pro-autophagic effects of trehalose to be observed.
We also define here the neuroanatomical basis for the autophagy-inhibiting and motor-dyscoordinating actions of THC. The CB 1 receptor is one of the most abundant metabotropic receptors in the striatum, where it is mainly expressed in D 1 R-MSNs, D 2 R-MSNs, GABAergic interneurons, and Animals were subsequently treated with THC (10 mg/ kg as a single i.p. injection) or its vehicle for 4 hr, and motor coordination was evaluated (RotaRod test, time to fall in seconds; n = 5 animals per group). **p<0.01 from vehicle-treated WT/post-treatment group, or ## p<0.01 from THC-treated WT/post-treatment group, by two-way ANOVA with Tukey's multiple comparisons test. Representative images of c-myc tag and phosphorylated ribosomal protein S6 staining in the dorsal striatum, together with their quantification (c-myc-positive cells relative to total cells, or phospho-S6-positive cells relative to total cells), are shown (n = 4 animals per group). **p<0.01 from WT group by unpaired Student t-test (c-myc immunofluorescence); *p<0.05, **p<0.01 from vehicle-treated/WT group, or ## p<0.01 from THC-treated/WT group, by two-way ANOVA with Tukey's multiple comparisons test (phospho-S6 immunofluorescence). (B) D 1 R-Cre mice and wild-type control littermates were injected stereotactically into the dorsal striatum with a CAG-DIO-p62 rAAV, and left untreated for 4 weeks. Animals were subsequently treated with THC (10 mg/kg as a single i.p. injection) or its vehicle for 4 hr, and motor coordination was evaluated (RotaRod test, time to fall in seconds; n = 5 animals per group). *p<0.05 from vehicle-treated WT/post-treatment group, or # p<0.05 from the respective WT/ pre-treatment group, by two-way ANOVA with Tukey's multiple comparisons test. Representative images of p62 staining in the dorsal striatum, together with their quantification (p62 fluorescence intensity relative to total cells), are shown (n = 4 animals per group). **p<0.01 from WT group by unpaired Student t-test. Raw numerical data and further statistical details are shown in Figure 6-source data 1.
The online version of this article includes the following source data and figure supplement(s) for figure 6: Source data 1. Source data for mTORC1 and p62 in D 1 R-MSNs participate in the THC-induced impairment of motor coordination in vivo. Figure 6 continued on next page astrocytes, as well as in glutamatergic terminals projecting from the cortex (Castillo et al., 2012;Davis et al., 2018;Uchigashima et al., 2007). This complex anatomical profile dictates an intricate repertoire of modulatory actions controlled by endocannabinoids through different CB 1 receptor pools, ranging from synaptic plasticity (Covey et al., 2017;Kreitzer, 2009) to astrocyte-neuron communication (Araque et al., 2017) and neuronal integrity (Chiarlone et al., 2014;Naydenov et al., 2014). Specifically, our data show that the pool of CB 1 receptors located on D 1 R-MSNs plays an indispensable role in cannabinoid-induced impairment of autophagy and motor coordination. Of note, it has been shown that this precise CB 1 receptor subpopulation is also necessary for cannabinoid-induced catalepsy, although not for overall cannabinoid-induced hypomotility (Monory et al., 2007), thus supporting that, in agreement with our data, it controls selected aspects of motor behavior. In addition, CB 1 receptors located on corticostriatal terminals, by controlling glutamatergic signalling, contribute to THC-induced hypolocomotion (Monory et al., 2007), participate in endocannabinoid-dependent long-term depression as evoked by D 1 R-MSNs (Bagetta et al., 2011;Wu et al., 2015) and protect D 1 R-MSNs from toxic insults (Ruiz-Calvo et al., 2018). Thus, endocannabinoid signalling fine-tunes the functions and viability of D 1 R-MSNs through a delicate armamentarium of CB 1 receptor pools located on both D 1 R-MSNs, presynaptic terminals impinging on them, and other surrounding cell types. We note that our work does not unveil the precise cellular and molecular mechanisms by which the CB 1 receptor-evoked inhibition of autophagy in D 1 R-MSNs affects brain functionality to change motor coordination. Neuronal communication is finely sensitive to proteostatic processes as autophagy, which, for example, clears dysfunctional proteins and fine-tunes the trafficking/recycling of membrane neurotransmitter receptors (e.g. ionotropic glutamate receptors; Birdsall and Waites, 2019). Neuronal activity is associated to the mTORC1 pathway and autophagy, and this could in turn participate in NMDA receptor-dependent synaptic plasticity and brain function (Shehata et al., 2012). Hence, the control of long-term depression exerted by CB 1 receptors on D 1 R-MSNs (Bagetta et al., 2011;Wu et al., 2015) might be mechanistically connected to the THC-evoked effects on mTORC1/autophagy reported here, perhaps by signaling in a cell-autonomous manner through the accumulation of the multifunctional scaffold protein p62 (Sánchez-Martín and Komatsu, 2018). These possibilities notwithstanding, our findings might also be relevant in other neurobiological processes that are known to be controlled by the striatum and impacted by cannabinoidsfor example cognition, affection and reward (Katona and Freund, 2008;Kreitzer, 2009;Lovinger, 2010). Moreover, from a translational point of view, it is tempting to speculate that D 1 R-MSNs, but not corticostriatal terminals, would constitute the neuroanatomical target of strategies aimed at managing some specific cannabis-induced behavioral alterations as catalepsy and dyscoordination. Continued on next page

Animals
We used conditional mutant mice, generated by the Cre-loxP technology, in which the CB 1 receptor gene (Cnr1) is absent either from D 1 R-expressing neurons (Cnr1 fl/fl mice bred with Drd1a Cre mice; referred to here as D 1 R-CB 1 R KO mice) (Monory et al., 2007) or from dorsal telencephalic glutamatergic neurons (Cnr1 fl/fl mice bred with Neurod6 Cre mice; referred to here as Glu-CB 1 R KO mice) (Monory et al., 2006), as well as their respective Cnr1 fl/fl (referred to here as CB 1 R-floxed) littermates. We also used BAC transgenic mice expressing the tdTomato and EGFP reporter genes under the control of the Drd1a gene promoter and Drd2 gene promoter, respectively (Drd1a-tdTomato; Drd2-EGFP mice; colony founders kindly provided by Rosario Moratalla, Cajal Institute, Madrid, Spain) (Suárez et al., 2014). Wild-type C57BL/6N mice were purchased from Harlan Laboratories (Barcelona, Spain). Animal housing, handling and assignment to the different experimental groups were conducted essentially as described before (Bagetta et al., 2011). Throughout the study, animals had unrestricted access to food and water. They were housed (4-5 mice per cage) under controlled temperature (range, 20-22˚C), humidity (range, 50-55%) and light/dark cycle (light between 8:00 am and 8:00 pm). Animals were habituated to housing conditions before the start of the experiments, were assigned randomly to the different treatment groups, and all experiments were performed in a blinded manner for genotype, pharmacological treatment and viral injection. All animals used in the experiments were male adults (ca. 8-week-old). Adequate measures were taken to minimize pain and discomfort of the animals, as well as the number of animals used in the experiments, on the basis of the 3Rs (replacement, reduction and refinement) principle. Mice were sacrificed either by intracardial perfusion with paraformaldehyde (and their brains subsequently excised for histological analyses) or by cervical dislocation (and their striata, or other brain regions, subsequently dissected for Western blot analyses). All the experimental procedures were performed in accordance with the guidelines and with the approval of the Animal Welfare Committee of Universidad  Figure 5-figure supplement 1), mice were treated with rimonabant (3 mg/kg as a single i.p. injection) or vehicle for 20 min, and, subsequently, with THC (10 mg/kg as a single i.p. injection) or vehicle for 4 hr. Temsirolimus (LC Labs, Woburn, MA, USA) was prepared fresh in DMSO just before the experiments. Mice were treated with temsirolimus (1 mg/kg as a single i.p. injection) or its vehicle for 20 min, and, subsequently, with THC (10 mg/kg as a single i.p. injection) or its vehicle for 4 hr. Trehalose (Merck-Calbiochem, Barcelona, Spain) was directly added to the drinking water of the animals. Mice were given trehalose (10 g/L) or plain water ad libitum for 24 hr, and, subsequently, were treated with THC (10 mg/kg as a single i.p. injection) or its vehicle for 4 hr. Under these conditions, the addition of trehalose did not affect the volume of water that was drunk by the animals. The doses of temsirolimus and trehalose used were selected from both previous studies (Puighermanal et al., 2013;Rodríguez-Navarro et al., 2010) and pilot experiments on motor behavior.

Viral vectors
The coding sequence of dominant-negative Raptor with a c-Myc tag (Myc-RaptorDCT) was generated by PCR-mediated deletion of 1293 base pairs at the C-terminus (Hara et al., 2002) of wild-type Myc-tagged Raptor (Addgene, Watertown, MA; plasmid #1859) (Koketsu et al., 2008). Myc-Rap-torDCT or human HA-tagged p62 (Addgene; plasmid #28027) was subcloned in a CAG-DIO rAAV vector, to allow the Cre-dependent expression of the transgene, by using standard molecular biology techniques. The vectors used were of an AAV1/AAV2-mixed serotype and were generated by calcium phosphate transfection of HEK-293T cells (American Type Culture Collection, Manassas, VA) and subsequent purification, as described previously (Bellocchio et al., 2016). Wild-type and D 1 R-Cre mice were injected stereotactically with the rAAV vector into the dorsal striatum. Each animal received one bilateral injection at the following coordinates (to bregma): antero-posterior +0.5, lateral ±2.0, dorso-ventral À3.0 (Bellocchio et al., 2016). Mice were left untreated for 4 weeks to attain transgene expression before being subjected to the behavioral tests.

Motor behavior
Motor coordination analysis (RotaRod test) was conducted with acceleration from 4 to 40 r.p.m. over a period of 600 s in an LE8200 device (Harvard Apparatus, Barcelona, Spain) (Blázquez et al., 2011). Any mouse remaining on the apparatus after 600 s was removed, and its time was scored as 600 s. RotaRod performance was evaluated in two phases. First, before any pharmacological treatment, naive mice were tested on three consecutive days, for three trials per day, with a rest period of 40 min between trials. Data from the three trials conducted on the first day were not used in statistical analyses, as they merely reflect the initial contact of the animal with the RotaRod device (Hockly et al., 2003). Data from the three trials conducted on the second day plus the three trials conducted on the third day were averaged for each animal, so constituting the herein referred to as 'pre-treatment' RotaRod performance. Second, on the day of the pharmacological experiment, 4 hr after vehicle or drug treatment, mice were tested for three trials, with a rest period of 40 min between trials. Data from these three trials were averaged for each animal, so constituting the herein referred to as 'post-treatment' RotaRod performance. Hence, for each animal, its 'post-treatment' RotaRod performance was compared with its 'pre-treatment' RotaRod performance. Motor activity analysis was conducted in an automated actimeter consisting of a 22.5 Â 22.5 cm area with 16 surrounding infrared beams coupled to a computerized control unit (ActiTrack; Panlab, Barcelona, Spain) (Blázquez et al., 2011). Four hours after vehicle or drug treatment, animals were recorded once for a period of 10 min, in which total distance travelled (cm), overall activity (counts), resting time (s), fast movements (counts) and stereotypic movements (counts) were measured.

Statistics
Unless otherwise specified, data are presented as mean ± SEM of the number of animals or independent neuronal preparations indicated in each case. Statistical comparisons were made by unpaired Student t-test, or by ANOVA followed by Tukey's multiple comparisons test, as indicated in each figure legend. For clarity, only p values lower than 0.05 were considered statistically significant. The source data files include all raw numerical data as well as further details of the statistical analyses, which were carried out with GraphPad Prism 8.0 software (San Diego, CA). Power analysis was conducted with IBM SPSS software (IBM France, Bois-Colombes, France).