Activation of GSK3β induced by recall of cocaine reward memories is dependent on GluN2A/B NMDA receptor signaling

Abstract

Glycogen synthase kinase-3β (GSK3β) is a critical regulator of the balance between long-term depression and long-term potentiation which is essential for learning and memory. Our previous study demonstrated that GSK3β activity is highly induced during cocaine memory reactivation, and that reconsolidation of cocaine reward memory is attenuated by inhibition of GSK3β. NMDA receptors and protein phosphatase 1 (PP1) are activators of GSK3β. Thus, this study investigated the roles of NMDA receptor subtypes and PP1 in the reconsolidation of cocaine contextual reward memory. Cocaine contextual memories were established and evaluated using cocaine conditioned place preference methods. The regulation of GSK3β activity in specific brain areas was assessed by measuring its phosphorylation state using immunoblot assays. Mice underwent cocaine place conditioning for 8 days and were tested for place preference on day 9. Twenty-four hours later, mice were briefly confined to the compartment previous paired with cocaine to reactivate cocaine-associated memories. Administration of the GluN2A- and GluN2B-NMDA receptor antagonists, NVP-AAM077 and ifenprodil respectively, immediately following recall abrogated an established cocaine place preference, while preventing the activation of GSK3β in the amygdala, nucleus accumbens, and hippocampus during cocaine memory reactivation. PP1 inhibition with okadaic acid also blocked the activation of GSK3β and attenuated a previously established cocaine place preference. These findings suggest that the dephosphorylation of GSK3β that occurred upon activation of cocaine-associated reward memories may be initiated by the activation of PP1 during the induction of NMDA receptor-dependent reconsolidation of cocaine mnemonic traces. Moreover, the importance of NMDA receptors and PP1 in reconsolidation of cocaine memory makes them potential therapeutic targets in treatment of cocaine use disorder and prevention of relapse.

Graphical Abstract

Evidence has shown that NMDA receptors can activate GSK3β through a process involving protein phosphatase 1 (PP1). In the present study, we find that function of this pathway is necessary for the reconsolidation of a cocaine reward mnemonic trace; inhibition of this signaling pathway at any step is sufficient to impair the maintenance of such memories, making these molecules potential therapeutic targets in treatment of cocaine dependence and the prevention of relapse.

Introduction

Craving elicited by exposure to environmental cues that were previously associated with use of addictive drugs is a primary contributor to relapse (O’Brien et al. 1990). Memories of drug-paired cues are highly engrained and difficult to erase in both humans (Ehrman et al. 1992; Gawin 1991) and experimental animals (de Wit & Stewart 1981). After reactivation, memories undergo a process of reconsolidation which strengthens the memory trace. However, the reactivation period is a critical time when memory traces become labile and can be manipulated behaviorally or pharmacologically by interfering with reconsolidation (Nader et al. 2000). Thus, targeting the reconsolidation process related to the maintenance of drug memory has therapeutic potential for the treatment of addiction and the prevention of cue-induced relapse.

NMDA receptors play a crucial role in reconsolidation of cocaine-related memory (Alaghband & Marshall 2013; Itzhak 2008; Milton et al. 2008), likely through bidirectional effects on synaptic plasticity, LTP and LTD (Sajikumar & Frey 2004). NMDARs are composed of GluN1 subunits, GluN2 subunits (A,B,C,D), and GluN3 subunits (Cull-Candy & Leszkiewicz 2004; Paoletti 2011; Traynelis et al. 2010). In the adult, GluN2A and GluN2B are the predominant subunits in the hippocampus, amygdala, striatum and cortex (Lopez de Armentia & Sah 2003; Monyer et al. 1992; Wenzel et al. 1995) and play a central role in synaptic function and plasticity. The subtypes of NMDA receptors that contribute to the maintenance of conditioned fear memories have been investigated. Administration of the GluN2B-selective NMDAR antagonist ifenprodil into the basolateral amygdala prior to fear memory reactivation prevents the destabilization of the conditioned-fear memory (Ben Mamou et al. 2006; Milton et al. 2013), whereas administration of the GluN2A-preferring NMDAR antagonist NVP-AAM077 prior to the fear memory reactivation session reduces conditioned freezing during subsequent testing (Milton et al, 2013). These findings suggest that GluN2B-NMDARs and GluN2A-NMDARs within the basolateral amygdala are required for fear memory destabilization and reconsolidation respectively (Milton et al, 2013). Although a pivotal role for NMDA receptors in the reconsolidation of drug-cue memories is appreciated (Alaghband & Marshall 2013; Brown et al. 2008; Kelley et al. 2007; Milton et al. 2008; Sadler et al. 2007; Wouda et al. 2010), the subtype(s) of NMDA receptors involved is not known and the signaling pathways remain under investigation.

Our previous study demonstrates that glycogen synthase kinase-3β (GSK3β) activity is enhanced in the nucleus accumbens, hippocampus, and prefrontal cortex during cocaine memory reactivation (Shi et al. 2014). Importantly, pharmacological inhibition of GSK3β blocks reconsolidation of cocaine reward memory and erases a previously established cocaine place preference indicating a critical role of GSK3β signaling in the reconsolidation of cocaine-reward memories (Shi et al. 2014). GSK3β has a wide influence over neural function and plasticity due to its ability to phosphorylate dozens of substrates including signaling proteins and transcriptional factors (Jope & Roh 2006; Kockeritz et al. 2006). GSK3β may contribute to memory reconsolidation by controlling the balance between long-term potentiation and LTD (Kimura et al. 2008). NMDA receptors can activate GSK3β through a process involving protein phosphatase 1 (PP1). PP1 dephosphorylates Ser9-GSK3β (Szatmari et al. 2005), thereby increasing GSK3β activity. In addition, PP1 indirectly activates GSK3β by dephosphorylating Thr308-Akt, thus reducing Akt activity (Peineau et al. 2008); Akt is a kinase that phosphorylates GSK3β. Therefore, during LTD, activation of PP1 can activate GSK3β both by direct dephosphorylation and indirectly through inhibition of Akt.

Our previous report proposed a model by which a protein phosphatase cascade, such as PP2B and PP1, is activated down-stream of NMDA receptors during LTD, resulting in the dephosphorylation of Akt and activation of GSK3β following recall of cocaine contextual memories (Shi et al. 2014). The present study sought to test the validity of the model and establish the mechanisms of regulation of GSK3β signaling that are involved in the maintenance of cocaine memories. The contributions of specific NMDA receptor subtypes and PP1 activity in the reconsolidation of cocaine contextual reward memories were investigated.

Materials and Methods

Animals

Male CD-1 mice (Charles River Laboratories, Wilmington, MA. RRID: SCR_003792), 8 weeks old on delivery (32–35g), were housed in group of 4 per cage (cage type: Mouse 750™ Ventilated Cage) without additional enrichment objects, under a 12-hour light/dark cycle (7:00/19:00). Standard chow and water were available ad libitum. Mice were housed for five days prior to experimental procedures and were weighed daily. Behavioral tests were conducted between 9:00–12:00 (Experiment 5–6), or 14:00–17:00 (Experiment 1–4). Mice were arbitrarily assigned to experimental groups; no randomization was performed. Animal testing was conducted in accordance with the National Institutes of Health guidelines for the Care and Use of Laboratory Animals and with an approved protocol from Temple University Institutional Animal Care and Use Committee (approval: ACUP 4600). The experimental protocol was not pre-registered. No sample size calculation was performed because the sample sizes were consistent with our previous reports (Shi et al. 2014) and those standard in the field. Six experiments were conducted in the present study. A total of 249 mice entered into the cocaine conditioned place preference procedure (described below). Three mice were removed from Experiment 6 because the cannulas became dislodged (N=2) or were unhealthy (N=1). 246 mice were tested for place preference on Day 9. Mice that did not have a place preference score of >180 sec on Day 9 did not continue in the study; 66 mice fulfilled this exclusion criteria and did not continue onto Day 10. 180 mice continued onto the second phase of the study. 32 mice were used for saline control experiments (Experiments 1 and 2) and received saline prior to being placed into both compartments of the conditioning chambers. In this case, the inclusion criteria was a place preference score of −359 sec to +359 sec. 24 mice fulfilled this criteria and continued to Day 10 of the study. Animal procedures were performed by skilled and experienced experimenters to ensure that no unnecessary pain or distress was caused as a result of the procedures.

Drugs

Cocaine hydrochloride was generously supplied by the National Institute on Drug Abuse Drug Supply Program, dissolved in sterile saline, and injected intraperitoneally (i.p.) in a volume of 3 ml/kg body weight. Okadaic acid (CAS#: 155751-72-7, 2014, 2019), NVP-AAM077 (CAS#:459836-30-7, 2014, 2017, 2018) and ifenprodil (CAS#: 23210-58-4, 2014, 2017, 2018) were purchased from Sigma-Aldrich (St. Louis, MO) and dissolved in sterile saline. An equal volume of saline served as the vehicle control for all drugs.

Cocaine Conditioned Place Preference

Conditioned place preference chambers were rectangular in shape (45×20×20cm) and consisted of two compartments, separated by a removable door. One compartment had a smooth floor with white and black vertical striped walls, whereas the other had a rough floor and white walls with black circles. Illumination in both compartments was equal. An unbiased conditioned place preference procedure was used, similar to that described in our previous study (Shi et al, 2014). Place conditioning occurred once daily for 8 days, wherein mice were injected with saline or cocaine (10 mg/kg, i.p.) and placed into alternate compartments of the conditioning chamber for 30 minutes, resulting in 4 pairings with saline and 4 pairings with cocaine. On Day 9, mice had access to both compartments for 30 minutes in a drug-free state and time in each compartment was recorded. Preference scores were calculated as time in cocaine-paired minus time in saline-paired compartments and reported in seconds. The investigators who performed the daily place conditioning sessions were not blinded to cocaine/saline administration. The experimenters who recorded the place preference scores were blinded to the experimental groups. Mice that showed cocaine place preference scores of >180 sec proceeded onto the reconsolidation phase of the studies as described in Experiments 1–6 below.

Protein Measurements by Immunoblotting

The nucleus accumbens, hippocampus, and amygdala from individual mice were sonicated in 100°C 1% sodium dodecyl sulfate with 1 mM NaF and 1 mM Na3VO4 as phosphatase inhibitors, boiled for 5 minutes, and stored at −80°C. Protein concentrations were determined by the Protein A280 method using a NanoDrop 2000c Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Protein extracts (12.5 μg) were separated on 7.5% Tris–HCl Bio-Rad Ready-gels (Cat#: 4561026). Bio-Rad Laboratories, Hercules, CA) and transferred onto nitrocellulose membranes. Membranes were blocked in Odyssey blocking buffer (LI-COR Biosciences, Lincoln, NE) and Tween-Tris-buffered saline, and incubated overnight at 4°C in the following primary antibodies; anti-phospho-ser9-GSK3β (1:1000, Cat#9323, RRID:AB_2115201, Cell Signaling, Beverly, MA) and anti-GSK3β (1:1000; Cat#9315, RRID:AB_490890, Cell Signaling). Membranes were washed in Tween-Tris-buffered saline and incubated with anti-rabbit (1:15,000, Cat# 926–32211, RRID:AB_621843) or anti-mouse (1:15, 000, Cat# 926–68070, RRID:AB_10956588) secondary antibodies conjugated to two different infra-red dyes (LI-COR Biosciences, Lincoln, NE). Membranes were visualized and proteins were quantified using the Odyssey infrared imaging system and software (Li-COR). Phosphorylated and total forms of GSK3β were detected simultaneously as the colors green and red respectively. Membranes were stripped of antibodies using the New Blot nitro stripping buffer (LI-COR) and re-probed with anti-β-tubulin (1:600,000; Cat#T8328, RRID:AB_1844090, Sigma-Aldrich, St. Louis, MO). Ratios of densities of phosphorylated GSK3β to total GSK3β; phospho-GSK3β to β-tubulin levels, and total GSK3β to β-tubulin were calculated. The experimenter performing the immunoblot assays was blind to animal treatment groups.

Intracranial implantation of injection cannulas

For experiments requiring intracerebroventricular (i.c.v.) drug infusions (Exp 5 and 6), indwelling cannula were implanted into the lateral ventricle under general anesthesia. A mixture of ketamine HCl (100 mg/kg Rompum; Fort Dodge Animal Health, Fort Dodge, IA) and xylazine (20 mg/kg AnaSed; Shenandoah, IA) was administered ip. This combination is considered to be a first choice in rodents when injectable anesthetics are used because it provides deep sedation, good immobilization with moderate analgesia (Flecknell et al. 1996). Anesthetized mice were mounted in a stereotaxic apparatus (Kopf Instruments, Tujunga, CA). Stainless steel guide cannulas (PlasticsOne, Roanoke, VA, USA, 26 gauge) were implanted into the lateral ventricle (unilateral) at the following coordinates from bregma: AP:−0.5mm, ML:1.0mm, DV:2.5mm (Paxinos & Franklin 2001), cemented in place and closed by stainless steel dummy cannulas of the same length (PlasticsOne, C315DC). Meloxicam (1mg/kg, sc) was administered pre-and post-surgical procedures for analgesia. Behavioral studies began 5–7 days after surgery.

Experiment 1: Effect of the GluN2A antagonist, NVP-AAM077, on the reconsolidation of cocaine reward memory.

The aim of Experiments 1 and 2 was to determine which subtype(s) of NMDA receptors contributes to the reconsolidation of cocaine-associated memories. In Experiment 1, 40 mice underwent cocaine conditioned place condition for 8 days and were tested for place preference on Day 9, as described above. The preference scores of 29 mice fulfilled the inclusion criteria and continued onto the second phase of the study. On Day 10, 24 hours following the test for cocaine place preference, 18 mice were arbitrarily assigned to three groups (N=6 mice/group) and were re-exposed to the compartment previously paired with cocaine for 10 min in order to reactivate the cocaine-paired memories. Memory reactivation was followed immediately by administration of the GluN2A-preferring antagonist NVP-AAM077 (0, 1 and 5 mg/kg, i.p.) (Fox et al. 2006; Laird et al. 2014). 11 mice were assigned to two non-reactivation control groups (N=5–6 mice/group). As such, they remained in their home cages and were injected with vehicle or NVP-AAM077 (5 mg/kg, i.p.) according to the same time schedule as the other three experimental groups. On Days 11 and 18, all mice were re-tested for place preference without further drug injections or conditioning sessions.

To test whether a single injection of NVP-AAM077 alone produces a place preference or aversion, 16 mice were conditioned with saline in both compartments for 8 days and tested on Day 9 for place preference. 12 mice showed a place preference score between −359 sec and +359 sec and continued onto the second phase of study. These mice were placed into the chamber for 10 min and received either vehicle or 5 mg/kg of NVP-AAM077 immediately thereafter (N=6 mice/group). Mice were re-tested for place preference without further drug injections or conditioning sessions on Days 11 and 18.

Experiment 2: Effect of the NMDAR GluN2B-selective antagonist, ifenprodil, on the reconsolidation of cocaine reward memory.

Experiment 2 was performed exactly as described in Experiment 1 except that mice were injected with the selective GluN2B antagonist ifenprodil (0, 1, 5 and 10 mg/kg, i.p.) (Liddie & Itzhak 2016; Suzuki et al. 1999) on Day 10 following re-exposure to the cocaine-paired chamber or in the home-cage. 47 out of 64 mice showed a cocaine place preference on Day 9 and were administered ifenprodil on Day 10 (N=6–9 mice/group). Likewise, the ability of a single injection of ifenprodil (0, 5 mg/kg, i.p.) to produce a place aversion was tested by the same methods described for in Experiment 1. Twelve of 16 mice had no place preference after saline conditioning and continued onto the second phase of the study (N=6 mice/group).

Experiment 3: Effect of the NMDAR GluN2A-preferring antagonist on the dephosphorylation of GSK3β induced by reactivation of cocaine reward memory.

GSK3β is activated in response to cocaine memory reactivation as evidenced by its dephosphorylation (Shi et al., 2014). Experiment 3 tested whether the regulation of GSK3β activity is dependent on GluN2A-containing NMDA receptors. 32 mice underwent cocaine conditioned place preference and testing on Day 9. 23 mice expressed a cocaine place preference and were included for study on Day 10. On Day 10, two groups of mice (N=5–7 mice/group) were injected with the GluN2A-preferring antagonist NVP-AAM077 (5 mg/kg) or vehicle 30 minutes prior to re-exposure to the compartment previously paired with cocaine. Two other groups (N=5–6 mice/group) were kept in their home cages and received NVP-AAM077 or vehicle on the same schedule. Immediately following the 10 minute reactivation session, mice were exposed to CO₂ for 15 seconds and decapitated. Brains were removed, and the amygdala, nucleus accumbens, and hippocampus were dissected and prepared for measurements of phosphoproteins by immunoblotting as described above.

Experiment 4: Effect of the NMDAR GluN2B-selective antagonist ifenprodil on the dephosphorylation of GSK3β induced by reactivation of cocaine reward memory.

32 mice underwent 8-day cocaine conditioned place conditioning and testing on Day 9; 26 mice showed cocaine place preference and continued onto Day 10. On Day 10, two groups of mice (N=6–7 mice/group) received the selective GluN2B antagonist ifenprodil (5 mg/kg) or vehicle 30 minutes prior to re-exposure to the compartment previously paired with cocaine. Two other groups of mice (N=6–7 mice/group) remained in their home cages and received ifenprodil or vehicle according to the same schedule. Immediately following the 10 minute reactivation session, mice were exposed to CO₂ for 15 seconds, decapitated and brains dissected on ice. The amygdala, nucleus accumbens, and hippocampus from individual mice were prepared for immunoblotting.

Experiment 5: Effect of PP1 inhibition on GSK3β phosphorylation during memory reactivation.

This experiment investigated the role of PPI in the activation of GSK3β in response to recall of cocaine-related memories using a pharmacological inhibitor of PP1, okadaic acid. Cannula were permanently implanted into the lateral ventricle of 38 mice as described above, and mice had 5–7 days recovery before place conditioning began. 26 mice expressed a cocaine place preference on Day 9 and were included in Day 10 of the study, assigned to four groups (N=6–7 mice/group). Experimental methods are exactly as described in Experiment 3 except that saline or okadaic acid (150 ng/3 μl, i.c.v.) was administered 30 min prior to the 10-min reactivation session or in the home cage. Immediately following the reactivation session, brains were removed and the nucleus accumbens and hippocampus were dissected and prepared for measurements of phosphoproteins by immunoblotting.

Experiment 6: Effect of the PP1 inhibitor, okadaic acid, on the reconsolidation of cocaine reward memory.

Cannulas were surgically implanted into the lateral ventricles of 43 mice. After 5–7 days recovery, mice underwent cocaine place conditioning and testing for place preference on Day 9. Injection cannula in two mice became dislodged during place conditioning, and one mouse was unhealthy, so these mice were removed from the study. 29 mice expressed cocaine place preference on Day 9 and continued on in the study. On Day 10, two groups of mice (N=7 mice/group) received saline or okadaic acid (150 ng/3 μl, i.c.v.) and 30 min later, were placed into the compartment previously paired with cocaine for 10 min. Two other groups (N=7–8 mice/group) were kept in their home cages and received saline or okadaic acid (150 ng/3 μl, i.c.v.) on the same schedule. On Day 11, mice were re-tested for place preference in order to determine if inhibition of PP1 prior to memory reactivation could attenuate the previously established cocaine place preference.

Statistical Analysis

Behavioral data (primary endpoint preference scores) were analyzed using two-way repeated measures analysis of variance (RM-ANOVA) with day and treatment factors (GraphPad Prism 5, La Jolla, CA). The Kolmogorov-Smirnov test (Prism 5) was used assess the normality of data. All data sets were normally distributed, thus parametric statistical tests were used. Protein data (p-GSK3β/total GSK3β) were analyzed by two-way ANOVA with exposure and treatment factors, followed by Bonferroni test for multiple comparisons. Grubb’s tests were applied to the protein data to identify potential outliers; 3 samples were identified as outliers and removed from the data set. The graphs are shown as standard box plots. The ends of the box indicate the upper and lower quartiles, the line across the box indicates the median (50^th percentile), and the whiskers from the two ends of the box extend to the minimum and maximum data points. The level of statistical significance was set at p<0.05 for all tests.

Results

Experiment 1: The GluN2A-preferring antagonist NVP-AAM077 disrupted an established cocaine place preference

Experiment 1 tested the role of the GluN2A-containing NMDA receptors in reconsolidation of cocaine-associated memories. As shown in Fig. 2A, administration of the GluN2A-preferring NMDAR antagonist NVP-AAM077 (NVP) after the memory reactivation session on Day 10 dose dependently reduced the previously established cocaine place preference in subsequent tests conducted 24 hours and 7 days later. Repeated measures two-way ANOVA of preference scores from Days 9, 11, and 18 (Fig 2A) revealed significant main effects of NVP-AAM077 treatment (F_2,30 = 8.04, p < 0.01) and test day (F_2,30 = 6.18, p < 0.01), as well as a significant treatment × day interaction (F_4,30 = 3.14, p < 0.05). Post hoc tests revealed that administration of NVP-AAM077 (5 mg/kg) immediately following reactivation of cocaine reward memories significantly attenuated preference for the cocaine-paired compartment when tested 24 h later (p < 0.01 vs. vehicle, Day 11) or one week later (p < 0.05 vs. vehicle, Day 18). No significant differences were found between the mice injected with the vehicle and 1 mg/kg of NVP-AAM077 (Fig. 2A). In the control experiment with no reactivation session on Day 10, NVP-AAM077 (5 mg/kg) administered in the home cage had no effect on the previously established cocaine place preference (Fig. 2B). Two-way RM-ANOVA of preference scores showed no significant effects of NVP-AAM077 treatment (F_1,18 = 1.20, p > 0.05), test day (F_2,18 = 0.43, p > 0.05), or treatment×day interaction (F_2,18 = 0.90, p > 0.05).

Figure 2.

Administration of GluN2A-NMDA preferring antagonist NVP-AAM077 after reactivation of a cocaine memory impaired reconsolidation and abrogated an established place preference. A. Mice conditioned with 10 mg/kg of cocaine showed a preference for the cocaine-paired chamber on Day 9. On Day 10, mice were confined to the cocaine-paired side for 10 minutes, followed by administration of the GluN2A antagonist, NVP-AAM077 (0, 1, 5 mg/kg, i.p.). Mice injected with 5 mg/kg NVP showed no preference for the cocaine-paired side when retested on Days 11 and 18. *** p<0.001 NVP vs saline. N=6 mice/group. B. An additional two groups of mice were tested for cocaine place preference on Day 9, and received vehicle or NVP-AAM077 (5 mg/kg, i.p.) in the home cage on Day 10. Cocaine place preference was maintained when tested on Days 11 and 18. N=5–6 mice /group. C. Two groups of mice were conditioned with saline in both compartments and showed no place preference on Day 9. Administration of NVP after re-exposure on Day 10 did not produce a significant place preference or aversion on Day 11 or 18. N=6 mice/group.

Since mice administered 5 mg/kg NVP-AAM077 on Day 10 after memory reactivation showed negative place preference scores when retreated on Days 11 and 18 (Fig 2A), a second control experiment was performed to test whether NVP-AAM077 alone could produce a place aversion under the conditions used in this study. To this end, two additional groups of mice underwent 8 days of conditioning with saline administered in both compartments, and both groups showed no place preference on Day 9 (Fig. 2C). Mice were re-exposed to the chamber for 10 min on Day 10 and received NVP-AAM077 (5 mg/kg) immediately thereafter. Place preference was re-tested on Days 11 and 18. Two-way ANOVA of preference scores showed no significant effects of NVP-AAM077 (F_1,20 = 0.02, p > 0.05), test day (F_2,20 = 0.54, p > 0.05), or treatment×day interaction (F_2,20 = 0.09, p > 0.05). Thus, NVP-AAM077 alone did not cause a significant place aversion but rather eliminated a place preference produced by prior cocaine conditioning.

Experiment 2: The selective GluN2B antagonist ifenprodil erased a previously established cocaine place preference

Experiment 2 tested the role of GluN2B-containing NMDAR in the reconsolidation of cocaine-associated memories using ifenprodil. Results demonstrate that administration of ifenprodil after memory reactivation dose-dependently attenuated the previous established cocaine place preference in subsequent tests conducted 24 hours and 7 days later (Fig. 3A). Two-way ANOVA showed significant main effects of ifenprodil treatment (F_3,54 = 7.83, p < 0.001) and test day (F_2,54= 17.58, p < 0.001), as well as a treatment×day interaction (F_6,54 = 8.13, p < 0.001). Post hoc tests revealed that administration of ifenprodil (5 and 10 mg/kg) immediately after the 10-min reactivation session significantly attenuated preference for the cocaine-paired compartment when tested 24 hours later and one week later (p < 0.01 vs. vehicle for 5 mg/kg; p < 0.001 vs. vehicle for 10 mg/kg, Days 11 and 18) as shown in Fig. 3A. In contrast, administration of ifenprodil (10 mg/kg, i.p.) in the home-cage environment had no effect on cocaine place preference (Fig. 3B); two-way ANOVA showed no significant effects of ifenprodil treatment (F_1,21 = 0.18, p > 0.05), test day (F_2,21 = 0.07, p > 0.05), or treatment×day interaction (F_2,21 = 0.06, p > 0.05) in the absence of memory reactivation (Fig. 3B). When ifenprodil was administered to mice that had undergone conditioning with saline in both compartments, no significant effects of ifenprodil treatment (F_2,20 = 0.004, p > 0.05), test day (F_2,20 = 0.55, p > 0.05), or treatment×day interaction (F_2,20 = 0.05, p > 0.05) were found (Fig. 3C). These results show that ifenprodil was effective in abrogating a cocaine place preference only when administered after memory reactivation, suggesting that reconsolidation of cocaine-associated reward memories involves GluN2B-containing NMDA receptors.

Figure 3.

Conditioned place preference was established with 10 mg/kg of cocaine, as shown on Day 9. A. Ifenprodil (0, 1, 5, 10 mg/kg. i.p.) administered after reactivation of the cocaine memory on Day 10, abrogated the established place preference when tested 24 hours (Day 11) and 7 days later (Day 18). ***p<0.001 ifenprodil vs saline. N=6–9 mice/group. B. Ifenprodil (10 mg/kg, i.p.) administered in the home cage on Day 10 did not affect cocaine place preference on Days 11 and 18. N=8 mice /group. C. Two groups of mice were conditioned with saline in both compartments and showed no place preference on Day 9. Administration of ifenprodil after re-exposure on Day 10 did not produce a significant place preference or aversion on Day 11 or 18. N=6 mice/group.

Experiment 3: Blockade of GluN2A-containing NMDA receptors prevented activation of GSK3β by cocaine memory recall

Our prior study demonstrated that GSK3β was activated after re-exposure to the environment previously paired with cocaine, as evidenced by a significant reduction in the level of GSK3β phosphorylation (Shi et al., 2014). Experiments 3 and 4 tested whether GSK3β activation was dependent on NMDA receptor activation and which NMDA receptor subtype might be involved. In experiment 3, mice underwent cocaine place conditioning and preference testing (Day 9). On Day 10, mice were pretreated with saline or NVP-AAM077 (5 mg/kg) 30 min prior to the 10-min memory reactivation session (i.e., exposure to the environment previously paired with cocaine) or in home cages (i.e., no exposure group). Brains were obtained immediately after 10 min reactivation to investigate the signaling pathway involved in cocaine memory reactivation. Representative immunoblots of amygdala (Fig. 4A), nucleus accumbens (Fig. 4B) and hippocampus (Fig. 4C) from mice with or without exposure to the environment previously paired with cocaine are presented. Two-way ANOVA of levels of phospho-GSK3β:GSK3β in amygdala (Fig. 4A) revealed significant main effects of NVP-AAM077 (F_1,19 = 4.63, p < 0.05) and environment (F_1,19 = 8.16, p < 0.05). Similarly, two-way ANOVA of levels of phospho-GSK3β:GSK3β also revealed a significant main effect of NVP treatment in the nucleus accumbens (F_1,16 = 15.74, p < 0.01) and hippocampus (F_1,19 = 6.03, p < 0.05), as well as a treatment×environment interaction in the nucleus accumbens (F_1,16 = 4.61, p < 0.05). Post hoc tests revealed that levels of phosphorylated GSK3β-Ser9 were lower following the reactivation of cocaine cue memories compared with no exposure in the amygdala, nucleus accumbens and hippocampus (all p < 0.05). Mice pretreated with NVP-AAM077 prior to re-exposure to cocaine-paired environment showed significantly higher levels of p-GSK3β in the amygdala (p < 0.05), nucleus accumbens (p < 0.01), and hippocampus (p < 0.05) compared with saline-injected exposed mice. These data indicate that reactivation of cocaine memory activates GSK3β (i.e., reduces phospho-GSK3β), which is dependent on the activation of GluN2A receptors. No differences in levels of total GSK3β:tubulin were found in any brain region (data not shown).

Figure 4.

Pretreatment with GluN2A-NMDA preferring antagonist NVP-AAM077 prior to the reactivation of cocaine-associated memory prevents the dephosphorylation of GSK3β in the amygdala (A), nucleus accumbens (B), and hippocampus (C). Top, representative immunoblots of p-GSK3β and total GSK3β in brain regions from mice pretreated with saline (Sal) or NVP-AAM077 (5 mg/kg) in home cage (i.e., no exposure, No E) or prior to exposure (E) to the environment previously paired with cocaine. Levels of p-GSK3β were significantly higher in the amygdala, nucleus accumbens and hippocampus of NVP-pretreated versus saline-pretreated mice after recall of cocaine-associated memories. Amy, amygdala; NAc, nucleus accumbens; Hippo, hippocampus. Data are shown as p-GSK3β/GSK3β integrated density ratios expressed as percent of the ratio in the saline + no-exposure control groups. *p<0.05 compared with saline + no exposure. #p<0.05, ## p<0.01 for NVP + exposure compared with saline + exposure. N=5–7 mice /group.

Experiment 4: Blockade of GluN2B-containing NMDA receptors prevented activation of GSK3β produced by cocaine memory recall

Mice underwent the same procedures as described in Experiment 3, except that on Day 10, mice were pretreated with saline or ifenprodil (5 mg/kg) 30 min prior to the 10-min memory reactivation session or remained in home cages (i.e., no exposure group). Brains were obtained immediately after the 10 min reactivation to investigate the regulation of GSK3β activity during cocaine memory reactivation. Representative immunoblots of the amygdala (Fig. 5A), nucleus accumbens (Fig. 5B) and hippocampus (Fig. 5C) from mice with or without exposure to the environment previously paired with cocaine are presented. Two-way ANOVA of levels of phospho-GSK3β in the amygdala (Fig. 5A) revealed significant main effects of environment (F_1,20 = 9.33, p < 0.01) and ifenprodil treatment×environment interaction (F_1,17 = 6.0, p < 0.05). Similarly, two-way ANOVA of levels of phospho-GSK3β also revealed significant main effects of ifenprodil treatment (F_1,20 = 6.40, p < 0.05), as well as ifenprodil treatment×environment interaction (F_1,17 = 4.55, p < 0.05) in the nucleus accumbens(Fig. 5B). A significant main effect of environment was found in the hippocampus (F_1,20 = 4.78, p < 0.05), as well as ifenprodil treatment×environment interaction (F_1,17 = 10.21, p < 0.01) (Fig. 5C). Post hoc tests revealed that levels of phosphorylated GSK3β-Ser9 were significantly lower following the reactivation of cocaine cue memories compared with no exposure in the amygdala, nucleus accumbens, and hippocampus (all p < 0.01). Mice pretreated with ifenprodil prior to re-exposure to cocaine-paired environment showed significantly higher levels of p-GSK3β in the amygdala (p < 0.05), nucleus accumbens (p < 0.01), and hippocampus (p < 0.01) compared with saline-injected exposed mice. These data indicate that reactivation of cocaine memory activates GSK3β (i.e., reduces p-GSK3β), which is dependent on activity of GluN2B receptors.

Figure 5.

Pretreatment with a GluN2B-NMDA selective antagonist ifenprodil prior to the reactivation of cocaine-associated memory prevents the dephosphorylation of GSK3β activity in the amygdala (A), nucleus accumbens (B), and hippocampus (C). Top, representative immunoblots of p-GSK3β and total GSK3β in brain regions from mice pretreated with saline or (Sal) ifenprodil (Ifen) (5 mg/kg) in home cage (i.e., no exposure, No E) or prior to exposure (E) to the environment previously paired with cocaine. Ifenprodil attenuated the dephosphorylation of GSK3β in the amygdala, nucleus accumbens and hippocampus after reactivation of cocaine memories. Amy, amygdala; NAc, nucleus accumbens; Hippo, hippocampus. Data are shown as p-GSK3β/GSK3β integrated density ratios expressed as percent of the ratio in the saline + no-exposure control groups. **p<0.01 compared with saline + no exposure. #p<0.05, ##p<0.01 for saline + exposure vs. ifenprodil + exposure. N=6–7 mice/group.

Experiment 5: Inhibition of PP1 blocked the activation of GSK3β produced by re-exposure to the cocaine-paired environment.

Experiment 5 investigated whether activation of GSK3β produced by recall of cocaine reward memories is dependent on protein phosphatase1 (PP1) function. Following cocaine place conditioning and preference testing (Day 9), two groups of mice were pretreated with saline or okadaic acid (OA; 150 ng/3ul, i.c.v) 30 min prior to the 10-min reactivation session on Day 10 (i.e., re-exposure groups). Two other groups remained in their home cages and received the same pretreatment without memory reactivation (i.e., no exposure groups). Brains were obtained immediately thereafter. Representative immunoblots of the nucleus accumbens (Fig. 6A) and hippocampus (Fig. 6B) from mice with or without exposure to the environment previously paired with cocaine are presented. Two-way ANOVA of levels of phospho-GSK3β in the nucleus accumbens (Fig. 6A) revealed significant main effects of okadaic acid treatment (F_1,21 = 22.04, p < 0.001) and exposure condition (F_1,21 = 7.71, p < 0.05). Similarly, two-way ANOVA of levels of phospho-GSK3β in the hippocampus also revealed significant main effects of okadaic treatment (F_1,21 = 23.77, p < 0.001) and exposure condition (F_1,21 = 10.53, p < 0.01). Post hoc tests revealed that okadaic acid significantly increased the levels of phospho-GSK3β in both the nucleus accumbens and hippocampus (p < 0.05 vs saline) under the ‘No Exposure’ condition. Reactivation of cocaine cue memories resulted in lower levels of phospho-GSK3β in the nucleus accumbens (p < 0.05; saline no exposure vs saline exposure) and hippocampus (p < 0.05). Compared with saline pretreatment, okadaic acid administration prior to exposure to cocaine-paired environment resulted in significantly higher levels of phospho-GSK3β in the nucleus accumbens (p < 0.01 saline exposure vs OA exposure) and hippocampus (p < 0.001 saline exposure vs OA exposure). These data suggest that PP1 regulates the phosphorylation of GSK3β, including during recall of cocaine-associated memories.

Figure 6.

Okadaic acid (OA) pretreatment followed by re-exposure of mice to the environment previously paired with cocaine reversed the dephosphorylation of GSK3β in the nucleus accumbens (A) and hippocampus (B). Top, representative immunoblots of GSK3β in brain regions from mice pretreated with saline or OA (150 ng/3ul, i.c.v.) in home cage (i.e., No Exposure, No E) or prior to exposure (E) to the environment previously paired with cocaine. Levels of p-GSK3β were significantly higher in the nucleus accumbens and hippocampus of OA-pretreated versus saline-pretreated mice under both exposure conditions. NAc: nucleus accumbens, Hippo: hippocampus. Data are shown as p-GSK3β/GSK3β integrated density ratios expressed as percent of the ratio in the saline + no-exposure control groups. *p<0.05 compared with saline + no exposure. ##p<0.01 for saline + exposure vs. OA + exposure. N=6–7 mice/group.

Experiment 6: Inhibition of protein phosphatase 1 abrogated a previously established cocaine-place preference.

Since the activation of GSK3β induced by re-exposure to an environment previously associated with cocaine was blocked by PP1 inhibition, the ability of okadaic acid to interfere with the reconsolidation of cocaine reward memories was investigated. Following an 8-day cocaine conditioning procedure, two groups of mice showed similar preferences for the cocaine-paired compartment of the conditioning chamber on Day 9 (Fig. 7). On Day 10, mice were pretreated with okadaic acid (150 ng/3ul, i.c.v) or saline 30 minutes before confinement to the prior cocaine-paired compartment. After the 10-minute re-exposure, mice were returned to the home cage and preference was tested again 24 hours later. Two-way RM-ANOVA of preference scores on Days 9 and 11 revealed significant main effects of okadaic acid treatment (F_1,22= 6.91, p < 0.05) and test day (F_1,22 = 6.0739, p < 0.05), and a treatment×day interaction (F_1,22 = 4.867, p < 0.05). Post hoc tests revealed that administration of okadaic acid prior to cocaine memory reactivation abolished the previously established preference for the cocaine-paired compartment when tested 24 h later (p < 0.01 vs. saline, Day 11. Fig. 7A). Two additional groups of mice underwent similar cocaine place conditioning and testing on Day 9. On Day 10, these mice were injected with vehicle or okadaic acid (150 ng/3ul, i.c.v) in their home cage without re-exposure to the cocaine-paired environment. When preference was re-tested 24 hours later, both groups of mice maintained their cocaine place preference at levels similar to Day 9 and were not significantly different from each other. Two-way RM-ANOVA of preference scores showed no significant effects of okadaic acide treatement (F_1,13 = 0.36, p > 0.05), test day (F1_,13 = 0.53, p > 0.05), or treatment×day interaction (F_1,13 = 0.003, p > 0.05) (Fig. 7B).

Figure 7.

Inhibition of PP1 with okadaic acid (OA) impaired the reconsolidation of cocaine associated memory. Conditioned place preference was established with 10 mg/kg of cocaine, as shown on Day 9 (A & B). A: On Day 10, mice were re-exposed to the cocaine-paired environment for 10 minutes following okadaic acid (150 ng/3ul, i.c.v) or saline administration. Mice treated with okadaic acid (OA) showed no preference for the cocaine-paired side when re-tested on Day 11. B: Mice that received vehicle or okadaic acid in the home cage environment maintained a place preference for the cocaine-paired side when tested on Day 11. **p<0.01 for OA vs. saline. N=7–8 mice/group.

Discussion

Our prior work demonstrated the requirement of GSK3β signaling in the reconsolidation of cocaine reward memories. GSK3β is activated by reactivation of cocaine-associated contextual memories, and inhibition of GSK3β after memory reactivation prevents reconsolidation and erases a previously established cocaine place preference (Shi et al., 2014). The present study sought to determine the upstream factors that control GSK3β activity and are critical for the maintenance of cocaine reward memory. Results presented herein demonstrate the importance of GluN2A and GluN2B NMDA receptor subtype activation during recall of cocaine reward memory to the regulation of GSK3β activity and the maintenance of these memories. Further, the data identify brain regions of potential significance for the reconsolidation of cocaine reward memory.

Previous studies have demonstrated the importance of NMDA receptors in the reconsolidation process (Alaghband & Marshall 2013; Itzhak 2008; Liddie & Itzhak 2016). NMDA receptors are heteromeric complexes composed of two GluN1 and two GluN2 subunits. There are four different GluN2 subunits, with GluN2A and 2B implicated in learning and memory (Ben Mamou et al. 2006; Milton et al. 2013; Shipton & Paulsen 2014). The present study investigated the roles of GluN2A- and GluN2B-containing NMDA receptors in reconsolidation of cocaine memories. Results demonstrate that blockade of GluN2A-containing NMDA receptors with the antagonist, NVP-AAM077, disrupted cocaine memory reconsolidation as evidenced by loss of a previously established cocaine place preference when tested 24 hours and 7 days later. Further, GluN2A receptors were found to regulate GSK3β activity during memory reactivation. Blockade of GluN2A receptors prior to cocaine memory reactivation prevented the dephosphorylation of GSK3β in the amygdala, nucleus accumbens and hippocampus, indicating that GluN2A-containing receptors are involved in the activation of GSK3β following reactivation of cocaine reward memory. Upon investigating the role of GluN2B-containing NMDA receptors in reconsolidation of cocaine reward memories, the selective GluN2B antagonist ifenprodil (Brittain et al. 2012) was found to attenuate a previously established cocaine place preference when administered immediately after memory reactivation. Ifenprodil also prevented the dephosphorylation of GSK3β in the amygdala, nucleus accumbens and hippocampus, indicating that GluN2B receptors are involved in the activation of GSK3β following cocaine reward memory reactivation.

The ability of NVP-AAM077 or ifenprodil to disrupt cocaine mnemonic traces only occurred when these agents were administered following memory reactivation, providing further support that they are specifically interfering with reconsolidation of cocaine-associated memories rather than impairing memory recall. Further support for interference with reconsolidation comes from the finding that there was no spontaneous recovery of place preference 7 days later. Our findings suggest that both GluN2A- and GluN2B-containing NMDA receptors are required for the reconsolidation of cocaine contextual memories. However, a potential caveat to this conclusion involves consideration of the selectivity of the GluN2A inhibitor. The relative selectivity of NVP-AAM077 for the GluN2A subunit over the GluN2B subunit is not as high as originally suggested (Auberson et al. 2002; Frizelle et al. 2006), and so NVP-AAM077 may be a preferential rather than selective GluN2A antagonist (Berberich et al. 2005). Thus, it cannot be ruled out that some of the effects of NVP-AAM077 on the reconsolidation of cocaine-associated memories reported here are mediated by GluN2B-containing receptors. The present findings with ifenprodil support a prior study demonstrating that ifenprodil or traxoprodil administered after reactivation of cocaine contextual memories disrupts an established place preferences (Liddie & Itzhak 2016), and so together demonstrate the importance of GluN2B-containing NMDA receptors in the reconsolidation of cocaine reward memories.

The regulation of GluN2A- and Glu2B-NMDA receptors during the reconsolidation of cocaine memories associated with cocaine self-administration has been investigated to a limited extent. Phosphorylation of GluN2A, but not GluN2B, in the dorsal hippocampus is higher during recall of cocaine contextual memories associated with the drug self-administration chamber (Wells et al. 2016). In addition, antagonism of GluN2A-containing NMDA receptors by NVP-AAM or PEAQX following cocaine-memory reactivation impairs subsequent drug context-induced cocaine-seeking (Wells et al., 2016). This report did not include a GluN2B antagonist, but it supports a role for GluN2A-containing receptors in reconsolidation of cocaine contextual memories as associated with cocaine self-administration.

In addition to reward memories, the involvement of subtypes of NMDA receptors in reconsolidation of aversive memory has been reported. GluN2A-containing (Milton et al, 2013), but not GluN2B-containing (Ben Mamou et al, 2006) NMDA receptors are required for fear memory reconsolidation, whereas GluN2B receptors are involved in destabilization of reactivated memories (Ben Mamou et al. 2006; Milton et al. 2013). However, inconsistencies in the literature exist. Nakayama and colleagues showed that GluN2B NMDA receptor signaling may be involved in both the restabilization of reactivated memories and their destabilization (Nakayama et al. 2016). The disparate results across studies may be due to differences in experimental design including site and timing of drug injection (Ben Mamou et al. 2006; Nakayama et al. 2016).

The present study also investigated PP1 which is known to be an activator of GSK3β via dephosphorylation of Ser9-GSK3β (Peineau et al. 2008; Szatmari et al. 2005). NMDA receptor stimulation in both cultured primary (cortical and hippocampal) neurons and adult mouse brain (striatum and hippocampus) rapidly activates GSK3β by dephosphorylation of Ser9-GSK3β (Szatmari et al, 2005). Further, dephosphorylation of GSK3β triggered by NMDA receptor stimulation is inhibited by 1μM okadaic acid, a concentration that inhibits both PP1 and PP2A (Cohen et al. 1989). At PP2A-selective concentrations, okadaic acid fails to affect NMDA regulation of Ser9-GSK3β (Szatmari et al, 2005), suggesting that PP1 is responsible for the dephosphorylation of GSK3β. Our previous study showed that levels of phospho-GSK3β were downregulated following reactivation of cocaine reward memory, and inhibition of GSK3 disrupted cocaine memory reconsolidation (Shi et al, 2014). In the present study, okadaic acid administered prior to cocaine memory reactivation blocked the dephosphorylation of GSK3β, indicating that activation of GSK3β following recall of cocaine memory is dependent on PP1. Okadaic acid increased the levels of Ser9-GSK3β phosphorylation in animals without memory reactivation, so PP1 may be involved in regulating phospho-GSK3β levels in this basal condition.

Conclusions

When established memories are reactivated by recall, they become labile and susceptible to disruption. This provides an opportunity to interfere with the maintenance of memory traces that contribute to pathological states, in this case drug-seeking behaviors. The data presented here suggest that following reactivation of cocaine reward memories, GluN2A- and GluN2B-containing NMDA receptors are stimulated, PP1 is activated and the activity of GSK3β is induced accordingly. Further, the function of this pathway is necessary for the reconsolidation of a cocaine reward mnemonic trace; inhibition of this signaling pathway at any step is sufficient to impair the maintenance of such memories. Post-reactivation inhibition of GluN2A- and GluN2B-containing NMDA receptor subtypes impairs reconsolidation and erases an established cocaine place preference, thus making them potential therapeutic targets in treatment of cocaine dependence and the prevention of relapse.

Figure 1.

Timeline for the experiments. Figure 1 top: Experiments 1, 2, and 6 involved 8 days of place conditioning followed by a test for place preference on Day 9. Mice were re-exposed to the conditioning chambers or remained in the home cage on Day 10, and administered a test agent immediately thereafter. Place preference was retested on Days 11 and 18. Figure 1 bottom: Experiments 3, 4, and 5 similarly involved 8 days of place conditioning followed by a test for place preference on Day 9. On Day 10, mice were re-exposed to the chamber or remained in the home cage and were euthanized immediately thereafter to obtain brains for protein measurements. Experiments 5 and 6 involved i.c.v. administration of okadaic acid; in these experiments, mice were surgically implanted with a guide cannula 5–7 days prior to the start of conditioning. Numbers of mice entering into each experiment on Day 1 are shown in black font, numbers of mice excluded after Day 9 testing due to failure to acquire a cocaine place preference are shown in blue font, and numbers of mice proceeding on to Day 10 testing are shown in red font. Animal numbers are described in detail in the Experimental Design section of Materials and Methods.

Abbreviations:

GSK3β

glycogen synthase kinase-3β

PP1

protein phosphatase 1

RM ANOVA

repeated measures analysis of variance

RRID

Research Resource Identifier

LTD

long-term depression

LTP

long-term potentiation

NMDAR

N-methyl-D-aspartate receptor