Introduction

Substance use disorder (SUD) is a chronic, relapsing disease in which patients often have uncontrolled and reckless drug abuse [1]. With the increasing number of drug abusers, drug abuse not only affects the health of individuals but also endangers society and imposes a huge economic burden on the state [2]. Evidence has revealed that repeated addictive drug exposure leads to enduring cellular, circuits and neuroplasticity alterations in mesocorticolimbic reward system, which includes prefrontal cortex (PFC), nucleus accumbens (NAc) and ventral tegmental area (VTA) [3,4,5,6,7,8]. Such long-lasting structural and functional modifications are thought to increase sensitivity to the motivational effects of addictive drugs, culminating in a loss of control over intake [9, 10].

The medial prefrontal cortex (mPFC), a critical hub for integrating different brain networks involved in reward processing, salience attribution, and inhibitory control, is proposed to precipitate bingeing and relapse, perpetuating the addiction cycle [11,12,13]. Accumulating evidence demonstrates that complex drug-induced neuroadaptations in mPFC are mediated largely through dynamic regulation of gene expression [14,15,16]. Epigenetic mechanisms, such as DNA methylation, histone modifications, and certain types of noncoding RNAs like circRNAs, are known to modulate expression of gene networks in mPFC and other drug reward-associated brain regions, contributing to the drug-induced structural, synaptic, and behavioral plasticity [17,18,19,20]. However, the precise molecular mechanisms driving these changes in the mPFC remain unclear.

Circular RNAs (circRNAs), vastly conserved non-coding RNAs and naturally occurring in a covalently closed loop structure, are produced by back-splicing [21, 22]. Numerous neuronal circRNAs are derived from synaptic gene locus, with expression levels changing in parallel with the synapse formation and neuroplasticity [23]. Evidence shows that circRNAs, like circHomer1, may play a key role in various neuropsychiatric disorders [24,25,26,27,28]. CircHomer1, derived from Homer protein homolog 1 (Homer1), is a neuronal-enriched circRNA abundantly expressed in the frontal cortex [29]. CircHomer1 knockdown in the mouse orbitofrontal cortex leads to specific deficits in cognitive flexibility [29]. And its inhibition appears to ameliorate methamphetamine-induced neuronal injury through inhibiting Bbc3 expression in HT-22 cells [30]. Moreover, circHomer1 regulates the expression of synaptic-related proteins and is involved in synaptic plasticity, learning, and memory, which are abnormal symptoms of patients with SUD [28,29,30,31]. Nonetheless, the specific influence of circHomer1 on drug-induced rewarding effects is largely unknown.

In this study, we utilized the conditioned place preference (CPP) model, self-administration model, viral-mediated circHomer1 expression, fluorescence in situ hybridization, and patch clamp to investigate the cell type-specific influence of circHomer1 on drug-induced behavioral and synaptic plasticity. We demonstrated that repeated cocaine exposure notably downregulated the expression of circHomer1 in the prelimbic cortex (PrL). However, recovering the expression of circHomer1 in the PrL reduced cocaine preference and intake, and repaired the hypoactivity of PrL neuron induced by cocaine. This indicated that circHomer1 may be a key modulator for the cocaine responses.

Material and methods

Animals

Male Sprague-Dawley rats (260 – 280 g) used in this study were obtained from Charles River Co., LTD. (Beijing, China). The rats were housed four to a cage in the animal care facility at a temperature (22 °C ~ 24 °C)- and humidity (40% ~ 60%)-controlled facility with a reverse 12 h-light/12 h-dark cycle (8:00–20:00) with food and water available ad libitum. The behavioral experiments were conducted during the dark phase of the cycle. All of the procedures were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and the experiments were approved by the Biomedical Ethics Committee for Animal Use and Protection of Peking University.

Our estimates of the number of animals that were required for behavioral testing (n ≥ 7 per group), Ex vivo electrophysiology (n ≥ 3 per group), immunofluorescence (n ≥ 3 per group), circHomer1 expression (n ≥ 5 per group), and spine analysis (n ≥ 3 per group) were based on our past experience in the laboratory and previous research. All rats were randomly assigned to the experimental and control groups. Experiments were not performed blindly in the research, but data collection and analyses were carried out blinded by using a computer-based recording system that was not influenced by human bias. Rats were excluded from the analysis based on the accuracy of injection site of viruses or guide cannulas location.

Drug treatment

Cocaine, methamphetamine, and morphine were purchased from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China) and were dissolved in 0.9% physiological saline. SCH23390 (SCH) (D054, Sigma, USA) and raclopride (RAC) (R121, Sigma, USA) were also dissolved in 0.9% physiological saline. Clozapine N-oxide (CNO) (C0832, Merck, USA) was dissolved in dimethylsulfoxide (DMSO) and then diluted with 0.9% physiological saline.

For chronic or acute cocaine exposure, cocaine (10 mg/kg) was administered via intraperitoneal (i.p.) injection respectively for repeated 14 days or one day. For the CPP test, cocaine (10 mg/kg or 5 mg/kg) were intraperitoneally injected, while methamphetamine (1 mg/kg) and morphine (10 mg/kg) were subcutaneously injected. SCH (2.0 μg/0.5 μl/side), RAC (5.0 μg/0.5 μl/side) and CNO (1.5 mg/kg, i.p.) were injected 30 min before cocaine injection.

Recombinant adeno-associated virus (rAAV)

To manipulate circHomer1 expression, recombinant adeno-associated virus serotype 9 (rAAV9) vectors expressing enhanced green fluorescent protein (eGFP) were employed. For ubiquitous overexpression, we used rAAV9 with a cytomegalovirus (CMV) promoter driving circHomer1 (rAAV-CMV-circHomer1-GFP, referred to as CMV-OE-circHomer1). For targeted knockdown, a vector expressing short hairpin RNA (shRNA) against circHomer1 under the U6 promoter was used (rAAV-U6-sh-circHomer1-GFP, referred to as sh-circHomer1). To achieve excitatory neuron-specific overexpression, circHomer1 was driven by the calcium/calmodulin-dependent protein kinase II alpha (CamKIIα) promoter (rAAV-CamKII-circHomer1-GFP, referred to as CamKII-OE-circHomer1). Corresponding control vectors included rAAV-CMV-GFP (referred to as CMV-GFP), rAAV-CamKII-GFP (referred to as CamKII-GFP), and rAAV-U6-sh-control-GFP (referred to as sh-control). All aforementioned circHomer1 manipulation and control vectors were obtained from Vigene Bioscience (Shandong, China).

For chemogenetic experiments, the rAAV9 virus that specifically activates D1 neurons [rAAV-D1-hM3D(Gq)-mCherry-WPREs] was purchased from BrainVTA, Wuhan, China.

Stereotaxic microinjection

The rats were anesthetized with isoflurane (4–5% for induction, 1–2% for maintenance) and placed in a stereotaxic frame (RWD, Shenzhen, China). The skull of the rat was then exposed and leveled. The rAAV was bilaterally delivered into the PrL at the following coordinates: anterior-posterior (AP): + 0.30 cm, medio-lateral (ML): ± 0.06 cm, dorso-ventral (DV): - 0.40 cm. The infusion rate was 0.06 μl/min for a total volume of 0.5 μl. The microinjector was left in the place for 10 min after microinjection to allow for the diffusion of the rAAV complexes. After surgery, penicillin sodium (0.02 mg/kg, i.p.) were injected to the rats once daily for five consecutive days to prevent infection. Rats were housed with free access to food and water and provided with standard care. Twenty-one days after surgery, rats that were microinjected with rAAV were used for behavioral tests.

Conditioned place preference (CPP)

The procedures for CPP training were based on previous studies with minor modifications [32]. The apparatus for CPP conditioning and testing consisted of three polyvinyl chloride boxes, which were identical except for their floors. The boxes had two large side chambers (27.9 cm long × 21.0 cm wide × 20.9 cm high) with a different type of floor (bar or grid), separated by a smaller chamber (12.1 cm long × 21.0 cm wide × 20.9 cm high) with a smooth polyvinyl chloride floor. In each box, the three chambers were separated by manual guillotine doors.

Before the CPP training, the rats were handled for seven days to acclimate to the experimenter’s touch. To determine baseline preference, rats were placed in the middle chamber and given 15 min to freely explore the three compartments with the door open. A computer recorded the time that each rat spent in each compartment to determine the baseline preference. Rats must spend approximately one-third of their time in each chamber; otherwise they will be excluded. Each rat was trained for 8 consecutive days with alternating injections of drug (cocaine, 10 mg/kg or 5 mg/kg, i.p.; methamphetamine, 1 mg/kg, s.c.; morphine, 10 mg/kg, s.c.) and saline (1 ml/kg, i.p. or s.c.) [32,33,34]. After each injection, the rats were placed in the corresponding conditioning chambers and then returned to their home cage 45 min later. The day after the last conditioning session, all rats were allowed to explore the three compartments freely for 15 min under conditions identical to those described in the baseline test. The CPP score was calculated as the time that each rat spent in the drug-paired chamber minus the time spent in the saline-paired chamber [35].

For SCH and RAC pre-treatment, guide cannulas (24 gauge; RWD, Shenzhen, China) were bilaterally implanted 1 mm above the PrL. The coordinates for PrL were the following: AP: +0.30 cm, ML: ± 0.06 cm, DV: −0.28 cm; 0.5 μl, 16° angle. SCH/RAC/vehicle was separately delivered into the PrL 30 min prior to cocaine injection. Other procedures were consistent with those described above. For CNO pre-treatment, CNO (1.5 mg/kg) was intraperitoneally injected 30 min before cocaine injection. All other procedures were identical to those described above.

Cocaine self-administration procedures

The procedures for cocaine self-administration (SA) training were based on previous studies with minor modifications [32]. Rats were anesthetized with isoflurane (4–5% for induction, 1–2% for maintenance). Catheters were inserted into the right jugular vein with the tip terminating at the opening of the right atrium as previously described. All rats were allowed to recover for 4–5 days after surgery. The operant chambers used (AniLab Software and Instruments) were equipped with two nose poke operandi (ENV-114M; Med Associates) that were 5 cm above the floor of the chambers. Nose pokes in the active operandum led to cocaine infusions that were accompanied by a 5-s tone-light cue. Nose pokes in the inactive operandum were also recorded but had no programmed consequences. The rats were trained to self-administer intravenous cocaine hydrochloride (0.75 mg/kg/infusion) during three 1-h daily sessions separated by 5 min over 7 days. A fixed-ratio one (FR1) reinforcement schedule was used. Each injection was accompanied by the illumination of a cue light above the active nose poke, followed by an additional 20 s timeout period when the cue and house lights were extinguished and additional nose poke responses had no programmed consequence. The number of drug infusions was limited to 20 per hour. At the end of the training phase, all rats underwent extinction. The conditions were the same as during training, except that drug was no longer available. The rats were given extinction training until responding on the active nose poke operandum decreased to less than 20% of the mean responding during the last three days of cocaine self-administration for at least two consecutive days.

Cocaine induced reinstatement test

Before the SA reinstatement test, the rats received a saline (1 ml/kg) or cocaine (10 mg/kg) injection. Conditions during the reinstatement test were the same as those during cocaine self-administration training.

RNA extraction, cDNA synthesis, and quantitative real-time PCR (qPCR)

Sample tissues were isolated, and total RNAs were extracted using TRIzol reagent (Invitrogen, Catalog no. 15596–026). RNA concentration and quality were determined using a NanoDrop spectrophotometer (Thermo Scientific, USA). 500 ng of total RNA from each sample was reverse transcribed into cDNA using the HiScript II 1st Strand cDNA Synthesis Kit (+gDNA wiper) (Vazyme Biotech, Catalog no. R212-01). The reaction parameters were 25 °C for 5 min, 50 °C for 50 min and 85 °C for 15 s. All cDNA samples were stored at −20 °C for further use. The qPCR was performed with SYBR Master Mix (QPK201, TOYOBO, Japan) in a QuantStudio 5 Real-Time PCR System (Applied Biosystems) under the following conditions: 95 °C for 2 min and 40 cycles of 95 °C for 30 s, 60 °C for 30 s. Relative expression levels were determined using the 2−ΔΔCt method [36]. GAPDH mRNA was used as reference genes to analyze Homer1a, Homer1b/c and circHomer1. Specific divergent primers designed at the backsplice junction of circHomer1 were used to detect its expression. All primers were obtained from previous study and listed in Supplementary Table 1 [23].

Basescope in situ hybridization

To detect the distribution and quantification of circHomer1 in different cells, we customized an RNA probe set that specifically targets the junction sites of circHomer1 (Cat No. 1055641-C1, Advanced Cell Diagnostics, USA). The procedures for Basescope in situ hybridization were based on previous studies with minor modifications [37, 38]. Twenty-four hours after the last cocaine injection, the rats were perfused with with 0.1 mol/L phosphate-buffered saline (PBS) and 4% paraformaldehyde (PFA; pH 7.4). The brains were then removed and post-fixed in 4% paraformaldehyde for 24 h. Afterward, the brains were then dehydrated in 30% sucrose (w/v) dissolved in 0.1 mol/L phosphate buffer. The brains were coronally sectioned at 12 μm using a sliding microtome, and five to six sections spanning the rostrocaudal axis of the PrL were collected and stained, with each section being taken from different rats. All sections were washed three times for 5 min each with phosphate buffered saline (0.1 mol/L PBS), and then treated with protease for 10 min. After washing off the protease, we incubated brain sections with the probe sets specifically targeting circHomer1 for 2 h at 40 °C in the HybEZTM oven. Following probe incubation, sections underwent a series of incubations with preamplifier probes, amplifier probes, and fluorescently labeled probes at 40 °C. We acquired fluorescent images using the fluorescence microscope (Olympus, Tokyo, Japan) with a 20× objective lens. Fluorescent dot numbers were obtained for each specific probe set (red dots defined as those for circHomer1).

Immunofluorescence

To detect the co-localization of circHomer1 and cell marker, the immunofluorescence based on previous studies with minor modifications was used [38, 39]. The sections hybridized with the probe were incubated with the primary antibody in PBST (PBS containing 0.05% triton-100) with 1% normal goat serum overnight at 4 °C. The following primary antibodies were used in our experiments: mouse anti-CamKⅡ (1:100, sc-13141, Santa Cruz Biotechnology), mouse anti-GAD67 (1:100, ab26116, Abcam), mouse anti-NeuN (1:100, MAB377, Millipore), rabbit anti-GFAP (1:500, ab23922, Abcam), rabbit anti-Iba1 (1:500, ab178847, Abcam), and rabbit anti-Dopamine Receptor D1 (D1R) (1: 100, ab279712, Abcam). The sections were then washed three times in PBST and incubated with the indicated secondary antibodies for 3 h at room temperature. The following secondary antibodies were used in our experiments: Alexa Fluor 488 goat anti-rabbit IgG (for GFAP, D1R and Iba1, 1:500; ab150077, Abcam), and Alexa Fluor 488 goat anti-mouse IgG (for CamKⅡ, GAD67, and NeuN, 1:500; ab150113, Abcam). In order to determine the specificity of CamKⅡ, we utilized Alexa Fluor 594 goat anti-mouse IgG (1:1000; ab150116, Abcam) as the secondary antibody. Finally, after three additional washes with PBST, the sections were mounted on Antifade Mounting Medium (S2110, Solarbio) and imaged using the fluorescence microscope (Olympus, Tokyo, Japan). Four to six images were randomly selected from individual animals for counting the co-localization of circHomer1 and cell markers. The co-localization of cells is defined as GFP cell markers surrounded by red circHomer1 dots.

Dendritic spine counting

Coronal sections (60 μm thick) were processed for imaging. All images were acquired at a resolution of 1,024 × 1,024. High-resolution z-stacks of GFP-positive cells were acquired using a 60× oil-immersion lens from each PrL section per rat, with a step size of 1 μm. High-resolution z-stacks of randomly selected secondary or tertiary dendritic branch segments from individual cells were acquired for spine counting. The acquisition was done with a 5× optical zoom for the dendritic branch segments. Four to six pyramidal neurons were randomly selected from each individual animal, and each neuron was scanned for spines counting on 4-6 secondary or tertiary dendrites. The dendritic spines are classified into four categories [40]: mushroom spines, long-thin spines, stubby spines and filopodia spines. Mushroom spine is defined as having a head diameter larger than 0.5 μm and a head-to-neck diameter ratio greater than 1:1 [40, 41]. The length of the spine is more than twice the average neck width, and the average neck width is less than or equal to the maximum width of the head, which are classified into long-thin spine [40]. The spine with the maximum width of the head is less than or equal to the average neck width is classified as filopodia and with a total length of less than 1 μm is classified as stubby [40].

Ex vivo electrophysiology

The experiments were performed as previously described [42]. The rats were deeply anesthetized with chloral hydrate and then decapitated. The brains were swiftly extracted and immersed in a frigid cutting solution consisting of the following composition: 87 mM NaCl, 26 mM NaHCO3, 3.0 mM KCl, 1.0 mM NaH2PO4, 1.3 mM MgSO4, 1.5 mM CaCl2, 20 mM D-glucose, and 75 mM sucrose saturated with a mixture of 95% O2 and 5% CO2. Coronal sections that encompassed the PrL region were delicately sliced at a thickness of 230 μm using a vibratome (Leica VT1200), while being immersed in the same ice-cold cutting solution enriched with 95% O2 and 5% CO2. The slices were subsequently transferred to a specialized incubation chamber, in which they were submerged in artificial cerebrospinal fluid (aCSF) comprising the following constituents: 124 mM NaCl, 26 mM NaHCO3, 3.0 mM KCl, 1.0 mM NaH2PO4, 1.3 mM MgCl2, 2 mM CaCl2, and 20 mM D-glucose. This aCSF solution was meticulously saturated with a blend of 95% O2 and 5% CO2, simulating the optimal physiological conditions. Initially, the slices were incubated at a precise temperature of 34 °C for a duration of 30 min. Subsequently, they were maintained at room temperature for a minimum of an additional 30 min before being employed for recording purposes.

Each individual slice was carefully transferred to a submerged chamber, where it experienced continual exposure to aCSF that had been thoroughly saturated with a precise blend of 95% O2 and 5% CO2, flowing at a controlled rate of 2 ml/min, regulated by a meticulous flowmeter. Initially, the slice was examined using a 4× objective, facilitating the precise localization of the PrL region, which was identified by its proximity to the forceps minor corpus callosum and the midline. Subsequently, under the illumination of near-infrared light, layer V of the PrL was observed using a high-resolution 40× water-immersion objective. The identification of layer V pyramidal cells was meticulously conducted based on their cellular morphology, size, and distinctive electrophysiological properties.

All experimental procedures were meticulously carried out under a controlled temperature of 32 °C. For the preparation of electrodes, thick-wall borosilicate glass was skillfully manipulated using a horizontal puller, resulting in electrodes with resistances ranging from 2.5 to 3.5 MΩ. Ensuring optimal electrical connectivity, the seal resistance exceeded the threshold of 1 GΩ. To capture cellular activity at the soma, whole-cell recordings were conducted using the advanced MultiClamp 700B amplifier.

For the current-clamp recordings, the finely crafted pipette solution consisted of a precise composition: 120 mM potassium gluconate, 10 mM KCl, 4 mM ATP-Mg, 0.3 mM GTP, 10 mM HEPES, 5 mM Na2-phosphocreatine, and 2 mM EGTA (with a pH value of 7.2 and an osmolarity of 270–280 mOsm, achieved through the addition of sucrose). To maintain neuronal membrane potentials around −60 mV, a gradual current was carefully applied. Series resistance was impeccably compensated for utilizing the bridge circuit integrated within the esteemed MultiClamp 700B amplifier. To evaluate the spike rate, the frequency of spikes was meticulously recorded in discreet 500 ms intervals. These results were then visually represented in relation to the intensity of the applied current. Within 20 seconds of establishing whole-cell configuration, the resting potentials of the neurons were accurately measured. Input resistance was methodically ascertained by administering hyperpolarizing current pulses of either −50 pA or −100 pA, inducing voltage shifts ranging from 5 to 15 mV below the resting membrane potential. The threshold, a key measurement, was determined as the precise moment when the slope of the rising membrane potential exceeded an impressive rate of 50 mV/ms. Additional measurements such as after-hyperpolarization, half-width, and overshoot were carefully estimated from all action potentials observed during the 200-pA current injection step. To conduct these experiments, the esteemed Clampex program (Molecular Devices) was employed. All data points were keenly digitized at an impressive rate of 20 kHz. For meticulous analysis, the esteemed Clampfit 10.7 software (Molecular Devices) was utilized.

Data collection and analysis of human brain circRNAs

An RNA sequencing raw data involving the dorsolateral prefrontal cortex (dlPFC) of patients with cocaine use disorder was obtained from GSE99349 in GEO database (http://www.ncbi.nlm.nih.gov/geo/). The procedures for identifying the expression of circHomer1 were based on our previous study [43]. In our analysis, we excluded participants with the age over 50 years old to align the human data with the age range of the rats used in our experiments. All including samples were listed in Supplementary Table 2 [44]. To shed light on the statistical significance of circHomer1 expression, we employed the Fragments Per Kilobase of exon model per Million mapped fragments (FPKM) in each participant as the basis of our rigorous analysis.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 8. Unless otherwise stated, normally distributed data was tested using one-way ANOVA followed by Tukey’s post hoc; repeated two-way ANOVA followed by Bonferroni’s post hoc or Holm-Sidak’s post hoc; two-way ANOVA followed by Tukey’s post hoc; three-way ANOVA followed by Tukey’s post hoc; and two-tailed, unpaired t test. Non-normally distributed data were tested using the Kruskal-Wallis test with Dunn’s post hoc and the two-tailed, unpaired Mann Whitney U test. The results are expressed as mean ± SEM. P < 0.05 were considered statistically significant. The detailed statistical analyses are showed in Supplementary Table 3.

Results

CircHomer1 in PrL is required for cocaine-induced rewarding effect

CircHomer1, an extensively preserved circRNA, originates from the exons 2 to 5 of Homer1 through the intricacies of backsplicing (Supplementary Fig. 1). To investigate the role of circHomer1 in cocaine response, we analyzed its expression dynamics across addiction-related brain regions following repeated cocaine exposure (Fig. 1a). Chronic cocaine administration induced region- and time-dependent alterations in circHomer1 expression. The prelimbic cortex (PrL) exhibited sustained circHomer1 reduction across all examined time points (30 min, 1 h, 6 h, 24 h, and 48 h) post-final cocaine injection (Fig. 1b). Other regions showed temporally restricted responses: significant downregulation occurred at 30 min in the central amygdala (CeA) and ventral tegmental area (VTA), at 1 h in the infralimbic cortex (IL) and basolateral amygdala (BLA), and at 6 h in the BLA (Fig. 1b). Notably, nucleus accumbens (NAc) core circHomer1 levels remained unchanged across all post-injection intervals (Fig. 1b). These region-specific temporal dynamics were further confirmed by BaseScope in situ hybridization, which validated the persistent circHomer1 reduction in PrL (Fig. 1c and d). Significantly, analysis of dorsolateral prefrontal cortex (dlPFC) samples from patients with cocaine use disorder revealed marked circHomer1 reduction (Fig. 1e), paralleling our observations in chronically cocaine-exposed rats. This cross-species conservation of circHomer1 dysregulation suggests its potential role as a therapeutic target in cocaine use disorder.

Fig. 1: CircHomer1 diminished after repeated cocaine exposure and was essential for conditioned place preference induced by cocaine in the prelimbic cortex.

a Timeline of the experiment. The rats were injected intraperitoneally with cocaine for 14 days and were decapitated to detect the expression of circHomer1 at 30 min, 1 h, 6 h, 24 h and 48 h after the last cocaine injection. b Detection of the expression of circHomer1 using qPCR in addiction-related brain regions following chronic cocaine exposure. First row (30 min): Unpaired t test, PrL: t9 = 2.94, P = 0.016; CeA: t9 = 3.34, P = 0.009; VTA: t9 = 2.44, P = 0.037. saline: n = 6/group, cocaine: n = 5/group. Second row (1 h): Unpaired t test, PrL: t14 = 2.54, P = 0.024; IL: t14 = 2.46, P = 0.028; BLA: t14 = 2.90, P = 0.012. PrL/IL/BLA/CeA/NAc core: n = 8/group; VTA: saline n = 7, cocaine n = 8. Third row (6 h): Unpaired t test, PrL: t13 = 2.87, P = 0.013; BLA: t14 = 2.83, P = 0.013. PrL/IL/VTA: saline n = 8, cocaine n = 7; BLA/CeA/NAc core: n = 8/group. Fourth row (24 h): Unpaired t test, PrL: t10 = 3.05, P = 0.012. n = 6/group. Fifth row (48 h): Unpaired t test, PrL: t12 = 2.21, P = 0.047. n = 7/group. c Visualization of the expression of circHomer1 in the PrL at 30 min, 1 h, 6 h, 24 h and 48 h after the last cocaine injection. Red dots indicate expressed circHomer1, while blue dots indicate the location of the cell nucleus. Scale bars: 50 μm. d Quantification of the circHomer1 expression using in situ hybridization in the PrL at 30 min, 1 h, 6 h, 24 h and 48 h after the last cocaine injection. Unpaired t test, 30 min: t10 = 2.66, P = 0.024; 1 h: t10 = 2.48, P = 0.033; 6 h: t10 = 4.73, P = 0.0008; 24 h: t10 = 7.71, P < 0.0001; 48 h: t10 = 3.36, P = 0.007. n = 6/group. e Decreased expression of circHomer1 in the dlPFC of patients with cocaine use disorder. Unpaired t test: t26 = 2.21, P = 0.037. control: n = 12; CUD: n = 16. f Timeline of the CPP. Before cocaine CPP training, the rats were bilaterally microinjected in the PrL with CMV-OE-circHomer1 to overexpress circHomer1 or sh-circHomer1 to knockdown circHomer1. Corresponding controls received CMV-GFP or sh-control. Following the rAAV expression, cocaine at dose of 10 mg/kg or 5 mg/kg was used to explore whether overexpression of circHomer1 could suppress cocaine CPP or knockdown of circHomer1 could enhance cocaine CPP, respectively. g GFP fluorescence indicates local expression of rAAV in the PrL of rats, as visualized under fluorescence microscopy. The injection site of rAAV (left), Scale bars: 2 mm. The expression of rAAV (right), Scale bars: 100 μm. h Elevated expression of circHomer1 in the PrL of rats injected with CMV-OE-circHomer1. Unpaired t test: t16 = 6.38, P < 0.0001. n = 9/group. i CircHomer1 overexpression resulted in a reduction in cocaine CPP. Repeated two-way ANOVA: the interaction of CPP training × the viral vector (F1, 18 = 5.54, P = 0.030). n = 10/group. j The expression of circHomer1 significantly decreased in the PrL of rats injected with sh-circHomer1. Unpaired t test: t16 = 9.49, P < 0.0001. n = 9/group. k Knockdown of circHomer1 enhanced cocaine CPP. Repeated two-way ANOVA: the interaction of CPP training × the viral vector (F1, 14 = 6.71, P = 0.021). n = 8/group. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ##P < 0.01. The data are expressed as mean ± SEM. CPP: conditioned place preference. CUD: cocaine use disorder. FPKM: Fragments Per Kilobase of exon model per Million mapped fragments.

Then, we examined expression changes in Homer1 linear transcripts following chronic cocaine exposure, focusing on three predominant isoforms: Homer1a and Homer1b/c. Homer1b and Homer1c were analyzed collectively due to their high structural and functional similarity, differing by only 36 base pairs [45, 46]. While Homer1a expression in PrL remained unchanged at both 30 min and 24 h post-cocaine injection, Homer1b/c showed significant upregulation at 24 h, but not at 30 min, following final cocaine administration (Supplementary Fig. 2). Given previous findings demonstrating circHomer1 binding to Homer1b in orbitofrontal cortex and its role in cognitive flexibility regulation [26], we hypothesized that circHomer1 might function through competitive sequestration of Homer1b/c, potentially explaining the transient Homer1b/c upregulation observed following repeated cocaine exposure.

Additionally, we detected the expression of circHomer1 after acute cocaine exposure, we found single cocaine exposure did not change the expression of circHomer1 (Supplementary Fig. 3). The reduction in circHomer1 levels induced by cocaine does not represent a mere pharmacological response, but rather signifies a lasting adaptive alteration within the PrL brought about by prolonged exposure to cocaine. This indicates that the enduring aberrant expression of circHomer1 may be the etiology and pathogenesis of cocaine addiction.

Next, we assessed whether manipulating circHomer1 expression in the PrL would impact cocaine-induced CPP. We first employed a recombinant adeno-associated virus (rAAV) with the CMV promoter to overexpress circHomer1 (CMV-OE-circHomer1), while using an empty vector (CMV-GFP) as a control (Fig. 1f). The expression of rAAV-mediated GFP was restricted to the PrL (Fig. 1g). The expression of circHomer1 was markedly increased in the PrL of rats in CMV-OE-circHomer1 group (Fig. 1h), which demonstrated that CMV-OE-circHomer1 induced significant overexpression of circHomer1. Before cocaine conditioning (baseline), both the CMV-OE-circHomer1 and CMV-GFP injected groups displayed comparable CPP scores, indicating no inherent preference (Fig. 1i). However, rats in CMV-OE-circHomer1 group showed a significantly decreased cocaine-induced CPP score compared to that of CMV-GFP rats after cocaine conditioning (10 mg/kg) (Fig. 1i). These findings suggest that overexpression of circHomer1 can mitigate the rewarding effect of cocaine.

We then evaluated whether inhibiting circHomer1 expression in the PrL could enhance cocaine-induced CPP. We utilized a rAAV vector with a U6 promoter to express a specific shRNA (sh-circHomer1) for targeted knockdown of circHomer1 expression in vivo (Fig. 1f). A scrambled sequence (sh-control) was used as a control. The expression of circHomer1 was notably decreased in the PrL of sh-circHomer1 group rats (Fig. 1j), indicating the effectiveness of shRNA-mediated suppression. The rAAV microinjection did not influence the preference of the rats, as both groups exhibited similar CPP score in the baseline. Post-cocaine conditioning (5 mg/kg), the sh-control group did not exhibit a place preference for the cocaine-paired compartment (Fig. 1k). However, the preference for the cocaine-paired compartment was developed in the sh-circHomer1 group rats (Fig. 1k). These results underscore the importance of circHomer1 in the PrL for cocaine-induced rewarding effects.

Collectively, these findings demonstrate that lasting adaptive alteration of circHomer1 expression within the PrL is essential for cocaine-induced rewarding effects.

Cocaine modulates circHomer1 expression in excitatory neurons of the PrL

To assess the predominant cell types expressing circHomer1 in the PrL, we used the BaseScope technique in conjunction with immunostaining of specific cell markers in naive rats. Analysis revealed that circHomer1 transcripts exhibited robust co-localization with the pan-neuronal marker NeuN (Supplementary Fig. 4a, 4b). In contrast, significantly less co-localization was observed with the astrocyte marker glial fibrillary acidic protein (GFAP) or the microglia marker ionized calcium-binding adapter molecule 1 (Iba-1), indicating neuronal enrichment of circHomer1 within the PrL. Then, we identified the specific types of neurons in which circHomer1 functioned after repeated cocaine exposure using the BaseScope method in conjunction with immunostaining of specific neuronal markers. As shown in Fig. 2, compared with saline-treated group, repeated cocaine exposure significantly decreased the proportion of circHomer1 co-labeled with CamKII (Fig. 2c–e), a marker of excitatory neurons, but not GAD67 (a marker of inhibitory neurons) (Fig. 2f–h) in the PrL. This results suggest that circHomer1 may modulate the rewarding effects