Increasing the GluN2A/GluN2B Ratio in Neurons of the Mouse Basal and Lateral Amygdala Inhibits the Modification of an Existing Fear Memory Trace

Abstract

Reconsolidation updating is a form of memory modification in which an existing memory can become destabilized upon retrieval and subsequently be modified via protein-synthesis-dependent reconsolidation. However, not all memories appear to destabilize upon retrieval and thus are not modifiable via reconsolidation updating approaches and the neurobiological basis for this remains poorly understood. Here, we report that auditory fear memories created with 10 tone-shock pairings are resistant to retrieval-dependent memory destabilization and are associated with an increase in the synaptic GluN2A/GluN2B ratio in neurons of the basal and lateral amygdala (BLA) compared with weaker fear memories created via one or three tone-shock pairings. To increase the GluN2A/GluN2B ratio after learning, we generated a line of mice that expresses an inducible and doxycycline-dependent GFP-GluN2A transgene specifically in α-CaMKII-positive neurons. Our findings indicate that increasing the GluN2A/GluN2B ratio in BLA α-CaMKII-positive neurons after a weak fear memory has consolidated inhibits retrieval-dependent memory destabilization and modification of the fear memory trace. This was associated with a reduction in retrieval-dependent AMPA receptor trafficking, as evidenced by a reduction in retrieval-dependent phosphorylation of GluR1 at serine-845. In addition, we determined that increasing the GluN2A/GluN2B ratio before fear learning significantly impaired long term memory consolidation, whereas short-term memory remained unaltered. An increase in the GluN2A/GluN2B ratio after fear learning had no influence on fear extinction or expression. Our results underscore the importance of NMDAR subunit composition for memory destabilization and suggest a mechanism for why some memories are resistant to modification.

Significance statement: Memory modification using reconsolidation updating is being examined as one of the potential treatment approaches for attenuating maladaptive memories associated with emotional disorders. However, studies have shown that, whereas weak memories can be modified using reconsolidation updating, strong memories can be resistant to this approach. Therefore, treatments targeting the reconsolidation process are unlikely to be clinically effective unless methods are devised to enhance retrieval-dependent memory destabilization. Currently, little is known about the cellular and molecular events that influence the induction of reconsolidation updating. Here, we determined that an increase in the GluN2A/GluN2B ratio interferes with retrieval-dependent memory destabilization and inhibits the initiation of reconsolidation updating.

Keywords: Glun2A; NMDAR; Pavlovian fear conditioning; amygdala; memory; reconsolidation.

Figure 1.

Auditory fear memories that are resistant to retrieval-dependent destabilization are associated with an increase in the GluN2A/GluN2B subunit ratio in the BLA. A, Mice trained with 10 TSPs froze significantly more than mice trained with TSPs during an LTM test, indicating that 10 TSP training results in stronger fear memory compared with three TSP training. B, Behavior timeline for a reconsolidation experiment using mice trained with three TSPs. Mice were auditory fear conditioned with three TSPs and, 2 d later, retrieval of the fear memory was induced by exposing the mice to a single tone presentation. After this, the mice were immediately infused with either anisomycin or vehicle into the BLA. One day later, the mice were reexposed to tones in a novel context during a PR-LTM test and freezing was measured. C, There were no differences in percentage freezing during retrieval, but the anisomycin group froze significantly less compared with the vehicle group during the PR-LTM test, indicating that these mice did exhibit retrieval-dependent memory destabilization. D, Behavior timeline for a reconsolidation experiment using 10 TSPs. Mice were auditory fear conditioned with 10 TSPs and, 2 d later, retrieval of the fear memory was induced by exposing the mice to a single tone presentation. After this, the mice were immediately infused with either anisomycin or vehicle into the BLA. One day later, the mice were reexposed to five tones in a novel context during a PR-LTM test and freezing was measured. E, During retrieval, mice trained with 10 TSPs did not exhibit differences in freezing between the vehicle and anisomycin groups. During PR-LTM, mice trained with 10 TSPs did not exhibit differences in freezing between the vehicle and anisomycin groups, indicating that these mice did not exhibit retrieval-dependent memory destabilization (*p < 0.05). F, Timeline for a Western blotting experiment performed with BLA PSD fractions to determine the levels of GluN1, GluN2A, and GluN2B levels in mice fear conditioned with one TSP, three TSPs, and 10 TSPs. GluN1, GluN2A, and GluN2B protein levels were normalized to PSD-95 levels to account for variability in protein loading and group means were reported. G–J, Western blot and quantification of protein levels in BLA PSDs extracted from mice 48 h after training. One-way ANOVA between groups that were trained with one TSP, three TSPs, or 10 TSPs revealed that there were no differences in GluN1 levels (G), GluN2A levels (H), or GluN2B levels. However, there was a trend for an increase in GluN2A levels and a trend for a decrease in GluN2B as TSP increased, respectively. J, The GluN2A/GluN2B ratio was significantly increased in mice trained with 10 TSPs compared with one TSP and three TSPs and no differences were observed between one TSP and three TSPs (*p < 0.05). Error bars indicate SEM.

Figure 2.

Spatiotemporal characterization of the GFP-GluN2A transgene expression in GFP-GluN2A+/tTA+ mice and its regulation using Dox. A, GFP-GluN2A transgene used to generate an inducible transgenic line expressing GFP-GluN2A+ from a TRE3G promoter. B–F, Representative coronal images depicting GFP-GluN2A transgene expression (GFP fluorescence) throughout the brain in double-transgenic mice (GFP-GluN2A+/tTA+) mice compared with controls (GFP-GluN2A+). Coronal images from bregma (−6.12; B) demonstrates no transgene expression in cerebellar cortex (Bi) and cerebellar nuclei (Bii). C, Bregma (−3.52) demonstrates positive transgene expression in auditory and temporal association cortex (Ci) and retrosplenial cortex (Cii), but not in thalamus (Ciii) or ventral hippocampus (Civ). D, Bregma (−1.82) demonstrates no transgene expression in dorsal hippocampus and central nucleus (Di), but positive transgene expression in basal and lateral amygdala (Dii). Ei, Bregma (−0.10) demonstrates no transgene expression in striatum. F, Bregma (+1.98) demonstrates positive transgene expression in olfactory cortex (Fi), medial prefrontal cortex (Fii), and motor cortex (Fiii). Scale bar, 100 μm (50×).

Figure 3.

Regulation of transgene expression in GFP-GluN2a+/tTA+ using Dox. A, Timeline for temporal regulation of transgene expression using Dox. B, Transgene expression in GFP-GluN2A+/tTA+ mice is tightly regulated using Dox. Representative coronal images containing BLA show no GFP-GluN2A transgene expression in GFP-GluN2A+/tTA+ mice that were On-Dox at 6 weeks (Bi) and 8 weeks (Biii), but Bii depicts robust transgene expression when Dox is removed from the mice diet for 2 weeks. Scale bars (left to right), 250 μm (20×), 100 μm (50×), 50 μm (100×).

Figure 4.

Overexpression of GFP-GluN2A alters NMDA currents in pyramidal neurons and leads to an increase in the GluN2A/GluN2B ratio within the BLA. A, Example traces of evoked EPSCs in control (left column) and GFP-GluN2A expressing neurons. Pharmacologically isolated NMDA currents (black traces) were recorded at +40 mV and then either the GluN2AR antagonist PEAQX (top row) or the GluN2BR antagonist Ro25-6981 (bottom row) was bath applied. The residual currents are shown in gray. B, Summary graph of the data shown in A. Cells from mice overexpressing GFP-GluN2ARs (gray bars) show a much larger reduction in current after PEAQX application than those from control littermates (black bars). GluN2BR-mediated currents, measured as a reduction after Ro25-6981 application, were not differentially affected between control and GFP-GluN2A+/tTA+ mice. C, Normalized sample traces of total NMDA (left) and isolated GluN2A currents (Ro25, top right) and GluN2B currents (PEAQX, bottom right). D, Expression of GluN2AR decreases decay time of the total NMDA current, but does not significantly affect currents in the presence of either Ro25-6981 or PEAQX. E–H, Western blot and quantification of GluN1, GluN2A, and GluN2B levels in BLA PSDs extracted from transgenic mice (GFP-GluN2A+/tTA+) expressing GFP-GluN2A compared with controls (GFP-GluN2A+). E, No differences in GluN1 levels were detected. F, G, A significant increase in GluN2A levels and a significant reduction in GluN2B levels were detected in GFP-GluN2A+/tTA+ mice. H, A significant increase in the GluN2A/GluN2B ratio was detected in GFP-GluN2A+/tTA+ mice. Error bars indicate SEM (*p < 0.05).

Figure 5.

An increase in the GluN2A/GluN2B ratio within BLA neurons inhibits the modification of an existing memory trace and reduces retrieval-dependent AMPAR trafficking. A, Behavior timeline for a reconsolidation experiment using control mice (GFP-GluN2A+ only) trained with three TSPs. Two days after training, half of the mice were removed from the Dox-containing diet. Fourteen days later, the mice were induced to retrieve the fear memory by exposing them to a single tone presentation and then the mice were immediately infused into the BLA with either anisomycin or vehicle. One day later, the mice were reexposed to five tones in a novel context during a PR-LTM test and freezing was measured. B, During retrieval, control mice did not exhibit a significant difference in freezing levels. During PR-LTM, there was a significant main effect of percentage freezing. The On-A group froze significantly less than the On-V group and the Off-V group. Further, the Off-A group froze significantly less than the Off-V group and the On-V group. However, there were no differences in percentage freezing due to Dox administration alone in the A or V groups (V = vehicle, A = anisomycin; *p < 0.05). C, Behavior timeline for a reconsolidation experiment using GFP-GluN2A+/tTA+ mice trained with three TSPs using a similar methodology as described in A. D, During retrieval, GluN2A+/tTA+ mice did not exhibit a significant difference in freezing levels. During PR-LTM, there was a significant main effect of percentage freezing. The On-A group froze significantly less than the On-V group, the Off-V group, and the Off-A group, but there were no differences between the Off-A and Off-V and On-V groups. There were also no differences in percentage freezing due to Dox administration alone in the vehicle group. E, Timeline for Western blotting experiment on BLA PSDs to determine the levels of pGluR1(Ser-845) and tGluR1 in control mice 1 h after retrieval (V = vehicle, A = anisomycin; *p < 0.05). F, In control mice, there was a significant effect of group on the pGluR1/tGluR1 ratio. The pGluR1/tGluR1 ratio was significantly higher in the On-Ret group compared with the On-No Ret and Off-No Ret groups. Further, the pGluR1/tGluR1 ratio was significantly higher in the Off-Ret group compared with the Off-No Ret and On-No Ret groups. However, there were no differences in the ratio due to Dox administration alone in the retrieval or no retrieval groups. G, Timeline for a Western blotting experiment on BLA PSDs to determine the levels of pGluR1(Ser-845) and tGluR1 in GFP-GluN2A+/tTA+ mice 1 h after retrieval (*p < 0.05). H, There was a significant effect of group on the pGluR1/tGluR1 ratio. The On-Ret group had significantly higher pGluR1/tGluR1 levels compared with the On-No Ret, Off-No Ret, and Off-ret groups. There were no significant differences in the pGluR1/tGluR1 ratio in the Off-Ret compared with Off-No Ret and the On-No Ret groups. In addition, there was no effect of Dox administration alone in the no retrieval groups (*p < 0.05). Error bars indicate SEM.

Figure 6.

An increase in the GluN2A/GluN2B ratio before fear learning impairs LTM consolidation. A, Behavior timeline for fear learning experiment. Wild-type, α-CaMKII-tTA+, GFP-GluN2A+, and GFP-GluN2A+/tTA+ mice were maintained on Dox, Dox was removed from their diet for 14 d, and then the mice were fear conditioned with three TSPs. Three hours after training, the mice conditioned fear responses to the auditory cue were examined in an STM test in which the mice were exposed to three tones. Twenty-one hours later, LTM was examined where the mice were exposed to five tones. B, For STM, there was no significant main effect of genotype on percentage freezing (wild-type, α-CaMKII-tTA+, GFP-GluN2A+, or GFP-GluN2A+/tTA+). C, For LTM, there was a significant main effect of genotype on percentage freezing. The GFP-GluN2A+/tTA+ group froze significantly less compared with the wild-type, α-CaMKII-tTA+, and GFP-GluN2A+ groups. However, there were no differences in percentage freezing among the wild-type, α-CaMKII-tTA, and GFP-GluN2A+ groups (wild-type vs α-CaMKII-tTA, α-CaMKII-tTA vs GFP-GluN2A+, GFP-GluN2A+ vs wild-type (*p < 0.05). Error bars indicate SEM.

Figure 7.

An increase in the GluN2A/GluN2B ratio after learning has no influence on fear memory extinction. A, Behavior timeline for fear expression and fear extinction experiment. GFP-GluN2A+/tTA+ mice were maintained on a diet containing Dox and then fear conditioned with three TSPs. Two days after training, half of the mice had Dox removed from their diet. Fourteen days later, the mice were exposed to 10 tones once per day over a period of 4 d. For expression of fear, freezing behavior was reported for the first three tones presented during LTM1. For fear extinction behavior, freezing behavior during LTMs 1, 2, 3, and 4 is presented. B, For fear expression, there was no significant effect of group on percentage freezing. C, For fear extinction, there was no significant effect of group on percentage freezing during LTM1, LTM2, LTM3, and LTM4. Error bars indicate SEM.