All procedures were approved by the Institute of Laboratory Animal Resources of Seoul National University (Korea). Male Sprague-Dawley rats (4–5 weeks old) were maintained with free access to food and water under an inverted 12/12 h light/dark cycle (lights off at 09:00 h). Behavioral training was done during the dark portion of the light/dark cycle. For fear conditioning, rats were placed in a conditioning chamber and left undisturbed for 2 min. Then, a neutral tone (30 s, 2.8 kHz, 85 dB) coterminating with an electrical foot shock (1.0 mA, 1 s) was presented three times at an average interval of 100 s. After fear conditioning, rats were returned to their home cages until preparation of brain slices. Rats in naïve groups stayed in their home cages until brain slices were prepared.

Sprague-Dawley rats (4–5 weeks old) were anesthetized with isoflurane and decapitated. Whole brains were isolated and placed in an ice-cold modified aCSF solution containing (in mM) 175 sucrose, 20 NaCl, 3.5 KCl, 1.25 NaH2PO4, 26 NaHCO3, 1.3 MgCl2, 11 D-(+)-glucose, and was gassed with 95% O₂/5% CO₂. Coronal slices (300 μm) including the LA were cut using a vibrating blade microtome (VT1200S, Leica Biosystems, Germany) and incubated in normal aCSF containing (in mM) 120 NaCl, 3.5 KCl, 1.25 NaH2PO4, 26 NaHCO3, 1.3 MgCl2, 2 CaCl2, 11 D-(+)-glucose, and was continuously bubbled at room temperature with 95% O₂/5% CO₂. Just before a given slice was transferred to the recording chamber, the cortex overlying the LA was cut away with a scalpel, so the addition of picrotoxin (100 μM; Abcam Plc., UK) would block cortical epileptic burst discharges from invading the LA.

We chose brain slices containing a well-isolated, sharply defined trunk (containing thalamic afferents) crossing the dorsolateral division of the LA where the somatosensory and auditory inputs converge. The sizes of the LA and the central amygdala were relatively constant in the utilized slices; when multiple trunks were observed, we used the closest trunk to the central nucleus of the amygdala. Unless otherwise noted, the thalamic afferents were stimulated at a frequency of 0.067 Hz using a concentric bipolar electrode (CBAEC75; FHC Inc., USA). The stimulation electrode was placed at the midpoint of the trunk between the internal capsule and the medial boundary of the LA. The regions and cells of interest for all recordings were located beneath the midpoint of the trunk spanning the LA horizontally.

Whole-cell recordings were made using an Axopatch 200A amplifier or Multiclamp 700A (Molecular Devices, Sunnyvale, CA, USA). For the whole-cell voltage-clamp recordings, the recordings were obtained using a Cs-based internal solution containing (in mM) 100 Cs-gluconate, 0.6 EGTA, 10 HEPES, 5 NaCl, 20 TEA, 4 Mg-ATP, 0.3 Na-GTP and 3 QX314, with the pH adjusted to 7.2 with CsOH and the osmolarity adjusted to approximately 297 mmol/kg with sucrose. The cells used were classified as principal neurons based on the pyramidal shape of their somata. We included picrotoxin (100 μM) in our recording solution to isolate excitatory synaptic transmission and block feed-forward GABAergic inputs to the principal neurons in the LA. The pipette resistances ranged from 3.5 to 4.5 Mohm. IR-DIC-enhanced visual guidance was used to select neurons that were 3–4 cell layers below the surface of the 300-μm-thick slices, which were held at 32 ± 1°C. The neurons were voltage-clamped at –70 mV except during paired pulse low-frequency stimulation (ppLFS-pairing), and the various solutions were delivered to the slices via gravity-driven superfusion at a flow rate of 1.4~1.5 ml/min. The pipette series resistance was monitored throughout each experiment, and the data were discarded if it changed by >20%. Whole-cell currents were filtered at 1 kHz, digitized at up to 20 kHz, and stored on a microcomputer (Clampex 9 software, Molecular Devices, Sunnyvale, CA, USA). EPSCs were monitored following stimulation at 0.067 Hz. One or two neurons were recorded per animal (a single neuron per slice). In the ppLFS-pairing protocol, stimulation for 3 min at 3 Hz was performed using paired pulses (50 ms interpulse interval) while the neuron was clamped at –50 mV, as described previously (Clem and Huganir, 2013). Blockade of CP-AMPARs and mGluR1 was performed using NASPM (50 μM; Sigma-Aldrich) and LY367385 [(S)- (+)-α-amino-4-carboxy-2-methylbenzeneacetic acid] (100 μM; Tocris Bioscience). All EPSC amplitudes were normalized to an average of the baseline responses for the first 10 min and were expressed as percentages of the average baseline response. The percent inhibition by drugs (or vehicle) indicated difference in the average percentage of the responses between before and after drug (or vehicle) treatment. Therefore, the percent inhibition by NASPM in this study could be used to compare the amount of synaptic CP-AMPARs between before and after induction of LTD or depotentiation. The periods used to calculate these average responses were the 5 min immediately preceding drug treatment and the final 5 min after drug treatment, respectively. To prevent bias, we performed experiments in a blinded fashion. For improved visualization, the running averages of four data points were applied to the time-lapse experiments.

Between-group comparisons of the data were performed using either an unpaired t-test or one-way ANOVA with subsequent Newman-Keuls post-hoc comparison. A paired t-test was used to determine whether the post-treatment responses differed significantly from the baseline responses (Figures 1E, 3D). A p-value < 0.05 was considered to be statistically significant. The data from each neuron/slice were treated as independent samples. In all experiments using behaviorally trained rats, the data included samples from three or more animals.

We first characterized the synaptic expression of CP-AMPARs according to the sensitivity of T-LA synaptic responses to NASPM, a CP-AMPAR antagonist, at various time points (12 and 48 h) after fear conditioning in rats. We measured AMPAR-mediated EPSCs at T-LA synapses in the presence of D-AP5 (50 μM) via whole-cell voltage-clamp recording in acute brain slices prepared from conditioned (or naïve) rats. A negligible level of CP-AMPARs was found in slices prepared from naïve rats, but a significant level of CP-AMPARs was detected 12 h after conditioning. The enhanced level of CP-AMPARs returned to baseline in slices prepared 48 h after conditioning (F_(2,13) = 4.787, p = 0.0319, one-way ANOVA; naïve, 4.70 ± 4.20%, n = 5; 12 h, 15.74 ± 1.6%, n = 4; 48 h, 4.52 ± 1.2%, n = 5; p < 0.05 for the 12 h group vs. the other groups, Newman-Keuls post-hoc test; Figures 1A–D), which was consistent with our previous study (Hong et al., 2013). Because NASPM has also been shown to inhibit kainate receptors (Koike et al., 1997; Cho et al., 2012), we examined whether UBP302 (10 μM), a specific antagonist of kainate receptors, blocked EPSCs at T-LA synapses when CP-AMPAR levels were elevated (12 h after conditioning). UBP302 displayed no significant effects on EPSCs at T-LA synapses in the presence of D-AP5 (50 μM) in slices prepared 12 h after fear conditioning (94.61 ± 4.10%, n = 5; p > 0.05, paired t-test; Figure 1E). In addition, we determined whether fear conditioning-induced synaptic potentiation was similar at those two time points (12 h vs. 48 h) after fear conditioning. There was no significant difference in the excitatory synaptic efficacy at T-LA synapses between the two groups (12 h after conditioning, 11.25 ± 1.57 pA/μA, n = 7; 48 h after conditioning, 11.84 ± 1.36 pA/μA, n = 9; p > 0.05, unpaired t-test; Figure 1F). Taken together, these results suggest that CP-AMPARs are transiently inserted into T-LA synapses after fear conditioning.

We have previously shown that DHPG, an agonist of group I mGluRs, induces the depotentiation of fear conditioning-induced synaptic potentiation at T-LA synapses; that is, DHPG produces synaptic depression in slices prepared after fear conditioning but not in slices prepared from naïve rats or other controls (Kim et al., 2007; Hong et al., 2011). To examine whether CP-AMPARs are removed after induction of depotentiation, we monitored the changes in the sensitivity to NASPM after fear conditioning. Application of DHPG (100 μM, 10 min) successfully induced synaptic depression in the presence of D-AP5 (50 μM) in slices prepared 12 or 48 h after fear conditioning. After DHPG-induced depression had been stabilized, NASPM (or vehicle) was applied. NASPM treatment inhibited EPSCs significantly more than vehicle treatment when the CP-AMPAR levels were elevated (12 h after conditioning), but it did not exert a significant effect on EPSCs when CP-AMPARs were minimally expressed (48 h after conditioning) (F_(2,16) = 29.74, p < 0.001, one-way ANOVA; vehicle group-12 h after conditioning, −0.42 ± 1.04%, n = 5; NASPM group-12 h after conditioning, 18.59 ± 2.15%, n = 6; NASPM group-48 h after conditioning, 1.78 ± 2.11%, n = 6; p < 0.0001 for NASPM group-12 h after conditioning vs. the other groups, Newman-Keuls post-hoc test; Figures 2A–D). Thus, when CP-AMPARs are expressed at synapses, a significant proportion of CP-AMPARs appears to be retained even after the induction of depotentiation.

Previous studies have shown that ppLFS also induces depotentiation or LTD (Kim et al., 2007; Hong et al., 2009). We performed a protocol of ppLFS-pairing which was used by Clem and Huganir (2013). Unlike their findings that ppLFS-pairing induced LTD regardless of fear conditioning (either before or after conditioning), ppLFS-pairing produced no significant depression in slices from naïve rats (96.84 ± 7.33%, n = 5; p > 0.05, paired t-test; Figure 3A). Alternatively, ppLFS-pairing produced synaptic depression in slices prepared 12 h after conditioning, confirming the previously reported depotentiation of fear conditioning-induced synaptic potentiation (Figures 3B,C; Kim et al., 2007). After ppLFS-pairing-induced depression had been stabilized, NASPM (or vehicle) was applied. NASPM treatment inhibited EPSCs relative to vehicle treatment (vehicle, 3.33 ± 0.74%, n = 4; NASPM, 17.45 ± 2.24%, n = 7; p = 0.0013, unpaired t-test; Figures 3B–D). Therefore, similar to the results using DHPG, pre-existing CP-AMPARs appear to be largely retained after depotentiation. Because this particular result contradicts with the previous findings by Clem and Huganir (2010, 2013), it was necessary to confirm that ppLFS-induced depression in the present study shared the same induction requirement with that in the two previous studies (i.e., mGluR1-dependency). Indeed, application of the mGluR1 antagonist, LY367385, completely blocked ppLFS-pairing-induced depression (vehicle group, 69.89 ± 8.54%, n = 4; LY367385 group, 97.69 ± 5.89%, n = 7; p = 0.0223, unpaired t-test, Figure 3E).

Comparing the results from Figures 1–3 revealed that regardless of prior depotentiation, the primary factor that contributes to NASPM sensitivity at LA synapses is the duration after fear conditioning. NASPM-induced inhibition was minimal in naïve or 48 h, slices whereas in 12 h slices, NASPM-induced inhibition was prominent both before and after depotentiation, regardless of the depotentiation protocol (NASPM at baseline, NASPM after DHPG treatment and NASPM after pp-LFS) when the CP-AMPAR levels were elevated (12 h after conditioning) (NASPM at baseline (naïve rats), 4.70 ± 4.20%, n = 5; NASPM at baseline (48 h after conditioning), 4.52 ± 1.24%, n = 5; NASPM at baseline (12 h after conditioning), 15.74 ± 1.61%, n = 4; NASPM after DHPG treatment (12 h after conditioning), 18.59 ± 2.15% n = 6; NASPM after ppLFS (12 h after conditioning), 17.45 ± 2.24%, n = 7; F_(4,26) = 7.699, p = 0.0005, one-way ANOVA; p > 0.05 between the three groups prepared 12 h after conditioning, p < 0.05 for these three groups vs. the other two groups, Newman-Keuls post-hoc test; Figure 3F). Thus, our data suggest that fear conditioning produces transient insertion of CP-AMPARs into T-LA synapses, but these CP-AMPARs do not appear to be removed by depotentiation. The observation that depotentiation can be induced regardless of the expression of CP-AMPARs also suggests that depotentiation does not require the pre-existence of CP-AMPARs at synapses.