Introduction

Behavioural sensitization is characterized by augmented locomotor activity in response to repeated and intermittent exposure to drugs [1]. Repeated exposure to drugs, such as cocaine, amphetamine, or morphine, induces behavioural sensitization in animals during the induction phase and causes more robust psychomotor sensitization following drug challenge after a withdrawal period (expression phase) [2]. Psychostimulant-induced behavioural sensitization is believed to greatly contribute to the reinforcing properties of drugs and plays important roles in relapse after cessation of drug abuse [2].

Studies on the cellular and molecular mechanisms underlying behavioural sensitization have focused on the mesolimbic dopamine system, which originates from ventral tegmental area (VTA) axons projecting primarily to the nucleus accumbens (NAc) [2, 3]. The VTA plays a role in the induction of behavioural sensitization, whereas the NAc plays a greater role in the expression of behavioural sensitization [4,5,6]. Repeated exposure to psychostimulants modulates neuronal activity, morphology and synaptic plasticity in the NAc, leading to the development of behavioural sensitization [6,7,8]. Motor execution of drug-seeking responses is closely associated with synaptic membrane trafficking of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) on NAc neurons. Repeated exposure to cocaine followed by a withdrawal period changes the surface expression and redistribution of neuronal AMPARs in the NAc [9]. Repeated exposure to drugs results in changes in synaptic strength and AMPAR surface levels in the NAc [6, 10]. Moreover, repeated contingent or noncontingent exposure to cocaine followed by an abstinence session triggers an increase in the AMPA/NMDA ratio and alters AMPAR surface expression in the NAc [11, 12]. However, the neuronal signals that modulate membrane trafficking of AMPARs during drug-induced behavioural sensitization have not been fully elucidated.

Mammalian target of rapamycin (mTOR) is a serine/threonine protein kinase that belongs to the PI3K-related kinase family. It is the catalytic subunit of two functionally distinct protein complexes, mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2); regulates the synthesis of synaptic proteins at dendrites; and regulates synaptic plasticity [13,14,15]. Disruption of NAc mTOR signalling suppresses drug-induced behavioural sensitization [16, 17], cocaine-evoked drug reinstatement [18], and morphine-induced conditioned place preference (CPP) [19] by regulating mTOR or p70 ribosomal S6 kinase (S6K) phosphorylation. In addition, upstream regulatory factors of mTOR or mTOR kinase, such as growth factors, amino acids and lipids, contribute to mTOR activation [13, 20, 21]. mTOR functions as an effector of PI3K-AKT signalling and is regulated by PI3K-AKT signalling [22, 23]. However, the role of PI3K in mTOR activation in cocaine-induced behavioural sensitization has not been addressed.

Synapse differentiation-induced gene I (SynDIG1) is a highly conserved type II integral membrane protein with a large intracellular N-terminal region and a single transmembrane domain [24, 25]. The majority of SynDIG1 overlaps with GluA2, an AMPAR subunit, at synapses, indicating that it plays a role in regulating the synaptic distribution of AMPARs [24]. Moreover, SynDIG1 induces synapse maturation by increasing AMPAR expression at synapses, and its extracellular C-terminal region is required for its association with AMPAR clusters as well as SynDIG1-mediated excitatory synaptic strength [24]. These findings suggest that SynDIG1 is an important regulator of synaptic strength because it directly regulates AMPAR expression at synapses.

In the present study, we demonstrated that SynDIG1 is a downstream effector of mTOR signalling and functions in maintaining synaptic surface expression of GluA2 through SynDIG1–GluA2 interaction. Suppression of mTOR signalling enhances the SynDIG1–GluA2 interaction in the NAc and increases the surface expression of GluA2, eventually attenuating cocaine-induced behavioural sensitization.

Materials and methods

Animals

Male C57BL/6J wild-type mice were purchased from Vital River Laboratory Animal Technology Co. Ltd (Beijing, China), and 8- to 12-week-old mice were used in this study. All mice were housed in the animal room with a standard 12-h light/12-h dark cycle at a constant temperature and provided food and water ad libitum. All experimental procedures and animal protocols were conducted in accordance with the guidelines established by the Association for Assessment and Accreditation of Laboratory Animal Care and the Institutional Animal Care and Use Committee of Sichuan University. All efforts were made to minimize the suffering of the mice.

Locomotor activity

Locomotor activity was measured as the distance travelled. Animals were acclimated to a chamber (48 cm × 48 cm) equipped with a camera for 15 min once a day for 3 consecutive days. Baseline locomotor activity was not significantly different between groups. The mice were given an intraperitoneal (i.p.) injection of cocaine (15 mg/kg; days 1~5) or an equal volume of saline, immediately placed in the chamber, and allowed to explore for 15 min. After 10 days of no injection (withdrawal session), the mice received a challenge dose of cocaine (15 mg/kg), and locomotor activity was measured (day 16). The distance travelled was measured daily, and automated tracking was performed with EthoVision 7.0 software (EthoVision 7.0; Noldus Information Technology, Leesburg, VA).

Drugs

Cocaine was purchased from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China) and dissolved in saline. Rapamycin (S1039) and LY294002 (S1105) were purchased from Selleck Chemicals (Houston, USA).

Stereotactic surgery and administration of inhibitor

Mice were anaesthetized with sodium pentobarbital (60 mg/kg) and mounted in a standard stereotaxic instrument (RWD Life Science). The scalp was incised to expose the skull, and permanent bilateral guide cannulas (RWD Life Science) were bilaterally implanted into the NAc (AP, +1.6 mm; ML, ±1.5 mm; and DV, −4.4 mm) with stereotaxic instruments. After recovering from surgery for 1 week, the mice were subjected to subsequent behavioural training. LY294002 was dissolved in saline containing 4% DMSO, 30% PEG300 and 5% Tween 80 at a final concentration of 5 μg/μL. Rapamycin was dissolved in saline containing 4% DMSO, 5% PEG300 and 3% Tween 80 at a concentration of 3 μg/μL.

Adeno-associated viral injection

A doxycycline-inducible adeno-associated virus (AAV), AAV2/9-SynDIG1-EGFP, was designed and produced by Vigene Biotechnology Co. Ltd (Shandong, China). Mice (8–12 weeks old) were anaesthetized with sodium pentobarbital (60 mg/kg) and placed on a stereotaxic apparatus to inject virus into the NAc (AP, +1.6 mm; ML, ±1.5 mm; and DV, −4,4 mm). Microsyringe needles were used to bilaterally infuse AAV2/9-SynDIG1-EGFP (0.5 μl, 0.05 μl/min) into the NAc. Mice were allowed to recover for at least 2 weeks before behavioural tests were performed. Mice were fed doxycycline (0.02% in 5% sucrose water) to switch on exogenous gene expression only during the withdrawal period (days 6–15).

Tissue isolation

Mice were sacrificed by rapid decapitation at the conclusion of the behavioural tests. The striatum, hippocampus and NAc were removed from the brain, snap-frozen on dry ice, and stored at −80 °C until assay.

Western blot analysis

Brain tissues were lysed, and proteins were extracted using a mammalian cell and tissue extraction kit (K269-500, BioVision). Total protein concentration was analyzed with a Bradford assay kit (P0006, Beyotime). Protein (20 μg) was loaded and separated in a 10% sodium dodecyl sulfate-polyacrylamide gel. Immunoreactivity was visualized using a chemiluminescence substrate (WBKLS0500, Millipore) with a chemiluminescence imaging system. Band optical density was quantified using Chemi analysis software (CLINX, Shanghai, China). The following antibodies were used for Western blotting: Rabbit anti-PI3K-p85 (1:1000; Cell Signaling Technology), Rabbit anti-p-PI3K-p85 (T 458/T199) (1:1000; Cell Signaling Technology), Rabbit anti-PI3K-p110α (1:1000; Abcam), Rabbit anti-AKT (1:1000; Abcam), Rabbit anti-p-AKT (S 473) (1:1000; Cell Signaling Technology), Rabbit anti-S6K (1:1000; Proteintech), Rabbit anti-p-S6K (T 389) (1:1000; Cell Signaling Technology), Rabbit anti- SynDIG1 (1:1000; Proteintech), Rabbit anti-GluA2 (1:1000; Abcam), Rabbit anti-β-actin (1:1000; Cell Signaling Technology), Mouse anti-GAPDH (1:2000; ZSGB-BIO), Goat anti-Mouse (HRP) (SAB), Goat anti-Rabbit (HRP) (SAB).

Synaptic protein fractionation

Synaptic protein fractions were extracted from NAc tissue using Syn-PER Synaptic Protein Extract Reagent (87793, Thermo Scientific). According to the manufacturer’s procedures, synaptic proteins were separated. Briefly, tissue was homogenized with Syn-PER Reagent (10 ml/mg) and centrifuged at 1200 × g for 10 min at 4 °C. Synaptosome pellets and cytosolic fractions were separated from the supernatant by centrifugation at 15000 × g for 20 min at 4 °C. The synaptosome pellet was lysed with a mammalian cell and tissue extraction kit (K269-500, BioVision) containing phosphatase inhibitors (4906845001, Roche).

Biotinylation of neuronal surface proteins

The NAc tissue was rinsed with cold artificial cerebrospinal fluid (ACSF) containing (in mM): NaCl 119.0, KCl 3.5, MgSO4 1.3, CaCl2 2.5, NaH2PO4 1.0, NaHCO3 26.2 and glucose 11.0 and incubated with ACSF containing 1.0 mg/mL sulfo-NHS-LC-biotin (21331, Thermo Scientific) for 30 min at 4 °C. Brain slices were rinsed in cold ACSF and incubated for 30 min with ACSF containing 100 mM glycine to neutralize unreacted biotinylation reagent, and subsequently, brain cells were lysed with a mammalian cell and tissue extraction kit (K269-500, BioVision). The protein concentration of each sample was quantified, and equal amounts of protein were incubated overnight with NeutrAvidin coupled-agarose beads (29201, Thermo Scientific). Biotinylated surface and total proteins were measured by Western blot analysis.

Co-immunoprecipitation analysis

Tissues were harvested for co-immunoprecipitation (co-IP) analysis by using the simplified and reliable Pierce™ Crosslink Magnetic IP/co-IP Kit (88805, Thermo Scientific). Briefly, primary anti-GluA2 antibody was bound to 50 μL of protein A/G magnetic beads (B23201, Bimake) for 15 min. After incubation, the protein supernatants were collected, and the protein concentration was measured using a Bradford assay kit. Each sample supernatant was incubated overnight with antibody-cross-linked beads at 4 °C. The following day, beads were washed twice with IP lysis/wash buffer. Elution buffer was used to elute the bound antigen, and neutralization buffer neutralized the low pH. The supernatants were collected for Western blot analysis.

Immunohistochemistry

Mice were anaesthetized with sodium pentobarbital (60 mg/kg) and perfused intracardially with phosphate-buffered saline (PBS), followed by ice-cold 4% paraformaldehyde in 0.1 M PBS (pH 7.4). Brains were carefully extracted from the skull, fixed in 4% PFA overnight, and then dehydrated with 30% sucrose at 4 °C. The brains were sectioned into 45-μm coronal slices using a freezing microtome and stored in 12-well plates filled with cryoprotectant solution at −20 °C. The sections were washed in TBS three times and anti-fade mounting medium was added with DAPI (H-1200, Vector). Confocal images were acquired with a laser confocal microscope (Nikon, Japan).

RNA sequencing

After the end of the locomotor sensitization test, mice were killed immediately, and bilateral NAc tissue was quickly collected for mRNA extraction. The purified cDNA fragments were sequenced by Majorbio Biotechnology Co. (Shanghai, China). The original paired-end readings were trimmed and quality controlled with SeqPrep (https://github.com/jstjohn/SeqPrep) and Sickle (https://github.com/najoshi/sickle) using default parameters. TopHat software (http://tophat.cbcb.umd.edu/, version 2.0.0) was utilized to compare the clean reads with the reference genome in directed mode. The new transcription region was defined as more than two consecutive windows without overlapping gene regions, where each window mapped at least two reads in the same direction. The abundances of the differentially expressed genes (DEGs) were used to quantify gene abundance by RSEM (http://deweylab.biostat.wisc.edu/rsem/). The statistical software package EdgeR (empirical analysis of digital gene expression in R) (Http://www.bioconductor.org/packages/2.12/bioc/html/edgeR.html) was used for differential gene expression analysis. Significance analysis (fold change (FC) > 1.5 and P < 0.05) of the results was performed to identify genes that were strongly up- or downregulated. Biological function and metabolic pathway analyses were performed to identify DEGs that were significantly enriched in a Kyoto Encyclopedia of Genes and Genomes (KEGG) (http://www.genome.jp/kegg/) enrichment analysis.

Untargeted lipidomic analysis with UPLC/QTOF MS/MS

The striatum, hippocampus and NAc were dissected from the brains, snap frozen in liquid nitrogen, and stored at −80 °C until assayed. Liquid–liquid methyl tert butyl ether (MTBE) extraction was based on a protocol used for lipidome analysis. Frozen tissue samples were added to 300 μL of ice-cold methanol and 450 μL of MTBE. The mixture was sonicated in an ice-cold water bath for 3 min, and 250 μL of 25% methanol was added to separate the phases. Samples were then centrifuged at 14,000 × g at 4 °C for 15 min. The upper organic phase was collected and dried under a gentle stream of nitrogen at room temperature. The residue was redissolved in 300 μL of an ACN: isopropanol:water (30:65:5) mixture. Chromatographic separation was performed by gradient elution using a Waters ACQUITY UPLC system (Waters Corp., USA) with a reversed-phase ACQUITY UPLC HSS T3 column (2.1 mm × 100 mm × 1.8 μm) maintained at 55 °C. Mobile phase A was acetonitrile-water (40:60, v/v), and mobile phase B was acetonitrile-isopropanol (10:90, v/v); both phases contained 10 mM ammonium acetate.

Mass spectrometric detection was performed in both positive ion mode and negative ion mode with electrospray ionization using a Xevo G2-S QTOF instrument (Waters Corp., USA). Data were collected in continuum mode from 50–1200 m/z. Leucine-enkephalin was applied to ensure the m/z accuracy of the mass spectrometer, and sodium formic solution was used for TOF mass spectrometer calibration. Data acquisition and processing were performed using MassLynx software (Waters Corp., USA).

Lipidomic data processing and analysis

Progenesis QI software (Waters) was used to analyze the lipidomic data acquired from UPLC/QTOF MS, and the lipids were identified according to their MS characteristics from the Lipid Maps Database (www.lipidmaps.org) and the Human Metabolome Database (http://www.hmdb.ca/). Datasheets in the Progenesis QI software were obtained, and absolute intensities of all identified compounds (normalized abundance) were recalculated to reflect the relative abundances of the lipid molecules (vs. The saline group molecules). The details of the normalization method can be found at the website (http://www.nonlinear.com/progenesis/qi/v2.3/faq/how-normalization-works.aspx). Pareto scaling was used for final statistical models. The data were processed by the supervised partial least-squares discriminate analysis methods to obtain group clusters. Lipid molecules with the highest impact on group clustering were identified in variable importance (VIP) plots (VIP > 1.5). An optimized false discovery rate approach was applied to control false positives.

Statistical analysis

All data were analyzed with GraphPad Prism 7 and are presented as the means ± SEM. Kolmogorov–Smirnov tests were performed to assess the distribution of the data. For simple comparisons, an unpaired two-tailed Student’s t test was used. For multiple comparisons, one-way or two-way ANOVA followed by Bonferroni’s post hoc test was utilized (see the specific experiments and figure legends). In all cases, n refers to the number of animals. For all the results, statistical significance was defined as P < 0.05.

Results

NAc mTOR participates in cocaine-induced behavioural sensitization in mice

Behavioural sensitization requires an initial induction phase (or development phase) followed by a period of drug withdrawal and then re-exposure to the drug to allow for manifestation of sensitization [16]. We first established a mouse behavioural sensitization model by administering 15 mg/kg cocaine daily for 5 continuous days (induction phase). After 10 days of withdrawal, the mice were injected with a single dose of cocaine (15 mg/kg) to assess behavioural sensitization [5] (Fig. 1a). The results showed that the locomotor activity of the mice was significantly increased during the cocaine injection session and that cocaine challenge on day 16 induced a marked increase in the locomotor response (Fig. 1b; treatment: F(1, 25) = 96.5, P < 0.0001; time: F(3.329, 78.80) = 24.60, P < 0.0001; interaction: F(6, 142) = 22.90, P < 0.0001).

Fig. 1: Repression of mTOR signalling attenuates cocaine-induced behavioural sensitization.
figure 1

a Schematic diagram of the experimental procedure of cocaine locomotor sensitization. b Locomotor activity of cocaine-treated mice and saline-treated mice (n = 13 per group; two-way repeated-measures ANOVA). c Timeline of the rapamycin administration procedure. d Pretreatment with rapamycin in cocaine withdrawal period inhibits cocaine-induced locomotor activity (n = 10 per group; two-way repeated-measures ANOVA). e Timeline of the rapamycin administration procedure. f Pretreatment with rapamycin in the induction phase of cocaine-induced behavioural sensitization does not suppress cocaine-induced locomotor activity (n = 10 per group; two-way repeated-measures ANOVA). g Schematic diagram of experimental procedure of rapamycin administration. h Pretreatment with rapamycin in withdraw fails to inhibit saline-exposure mice behavioural sensitization (n = 9 per group; two-way repeated-measures ANOVA). i Schematic diagram showing the experimental procedure of rapamycin administration. j Pretreatment with rapamycin in withdraw fails to inhibit cocaine CPP-induced reinstatement (n = 9 per group; two-way repeated-measures ANOVA). All the data are presented as mean ± SEM. **P < 0.01, ***P < 0.001 and ***P < 0.0001. SV saline + vehicle, SR saline + rapamycin, CV cocaine + vehicle, CR  cocaine + rapamycin.

Full size image

To explore the role of mTOR in behavioural sensitization to cocaine, we tested the effect of rapamycin, a specific inhibitor of mTORC1, on cocaine-induced behaviour. During the final 6 days of the cocaine withdrawal period, the mice were administered rapamycin (10 mg/kg) i.p. daily and then challenged with a single dose of cocaine. Importantly, we found that rapamycin significantly suppressed cocaine challenge-induced hyperlocomotor activity (Fig. 1c, d; treatment: F(3, 36) = 213.9, P < 0.0001; time: F(3.669, 128.4) = 53.75, P < 0.0001; interaction: F(18, 210) = 19.92, P < 0.0001; Bonferroni post hoc test, P < 0.0001). To determine whether rapamycin treatment in the induction phase of behavioural sensitization has a similar effect, mice were injected with rapamycin (10 mg/kg) 1 h before each cocaine injection for 5 consecutive days (Fig. 1e). However, rapamycin had no effect on cocaine-induced locomotor sensitization (Fig. 1f). Next, we also sought to determine whether the role of rapamycin was specific to cocaine-exposed mice. As expected, mice were exposed to saline in the development session (Fig. 1g), and rapamycin did not disrupt the response of mice to cocaine-induced locomotor activity (Fig. 1h). We continued to explore whether administration of rapamycin in withdrawal phase can disrupt cocaine CPP-induced reinstatement. After CPP training, systemic delivery of rapamycin in the withdrawal period failed to decrease the CPP score after cocaine priming (Fig. 1i, j). Collectively, these data indicated that cocaine locomotor sensitization can be repressed by mTOR inhibition in the withdrawal phase but not in the induction phase.

S6K is a substrate downstream of activated mTOR, and phosphorylation of S6K at Thr389 is a marker of mTORC1 activation [6, 16]. To determine whether mTOR is activated during cocaine-induced sensitization, we performed Western blotting to measure the phosphorylation of S6k (Thr389) in the NAc. We found that S6k phosphorylation was significantly increased in the expression phase of cocaine-induced behavioural sensitization (Fig. 2a; p-S6K: t (6) = 2.561, P < 0.05), suggesting that mTOR is involved in this process. We continued to investigate whether rapamycin can directly modulate cocaine-induced behavioural sensitization. The NAc in mice was infused with rapamycin (3 μg per side) for 6 continuous days during the cocaine withdrawal period and then challenged with cocaine (Fig. 2b). Notably, injection of rapamycin into the bilateral NAc significantly attenuated cocaine-induced behavioural sensitization (Fig. 2c; treatment: F(3, 32) = 87.26, P < 0.0001; time: F(6, 186) = 36.41, P < 0.0001; interaction: F(18, 186) = 16.00, P < 0.0001). In addition, rapamycin markedly suppressed cocaine-induced S6K phosphorylation in the NAc (Fig. 2d; p-S6K: F(3, 20) = 9.718, P < 0.001). Next, to determine whether mTOR activation occurred in the withdrawal phase, we detected the levels of S6K and S6k phosphorylation in the induction phase and found that S6K and S6k phosphorylation levels were not changed (Fig. 2e, f). These data showed that cocaine activates NAc mTOR signalling in the withdrawal phase, which may mediate cocaine-induced behavioural sensitization.

Fig. 2: mTOR signalling in the NAc participates in the behavioural sensitization to cocaine.
figure 2

a The levels of S6K and p-S6K were measured by Western blot (n = 4 per group; unpaired t test). b Timeline of rapamycin administration procedure. c Intra-NAc administration of rapamycin (3 μg/side, respectively) attenuates cocaine-induced locomotor activity (n = 9 per group; two-way repeated-measures ANOVA). d The expression levels of S6K and p-S6K were measured by Western blot (n = 6 per group; one-way ANOVA). e Timeline of the experimental procedure. f The levels of S6K and p-S6K were measured by Western blot (n = 4 per group; unpaired t test). All the data are presented as mean ± SEM. *P < 0.05 and **P < 0.01. p phosphorylated, SV saline + vehicle, SR saline + rapamycin, CV cocaine + vehicle, CR  cocaine + rapamycin.

Full size image

Phosphatidylinositol-4-monophosphate (PIPs) may contribute to mTOR activation

Lipids, such as inositol polyphosphate 3 (IP3) and 2-arachidonoyl-sn-glycerol, play pivotal roles in signal transmission since they can act as signalling molecules to activate signalling pathways [21]. To identify the lipid molecules potentially involved in mTOR activation in the NAc in cocaine-sensitized mice, we performed lipid profiling with liquid chromatography coupled with tandem mass spectrometry to analyze lipid alterations in three different brain regions. Well-fitted orthogonal projections to latent structures-discriminant analysis models were constructed, and clear separation in the different brain regions of each treatment group was identified in positive and negative ionization modes. In addition, the results were validated by measuring the goodness of fitness (R2) and the predictability (Q2) of the original model based on the data. Q2 > 0.5 and R2 > 0.9 implied good and excellent predictive abilities in both positive and negative ionization modes (Fig. S1a and S2a, f).

All significantly changed lipids were classified into six groups according to lipid class: sphingolipids (SPs), sterol lipids (SLs), glycerolipids (GLs), glycerophospholipids (GPs), fatty acyls (FAs) and other lipids (others). Interestingly, the greatest proportion of lipids that were altered after cocaine-induced behavioural sensitization in the striatum, hippocampus and NAc were GPs (Figs. S1b, c; S2b, c; S2g, h). Significantly modified molecules were detected in both positive and negative ionization modes (variable importance (VIP) > 1.5 and P < 0.05 were regarded as significantly modified lipids); the NAc exhibited the most obvious lipid alterations, with changes in 186 lipids in 22 subclasses (Fig. S1d), while the striatum and hippocampus presented fewer lipid alterations (Fig. S2d, i). A heat map was used to visualize the degree of change in these modified lipids of the striatum, hippocampus and NAc (Figs. S2e, j; S3a–e).

Using MetaboAnalyst 4.0 software, we analyzed all the altered lipids and found that those altered lipids in the NAc were enriched in SP and GP metabolism (Fig. S3f). Compared with the control, cocaine caused a profound imbalance in lipid metabolism, inducing markedly increased levels of SPs and GPs but decreased levels of GLs (Fig. S3g). Interestingly, the levels of phosphatidylinositol-4-monophosphates (PIPs), including phosphatidylinositol-4-monophosphate (PIP), phosphatidylinositol 4,5-bisphosphate (PIP2) and phosphatidylinositol-3,4,5-trisphosphate (PIP3), were significantly increased in the NAc in cocaine-treated mice (Fig. S3h; PIP (16:0/20:1): t(16) = 5.743, P < 0.001; PIP2(16:0/18:2): t(16) = 3.095, P < 0.01; PIP2(16:0/20:1): t(16) = 3.194, P < 0.01; PIP3(17:0/20:4): t(16) = 3.438, P < 0.01) and exhibited NAc-specific changes (Fig S3h). Since PIP3, a lipid messenger, is able to activate the PI3K-AKT pathway [26], our findings suggested that the PI3K-AKT pathway may be the upstream activator of mTOR in cocaine-induced behavioural sensitization.

Inhibition of PI3K-AKT attenuates cocaine-induced behavioural sensitization

To investigate whether the PI3K-AKT pathway is involved in cocaine-induced behavioural sensitization, we measured the expression levels of p-PI3K-p85, PI3K-p110α and p-AKT in the NAc by Western blotting. The data showed that the expression levels of p-PI3K-p85, PI3K-p110α and p-AKT were significantly increased in cocaine-sensitized mice compared with saline-treated mice (Fig. 3a; p-PI3K-p85: t(6) = 2.654, P < 0.05; PI3K-p110α: t(6) = 2.462, P < 0.05; p-AKT: t(6) = 2.892, P < 0.05), suggesting that PI3K-AKT signalling is activated in response to cocaine. To elucidate the role of PI3K-AKT in locomotor sensitization to cocaine, we used LY294002, an AKT-specific inhibitor, to pharmacologically suppress the PI3K-AKT pathway in the NAc. LY294002 was microinfused into the bilateral NAc (5 μg per side) through an implanted cannula before cocaine challenge (Fig. 3b). Notably, LY294002 significantly attenuated cocaine-induced locomotor activity (Fig. 3c; treatment: F(3, 29) = 110.4, P < 0.0001; time: F(4.449, 121.6) = 43.21, P < 0.0001; interaction: F(18, 164) = 21.62, P < 0.0001). Moreover, the expression levels of PI3K-p110α and phosphorylated AKT were decreased by LY294002 (Fig. 3d; PI3K-p110α: F(3, 20) = 5.005, P < 0.05; p-AKT: F(3, 20) = 5.912, P < 0.05). We then performed a new experiment to determine whether infusion of LY294002 into the NAc in the induction phase can reduce cocaine-induced locomotor activity. The results showed that LY294002 failed to significantly attenuate cocaine-induced locomotor activity when infusion of LY294002 was performed in the induction phase (Fig. 3e, f). Taken together, our results showed that the PI3K-AKT pathway may be activated in the withdrawal phase and that the PI3K-AKT pathway may participate in cocaine-induced behavioural sensitization.

Fig. 3: PI3K-AKT pathway in the NAc is involved in behavioural sensitization to cocaine.
figure 3

a Immunoblotting for PI3K-p85, p-PI3K-p85, PI3K-p110α, AKT and p-AKT expression in the NAc (n = 4 per group; unpaired t test). b Timeline of the LY294002 administration procedure. c Pretreatment with LY294002 decreases the response to a cocaine challenge (n = 8 per group; two-way repeated-measures ANOVA). d Immunoblotting for PI3K-p85, PI3K-p110α, AKT and p-AKT expression in the NAc (n = 6 per group; one-way ANOVA). e Timeline of the LY294002 administration procedure. f Induction phase LY294002 administration failed to attenuate the response to a cocaine challenge (n = 8 per group; two-way repeated-measures ANOVA). All the data are presented as mean ± SEM. *P< 0.05 and **P < 0.01. p phosphorylated, SV saline + vehicle, SL saline + LY294002, CV cocaine + vehicle, CL cocaine + LY294002.

Full size image

Identification of the downstream target of mTOR in the NAc through transcriptome analysis

To determine the downstream target of mTOR in cocaine-induced behavioural sensitization, we used RNA sequencing to assess the alterations in gene expression in the NAc in the presence or absence of rapamycin treatment. For the RNA-sequencing analysis, we used the criteria of FC > 1.5 and P < 0.05 to identify DEGs. The results showed that the expression of 309 genes was significantly changed in cocaine-sensitized mice compared with saline-treated mice; moreover, the expression of 371 genes was significantly changed in cocaine + rapamycin-treated mice compared with cocaine-sensitized mice. Among these altered genes, 258 and 320 genes showed unique expression in cocaine-sensitized mice and cocaine + rapamycin-treated mice, respectively, and 51 genes overlapped (Fig. S4a). A heat map was used to visualize the degree of change in the expression of these 51 overlapping genes (Fig. S4b, c). Through bioinformatics analysis, we identified the cellular processes and biological functions in which these genes may be involved. The data showed that these altered genes were mainly enriched in GnRH signalling, the MAPK signalling pathway and morphine addiction (Fig. S4d). Interestingly, GO analysis revealed that four significantly altered genes, namely, SynDIG1, NPAS4, DNAH11 and EOMES, play important roles in nervous system development, synaptic plasticity and/or postsynaptic maturation. Although the gene expression levels of the SynDIG1, DNAH11 and EOMES genes were decreased by cocaine, only the change in SynDIG1 gene expression was significantly reversed by rapamycin. Considering that SynDIG1, a brain-specific transmembrane protein, is necessary for excitatory synapse development and maturation via alterations in the level of AMPARs at excitatory synapses [27], we focused on the potential function of SynDIG1 in mediating cocaine-induced behavioural sensitization.

NAc SynDIG1 is involved in cocaine-induced locomotor sensitization

Western blot analysis revealed that the expression level of SynDIG1 was significantly reduced during the expression of cocaine-induced locomotor sensitization (Fig. 4a, b; t (6) = 2.475, P < 0.05), implying that cocaine has a regulatory effect on SynDIG1 expression in the NAc. However, the level of NAc SynDIG1 was not altered at the end of the induction phase (Fig. 4c, d) or in the early withdrawal phase (days 5–10) of cocaine-induced behavioural sensitization (Fig. 4e, f). To investigate whether the SynDIG1 level in the NAc is regulated by mTOR, we pharmacologically inhibited NAc mTOR signalling through systemic (Fig. 5a) or intra-NAc delivery of rapamycin into cocaine-sensitized mice (Fig. 5c). Notably, SynDIG1 expression was markedly reduced in the cocaine-sensitized mice, whereas this reduction was reversed by either intraperitoneal (Fig. 5b; F(3, 36) = 4.734, P < 0.05) or intra-NAc administration of rapamycin during the cocaine withdrawal period (days 10–15) (Fig. 5d; F(3, 20) = 4.571, P < 0.05), suggesting that rapamycin may suppress cocaine-induced behavioural sensitization by regulating SynDIG1 levels. We further tested the effect of SynDIG1 overexpression in the NAc on cocaine-induced neurobehaviours. The Tet system is one of the most widely used drug-inducible transgene expression systems. It is based on the release of the Escherichia coli Tet repressor protein (TetR) from its bound tet operator (TetO) sequence upon the addition of tetracycline or its derivative dox. The addition of doxycycline triggers reverse-tTA (rtTA) binding to the TetO-containing promoter and switches on target gene expression [28]. Doxycycline-inducible AAV2/9-SynDIG1-EGFP was bilaterally infused into the mouse NAc (0.5 μL/side, 0.05 μL/min). After 2 weeks of recovery, the mice were administered cocaine daily for 5 days and then fed doxycycline (0.02% in 5% sucrose water) during the withdrawal period (days 6–15) (Fig. 6a). By directly visualizing EGFP expression in the NAc, we found that doxycycline successfully switched on exogenous SynDIG1 gene expression and induced SynDIG1 overexpression (Fig. 6b). Notably, overexpression of SynDIG1 in the NAc significantly decreased the responses of the mice to cocaine challenge (Fig. 6c; treatment: F(3, 36) = 279.0, P < 0.0001; time: F(4.116, 148.2) = 29.53, P < 0.0001; interaction: F(18, 216) = 14.25, P < 0.0001; Fig. 6d; F(3, 20) = 12.11, P < 0.0001), indicating that SynDIG1 plays a negative modulatory role in cocaine-induced motor sensitization.

Fig. 4: mTOR regulates SynDIG1 expression in the NAc in the cocaine-sensitized mice.
figure 4

a, c and e Timeline of the experimental procedure. b, d and f Immunoblotting for SynDIG1 expression in the NAc (n = 4 per group; unpaired t-test). All the data are presented as mean ± SEM. *P < 0.05.

Full size image
Fig. 5: NAc SynDIG1 is involved in the cocaine-induced behavioural sensitization.
figure 5

a and c Experimental schedule for systemic or intracranial administration of rapamycin. b and d Representative immunoblotting of SynDIG1 in the NAc with rapamycin administration (n = 6 per group; one-way ANOVA). All the data are presented as mean ± SEM. *P < 0.05. SV saline + vehicle, SR saline + rapamycin, CV cocaine + vehicle, CR  cocaine + rapamycin.

Full size image
Fig. 6: SynDIG1 overexpression in withdrawal period suppresses the response to cocaine challenge.
figure 6

a Experimental schedule for AAV-based SynDIG1 overexpression in the cocaine withdrawal period. b Representative images of SynDIG1 expression in the NAc (green); DAPI, nucleus (blue). c Overexpression of SynDIG1 in withdrawal period attenuates the response to a cocaine challenge (n = 10 per group; two-way repeated-measures ANOVA). d Immunoblotting for SynDIG1 expression in the NAc (n = 6 per group; one-way ANOVA). e Western blot analysis shows the expression of GluA2 in the NAc or synaptic compartment of NAc (n = 6 per group; one-way ANOVA). f The effect of SynDIG1 overexpression on the surface or intracellular distribution of GluA2 in the NAc (n = 4 per group; one-way ANOVA). g Endogenous GluA2 was immunoprecipitated with antibodies against GluA2 from NAc tissue lysates (n = 3 per group). Rabbit immunoglobulin G (IgG) was used as control. All the data are presented as mean ± SEM. *P < 0.05 and **P < 0.01. SA saline + AAV-EGFP, SO saline + overexpression of SynDIG1, CA cocaine + AAV-EGFP, CO cocaine + overexpression of SynDIG1.

Full size image

SynDIG1 modulates cocaine-induced behavioural sensitization by interacting with GluA2

GluA2-containing Ca2+-impermeable AMPARs are critical for excitatory transmission onto medium spiny neurons, which constitute the major neuron type in the NAc [29, 30]. Previous studies have shown that neuronal surface expression of GluA2 is involved in locomotor sensitization to cocaine [6, 9] and that the majority of GluA2 molecules interact with SynDIG1 at the synapse [24]. We first measured the expression level of GluA2 in the NAc and the synaptic compartment in the NAc. Western blot analysis showed that cocaine treatment had no effect on GluA2 expression in either the NAc or NAc synaptic compartment (Fig. 6e). In addition, AAV2/9-mediated overexpression of SynDIG1 did not change the expression level of GluA2 in either the NAc or the NAc synaptic compartment (Fig. 6e). Although total GluA2 expression in the NAc was not altered by cocaine, we wondered whether cocaine treatment can affect the membrane trafficking of GluA2 in the NAc. To this end, we measured the surface expression of GluA2 by labelling cell-surface receptors with biotin. Importantly, we found that the surface level of GluA2 was significantly decreased, whereas the intracellular level of GluA2 was increased in cocaine-sensitized mice. Furthermore, this increase in the intracellular level of GluA2 was markedly reversed by AAV-mediated overexpression of SynDIG1 in the NAc (Fig. 6f; surface: F(3, 12) = 12.66, P < 0.001; intracellular: F(3, 12) = 14.23, P < 0.001), indicating that SynDIG1 rescued the cocaine-induced reduction in the surface expression of GluA2. We continued to investigate whether cocaine-induced transport of GluA2 to the cell surface is mediated by the interaction of GluA2 with SynDIG1. Co-IP revealed that cocaine significantly decreased the SynDIG1–GluA2 interaction during cocaine-induced locomotor sensitization, whereas this effect was attenuated by overexpression of SynDIG1 (Fig. 6g; F(3, 8) = 17.22, P < 0.001). Collectively, these results suggested that SynDIG1 may play a role in maintaining proper expression and synaptic trafficking of GluA2 by interacting with GluA2. Cocaine downregulates the expression of SynDIG1, decreases the interaction of SynDIG1 with GluA2, and promotes the intracellular translocation of GluA2 in the NAc, which may eventually contribute to cocaine-induced locomotor sensitization (Fig. S5).

Discussion

Sensitization is a process by which repeated exposure to the same stimulus results in greater responses over time and is believed to strongly contribute to relapse [31, 32]. Here, we show that mTOR signalling plays an important role in the expression of behavioural sensitization to cocaine. By performing a lipidomic analysis, we discovered that cocaine significantly modified the lipid profile in the NAc and activated the PI3K-AKT pathway. Cocaine markedly decreased the expression of NAc SynDIG1, a downstream target of mTOR signalling. The decrease in the SynDIG1–GluA2 interaction promoted intracellular translocation of GluA2, which mediated cocaine behaviour sensitization. These findings reveal that NAc mTOR signalling participates in cocaine-induced behaviour sensitization by modulating the synaptic surface trafficking of GluA2. Thus, selective targeting of the PI3K-AKT-mTOR signalling pathway or the SynDIG1–GluA2 interaction may be a potential therapeutic strategy for cocaine addiction.

mTOR signalling initiates translational machinery by directly phosphorylating S6K at its hydrophobic motif site, Thr389, and activates several substrates that promote mRNA translation initiation [13]. Intracranial administration of rapamycin inhibits cue-induced reinstatement in mice subjected to cocaine self-administration by reducing S6K phosphorylation in the NAc [18]. In the morphine CPP paradigm, the phosphorylation of both mTOR and S6K is increased in the hippocampal CA3 region, and it is suppressed by rapamycin, which induces a decrease in the levels of phosphorylated mTOR and S6K [19]. These studies suggest a role for mTOR signalling in drug addiction and reward. In the present study, cocaine activated NAc mTOR signalling during behavioural sensitization, and suppression of mTOR activation by either systemic or intracranial administration of rapamycin blocked the expression of locomotor sensitization; both of these results indicate an important role for mTOR in cocaine-induced locomotor sensitization. This notion is supported by previous studies showing that rapamycin affects the development of cocaine-induced behaviour sensitization and that mTOR signalling activation in the NAc is critical for locomotor sensitization [16, 17].

The PI3K-AKT pathway contributes to mTOR activation in cocaine-induced behavioural sensitization

mTOR signalling is activated by several upstream signalling factors that are involved in both physiological and pathological events, such as growth factors and energy and metabolic pathways. For instance, insulin-like growth factor 1 activates mTOR through the PI3K-AKT signalling pathway in ovarian cancer cells [33], and amino acid intermediates involved in lipid and ATP synthesis are necessary for mTOR activity[13]. Moreover, mTOR is activated by nicotine via nicotinic acetylcholine receptors or by dopamine neurotransmitters in behaviourally sensitized mice [34, 35]. Our data revealed that, among brain regions, the NAc showed the most robust alteration in lipid profiles, supporting the idea that NAc lipid metabolism may participate in drug-induced behaviour sensitization. Importantly, among the lipid molecules modified in the three analyzed brain regions, PIP, PIP2 and PIP3 were significantly upregulated only in the NAc in cocaine-sensitized mice. PIP3, a lipid signalling messenger, binds AKT and induces the translocation of AKT from the cytoplasm to the cell membrane, further activating AKT through AKT self-phosphorylation at Thr308 and Ser473 [26]. We demonstrated that the expression of cocaine-induced sensitization was associated with parallel promotion of PI3K-AKT signalling in the NAc and that a PI3K-AKT signalling pathway inhibitor was able to attenuate the manifestations of cocaine sensitization [8, 36]. Collectively, cocaine-induced upregulation of PIP3 expression may trigger PI3K-AKT signalling pathway activation and further activate mTOR signalling. PIP3 is generated by the phosphorylation of PIP2, a reaction primarily catalyzed by four class I catalytic isoforms (PIK3CA, PIK3CB, PIK3CG, and PIK3CD) [37]. How cocaine regulates PIP3 production in behavioural sensitization remains an important question to be answered. Nevertheless, our data provided a clue suggesting that phospholipase A2 (PLA2) may be involved in this process. We discovered that among all the lipid molecules altered in the NAc in cocaine-treated mice, SP and GP were enriched, indicating an alteration in lipids metabolism. Interestingly, the expression of PI, PIP2 and PIP3 was upregulated, but that of PC and LysoPC was downregulated. Considering that PLA2 hydrolyses PC to LysoPC [38], we inferred that PLA2 may be inactivated by cocaine during the development of behavioural sensitization. This notion is supported by a previous study showing that PLA2 expression is reduced in dopamine-rich brain regions in individuals addicted to cocaine or methamphetamine [39] and that abnormal GP metabolism is evident in the NAc in cocaine-addicted mice [40]. We propose that increased levels of PIP3 may trigger PI3K-AKT signalling pathway activation in the NAc during cocaine-induced behaviour sensitization, contributing to the activation of mTOR.

SynDIG1–GluA2 interaction modulates behavioural sensitization to cocaine

Drug-triggered synaptic plasticity and AMPAR signalling have been causally linked to behavioural sensitization [41], and enhanced AMPAR transmission can abolish locomotor sensitization and cue-induced drug seeking [42]. AMPARs are composed of four subunits, GluA1 to GluA4. Since GluA2 is present at most synapses in the adult brain, its interacting proteins may regulate endogenous AMPAR subunit composition, which plays an important role in mediating synaptic plasticity [6]. SynDIG1 participates in the development and maturation of excitatory synapses by modulating AMPAR expression at excitatory synapses [27]. Interestingly, GluA2 and SynDIG1 largely overlap at synaptic sites, suggesting that SynDIG1 and AMPARs may be trafficked to synapse together [24, 27]. In addition, the loss of SynDIG1 results in decreased density of GluA2 at the surface [24]. We observed that repeated exposure to cocaine followed by withdrawal markedly decreased the surface expression of AMPARs in the NAc; nevertheless, overexpression of SynDIG1 in the NAc significantly restored the surface expression of GluA2 by promoting the SynDIG1–GluA2 interaction and attenuating cocaine-induced behavioural sensitization. This finding is in line with a previous finding showing that SynDIG1 is required for the surface expression of AMPARs and that SynDIG1 regulates AMPAR content at synapses [24]. We propose that maintenance of surface GluA2 expression through the interaction of GluA2 with SynDIG1 at excitatory synapses may confer protection against cocaine-induced sensitization. A recent study reported that intra-NAc injection of a peptide (Tat-GluR23Y) specifically inhibits the endocytosis of postsynaptic GluA2 and attenuates behavioural sensitization to amphetamine [43]. Similarly, prevention of GluA2 removal from synapses or disruption of membrane trafficking of GluA2 through suppression of PICK1 expression in the NAc decreases the reinstatement of cocaine seeking [44, 45]. These results are consistent with our findings that maintaining the surface expression of GluA2 may suppress cocaine-induced behavioural sensitization.

Collectively, our data support the hypothesis that cocaine-induced disruption of the SynDIG1–GluA2 interaction in the NAc may be a novel mechanism underlying cocaine-induced behavioural sensitization and suggest that the SynDIG1–GluA2 interaction can serve as a potential therapeutic target for cocaine-induced locomotor sensitization.