Glutamate receptor plasticity in fear memory
reconsolidation
ABSTRACT
The reactivation of a consolidated memory transiently destabilizes it, after which a reconsolidation process can strengthen or modify its content. During reconsolidation of contextual fear, AMPA receptor (AMPAR) plasticity in the dorsal hippocampus (DH) orchestrates the reorganization of fear memory at the synaptic level. The NMDA type of glutamate receptor (the NMDAR) is also an important regulator of synaptic plasticity. However, it is unclear if and how the NMDAR contributes to contextual fear memory reconsolidation. Here, we aimed to investigate the specific role of NMDARs in the DH during fear memory reconsolidation and their relation to the expression of contextual fear. We used a competing ligand, in the form of a mimetic peptide of the C-terminal tail of the GluN2A subunit, to prevent the previously observed upregulation of GluN2A-containing NMDARs in the synaptic membrane in the DH 7 h after retrieval. To investigate whether this process also occurs in the ventral hippocampus (VH), we compared the expression levels of AMPARs and NMDARs in the DH and VH at this time point. Although the ligand failed to reduce retrieval-induced GluN2A expression or alter the expression of fear, we confirmed enhanced GluN1 and GluN2A expression 7 h after retrieval. Moreover, enhanced expression of GluN2B was found at this time point. In addition, we report minor differential expression patterns of ionotropic glutamate receptors in the VH compared with the DH.
Author: Koen Seignette (6042899)
Supervisor: Sabine Spijker
Daily supervisor: Leanne J.M. Schmitz
UvA representative: Harm Krugers (also co-assessor)
Research Institute: Center for Neurogenomics and Cognitive research (CNCR)
Study: Research Master Brain and Cognitive Sciences (RMBCS)
Track: Neuroscience
University: University of Amsterdam
Period: January - July 2013
INTRODUCTION
The formation and survival of new memories depends on two distinct processes. After its initial acquisition newly learned information becomes stabilized over time through specific network modifications, a process generally referred to as consolidation. The last few decades it has become clear that consolidated memories are not necessarily stabile, but can become dynamic and sensitive to modifications after their retrieval1. Such reorganization of memory content is termed
reconsolidation and its mechanisms have received much interest in recent years. It is evident that both protein synthesis and synaptic protein degradation are crucial for the reorganization of reactivated memories2-6. These findings are in line with the synaptic plasticity hypothesis of learning
and memory, where long-term depression (LTD) and long-term potentiation (LTP) serve to bring about long lasting changes in synaptic efficacy7.
The modifications in synaptic strength that occur during LTD and LTP depend on the synthesis, degradation and trafficking of many intracellular proteins, including postsynaptic density (PSD) proteins and several membrane receptor complexes8-10. Particularly important for
hippocampus-dependent memory storage, are the ionotropic type of glutamate receptors (iGluRs). Glutamate receptors mediate excitatory neurotransmission throughout the brain and their synaptic insertion is crucial for the induction of LTP11. The three defined classes – the N-methyl-D-aspartate type (NMDA),
α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid type (AMPA) and Kainate type (KA) – are all formed by the assembly of different subunits, each coded by their own specific gene12,13.
The tetrameric NMDA receptor (NMDAR) mainly assembles as two copies of the obligatory GluN1 subunit (NR1) with either two copies of the GluN2A (NR2A) type, two copies of the GluN2B (NR2B) type, or one copy of GluN2A and one copy of GluN2B. Specific assemblies in the membrane not only differ in stabilization, internalization and surface mobility14-18, but also vary in physiological
properties10,12,13,26. For instance, in Human Embryonic Kidney 293 cells (HEK293 cells) that express
NMDAR subunit complementary DNAs (cDNAs), GluN2A-containing NMDARs show faster deactivation kinetics than GluN2B-containing NMDARs41. Likewise, a more recent study on HEK293 cells showed
higher open probabilities of single channel GluN2A-containining NMDARs compared to GluN2B-containing NMDARs upon glutamate application42. In addition, a developmental switch in NMDAR
subunits expression has been suggested in thalamic and cortical synapses, where GluN2B-containing NMDARs are gradually replaced by GluN2A-containing NMDARs44. Thus, physiological properties also
change during development.
Like the NMDAR, the AMPA receptor (AMPAR) is composed of four subunits, but appears as a homodimer or heterodimer. In the rat hippocampus, GluA2 predominantly assembles with GluA1 and GluA319. The GluA2-containing AMPAR is Ca2+-impermeable (CI-AMPARs) and relatively stable at the
synapse, whereas GluA1- and GluA3-containing AMPARs conduct Ca2+ (CP-AMPARs) and therefore
appear to be important in fast synaptic potentiation20. The significance of AMPARs in synaptic
plasticity is shown by experiments where induction of LTD coincides with the endocytosis of AMPARs, whereas LTP often strongly increases AMPAR levels in the potentiated synapse, as reviewed by Malinow & Malenka (2002)21.
where a specific context becomes associated with an aversive stimulus (e.g. foot shock). Similarly, during amygdala-dependent auditory fear conditioning, foot shocks are paired to an auditory cue (tone). Re-exposure (retrieval) to the conditioned stimulus (context or tone, respectively) induces freezing which is the lack of movement that is used as readout of fear. Several studies indicate that contextual fear conditioning and auditory fear conditioning induce AMPAR plasticity in the hippocampus and amygdala, respectively. For instance, contextual fear conditioning generates a spine type specific recruitment of overexpressed GluA1-containing AMPARs in the hippocampus 24 h after training33, whereas in the lateral amygdala (LA) a GluA1-induced synaptic potentiation has been
observed after auditory fear conditioning. Blocking this latter process effectively abolishes the auditory fear memory34. Moreover, the erasure of fear caused by timed extinction training seems to
depend on specific removal of synaptic GluA1- and GluA3-containing AMPARs in the amygdala after retrieval of conditioned memories35. Specific timing within the reconsolidation period therefore
seems to be highly important for the disruption of unwanted memories, an observation with strong implications for the treatment of post-traumatic stress disorder (PTSD).
These studies show that AMPAR trafficking during consolidation and reconsolidation dynamically regulates potentiation of synapses and that this translates to the expression of fear. Indeed, pharmacological manipulation of AMPAR dynamics seems to interfere with synaptic plasticity and can even alter the formation and content of memory. For instance, showed that the retrieval of hippocampus-dependent contextual fear memory initiates a biphasic wave of AMPAR plasticity. During the first phase (measured 1-4 h after retrieval), AMPAR endocytosis in the dorsal hippocampus (DH) leads to a reduction in synaptic AMPARs that marks a depotentiated state of the synapse. The specific reinsertion of GluA2-containing AMPARs during the second phase (7 h after retrieval) subsequently causes synaptic potentiation. Interestingly, blocking the hippocampal GluA2-containing AMPAR endocytosis prevents these plasticity waves and reinforces contextual fear memories5. These
results suggest that the DH acts as an inhibitory constraint on the expression of fear. Another recent study showed that a similar process occurs in the lateral amygdala (LA), where retrieval of auditory fear resulted in the exchange of CI-AMPARs to CP-AMPARs22. Shortly hereafter the exchange reversed
and the GluA2-AMPARs were reinserted back into the synapse. These findings indicate that similar processes of AMPAR plasticity, albeit with different time frames, in different brain regions are required for reconsolidation.
In addition to the AMPAR, we also observed a reconsolidation-dependent increase in NMDAR expression (see figure 1a,b, Rao-Ruiz et al., unpublished data). More specifically, the retrieval of contextual fear initiated the insertion of GluN1- and GluN2A-, but not GluN2B-containing NMDARs, 7 h after memory reactivation. This GluN2A-containing NMDAR regulation coincided with a decrease in NMDAR-dependent miniature EPSC decays (figure 1e,f). As stated before, GluN2A-containing NMDARs generally demonstrate faster deactivation kinetics and, therefore, shorter current decays10.
The reason for this apparent discrepancy is unclear and remains to be elucidated.
Blocking AMPAR endocytosis did not only prevent the AMPAR biphasic wave, but also the GluN2A-containing NMDAR increase (figure 1c,d), suggesting that NMDAR regulation at least partly depends on the initial AMPAR plasticity. However, it is unclear if and how these receptors interact to regulate behavioral fear and whether the NMDAR regulation is a direct effect of the AMPAR plasticity or an
indirect result of other downstream processes that depend on AMPAR trafficking. We therefore aimed to investigate the specific role of NMDARs in fear memory reconsolidation and their relation to the expression of contextual fear. To study this, we set out to use mimetic peptides that could prevent the observed NMDAR upregulation in vivo. A mimetic peptide that prevents membrane anchoring of NMDARs has been used previously in vitro15. This peptide specifically blocks the interaction between
GluN2A and postsynaptic density protein 95 (PSD-95), a membrane-associated guanylate kinase (MAGUK) scaffolding protein that anchors receptors in the synaptic membrane15. In this study, Bard et
al. (2010) used Quantum Dot (QD) tracking on NMDAR subunits in hippocampal cultured neurons to investigate their synaptic expression. Incubation with the mimetic peptide strongly reduced the amount of GluN2A subunits in the synaptic membrane on the time scale of minutes, indicating that the ligand had disrupted the interaction between GluN2A and PSD-95.
PSD-95 reduces surface trafficking and diffusion of the receptor, thereby preventing its internalization. The described competing ligand should also prevent this interaction in vivo and therefore we hypothesize that this peptide could decrease the retrieval-induced surface expression of GluN2A-containing NMDARs (see Methods). We also hypothesized that NMDAR plasticity in the DH after retrieval contributes to the expression of contextual fear. Blocking this process should reduce GluN2A-containing NMDAR synaptic membrane expression and therefore the inhibitory constraint on the expression of fear.
A second goal of this study was to investigate whether the plasticity mechanisms as described above also occur in the ventral hippocampus (VH). It is evident from lesion studies that the dorsal and ventral parts of the hippocampus serve different functions and contribute differentially to performance in spatial learning tasks such as the Morris water maze23,24. Indeed, the VH is thought to
be important in fear and anxiety, while the DH is often associated with spatial learning45. For instance,
VH lesions increased open arm entries in the elevated plus maze in rats, while DH lesions had no such effect48. However, pre- and post-training lesions of the VH in rats have been shown to impair auditory
trace fear, suggesting an important role for the VH in fear memory50. In addition, completely blocking
neuronal activity in the VH using Tetrodotoxin (a sodium channel blocker) before conditioning impaired both tone and contextual fear conditioning in rats. Clearly, the VH is an important structure for mediating conditioned fear memory. The exchange of AMPAR subunits during reconsolidation of contextual fear has been observed in both the DH and LA. These findings imply that this plasticity mechanism generalizes to other brain areas than the hippocampus. To investigate whether this process also takes place in the VH we therefore compared the expression levels of AMPARs and NMDARs in the dorsal and ventral hippocampus 7 h after the retrieval of contextual fear using western blot analyzes.
In short, we found increased NMDAR expression levels at 7 h after memory reactivation, along with small increases in GluA2-AMPARs. Our competing ligand failed to decrease retrieval induced upregulation of GluN2A levels, but significantly reduced a previously unobserved upregulation of GluN2B. We report small differences between the DH and VH, with the former showing increased regulation during reconsolidation. These data confirm important AMPA and NMDA receptor dynamics in the reconsolidation of contextual fear, with a more profound role for the DH as compared to the VH.
Figure 1 NMDAR plasticity in reconsolidation of contextual fear (unpublished data). (a) Timeline indicates the
experimental setup: NSR and USR animals underwent training (day 1) and retrieval (day 2). Dorsal hippocampus
collection took place 7 h after retrieval. (b) Quantification of synaptic membrane fraction NMDA receptor subunit
expression as a percentage of NSR values. Representative blots from samples on the same gel are shown.
Upregulation of GluN2A and GluN1, but not GluN2B, due to
conditioning was observed. (c) Experimental design with 3
groups. NS-3Y mice were exposed to context only and received dorsohippocampal injection of AMPAR endocytosis blocker (3Y) 1 h before retrieval. Shocked mice received dorsohippocampal injection of scrambled peptide (S-3A) or the AMPAR endocytosis blocker (S-3Y). (d) Blocking AMPAR endocytosis reduced GluN2A, but not GluN1, expression 7 h after retrieval. (e) Example trace of AMPAR-mediated and AMPAR+NMDAR-mediated miniature excitatory postsynaptic currents (mEPSCs) 7 h after retrieval, showing different group effects on decay of mEPSCs due to APV treatment (NMDAR antagonist). (f) Bar graphs of mEPSCs decay decrease due to APV treatment: stronger effects of APV were observed in conditioned mice that received scrambled 3A. This was prevented by blocking AMPAR endocytosis, indicating that the NMDA kinetics were back to control (NS) levels, confirming the biochemical data as shown in (d). All data points show mean ± s.e.m.; significant P-values and are indicated.
METHODS
Mice
All experiments were carried out in accordance to the Animal User Care Committee of Vrije Universiteit. Adult male C57BL/6J mice were individually housed at 20 °C at a 12 h light/dark cycle with ad libitum access to food and water. Experiments were performed during the light phase. All mice were 12 weeks of age during testing. The number of mice used for testing is indicated in each figure. In all experiments, brains were removed after cervical dislocation.
Cannula operations
Mice were anesthetized by intraperitoneal injection of Tribomethanol (Avertin, 0.02 ml g-1) and kept
warm on a 37°C blanket until the start of the operation. To reduce post-operational pain, mice received a subcutaneous injection of Buprenorphine analgesic (Temgesic®, 0.1 mg kg-1). After
exposing and cleaning the skull, double guide cannulas (Plastics One) were implanted in the dorsal hippocampus and chronically fixed to the skull with dental cement using a high-precision stereotaxic system. Coordinates were determined using the mouse brain atlas25 ; anterior-posterior coordinates
relative to bregma were 1.6 mm, and lateral coordinates relative to the midsagittal suture line were 1.03 mm. Mice were allowed to recover from the operation for at least 7 days before further procedures commenced.
Intrahippocampal injection of synthetic GluN2A derived peptide
To specifically prevent the GluN2A upregulation 7 h after retrieval, we used a biomimetic divalent competing ligand derived from the last 15 amino acids of the GluN2A carboxyl terminal ([GluN2A]2:
-NNRVYKKλPSIESDV-COOH) and a scrambled control peptide ([GluN2Scr]2: -YSLHANTANRRTRPR-CONH2)
which contained a nonsense amino acid sequence15.
A dose of 25 pmol per side delivered in a volume of 0.25 μl artificial cerebrospinal fluid (ACSF) was bilaterally infused into the dorsal hippocampus using a microinjection pump (CMA/100, CMA/Microdialysis) at a flow rate of 0.33 μl min-1 during a 90-s isoflurane (Forene; Abbott) inhalation
anesthesia. The injector remained in place for 30 s after injections to prevent back flow into the double guide cannulas. The peptides were injected exactly 5.5 h after the first retrieval test to allow optimal spread of the peptide in the DH. For behavioral experiments saline injections served as a negative control.
Contextual fear conditioning and retrieval
All behavioral experiments were carried out using a TSE Fear Conditioning System. Training and testing sessions were performed in a Plexiglas chamber with a stainless steel grid floor with constant illumination (100-500 lx) and continuous background noise (68 dB sound pressure level), situated in a gray box to shield it from the outside. The grid floor and Plexiglas chamber were cleaned with 70% ethanol before each individual session. For training, mice were placed in the test chamber where they were allowed to explore the environment for 3 minutes. USR mice then received a mild foot shock (0.7 mA, 2 s) through the grid floor. NSR mice did not receive the shock but were exposed to the context only. After an additional 30 s the animals were returned to their home cage.
During all retrieval tests, mice were re-exposed to the context only (conditioned stimulus) for 3 min during which freezing behavior was automatically assessed. Freezing was defined as lack of any movement besides respiration and heart beat during 2 s intervals and is presented as a percentage of the total test time.
Synaptic membrane isolation
Bilateral hippocampi samples were dissected on ice and stored at -80 °C. The distinction between the dorsal (septal) and ventral (temporal) part of the hippocampus was made as such that both areas comprised 50% of total hippocampal tissue. The hippocampus samples were pooled from two mice (n = 4 hippocampi, n = 5 pooled samples per group) and homogenized on ice in 0.32 M sucrose homogenization buffer (pH = 7.4) with protease inhibitors (Roche; cOmplete, mini, EDTA-free). In addition, buffers and samples were kept on ice to minimize proteolysis throughout the isolation procedure. Tissue homogenate was purified from cell debris by low speed centrifugation (1,000g for 10 min at 4 °C) and the resulting supernatant was transferred to a centrifuge tube on top of a 0.85-1.2 M sucrose step gradient. Synaptosomes were separated from remainder cell contents by ultracentrifugation (30,000 rpm for 2 h at 4 °C). The synaptosome disk was collected from the 1.2-0.85 M sucrose interphase and added to 2 ml of 5 mM HEPES buffering agent. Samples were balanced with homogenization buffer and high speed centrifuged (25,000 rpm for 30 m at 4 °C). The pellets were transferred to HEPES buffer to induce osmotic shock on ice for 15 minutes to remove loosely bound proteins. The synaptic membranes (SMs) were isolated using the same sucrose gradient and centrifugation step (30,000 rpm for 2 h at 4 °C) as described above. The SM disk on the sucrose interphase was collected and centrifuged with HEPES (25,000 rpm for 30 m at 4 °C) to obtain synaptic membrane pellets. These were resuspended and subsequently stored in 70 µl HEPES (+ Roche) at -80 °C.
Protein determination and immunoblotting analysis
We used a Bradford assay (Bio-Rad) to determine the total protein concentration in the synaptic membrane fractions. For all groups, 5 µg protein per sample was dissolved in 5X SDS loading b ffer and boiled for 5 minutes. We resolved the samples (groups alternated) on a 4-12% SDS-polyacrylamide pre-cast gel (Bio-Rad Criterion) using electrophoresis (SDS-PAGE; 120 V, 1.5 h), and transferred the proteins to a nitrocellulose membrane (Bio-Rad, specifics) (40 V, overnight). After transfer, membranes were blocked for 2 hrs in 5% nonfat dried milk in 1X Tris-buffered saline and 20% Tween (TBS-T). After washing in TBS-T, membranes were incubated overnight at 4 °C with specific primary antibodies in 3% nonfat dried milk in TBS-T. We used antibodies against GluN1 (Neuromab, 1:1,000), GluN2A (Epitomics, 1:10,000), GluN2B (Neuromab, 1:1,000), GluA1 (Epitomics, 1:20,000), GluA2 (Neuromab, 1:1,000) and PSD-95 (Neuromab, 1:10,000). After substantial washing in TBS-T, membranes were incubated for 1 h at room temperature with horseradish peroxidase (HRP)-conjugated secondary antibodies in TBS-T. Antibodies were detected with SuperSignal West Femto Chemiluminescent Substrate using ODYSSEY® Fc imaging system. To correct for input differences, we compared the total amount of protein of each sample (see supplementary figure S1), quantified using Bio-Rad analysis software (Image Lab v3.0). Membrane background staining was subtracted from individual band intensities.
Statistics & analyses
Data from immunoblot experiments concerning pharmacological treatment were analyzed using One-Way ANOVA (NSR/USR-Scr/USR-2A, significance set at P < 0.05). For multiple comparisons, significant group effects were further analyzed with Fisher’s least significant difference (LSD) post-hoc test. Independent samples t-tests were used to analyze the effect of unconditioned stimulus presentation (NSR/USR) in both DH and VH samples. Fear conditioning data were analyzed using a repeated-measures ANOVA test for retrieval 1 (day 2) and retrieval 2 (day 3) to analyze the effects of conditioning on fear expression. Univariate ANOVA was used to analyze effects of pharmacological treatment.
RESULTS
Preventing recall-induced NMDAR regulation
To investigate the role of NMDAR regulation during reconsolidation mice were trained in a hippocampus-dependent contextual fear conditioning paradigm and re-exposed to the context 24 h later. To attenuate the expression GluN2A-containing NMDARs in the synaptic membrane we applied a mimetic peptide bilaterally in the DH. We chose to inject the ligand 5.5 h after memory reactivation to specifically prevent the earlier observed GluN2A upregulation 7 h after retrieval. We reasoned that this timing would allow for optimal spread through the hippocampus and sufficient time for the peptide to exert its effect. We used a scrambled peptide to control for possible effects of the injection. Both peptides were developed to be cell-permeable by fusing each to the cell-membrane transduction domain of the HIV-1 TAT protein. In vitro, TAT-[GluN2A]2, but not TAT-[GluN2Scr]2, has
been shown to specifically perturb the interaction between the GluN2A subunit and the PDZ protein PSD-9515 (figure 2). The peptide therefore inhibits the anchoring of GluN2A at the level of the
synaptic membrane. This destabilization theoretically prevents the expression of the GluN2A-containing NMDARs into the synapse.
Figure 2 Theoretical working model of TAT-[GluN2A15]2. In native conditions, GluN2A-containing NMDARs
interact with MAGUK proteins such as PSD-95 to ensure their anchoring in the synapse. TAT-[GluN2A]2
specifically competes with GluN2A-containing NMDARs for their interaction with PSD-95, which destabilizes GluN2A-containing NMDARs in the membrane, allowing for their removal away from the synaptic site. NR2A, GluN2A; NR1, GluN1. Adapted from Bard et al., (2007).
Figure 3 shows the experimental setup and synaptic membrane expression levels of the analyzed
receptors. An effect of conditioning was found for both the GluN2A (F2,12 = 6.483, P = 0.012) and
TAT-[GluN2A]2 showed higher expression levels compared to the no-shock controls (figure 3b). The GluN1
subunit showed a trend for effect of treatment (F2,12 = 3.869, P = 0.051). Again, TAT-[GluN2Scr]2
-injected and TAT-[GluN2A]2-injected USR mice showed higher expression levels than the no-shock
controls. Surprisingly, TAT-[GluN2A]2 did not affect the expression profile of either the GluN2A
(+10.96%) or GluN1 (+8.08%) subunits, but significantly downregulated GluN2B compared with TAT-[GluN2Scr]2 (-24.57%, P = 0.044). For the AMPAR, no main effects of treatment were found, although
GluA2 showed minor – but not significant – expression increases in both TAT-[GluN2Scr]2-injected
(+21.12%) and TAT-[GluN2A]2-injected (+17.37%) mice (figure 3c).
Synaptic potentiation can lead to an overall increase in spine and PSD size and volume. Hence, using western blotting on synaptic membrane fractions, we not only analyzed AMPARs and NMDARs, but also PSD-95 expression. This allowed us to control for overall growth of the synapse and PSD to ensure that our effects were specific for the analyzed receptors. We found no main group differences, indicating that neither the memory retrieval, nor the peptide treatment had any effect on its expression (figure 3c).
Our observation that – in addition to GluN2A – GluN2B is also regulated 7 h after retrieval and that our peptide significantly reduced GluN2B, but not GluN2A, suggests that GluN2B-containing NMDARs might also contribute to the expression of fear memory. Therefore, next we examined whether TAT-[GluN2A]2 was able to increase contextual fear expression at the behavioral level (figure 4).
Conditioned mice received a dorsohippocampal injection of TAT-[GluN2A]2 or saline 5.5 h after
retrieval. The expression of fear was tested on day 3. No treatment effect was found, indicating that TAT-[GluN2A]2 had failed to alter contextual fear.
Figure 3 Retrieval induces an increase in synaptic NMDARs. (a) Timeline indicates the experimental setup: NSR and USR animals undergo training (day 1) and retrieval (day 2). USR Animals receive a peptide injection at 5.5 h after retrieval (1: TAT-[GluN2A]2 ,
2: TAT-[GluN2Scr]2). Dorsal
hippocampi are collected at 7 h after retrieval. (b,c) Quantification of synaptic membrane fraction NMDA receptor subunit expression as a percentage of NSR values. Representative blots from samples on the same gel are shown. Approximate molecular weights are indicated. For input sample used for normalization, see Supplementary figure S1. (b) All NMDAR subunits showed increased membrane expression 7 h after retrieval for TAT-[GluN2Scr]2-injected
mice (GluN1, +35.91%; GluN2A, +51.48%; GluN2B,
+62.30%). For TAT-[GluN2A]2-injected mice, both GluN2A (+40.52%) and GluN2B (+37.73%) were significantly
upregulated compared to NSR mice (P-values shown), while these mice showed a trend of upregulation (+27.82%) for GluN1. GluN2B expression was significantly lower in USR-2A compared to USR-Scr mice (- 24.57%) (c) No significant differences were found for GluA1, GluA2 and PSD-95 expression. All data points show mean ± s.e.m.; significant and trend-like P-values and are indicated.
Figure 4 TAT-[GluN2A]2
does not increase the expression of contextual fear. The timeline
indicates the
experimental design. All mice undergo a training session at day 1, and 5.5 h after the retrieval session at day 2 they receive either a 1) TAT-[GluN2A]2 (n=8), or 2) a
saline control (n=8) injection. The effect of treatment is determined by measuring freezing behavior during the retrieval session at day 3
(see methods). No differences were found between TAT-[GluN2A]2 and Saline treated mice: F1,14 = 0.525, P =
0.481, interaction of time x treatment: F2,28 = 0.989, P = 0.384. 2A, TAT-[GluN2A]2; Sal, Saline. All data points
show mean ± s.e.m.
Glutamate receptor plasticity in the dorsal and ventral hippocampus
To distinguish between the effects of memory reactivation on glutamate receptor plasticity in the DH and VH we analyzed AMPAR and NMDAR expression in both areas separately at 7 h after retrieval (figure 5a). We found no significant group differences for membrane receptor expression levels in the VH (GluN1, +2.14%; GluN2A, +14.90%; GluN2B, +6.03%; GluA1, +2.00%; GluA2, +0.68%). This suggests that retrieval of a contextual fear memory does not lead to substantial changes in glutamate receptor plasticity at this time point in the ventral part of the hippocampus (figure 5b). In addition, no significant regulation was observed in the DH (figure 5c), but all analyzed subunits besides GluA1 (-1.64%), which is known not to be regulated at this time point5, showed consistent increases in
expression in the USR compared to NSR mice (GluN1, +19.74%; GluN2A, +25.74%; GluN2B, +10.14%; GluA2, +14.20%). This effect was not observed in the VH samples.
Figure 5 Glutamate receptor plasticity differences in DH and VH. (a) The experimental design was as follows: training (day 1), retrieval (day 2) and collection (7 h after retrieval) of dorsal hippocampi (sample size is indicated in figure b,c). (b,c) Quantification of ventral (b) and dorsal (c) hippocampus synaptic membrane fraction AMPAR and NMDAR subunit expression as a percentage
of NSR values.
Representative blots from samples on the same gel are shown. Approximate molecular weights are
indicated. See
Supplementary figure S1 for input gel scans. (b) No group differences were found for all analyzed NMDAR and AMPAR subunits in the VH. (c) DH analyses showed
non-significant stimulus induced upregulations for all analyzed NMDAR subunits and for the AMPAR GluA2 subunit. All data points show mean ± s.e.m. P-values are indicated.
DISCUSSION
This study aimed to investigate the specific role of NMDARs in fear memory reconsolidation and their relation to the expression of contextual fear. We used a contextual fear conditioning paradigm in combination with timed dorsohippocampal injections of a divalent competing ligand to intervene with the NMDAR expression profile 7 h after memory reactivation. We found that 7 h after retrieval control TAT-[GluN2Scr]2-injected mice showed enhanced synaptic GluN1 and GluN2A expression,
along with small but not significant increases in GluA2-containing AMPARs. In addition, we found a previously unobserved increase in GluN2B subunits at this time point. The TAT-[GluN2A]2 peptide used
to inhibit GluN2A-containing NMDAR anchoring failed to prevent GluN2A upregulation, but significantly reduced the amount of GluN2B subunits in the synaptic membrane. In addition, TAT-[GluN2A]2 did not show any effect on freezing behavior. The second objective of our study was to
investigate whether retrieval induced changes in membrane expression of NMDARs and AMPARs observed in the DH generalize to the VH. We reported differential expression patterns of ionotropic glutamate receptors in the VH compared to the DH. While we do observe subtle retrieval induced changes in AMPAR and NMDAR membrane expression in the DH, these receptors remain at baseline levels in the VH synaptic membranes.
NMDAR subunit regulation
Our results indicate that 7 h after retrieval GluN1, GluN2A and GluN2B subunits are upregulated in the DH. This observation points to an important role for the NMDAR in reconsolidation, a finding that is in line with previous research. For instance, a study in rats showed that systemic injections of the NMDAR antagonist MK-801 shortly before retrieval reduced freezing levels 24 h later29. Although in
that study any treatment effects on the retrieval itself cannot be ruled out and no retrieval-induced endogenous effects on NMDAR functioning were tested, these results also suggest the contribution of NMDARs in reconsolidation.
NMDARs are mainly composed of GluN1, GluN2A and GluN2B subunits. Although this implies that the obligatory GluN1 is the most abundant subunit in functional NMDARs, we find a relatively small increase (35.91%) compared to GluN2A (51.48%) and GluN2B (62.30%). This observation could suggest that NMDARs as present in naïve DH – or our NSR mice – could assemble from other subunit types such as GluN2C, GluN2D and GluN3A-D30. Indeed, GluN2C and GluN2D are known to localize at
postsynaptic membranes of rat hippocampal mossy fibres40. Perhaps the retrieval-induced regulation
as observed here not merely reflects the insertion of new receptors but also partly a subunit exchange. This is also suggested by the electrophysiology experiments as described earlier, since no amplitude differences were observed between groups (figure 1). As such, less abundant subunits could be replaced by the more dominant GluN2A and GluN2B. We did not analyze all known NMDAR subunits, but it is tempting to speculate about a possible retrieval-induced downregulation of these more infrequent assemblies.
Our data reveal that in addition to GluN1 and GluN2A (figure 1), GluN2B also shows increased retrieval-induced expression. These results might explain the discrepancy between the previously observed retrieval-induced NMDAR kinetics and those described in literature. The previous
subunits coincide with increases in, thus slower, NMDA-dependent current decays counteracts the widely accepted notion that GluN2A-containing NMDAR kinetics are faster than those of GluN2B-containing NMDARs. Our experiments reveal that GluN2B upregulation might have also contributed to this effect on mEPSCs as shown in figure 1. This is in accordance with a recent study in which drug-induced increases in GluN2A- and GluN2B-containing NMDARs coincided with slower NMDA decays, as measured in the brain31.
NMDARs and TAT-[GluN2A]
2Here we found no evidence that the mimetic peptide successfully suppressed upregulation of GluN2A in the synaptic membrane, for which there might be several reasons. First, the described competing ligand has been shown to work in vitro15, but has not yet been tested in vivo. Factors such as dosage,
concentration and timing are still experimental and can be crucial for the overall effect.
Second, the treatment could induce compensatory mechanisms that counteract the effect of the peptide. In hippocampal cultured neurons, the TAT-[GluN2A]2-induced reduction in
GluN2A-containing NMDARs is paralleled by a marked increase in GluN2B-GluN2A-containing NMDARs15. In these
experiments, the overall excitatory postsynaptic current (EPSC) amplitude remained unchanged, thus confirming the compensatory effect at the cellular level. However, we find no such compensatory increase in GluN2B-containing NMDARs. Rather, we find specifically a decreased expression of GluN2B, as if the peptide was specific for GluN2B instead of GluN2A. The reason for this discrepancy is unclear, but the resulting GluN2A/GluN2B ratio change can have major implications for synaptic functioning (see below). Furthermore, we rule out a possible compensatory regulation of the PSD-95 PDZ containing protein, since the immunoblot analyzes showed that the expression of this protein remained unaltered after peptide infusion (see figure 2). Interestingly, in cultured rat cerebellar granule cells (CGCs) PDZ-lacking GluN2A subunits, in contrast to PDZ-lacking GluN2B subunits, can still localize to the synapse, indicating that the binding of GluN2A to PDZ proteins such as PSD-95 is not necessarily required for synaptic localization32. Our peptide seemed to be non-specific for GluN2A and
GluN2B but only reduced GluN2B expression. This can possibly be explained by the observation that GluN2A – but not GluN2B – can still localize in the synapse without a PSD-95 interaction. Yet another mechanism to neutralize the actions of the competing ligand could be a reduction in GluN2A-containing NMDAR internalization. The machinery that regulates NMDAR endocytosis discriminates between GluN2A and GluN2B16,17 and could therefore dynamically regulate their internalization.
However, our methods did not include the analyses to study the proteins involved in these processes. Third, and in line with the point made above, if the anchoring mechanisms are indeed disrupted by our peptide this would allow for lateral mobility and surface diffusion of the affected subunit15.
Using single-particle tracking in cultured rat hippocampal neurons, one study showed that GluN2A-containing NMDARs are more stable at the synapse and show less membrane diffusion compared with GluN2B-containing NMDARs18. This result, along with the observation that GluN2A does not
require PSD-95 interactions to localize at the synapse in a stable manner, suggest that it might be impossible to prevent the incorporation of GluN2A-containing NMDARs with the here used ligand in the synaptic membrane in vivo.
Lastly, even if the receptors were laterally transported away from the active synaptic site, they could have remained within the synaptic membrane without contributing to synaptic communication.
Using the synaptic membrane fraction analyses might have failed to differentiate between the receptors that actively contribute to synaptic signaling and those that merely await their endocytosis or degradation. In addition, differences in presynaptic NMDAR expression could have contributed to the observed effects. Perhaps retrieval also induced changes in presynaptic NMDAR expression. Furthermore, our ligand might also have exerted an effect on these possible presynaptic NMDARs. It has been suggested that presynaptic NMDARs can mediate signaling properties such as release probability through various calcium-mediated signaling pathways (reviewed by Corlew et al., 2008)43.
Analyzing fractions that contain the isolated PSD rather than pre- and postsynaptic membranes as a whole could be one way to circumvent such issues.
Although it is unclear why GluN2B-containing NMDAR levels are low in the TAT-[GluN2A]2-treated
mice, the increased GluN2A/GluN2B ratio that resulted from this could have important implications for synaptic functioning. It has been proposed that the GluN2A/GluN2B ratio controls the threshold for the induction of LTD and LTP26. Following the developmental switch from mainly
GluN2B-containing NMDARs to GluN2A-GluN2B-containing NMDARs, increased GluN2A/GluN2B ratios raise the threshold to induce LTP (stronger stimulation is required). Thus, a decreased GluN2A/GluN2B ratio favors LTP, probably through interaction with the AMPAR and several kinases.
Apart from these theoretical speculations about the molecular consequences, it is important to note that the fear conditioning setup did not function properly during the training of the mice used for protein analysis. Overall freezing levels were relatively low, and several animals in the USR-2A group did not learn to associate the context with the foot shock at all, due to a poor foot shock delivery. It is therefore important to keep in mind that the competing ligand possibly exerted its effects (if any) on unconditioned animals that showed similar freezing behavior as NSR control mice (data not shown). In our experiment we did not study possible effects of the ligand on unconditioned mice, and therefore results should be interpreted carefully. However, it is clear that despite the low freezing levels, USR-2A mice showed similar expression levels for all analyzed subunits, except for GluN2B. It is unclear whether this is due to endogenous regulation, or an injection-induced increase of the membrane expression levels as shown in figure 3.
The problems described above prevent us from drawing reliable conclusions from our molecular experiments. It therefore remains unclear whether the variation between individual synaptic membrane samples results from the variability in learning behavior or the effects of the competing ligand (Supplementary figure S2).
Using the mimetic peptide, we aimed to investigate whether GluN2A expression relates to the expression of contextual fear. Even though the molecular data are difficult to interpret, our behavioral experiment (figure 4) clearly demonstrates that TAT-[GluN2A]2 did not alter the expression of fear. As
described above, it could be that the competing ligand failed to induce retrieval-induced GluN2A upregulation. However, in contrast with our hypothesis, the expression of fear as measured on day 3 might not be dependent on NMDAR expression directly. Perhaps the NMDAR regulation is merely a consequence of the preceding AMPAR plasticity with no behavioral effects. In addition, molecular changes do not inevitably alter the behavior itself, and it could be that small adaptations in synaptic functioning are not noticeable in our behavioral analyses.
Dorsal vs. ventral hippocampus
Lastly, we investigated whether retrieval induced changes in membrane expression of NMDARs and AMPARs observed in the DH generalize to the VH. Although our data reveals dissimilarities between the expression profile of the DH and VH to some degree, the lack of significant differences within the DH weakens the conclusions drawn from this experiment. Our lab has previously shown that retrieval mediates the synaptic enrichment of specific AMPAR and NMDAR subunits in the DH 7 h after reactivation5. The reason why the results here are less reproducible and convincing is therefore likely
to be of technical nature (e.g. membrane isolation or sample preparation). This is also stressed by the fact that we observe a relatively high variation in the NSR DH samples (figure 5c). Nevertheless, the dorsal versus ventral comparison suggests that the dorsal part – because of its stronger response to retrieval – is a more interesting target candidate for studying synaptic plasticity in contextual fear conditioning. Although not shown here, comparison of DH versus VH samples on the same blot indicated lower overall expression levels of AMPARs and NMDARs in the ventral compared to the dorsal synaptic membrane fractions. These observations are in line with previous work showing reduced baseline AMPA and NMDA receptor mRNA and protein levels in the VH compared to the DH38. However, we only analyzed expression levels at 7 h after retrieval, so it remains unclear what
specifically happens in the VH earlier on in the reconsolidation window. Indeed, the AMPAR exchange as observed in the amygdala occurs shortly (0-3 h) after retrieval22. Since the amygdala only receives
input from the VH36 (and reviewed by Pitkanen et al., 200637), but not the DH, it is comprehensible
that if the glutamate receptor waves also occur in the VH they more closely resemble the time pattern of the amygdala, rather than the DH. Future VH analyses on early reconsolidation (1-4 h after retrieval) should shed light on this matter.
AMPAR and NMDAR regulation during late reconsolidation
After reactivation of contextual fear, the biphasic AMPAR wave mediates the reorganization of hippocampus-dependent aversive memory15. At first, AMPAR subunits are internalized (measured at
1-4 h) and the synapses involved are depotentiated. Second, the GluA1 subunit is reinserted into the membrane (4 h after retrieval). This initial potentiation is reinforced by marked increases in GluA2-and GluA3-containing AMPARs in the synaptic membrane (measured at 7 h). The NMDAR subunit composition does not change during the first AMPAR wave, but parallels the AMPAR-mediated synaptic potentiation 7 h after retrieval. Here we show that this includes all three main NMDAR subunits: GluN1, GluN2A, and GluN2B. Since our manipulation failed to isolate the specific role of this regulation, it remains unclear if and how these receptors interact during reconsolidation of contextual fear. The first AMPAR wave is necessary for the second to occur, and blocking this early endocytosis also prevents the NMDAR wave. These observations indicate several possible pathways through which glutamate receptors can regulate synaptic strength during reconsolidation.
1) GluA1 trafficking triggers the reinsertion of GluA2 and GluA3 subunits. The NMDAR responds to this potentiation by increasing and redistributing its subunits in the membrane. Indeed, GluA1 insertion is activity-dependent, while GluA2 and GluA3 subunits often take in the place of existing receptors46. This suggests that during LTP GluA1 subunits are generally
2) GluA1 trafficking drives the incorporation and reorganization of the NMDAR. GluA2 and GluA3 follow and consequently boost synaptic potentiation.
3) GluA1 trafficking initiates downstream cascades that subsequently regulates the synaptic AMPA and NMDA receptor content.
How can GluA1 insertion trigger the reorganization of synaptic content? As stated before, GluA1-AMPARs are Ca2+-permeable, so they presumably contribute to Ca2+ flow into the cell. The rise in
intracellular calcium may trigger several molecular events needed for reconsolidation, including the integration of many membrane proteins and phosphorylation-dependent receptor insertion, mediated by Protein Kinase A (PKA) and/or Ca2+/calmodulin-dependent Kinase II (CaMKII)48. The
importance of CP-AMPARs for reconsolidation also appears from work on different brain areas. Indeed, blocking their activity in the lateral amygdala shortly after retrieval impairs contextual fear memory 24 h later, showing their crucial role in the reconsolidation phase22. This study presented
additional evidence that blocking AMPAR endocytosis in the amygdala shortly before retrieval does not alter freezing levels 24 h later. In contrast, the same manipulation in the DH increases contextual fear. In addition, a reduction in CP-AMPARs in the LA 2 h after retrieval of tone-conditioned fear was found35. This points to another difference in AMPAR plasticity between the LA and DH, since at the
same time point the DH shows a reduction in CI-AMPARs after contextual fear conditioning5. Taken
together, these results stress that despite their similar response to retrieval (as shown by the AMPAR plasticity) and strong involvement in fear memory reconsolidation, the plasticity processes in the amygdala and DH are functionally different and dependent on the behavioral paradigm used.
Our divalent competing ligand proved to be ineffective in preventing synaptic upregulation of GluN2A-containing NMDARs in vivo. However, due to technicalities during conditioning conclusions from our molecular experiments should be drawn carefully. Repeating the experiment with proper controls (NSR-Sal, USR-Sal, USR-Scr, USR-2A) would be necessary to determine the efficacy of TAT-[GluN2A]2. Our results suggest that TAT-[GluN2A]2 more strongly affects GluN2B localization, while
this does not alter the expression of fear. Perhaps GluN2B does not play a crucial role in this translation. However, whether GluN2A is more important remains an open question. One possible way to dissect the role of NMDAR activity after retrieval is timed pharmacological manipulation using subunit specific agonists and antagonists in vivo. This approach could reveal subunit-specific retrieval effects and their contribution to the dorsal hippocampus-dependent inhibitory constraint on fear memory expression.
The data presented here can be implemented in the model as described by Rao-Ruiz et al., (2011)5. The biphasic AMPAR wave is paralleled by the NMDAR regulation as measured at 7 h after
retrieval (Supplementary figure S3). Taken together, even though our manipulation did not yet prove to be successful, our results point to an important role for the NMDAR in dorsal hippocampus-dependent contextual learning, stressing the need for tools to study their mechanisms in vivo.
REFERENCES
1. Alberini, C.M. (2005) Mechanisms of memory stabilization: are consolidation and reconsolidation similar or distinct processes? TRENDS in Neurosciences, 28, 51-56.
2. Lee, S.-H., Choi, J.-H., Lee, N., Lee, H.-R., Kim, J.-I., Yu, N.-K., Choi, S.-L., Lee, S.-H., Kim, H. & Kaang, B.-K. (2008) Synaptic protein degradation underlies destabilization of retrieved fear memory. Science, 319, 1253-1256.
3. Nader, K., Schafe, G.E. & LeDoux, J.E. (2000) Fear memories require protein synthesis in the amygdala for reconsolidation after retrieval. Nature, 406, 722-726.
4. Debiec, J., LeDoux, J.E. & Nader, K. (2002) Cellular and systems reconsolidation in the hippocampus.
Nature, 36, 527-538.
5. Rao-Ruiz, P., Rotaru, D.C., van der Loo, R.J., Mansvelder, H.B., Stiedl, O., Smit, A.B. & Spijker, S. (2011) Retrieval-specific endocytosis of GluA2-AMPARs underlies adaptive reconsolidation of contextual fear.
Nature Neuroscience, 14, 1302-1308.
6. Suzuki, A., Josselyn, S.A., Frankland, P.W., Masushige, S., Silva, A.J. & Kida, S. (2004) Memory reconsolidation and extinction have distinct temporal and biochemical signatures. The Journal of
Neuroscience, 24, 4787-4795.
7. Malenka, R.C. & Bear, M.F. (2004) LTP and LTD: an embarrassment of riches. Neuron, 44, 5-21.
8. Colledge, M., Snyder, E.M., Crozier, R.A., Soderling, J.A., Jin, Y., Langeberg, L.K., Lu, H., Bear, M.F. & Scott, J.D. (2003) Ubiquitination regulates PSD-95 degradation and AMPA receptor surface expression. Neuron, 40, 595-607.
9. Carroll, R.C. & Zuking, R.S. (2002) NMDA-receptor trafficking and targeting: implications for synaptic transmission and plasticity. TRENDS in Neurosciences, 25, 571-577.
10. Lau, C.G. & Zukin, R.S. (2007) NMDA receptor trafficking in synaptic plasticity and neuropsychiatric disorders. Nature Reviews Neuroscience, 8, 413-426.
11. Lu, W.-Y., Man, H.-Y., Ju, W., Trimble, W.S., MacDonald, J.F. & Wang, Y.T. (2001) Activation of synaptic NMDA receptors induces membrane insertion of new AMPA receptors and LTP in cultured hippocampal neurons.
Neuron, 29, 243-254.
12. Dingledine, R., Borges, K., Bowie, D. & Traynelis, S.F. (1999) The glutamate receptor ion channels.
Pharmacological Reviews, 51, 7-61.
13. Stephenson, F.A., Cousins, S.L. & Kenny, A.V. (2008) Assembly and forward trafficking of NMDA receptors (review). Molecular Membrane Biology, 25, 311-320.
14. Cousins, S.L., Kenny, A.V. & Stephenson, F.A. (2009) Delineation of additional PSD-95 binding domains within NMDA receptor NR2 subunits reveals differences between NR2A/PSD-95 and NR2B/PSD-95 association. Neuroscience, 158, 89-95.
15. Bard, L., Sainlos, M., Bouchet, D., Cousins, S., Mikasova, L., Breillat, C., Stephenson, F.A., Imperiali, B., Choquet, D. & Groc, L. (2010) Dynamic and specific interaction between synaptic NR2-NMDA receptor and PDZ proteins. PNAS, 107, 19561-19566.
16. Lavezzari, G., McCallum, J., Dewey, C.M. & Roche, K.W. (2004) Subunit-specific regulation of NMDA receptor endocytosis. The Journal of Neuroscience, 24, 6383-6391.
17. Scott, D.B., Michailidis, I., Mu, Y., Logothetis, D. & Ehlers, M.D. (2004) Endocytosis and degradative sorting of NMDA receptors by conserved membrane-proximal signals. The Journal of Neuroscience, 24, 7096-7109. 18. Groc, L., Heine, M., Cousins, S.L., Stephenson, F.A., Lounis, B., Cognet, L. & Choquet, D. (2006) NMDA
19. Wenthold, R.J., Petralia, R.S., Il, J.B. & Niedzielski, A.S. (1996) Evidence for multiple AMPA receptor complexes in hippocampus CA1/CA2 neurons. The Journal of Neurocience,16, 1982-1989.
20. Derkach, V.A., Oh, M.C., Guire, E.S. & Soderling, T.R. (2007) Regulatory mechanisms of AMPA receptors in synaptic plasticity. Nature Review Neuroscience, 8, 101-113.
21. Malinow, R. & Malenka, R.C. (2002) AMPA receptor trafficking and synaptic plasticity. Annual Review
Neuroscience, 25, 103-126.
22. Hong, I., Kim, J., Kim, J., Lee, S., Ko, H.-G., Nader, K., Kaang, B.-K., Tsien, R.W. & Choi, S. (2013) AMPA receptor exchange underlies transient memory destabilization on retrieval. PNAS, 110, 8218-8223.
23. Fanselow, M.S. & Dong, H.-W. (2010) Are the dorsal and ventral hippocampus functionally distinct structures? Neuron, 65, 1-25.
24. Moser, E., Moster, M.-B. & Andersen, P. (1993) Spatial learning impairment parallels the magnitude of dorsal hippocampal lesions, but is hardly present following ventral lesions. The Journal of Neuroscience, 13, 3916-3925.
25. Paxinos, G. & Franklin, K.B.J. (1997) The Mouse Brain in Stereotaxic Coordinates (Academic, San Diego). 26. Yashiro, K. & Philpot, B.D. (2008) Regulation of NMDA receptor subunit expression and its implication for
LTD, LTP, and metaplasticity. Neuropharmacology, 55, 1081-1094.
27. Tsien, J.T., Huerta, P.T & Tonegawa, S. (1996) The essential role of hippocampal CA1 NMDA receptor-dependent synaptic plasticity in spatial memory. Cell, 87, 1327-1338.
28. Goosens, K.A. & Maren, S. (2004) NMDA receptors are essential for the acquisition, but not expression, of conditional fear and associative spike firing in the lateral amygdala. European Journal of Neuroscience, 20, 537-548.
29. Lee, J.L.C., Milton, A.L. & Everitt, B.J. (2006) Reconsolidation and extinction of conditioned fear: potentiation and inhibition. The Journal of Neuroscience, 26, 10051-10056.
30. Traynelis, S.F., Wollmuth, L.P., McBain, C.J., Menneti, F.S., Vance, K.M., Ogden, K., Hansen, K.B., Yuan, H., Myers, S.J. & Dingledine, R. (2010) Glutamate receptor ion channels: structure, regulation, and function.
Pharmacological Reviews, 62, 405-496.
31. Gipson, C.D., Reissner, K.J., Kupchik, Y.M., Smith, A.C.W., S, N., Hensley-Simon, M.E. & Kalivas, P.W. (2013) Reinstatement of nicotine seeking is mediated by glutamatergic plasticity. PNAS, 110, 9124-9129.
32. Prybylowski, K., Chang, K. & Sans, N. (2005) The synaptic localization of NR2B-containing NMDARs is controlled by interactions with PDZ proteins and AP-2. Neuron, 47, 845-857.
33. Matsuo, N., Reijmers, L. & Mayford, M. (2008) Spine-type – specific recruitment of newly synthesized AMPA receptors with learning. Science, 319, 1104-1107.
34. Rumpel, S., LeDoux, J., Zador, A. & Milanow, R. (2005) Postsynaptic receptor trafficking underlying a form of associative learning. Science, 308, 83-88.
35. Clem, R.L. & Huganir, R.L. (2010). Calcium-permeable AMPA receptor dynamics mediate fear memory erasure. Science, 330, 1108-1112.
36. Maren, S. & Fanselow, M.S. (1995) Synaptic plasticity in the basolateral amygdala induced by hippocampal formation stimulation in vivo. The Journal of Neuroscience, 15, 7548-7564.
37. Pitkanen, A., Pikkarainen, M., Nurminen, N. & Ylinen A. Reciprocal connections between the amygdala and the hippocampal formation, perirhinal cortex, and postrhinal cortext in rat. (2006) Annals of the New York
Academy of Sciences, 911, 369-391.
38. Pandis, C., Sotiriou, E., Kouvaras, E., Asprodini, E., Papatheodoropoulus, C. & Angelatou, F. (2006) Differential expression of NMDA and AMPA receptor subunits in rat dorsal and ventral hippocampus.
Neuroscience, 140, 163-175.
40. Berg, L.K., Larsson, M., Morland, C. & Gundersen, V. (2013) Pre- and postsynaptic localization of NMDA receptor subunits at hippocampal mossy fibre synapses. Neuroscience, 230, 139-150.
41. Vicini, S., Wang, J.F., Li, J.H., Zhu, W.J., Wang, Y.H., Luo, J.H., Wolfe, B.B. & Grayson, D.R. (1998) Functional and pharmacological differences between recombinant N-Methyl-d-aspartate receptors. Journal of
Neurophysiology, 79, 555-566.
42. Erreger, K., Dravid, S.M., Banke, T.G., Wyllie, D.J.A. & Traynelis, F.S. (2005) Subunit-specific gating controls rat NR1/NR2A and NR1/NR2B NMDA channel kinetics and synaptic signaling profiles. Journal of Physiology, 563, 345-358.
43. Corlew, R., Brasier, D.J., Feldman, D.E. & Philpot, D. (2008) Presynaptic NMDA receptors: newly appreciated roles in cortical synaptic function and plasticity. Neuroscientist, 14, 609-625.
44. Liu, X.-B., Murray, K.D. & Jones, E.G. (2004) Switching of NMDA receptor 2A and 2B subunits at thalamic and cortical synapses during early postnatal development. The Journal of Neuroscience, 24, 8885-8895. 45. Bannerman, D.M., Rawlins, J.N.P., McHugh, S.B., Deacon, R.M.J., Yee, B.K., Bast, T., Zhang, W.-N., Pothuizen,
H.H.J. & Feldon, J. (2004) Regional dissociation within the hippocampus – memory and anxiety.
Neuroscience and Biobehavioral reviews, 28, 273-283.
46. Shi, S.-H., Hayashi, Y., Esteban, J.A. & Malinow, R. (2001) Subunit-specific rules governing AMPA receptor trafficking to synapses in hippocampal pyramidal neurons. Cell, 105, 331-343.
47. Henley, J.M. & Wilkinson, K.A. (2013) AMPA receptor trafficking and the mechanisms underlying synaptic plasticity and cognitive aging. Dialogues in Clinical Neuroscience, 15, 11-27.
48. Derkach, V.A., Oh, M.C., Guire, E.S. & Soderling, T.S. (2007) Regulatory mechanisms of AMPA receptors in synaptic plasticity. Nature Review Neuroscience, 8, 101-113.
49. Kjelstrup, K.G., Tuvnes, F.A., Steffenach, H.-A., Murison, R., Moser, E.L. & Moser, M.-B. (2002) Reduced fear expression after lesions of the ventral hippocampus. PNAS, 10825-10830.
50. Yoon, T. & Otto, T. (2007) Differential contributions of dorsal vs. ventral hippocampus to auditory trace fear conditioning. Neurobiology of Learning and Memory, 87, 464-475.
51. Bast, T., Zhang, W.-N. & Feldon, J. (2001) The ventral hippocampus and fear conditioning in rats: Different anterograde amnesias of fear after tetrodotoxin inactivation and infusion of the GABAA agonist muscimol.
SUPPLEMENTARY FIGURES
Supplementary figure S1 Examples of blots with their corresponding gels. Figure shows representative blots with their matching input lanes used for normalization of total amount of protein for each sample. All shown blots are identical to those referred to in main text (main text figure number is indicated on the left); approximate molecular weight is indicated. NSR, no shock + retrieval; USR, shock + retrieval; USR-scr, shock + retrieval + dorsohippocampal injection of TAT-[GluN2Scr]2; USR-2A, shock + retrieval + dorsohippocampal injection of TAT-[GluN2A]2; VH, ventral hippocampus; DH, dorsal hippocampus.
Supplementary figure S2 Expression variation between individual synaptic membrane samples for all analyzed NMDAR subunits. Plots reveal that one sample of TAT-[GluN2A]2-injected animals shows a relatively high
GluN2A expression compared to the other samples (circled dot). This could indicate 1) that animals represented by this sample were the only ones that received a proper shock, which would mean that the ligand was unable to reduce the GluN2A expression, or 2) that these animals were the only ones where the ligand did not reduce the GluN2A expression (unsuccessful infusions), while it did exert an effect on the other samples. The technicalities described in the main text prevent us from drawing reliable conclusions from this.
Supplementary figure S3 Revised model for fear memory reorganization. In addition to the AMPAR biphasic wave as proposed elsewhere5, NMDAR regulation including GluN1, GluN2A and GluN2B subunits as observed
here might also contribute to the reconsolidation of contextual fear memory. Adapted and adjusted from Rao-Ruiz et al., (2011).
