Inhibition of Glycogen Synthase Kinase 3β Activity in the Basolateral Amygdala Disrupts Reconsolidation and Attenuates Heroin Relapse

Exposure to a heroin-associated conditioned stimulus can reactivate drug reward memory, trigger drug cravings, and induce relapse in heroin addicts. The amygdala, a brain region related to emotions and motivation, is involved in processing rewarding stimulus. Recent evidence demonstrated that disrupting the reconsolidation of the heroin drug memories attenuated heroin seeking which was associated with the basolateral amygdala (BLA). Meanwhile, neural functions associated with learning and memory, like synaptic plasticity, are regulated by glycogen synthase kinase 3 beta (GSK-3β). In addition, GSK-3β regulated memory processes, like retrieval and reconsolidation of cocaine-induced memory. Here, we used a heroin intravenous self-administration (SA) paradigm to illustrate the potential role of GSK-3β in the reconsolidation of drug memory. Therefore, we used SB216763 as a selective inhibitor of GSK-3β. We found that injecting the selective inhibitor SB216763 into the BLA, but not the central amygdala (CeA), immediately after heroin-induced memory retrieval disrupted reconsolidation of heroin drug memory and significantly attenuated heroin-seeking behavior in subsequent drug-primed reinstatement, suggesting that GSK-3β is critical for reconsolidation of heroin drug memories and inhibiting the activity of GSK-3β in BLA disrupted heroin drug memory and reduced relapse. However, no retrieval or 6 h after retrieval, administration of SB216763 into the BLA did not alter heroin-seeking behavior in subsequent heroin-primed reinstatement, suggesting that GSK-3β activity is retrieval-dependent and time-specific. More importantly, a long-term effect of SB216763 treatment was observed in a detectable decrease in heroin-seeking behavior, which lasted at least 28 days. All in all, this present study demonstrates that the activity of GSK-3β in BLA is required for reconsolidation of heroin drug memory, and inhibiting GSK-3β activity of BLA disrupts reconsolidation and attenuates heroin relapse.


INTRODUCTION
Opioid use disorder is a chronic recurrent brain disease caused by abnormal learning and memory patterns. And the symptom is the loss of substance use control. Drug-associated cues are important factors to promote the relapse to drug use. In addiction animal models, cues related to drug abuse can promote relapse (Davis and Smith, 1976;Dymshitz and Lieblich, 1987;Di Ciano and Everitt, 2004;Anker and Carroll, 2010). Once drug use occurs, it is possible to form the association between cues and reward-related memory that is not easily disrupted.
Same as other types of memories (Nader et al., 2000;Milekic and Alberini, 2002;Morris et al., 2006;Piva et al., 2019), drug reward memory experiences the process of acquisition, consolidation, retrieval, and reconsolidation. Once a consolidated drug memory is reactivated, it becomes unstable, allowing for modification or destruction by different treatments (Miller and Marshall, 2005;Lee et al., 2006;Wang et al., 2008;Lee, 2009;Li et al., 2010). The role of reconsolidation is critical in stabilizing the reactivated memory, involving de novo protein synthesis (Nader et al., 2000). Reconsolidation provides a time window, during which is expected to modify or even eliminate drug memory. Thus, disrupting reconsolidation of addiction memory is considered to be an effective measure of preventing relapse and drug seeking.
A large amount of evidence shows that pharmacological interventions disrupting reconsolidation of drug-reward memory are effective in relapse prevention. Using a conditioned place preference (CPP) or self-administration (SA) animal model, propranolol or rapamycin has been shown to effectively interfere with reconsolidation of drug memory and attenuate drug-seeking behavior (Lin et al., 2014;Xue et al., 2017;Chen et al., 2021;Zhang et al., 2021).
Glycogen synthase kinase-3 (GSK-3), a serine/threonineprotein kinase expressed widely in the mammalian brain, has essential roles in physiological activities like development, cell cycle, or apoptosis (Jope and Johnson, 2004;Medina et al., 2011). GSK-3β is a isoform of GSK-3, extensively involved in memory processing. The regulation of GSK-3β can affect neural functions like synaptic plasticity, which is the foundation of learning and memory (Banach et al., 2022;Li et al., 2022;Marosi et al., 2022). Also, GSK-3β regulates the structural and functional synaptic plasticity. GSK-3β deficient mice marked that memory reconsolidation and their ability to form long-term memories were impaired, suggesting that GSK-3β is important for normal brain function (O'Brien et al., 2004;Kimura et al., 2008;Kaidanovich-Beilin et al., 2009;Maurin et al., 2013). It has been detected that GSK-3β Ser21/9 phosphorylation levels changed during long-term potentiation (LTP) or longterm depression (LTD), which were essential for memory (Hooper et al., 2007;Peineau et al., 2007). For example, GSK-3β affects long-term memory formation as it promotes LTD by inhibitory phosphorylation of Serine-9 in inhibitory avoidance and novel object recognition test (Dewachter et al., 2009). The removal of GSK-3β in excitatory neurons of dentate gyrus reduced the synaptic transmission of hippocampus and decreased the expression of synaptic proteins like N-methyl-Daspartate receptor (NMDAR) and anti-alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropinoic receptor (AMPAR), impairing the formation of spatial and fear memories (Liu et al., 2017). Cocaine administration for 14 days could obviously reduce the phosphorylation of GSK-3β in the amygdala (Perrine et al., 2008). Hippocampal GSK-3β was activated during memory retrieval progress in the passive avoidance task (Hong et al., 2012). Memory retrieval activates hippocampal GSK-3β, and memory reconsolidation is impaired by a GSK-3 inhibitor systemically administration before memory retrieval (Kimura et al., 2008;Hong et al., 2012). Knockdown GSK-3β of ventral hippocampal displays a developmental reduction in cocaine-CPP, while not in morphine-CPP (Barr et al., 2020). Wu et al. (2011) directionally inhibited GSK-3β in the BLA immediately after the retrieval of drug cue memories to reduce the subsequent cocaine-seeking behavior (Wu et al., 2011). It has been shown that GSK-3β signaling pathway participated in the reconsolidation of cocaineinduced CPP.
However, whether the inhibition of GSK-3β in BLA could disrupt reconsolidation and prevent drug seeking and relapse in heroin SA animal model remains unknown. In our present study, we inhibited the activity of GSK-3β in the BLA to identify the effects of GSK-3β on reconsolidation of heroin cue memory in SA animal model. Amygdala plays a critical role in both cueassociated learning and the expression of cue-induced relapse of drug-seeking behavior (Luo et al., 2013), and also mediates the reconsolidation of aversive or appetitive memories (McGaugh, 2004;Díaz-Mataix et al., 2013;Nader, 2015;Björkstrand et al., 2016;Haubrich et al., 2020;Higginbotham et al., 2021;Yan et al., 2021). So we chose it as a targeted brain region to deliver SB216763. The role of GSK-3β activity in amygdala on reconsolidation of heroin cue memory was assessed by the heroin self-administration paradigm. More importantly, the long-term inhibitory effects of GSK-3β activity inhibition on reconsolidation of heroin cue memory were also tested.

Subjects
We housed the Sprague Dawley rats (male, 7-8 weeks of age on arrival) five-per-cage in a 26 • C and 60% humidity room and the cycle was 12-h-light/dark (8 a.m.-8 p.m.). Food and water were provided ad libitum. For the animal better adapted to operator, we performed a 5-day grasp-stroking adaptation procedure 3 min per day on experimental animals before the surgeries. All animal procedures and operations were carried out with the approval of the Xiangya Hospital Ethics Committee, Xiangya Hospital (Changsha, China). At the dark phase (8 a.m.-8 p.m.), the experiments were performed.

Intravenous Surgery
Surgery was performed when the rats' weight reaches 300-320 g. We used the sodium pentobarbital (60 mg/kg, i.p.) to anesthetize rats. The right jugular was exposed by surgery, then inserted an aseptic catheter into it (Lu et al., 2005;Ambroggi et al., 2008).
The catheter passed under the skin of the neck and through the skin of the head and was fixed to the rat's skull with dental acrylic. Heparinized saline (30 USP heparin/saline; Hospira) was infused into the intravenous catheters per 2 days to prevent clogging. After surgery, rats undergo a 7-day recovery period and their weight should remain constant.

Cannulae Implantation
Guide cannulae was implanted 1 mm above the BLA or CeA bilaterally after the rats (300-320 g) were anesthetized by using the sodium pentobarbital (60 mg/kg, i.p.). And the coordinate of BLA (Wu et al., 2011) were the following: anterior/posterior: -2.9 mm and medial/lateral: ± 5.0 mm from bregma, dorsal/ventral: -8.5 mm from the surface of the skull. The coordinates for the CeA were the following: anterior/posterior: -2.9 mm and medial/lateral: ± 4.2 mm from bregma, dorsal/ventral: -7.8 mm from the surface of the skull. And the guide cannulae was anchored by the stainless steel screws and dental cement. After surgery, rats undergo a 5-7 days' recovery period.

Intracranial Injections
The injection of SB216763 was based on a previous report by some minor changes (Xu et al., 2009). The BLA or CeA (0.5 µL/side) was received drug delivery by a microinjection pump with a rat of 0.5 µL/min. The injection time was no more than one minute. To ensure that the drug is fully administered and fully diffused, wait for more than 1 min to withdraw the needle after administration. Nissl staining was used to verify the cannula placements. A schematic diagram of the injection site in the BLA or CeA is shown in Figure 1.

Intravenous Heroin Self-Administration (SA) Training
Following the study of Ye et al. (2017), we established an intravenous SA training session experimental conditions were slightly modified. The operant chambers (AniLab Scientific Instruments Co., Ltd., China) used in this experiment were equipped with two nosepoke sensors, which were 5 cm high from the installation floor. The nosepoke sensors could record the number of animals' nosepokes. The number of nosepoke recorded by the sensor in the left side operandum was defined as "active." Nosepokes in "active" would lead to intravenous heroin administration with a 5 s tone-light cue synchronization. Corresponding to "active, " the sensor operandum recorded the number of "inactive" nosepoke in the right side. "Inactive" nosepokes had no programmed consequences.
We trained rats to adapt the intravenous heroin selfadministration (0.05 mg/kg/infusion) for 10 days. In the training sessions, rats have been trained for three 1-h daily training sessions and every training session was separated by 5 min. The training of rats using a fixed ratio 1 (FR1) reinforcement program up to 40 s. In 10-day training program, rats were received three 1-h heroin infusions intravenously (0.05 mg/kg/infusion) per day, each infusion separated 5 min. All heroin infusions were completed by an injection pump loading a 10 ml syringe. Each training companied a house light illuminated until the end. When rats were at the left nosepoke (active), the maneuver results in an intravenous heroin infusion with a 5 s tone-light cue synchronization. However, the inactive nosepokes had no consequences. To prevent animals' from death due to overdose, the drug infusions degree was restricted to 20 times/h (Xue et al., 2012;Luo et al., 2015). Some rats (n = 11) were excluded from the experiments: five rats died in intravenous surgery and six rats can't form a stable heroin self-administration.

Nosepoke Extinction
After the drug self-administration session, a 3-h daily nosepoke extinction training was followed. In extinction training (Experiments 1-4), the nosepoke behavior of rats to the sensor resulted in no programmed consequences like heroin intravenous delivery, tone-light cues. When at the last three self-administration sessions, the number of active nosepoke responding decreased at least 80% compared with the self-administration, the nosepoke extinction session was ended.

Heroin Memory Retrieval Trial
In Experiments 1, 2, and 4, the heroin-associated memories were reacted by performing a 15 min retrieval trial. The conditions of the retrieval session were the same as the SA training session, except that when rats got the "active" sensor, no reward heroin was infusion.

Cue Extinction
In Experiments 1, 3, and 4, a cue extinction session, which 3h per day, was performed on rats. The conditions of the cue extinction session were the same as that SA training session. But after the cue (tone/light) rendering, there was no heroin intravenous delivery.

Cue-Induced Reinstatement of Drug-Seeking (Experiments 1-4)
After the SB216763 or vehicle administration (intracranial injections into BLA or CeA), rats got rest for 24 h and then returned to the SA training context and the reinstatement test was performed. In this session, the number of active and inactive nosepokes was recorded for 1 hr. The test condition was the same as the heroin memory retrieval trial.

Heroin-Induced Reinstatement Test (Experiments 1, 3, and 4)
A low dose of heroin (0.25 mg/kg, s.c.) was delivered to rats 5 min earlier before the session started, then put rats in the SA training context. Through the two sensors, the number of two type nosepokes ("active" and "inactive") was recorded during the test. The reinstatement test lasted for 1 h. The test condition was the same as the heroin memory retrieval trial.

Spontaneous Recovery Test (Experiment 2)
In this test, after 28 days of the withdrawal phase, two types of nosepokes ("active" and "inactive") were recorded for 1 h. The test condition was the same as the heroin memory retrieval trial.

Experimental Design
Experiment 1: The Effect of Immediate Post-CS Retrieval SB216763 Treatment on Subsequent Heroin-Seeking Behavior Through 10 days of heroin SA training, nosepoke extinction training was followed in the same apparatus for 10 consecutive days. After nosepoke extinction training, rats were allowed a rest for 24 h, then rats were received a 15-min conditioned stimulus (CS) retrieval session. After CS exposure, the rats were divided into four groups: (1) Intracranial injection of vehicle into BLA (0.5 µL/side) immediate after the retrieval trial (BLA+vehicle); (2) Intracranial injection of SB216763 into BLA (0.5 µL/side) immediate after a 15 min retrieval test (BLA+SB216763); (3) Intracranial injection of vehicle into CeA (0.5 µL/side) immediate after a 15 min retrieval test (CeA+vehicle); (4) Intracranial injection of SB216763 into CeA (0.5 µL/side) immediate after a 15 min retrieval test (CeA+SB216763). On Day 23, the rats were performed a cue-induced reinstatement test to explore the effect of SB216763 on heroin drug memory. Subsequently, the two-day cue extinction session was carried out. On Day 26, heroin priming-induced reinstatement was tested in rats (Figure 2A).

Experiment 2: The Lasting and Long-Term Effect of Immediate Post-CS Retrieval SB216763 Treatment on Cue-Induced Reinstatement Test and Spontaneous Recovery 28 Days Later
After CS exposure, the rats were divided into four groups: (1) Intracranial administration of vehicle into BLA (0.5 µL/side) immediate after a 15 min retrieval trial (BLA+vehicle); (2) Intracranial injection of SB216763 into BLA (0.5 µL/side) immediate after a 15 min retrieval trial (BLA+SB216763); (3) Intracranial injection of the vehicle into CeA (0.5 µL/side) immediate after a 15 min retrieval trial (CeA+vehicle); (4) Intracranial injection of SB216763 into CeA(0.5 µL/side) immediate after a 15 min retrieval (CeA+SB216763). Rats were received a cue-induced reinstatement test after a 24 h rest. After 28 days of abstinence, to access the long-term effect of SB216763 on heroinseeking behavior, a spontaneous recovery test was carried out ( Figure 3A).

Experiments 3 A and B: The Effect of SB216763 Treatment on Subsequent Heroin-Seeking Behavior Without CS Exposure
The experimental procedure was the same as Experiment 1, except that no CS exposure session to reactivate heroin cue memories ( Figure 4A).

Experiment 4: The Effect of Delayed Post-CS Retrieval SB216763 Treatment
The rats were received treatment of SB216763 delayed for 6 h after the retrieval session, and other parts of the experiment were the same as Experiment 1 (Figure 5A).

Statistical Analysis
We used the repeated-measures ANOVAs to analyze the data with the between-subjects factor of treatment condition and within-subjects factor of test condition, followed by Tukey's post-hoc test in each experiment (see Results). All values of experiments were presented as mean ± SEM and data analysis was done on GraphPad, v.9.0. p values < 0.05 were considered statistically significant. We tested the influences of post-retrieval BLA and CeA SB216763 injection on reinstatement of heroin seeking which was cue-induced or heroin-induced by using four groups of rats (Figure 2A). In the acquisition session, there were no significant differences between groups of BLA injected with vehicle (N = 10) or SB216763 (N = 11) and groups of CeA treated with vehicle (N = 10) or SB216763 (N = 11) which were shown by the heroin infusion numbers [main effect of acquisition time: F (9,342) = 81.35, p < 0.0001; administration condition: F (3,38) = 0.2803, p = 0.8393; acquisition time × administration condition interaction: F (27,342) = 1.443, p = 0.0744; Figure 2B]. As the numbers of nosepokes showed, groups did not significantly differ from each other in the extinction session [main effect of extinction time: F (9,342) = 95.68, p < 0.0001; administration condition: F (3,38) = 0.092, p = 0.9640; extinction time × administration condition interaction: F (27,342) = 0.8897, p = 0.6274; Figure 2C].   The results suggest that injecting SB216763 in BLA but not the CeA reduced cue-induced and heroin-primed heroin seeking reinstatement immediately after heroin cue retrieval. We tested the influences of immediate post-retrieval SB216763 injection in BLA and CeA on heroin seeking reinstatement which was cue-induced or heroin-induced besides the influences of long-term by using four rats groups (Figure 3A). In the heroin self-administration acquisition session, groups of BLA administrated vehicle (N = 10) or SB216763 (N = 10) and of CeA administrated vehicle (N = 10) or SB216763 (N = 10) had no significant differences that were shown by the similar heroin infusion numbers [main effect of acquisition time: F (9,324) = 48.79, p < 0.0001; administration condition: F (3,36) = 0.1662, p = 0.9185; acquisition time × administration condition interaction: F (27,324) = 0.5552, p = 0.9663; Figure 3B]. Analogously, no significant differences of group were found in 10-day extinction training [main effect of extinction time: F (9,324) = 207.2, p < 0.0001; administration condition: F (3,36) = 0.5019, p = 0.6833; extinction time × administration condition interaction: F (27,324) = 0.9528, p = 0.5354; Figure 3C].
Consistent with the results of Experiment 1, in the cueinduced reinstatement test, we found that nosepokes of groups differed from each other significantly in active side of BLA+vehicle and BLA+SB216763 groups [main effect of trial condition: F (1,18) = 67.92, p < 0.0001; administration condition: F (1,18) = 21.07, p = 0.0002; trial condition × administration condition interaction: F (1,18) = 22.67, p = 0.0002]. In the cueinduced reinstatement test, the post-hoc test shown a significant reduction in active side nosepokes of BLA+SB216763 group compared to that BLA+vehicle (p < 0.001) (Figure 3D  Thus, the results suggest that immediate post-CS BLA SB216763 treatment rather than CeA can reduce subsequent cue-induced heroin seeking and last at least 28 days.
Thus, the results indicate that the CS retrieval of drugassociated memory determines the effect of SB216763 on subsequent heroin seeking and when SB216763 is used alone, there will be no influence on the heroin seeking of the subsequent cue-induced and heroin-primed reinstatement test.

Experiment 4: SB216763 Treatment 6 h After Retrieval Does Not Affect the Heroin Seeking in the Following Cue-Induced and Heroin-Primed Reinstatement
Finally, we employed whether SB216763 treated out of the reconsolidation time window would suppress the following heroin seeking (Figure 5A) The results suggest that the effect of SB216763 on the heroin seeking of the following cue-induced and heroin-primed reinstatement test is time-specific.

DISCUSSION
In our present study, we examined the effects of SB216763, a GSK-3β inhibitor, on heroin-seeking behavior and relapse via an intravenous heroin SA procedure. The main findings are following: (1) Immediate post-CS SB216763 treatment in BLA rather than CeA can reduce subsequent cue-induced and heroin-primed reinstatement of heroin seeking. (2) The effect of SB216763 (intracranial injection into BLA) immediately after retrieval session, which reactivated the heroin cue memory, on cocaine-seeking behavior continued for at least twenty-eight days; (3) SB216763 administration into BLA delayed 6 h or without the retrieval have no effect on the cue-induced and heroin priming-induced reinstatement test, suggesting that the effects of SB216763 on drug-seeking behavior are retrieval-dependent and time-specific. The results demonstrated that the activity of GSK-3β in BLA is required for reconsolidation and intra-BLA injection of SB216763 disrupts memory reconsolidation, suggested that inhibiting GSK-3β activity of BLA disrupts reconsolidation and attenuates heroin relapse.
Existing evidence has shown that consolidated drug-induced memory could be unstable after retrieval by conditioned stimulus (drug-associated cues) or unconditioned stimulus (the drugs). Disrupting the reconsolidation after drug memory retrieval effectively blocks the association between drug-related cues and drugs, thus producing a forgetting effect and reducing drugseeking behavior. Memory retrieval activates GSK-3β in the BLA, and reconsolidation was impaired by systemical administration of a GSK-3 inhibitor before memory retrieval in rats (Wu et al., 2011). We have demonstrated that SB216763, as a GSK-3β inhibitor, reduced heroin seeking and relapse by disrupting reconsolidation of heroin drug memory. Nevertheless, without CS exposure or with a 6h delay administration of SB216763 has no effect on heroin seeking behavior, which is consistent with the "reconsolidation theory" that allowing for modifications during reconsolidation in a putative time window. Indeed, GSK-3β expressed in multiple brain regions (e.g., NAc, PFC, amygdala) (Winder et al., 2002;Morgane et al., 2005;Robbins et al., 2008), regulates the dopamine-associated behaviors (Beaulieu et al., 2007a). The role of GSK-3β in drug addiction has also been confirmed by several studies. For example, in a cocaine-induced CPP model, the reconsolidation of cocainecue memories was impaired by systemic injection of lithium chloride (a GSK-3β inhibitor) after memory retrieval, and SB216763 administration into BLA can result in a long-term effect on cocaine cue memories via disrupting reconsolidation (Wu et al., 2011).
Our present study complements the role of GSK-3β in heroin self-administration model. We show that inhibiting GSK-3β activity by injection of SB216763 into the BLA, but not the CeA, immediately after the retrieval reduced the drugseeking behavior. In addition, some studies have shown that the GSK-3β in the hippocampus regulates learning and memory ability. Targeted downregulation of GSK-3β in the ventral hippocampus disrupting the formation of cocaine CPP and the spatial memory (Barr et al., 2020). Overexpression of GSK-3β in the hippocampus impaired the spatial memory . And both the amygdala, hippocampus, are the important brain regions in drug addiction (Alizamini et al., 2022). Previous study found that inhibited the activity of GSK-3β in the BLA immediately after the retrieval could disrupte cocaine context memory (Wu et al., 2011). Our findings are consistent with these results and showed that inhibiting GSK-3β in the BLA in the reconsolidation time window decreased drug seeking and relapse in heroin SA rats.
The amygdala, associated with emotion and motivation, plays roles in processing rewarding environmental stimulus (Janak and Tye, 2015). Recent evidence demonstrates that disrupting the reconsolidation of the cocaine drug memories attenuates cocaine seeking which was associated with the BLA (Wu et al., 2011). In our present study, we discussed the role of GSK-3β in a heroin self-administration rat model. Inhibiting activity of GSK-3β in the BLA, but not the CeA, effectively disrupted reconsolidation of heroin cue memory and blocked subsequent drug-seeking behaviors. Many studies have demonstrated that BLA is involved in memory reconsolidation (Wells et al., 2013;Yuan et al., 2020). However, the spcefic mechanism of GSK-3β in drug memory reconsolidation remains to be elucidated.
GSK-3β is widely expressed in the brain and involved in fundamental brain functions like neurogenesis, neurotransmitter signaling (Beaulieu et al., 2004(Beaulieu et al., , 2007bLi et al., 2004), circadian rhythms (Yin et al., 2006), and memory process (Hooper et al., 2007;Kimura et al., 2008;Hong et al., 2012). GSK-3β has been shown to regulate many transcription factors (e.g., β-catenin, NF-κB, activator protein-1), and memory-associated proteins which have been implicated in fundamental brain functions (Frame and Cohen, 2001). Specifically, in Liu et al.'s study, the role of GSK-3β/β-catenin signaling pathway on memory consolidation in rats trained in a Morris water maze task was proved  and the consolidation of fear memory need the β-catenin in the amygdala (Maguschak and Ressler, 2008). As a study revealed, the Akt/GSK-3/mTORC1 signaling pathway, which has been found in the hippocampus and nucleus accumbens, participates in the reconsolidation progress of cocaine memory (Shi et al., 2014). Psychostimulant activates GSK-3β by inactivating protein kinase B (Akt) and reducing its inhibition of serine-phosphorylation . These findings indicate that during the reconsolidation process, GSK-3β may be required to be activated, but the role of GSK-3β on reconsolidation of heroin-SA and the possible molecular mechanisms involved in this process have not yet known. Our present study showed that intra-BLA infusion of the GSK-3β inhibitor SB216763 disrupted the reconsolidation of heroin reward memory and the reduction of heroin seeking behaviors lasted for a long time (28 days). And these effects were BLA-specific and time-specific. Studies have reported that the activity of GSK-3β was involved in the LTP and LTD which both depended on the NMDA receptor and also regulated the interreaction between them (Peineau et al., 2007(Peineau et al., , 2008(Peineau et al., , 2009. But there is less study that reported the relationship between GSK-3β and NMDA in addiction. Furthermore, reconsolidation is associated with de novo protein synthesis and synaptic plasticity alteration (Nader et al., 2000). Thus, our future study will focus on the relationship between GSK-3β and NMDA in reconsolidation of heroin reward memory and investigate the synaptic plasticity alteration during this process by electrophysiological methods. Compared with Wu et al. (2011) study, our present study is the extension of previous studies, and we used a drug self-administration paradigm, which can simulate craving and relapse in addicts and is more suitable for studying craving and relapse (Spealman and Goldberg, 1978;Schindler et al., 2002).
In summary, the present study demonstrated the effects of GSK-3β on the reconsolidation of drug reward memory via a rats heroin SA paradigm. After heroin cue memory retrieval, injection of SB216763 into BLA impairs reconsolidation of the heroin cue memory, effectively reduces the heroin seeking behavior in rats, and these effects lasted at least 28 days. Moreover, the reconsolidation window may be an important determinant to reduce heroin seeking and relapse. Our present study identifies that GSK-3β inhibitors may have the potential therapeutic value in the treatment of heroin addiction.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by the Xiangya Hospital Ethics Committee, Xiangya Hospital (Changsha, China).