The Relation Between Long-Term Synaptic Plasticity at Glutamatergic Synapses in the Amygdala and Fear Learning in Adult Heterozygous BDNF-Knockout Mice

The Relation Between Long-Term Synaptic Plasticity at Glutamatergic Synapses in the Amygdala and... Abstract Brain-derived neurotrophic factor (BDNF) heterozygous knockout mice (BDNF+/− mice) show fear learning deficits from 3 months of age onwards. Here, we addressed the question how this learning deficit correlates with altered long-term potentiation (LTP) in the cortical synaptic input to the lateral amygdala (LA) and at downstream intra-amygdala synapses in BDNF+/- mice. Our results reveal that the fear learning deficit in BDNF+/− mice was not paralleled by a loss of LTP, neither at cortical inputs to the LA nor at downstream intra-amygdala glutamatergic synapses. As we did observe early fear memory (30 min after training) in BDNF+/− mice while long-term memory (24 h post-training) was absent, the stable LTP in cortico-LA and downstream synapses is in line with the intact acquisition of fear memories. Ex vivo recordings in acute slices of fear-conditioned wildtype (WT) mice revealed that fear learning induces long-lasting changes at cortico-LA synapses that occluded generation of LTP 4 and 24 h after training. Overall, our data show that the intact LTP in the tested amygdala circuits is consistent with intact acquisition of fear memories in both WT and BDNF+/− mice. In addition, the lack of learning-induced long-term changes at cortico-LA synapses in BDNF+/− mice parallels the observed deficit in fear memory consolidation. amygdala, BDNF, fear conditioning, field recording, long-term potentiation Introduction The neurotrophin brain-derived neurotrophic factor (BDNF) and its corresponding receptor TrkB (tropomyosin-related kinase B) are known to be essential for learning and memory as well as for regulation of synaptic strength and synaptic plasticity (Gottmann et al. 2009; Cowansage et al. 2010; Yoshii and Constantine-Paton 2010; Park and Poo 2013; Edelmann et al. 2014). Several recent studies demonstrated a critical role for BDNF signaling in amygdala-dependent fear learning. Thus, overexpression of a nonfunctional, truncated TrkB receptor (tTrkB) in the amygdala led to impaired fear learning (Rattiner et al. 2005). Furthermore, acute inhibition of BDNF/TrkB signaling blocked the acquisition and prevented the formation of long-term fear memories, depending on the time of application of TrkB inhibitors (Ou and Gean 2006; Ou et al. 2010). Fear learning was also impaired by a point mutation in the Y816F or Y515F phosphorylation site of the TrkB receptor, respectively (Musumeci et al. 2009). Fear learning relies on synaptic plasticity at thalamic and cortical inputs to the dorsal part of the lateral amygdala (LA; Sigurdsson et al. 2007; Ehrlich et al. 2009; Pape and Paré 2010). In addition, recent data demonstrated that synaptic plasticity in the LA depends on intact BDNF/TrkB signaling (Li et al. 2011; Daftary et al. 2012; Meis et al. 2012). In this respect, long-term potentiation (LTP) at thalamic afferents to the LA is prevented by acute inhibition of BDNF/TrkB signaling (Meis et al. 2012). In addition, chronic BDNF deficiency in heterozygous BDNF-knockout mice (BDNF+/− mice) abolished LTP at thalamo-LA synapses already at an age of 1 month. In contrast, LTP at cortico-LA synapses was unaffected (Meis et al. 2012). Interestingly, we could recently demonstrate that these mice exhibit deficits in fear learning when they are 3 months of age or older (Endres and Lessmann 2012). As LTP at thalamo-LA afferents is already impaired in 1-month-old animals while fear conditioning is still functional up to 2 months, intact fear learning at 2 months might be enabled by synaptic plasticity at cortico-LA synapses. However, LTP at this synapse may become incapable to contribute to fear memory storage at older ages. This idea is supported by several studies demonstrating postpubertal changes in synaptic properties of thalamo-LA and cortico-LA synapses as well as in network morphology (e.g., Cunningham et al. 2002, 2008; Pan et al. 2009a; Gambino et al. 2010). To test our hypothesis, we performed in vitro field potential recordings in amygdala slices of adult (≥3 months) BDNF+/− mice and WT littermates. Surprisingly, LTP at cortico-LA synapses, as well as at different intra-amygdala glutamatergic synapses, was unaffected in BDNF+/− mice. Therefore, we analyzed in a second series of experiments whether cued fear learning alters plasticity of cortico-LA synapses. As it has been shown for several brain regions—including the amygdala—learning results in occlusion of LTP at synapses involved in memory formation (see e.g., Rioult-Pedotti et al. 2000; Tsvetkov et al. 2002; Whitlock et al. 2006). Therefore, we tested for occlusion of LTP in slices of previously fear-conditioned mice and indeed observed occlusion of LTP in WT mice 4 and 24 h after fear conditioning. This indicates fear learning-related long-term changes at these synapses. However, this learning-induced occlusion of LTP was absent in BDNF+/− mice at both time points (4 and 24 h). As we observed on the one hand intact LTP and on the other hand a lack of learning-related long-term changes at cortico-LA synapses in BDNF+/− mice, we performed a detailed analysis of the early fear memory (i.e., 0.5–6 h after fear conditioning) in BDNF+/− and WT mice. Here, we observed robust early fear memory also in BDNF+/− mice that gradually declined in precision over time. These results suggest that acquisition of fear memory is still intact in adult BDNF+/− mice, but due to a defective fear memory consolidation no stable long-term fear memory is established in these mice. Overall, our results suggest that cued fear learning results in long-lasting changes at cortico-LA synapses in WT mice. These learning-induced changes in plasticity are absent in BDNF+/− mice 4 h after fear conditioning, probably resulting in the observed deficit in fear memory consolidation. Materials and Methods Animals We used 3- to 5-month-old male BDNF+/− mice (Korte et al. 1995), which were outbred on a C57BL/6 J (Charles River) genetic background. In addition, 2-month-old animals were utilized in 2 sets of experiments as stated in the Results section. Wildtype (WT) littermates served as controls. The animals were housed in groups of 4 animals and had free access to food and water. All experiments were carried out in accordance with the European Committees Council Directive (86/609/EEC) and were approved by the local animal care committee (Landesverwaltungsamt Sachsen Anhalt, IPHY/G/01-872/08 & IPHY/G/01–1191/13). Slice Preparation Standard procedures were used to prepare coronal slices from male BDNF+/− mice or their WT littermates, respectively. Mice at postnatal days P86–P140 were deeply anesthetized by inhalation of 4% isofluran (Kulisch et al. 2011) and killed by decapitation. A block of tissue containing the amygdala was rapidly removed and placed in standard artificial cerebrospinal fluid (ACSF) containing (in mM): NaCl, 125; KCl, 2.5; NaH2PO4, 0.8; NaHCO3, 25; MgCl2, 1; CaCl2, 2; glucose, 10; bubbled with 95% O2/5% CO2. Coronal slices (400 μm thick) were prepared on a vibratome (Model 1000, The Vibratome Company), and were incubated in ACSF in an interface chamber for at least 2 h before recording. Field Potential Recordings Field potential recordings were performed in an interface chamber at 32 ± 1 °C (Matthies et al. 1997). Recording pipettes were pulled from borosilicate glass (GC150TF-10, Clark Electromedical Instruments), filled with ACSF (3–4 MΩ) and positioned in the LA (see insets in respective figures). A concentric bipolar electrode (FHC Inc.) was placed on the surface of the slice above the external capsule when stimulating cortico-LA synapses. For assessment of LTP at intra-amygdala afferents, the stimulation electrode was positioned within the basal amygdala, and the recording pipette was placed in the basal or central medial amygdala (CeA). For ex vivo recordings in fear-conditioned or pseudoconditioned mice (see below), animals were sacrificed 2 or 24 h after fear conditioning, and recordings started 4–8 h or 26–30 h after training, respectively. Field potentials were evoked by stimuli of 100 µs duration delivered by a stimulus isolator (Isoflex, AMPI) at 0.016 Hz. Stimulus intensity was adjusted to evoke responses of halfmaximal amplitude. Signals were amplified by a DAM-80 amplifier (WPI) and digitized with a CED 1401plus interface (Cambridge Electronic Design), controlled by a custom-made software (Reymann and Frey, LIN Magdeburg). Since in the amygdala the recorded field potentials consist of a summation of excitatory postsynaptic potentials (EPSPs) and synchronized action potentials, the analysis of the field potential amplitudes, instead of slopes, is more reliable. In addition, EPSP amplitudes are less sensitive to variability and noise than the slope (compare discussion in Drephal et al. 2006). Therefore, in keeping with many other LTP studies in the amygdala (see e.g., Watanabe et al. 1995; McKernan and Shinnick-Gallagher 1997; Rogan et al. 1997; Doyère et al. 2003) we also analyzed EPSP amplitudes in our recordings. Signal amplitude was measured as the sum of 1) the difference between onset and peak of the negative voltage deflection and 2) the difference of the peak of the negative voltage deflection and the succeeding positive peak, divided by 2 (Drephal et al. 2006; Kulisch et al. 2011). To induce LTP, we tested different types of stimulation patterns because, based on previous literature, specific induction paradigms may rely on distinct signaling pathways (Huang et al. 2000; Bauer et al. 2002). A high-frequency stimulation (HFS) pattern was composed of 3 trains of 100 stimuli at 100 Hz separated by 30 s. A second type of HFS consisted of 4 trains of 100 stimuli at 100 Hz separated by 5 min. Theta-burst stimulation (TBS) comprised 2 trains of 4 stimuli at 100 Hz, repeated 10 times at 5 Hz, separated by 20s. Stimulation protocols were executed at time point zero. For comparison of LTP between genotypes, LTP was quantified by normalizing and averaging field potentials during the last 5 min of experiments (i.e., 55–60 min or 115–120 min after LTP induction) relative to 30 min baseline. For statistical analysis of successful LTP induction, we compared averaged field potential amplitudes 5 min before LTP induction with the respective amplitudes during the last 5 min of recordings. Patch-Clamp Recordings Whole-cell patch-clamp recordings of IPSCs were done as described previously (Meis et al. 2008). Briefly, single slices were transferred to a submerged chamber. Recordings were made using a patch-clamp amplifier (EPC-9, Heka). Patch pipettes were pulled from borosilicate glass (GC150TF-10, Clark Electromedical Instruments) to resistances of 2–3 MΩ, and filled with (in mM): Csgluconate, 107; CsCl, 13; MgCl2, 1; CaCl2, 0.07; EGTA, 11; HEPES, 10; MgATP, 3, NaGTP, 0.5 (pH 7.2 with KOH). A liquid junction potential of 10 mV of the pipette solution was corrected for. After obtaining the whole-cell configuration, neurons were held at 0 mV. Drugs All chemicals were obtained from Sigma, except for K252a (Alomone). Assessment of Fear Learning For fear conditioning, we used an automated fear conditioning setup (TSE-Systems). The animals were placed in a cubic box (23 × 23 cm2), located in a sound attenuating chamber. The floor of the test box consisted of a grid floor, by which the unconditioned stimulus (US, 1 s, 0.7 mA, scrambled foot shock) was delivered. The conditioned stimulus (CS, 30 s, 8 kHz sine tone, 70 dB sound pressure level [SPL]) was presented by a loudspeaker located at the ceiling of the test box. An array of infrared light beams around the arena enabled tracking the activity of the animals. To provide different contexts between fear conditioning and fear memory retrieval tests, the color of the test boxes (black or transparent) as well as the cleaning agents (70% ethanol or Deskosept; Dr Schumacher GmbH) were randomly changed to compose distinct contextual environments. For fear conditioning training, the animals were permitted to explore the test chamber for the first 2 min to become habituated. Then, the CS and US were presented 3 times with random interstimulus intervals (90–240 s) in a paired manner, that is, the CS (30 s) coterminated with the US (1 s). This protocol is identical to the one used in our previous studies, resulting in impaired fear learning in BDNF+/− at 3 months of age and beyond (Endres and Lessmann 2012). As a control for the occlusion experiments, we presented the CS and US in a random unpaired manner. To assess the cued fear memory of the animals, we exposed the animals 5 times to the CS in a different context (as described above). Besides analyzing freezing behavior during the CS periods (5× 30 s), we also quantified the average freezing during the 30-s periods before each of the 5 CS presentations (pre-CS freezing). In order to get more detailed insights into the characteristics of the learning deficit, we performed the fear memory test in different groups at different time points, that is, 0.5, 2, 4, 6, or 24 h, after fear conditioning. To quantify the precision of early fear memory, we performed a tone frequency-dependent discriminative fear learning task. To this aim, one tone (CS−, 30 s, 2 kHz sine tone, 75 dB SPL) was presented in the absence of a coterminating foot shock, while another tone (CS+, 30 s, 8 kHz sine tone, 75 dB SPL) was followed by a foot shock (1 s, 0.7 mA). Both CS+ and CS− were presented 3 times at random sequence during the conditioning session with random interstimulus intervals. Three to four hours later, we tested the fear memory by presenting CS+ and CS− in a random sequence in a novel context (identical to the classical fear conditioning procedure). Data Analysis Nonparametric data were analyzed by Wilcoxon signed rank test, or Mann–Whitney U-test, by using Graph Pad Prism software. Normally distributed data were analyzed by performing an analysis of variance (ANOVA, JMP 8, SAS Institute), followed by post hoc Tukey comparisons. All electrophysiological recordings were normalized to average baseline amplitudes and analyzed with Origin 8.0. (OriginLab Corporation). To statistically analyze the fear conditioning experiments, repeated measure ANOVAs were performed using “phase of the experiment” (habituation, pre-CS, CS, or pre-CS+, pre-CS−, CS+, CS−) as within-subject factor and “genotype” or “time of testing” as between-subject factors. All data are presented as mean ± standard error of the mean (SEM). Differences were considered statistically significant at P < 0.05. For electrophysiological recordings, the lowercase n (“n”) indicates the number of slices tested, while the capital n (“N”) refers to the number of animals from which these slices were obtained from. Results HFS Induced LTP in Cortico-LA Afferents To correlate LTP at LA inputs with fear learning, we focused in this study on LTP at cortico-LA synapses in adult mice (≥3 months). LTP was induced by an HFS pattern composed of 3 trains of 100 stimuli at 100 Hz, separated by 30 s. In slices from adult WT mice, the average field potential amplitude was significantly increased 60 min after the induction when compared with pretetanus values (125.5 ± 9.5%, n = 12 slices from N = 7 mice; P = 0.009; Fig. 1A). A similar significant increase in field potential amplitudes in response to LTP induction was observed in BDNF+/− mice of analogous age (114.9 ± 7.6%, n = 14 slices/N = 10 mice, P = 0.02). This potentiation in BDNF+/− mice was not significantly different from WT slices (P = 0.2472, Fig. 1A). Thus, HFS-induced LTP at this input structure is not hampered in adult BDNF+/− mice. To verify that the LTP paradigm we used was nevertheless dependent on BDNF signaling, we performed a control experiment in WT mice (≥ 3 months) in the continuous presence of the tyrosine kinase inhibitor K252a (100 nM, preincubation time ≥2 h). In WT slices exposed to the same concentration of DMSO as used to dissolve K252a (0.1%), the average field potential amplitude was significantly increased 60 min after LTP induction (160.4 ± 9.7%, n = 8 slices/N = 5 mice; P = 0.0078; Fig. 1B). In contrast, LTP was not induced in the presence of K252a (97.1 ± 2.9%, n = 9 slices/N = 6 mice; P = 0.2031). The difference between the 2 data sets was highly significant (P < 0.0001, Fig. 1B). Likewise, LTP was also prevented in the presence of K252a in BDNF+/− mice (P = 0.0043, Supplementary Fig. 1). These results suggest that the LTP induced by our tetanic stimulation is dependent on BDNF/TrkB signaling. However, BDNF levels in BDNF+/− mice do not seem to be reduced to a level that prevents LTP expression. Figure 1. View largeDownload slide BDNF dependency of HFS-LTP at cortico-LA synapses in adult BDNF+/− mice. LTP was induced by 3 trains of 100 stimuli at 100 Hz separated by 30 s intervals. (A) Time course of averaged evoked field potentials in response to stimulation of cortical afferents in all slices recorded from WT and BDNF+/− mice. In both genotypes, LTP could be reliably induced. Insets at the right depict averaged field potential recording traces 5 min before LTP induction and during the last 5 min of recordings for WT and BDNF+/− mice, for 2 representative individual slices. WT: n = 12 slices from N = 7 animals, BDNF+/−: n = 14 slices from N = 10 animals, ns: not significantly different. (B) Time course of averaged evoked field potentials in response to stimulation of cortico-LA afferents in slices obtained from WT animals with and without application of the tyrosine kinase inhibitor K252a, labeled as white and black data points, respectively. LTP was abolished by application of 100 nM K252a (preincubation ≥2 h). Insets at the right depict averaged field potential recording traces 5 min before LTP induction and during the last 5 min of recordings, for 2 representative individual slices. Control, 0.1% DMSO: n = 8 slices from N = 5 animals, K252a: n = 9 slices from N = 6 animals, ***: significant, P < 0.0001. Insets at the left depict positions of recording and stimulation electrodes (modified from Paxinos and Franklin 2001). Figure 1. View largeDownload slide BDNF dependency of HFS-LTP at cortico-LA synapses in adult BDNF+/− mice. LTP was induced by 3 trains of 100 stimuli at 100 Hz separated by 30 s intervals. (A) Time course of averaged evoked field potentials in response to stimulation of cortical afferents in all slices recorded from WT and BDNF+/− mice. In both genotypes, LTP could be reliably induced. Insets at the right depict averaged field potential recording traces 5 min before LTP induction and during the last 5 min of recordings for WT and BDNF+/− mice, for 2 representative individual slices. WT: n = 12 slices from N = 7 animals, BDNF+/−: n = 14 slices from N = 10 animals, ns: not significantly different. (B) Time course of averaged evoked field potentials in response to stimulation of cortico-LA afferents in slices obtained from WT animals with and without application of the tyrosine kinase inhibitor K252a, labeled as white and black data points, respectively. LTP was abolished by application of 100 nM K252a (preincubation ≥2 h). Insets at the right depict averaged field potential recording traces 5 min before LTP induction and during the last 5 min of recordings, for 2 representative individual slices. Control, 0.1% DMSO: n = 8 slices from N = 5 animals, K252a: n = 9 slices from N = 6 animals, ***: significant, P < 0.0001. Insets at the left depict positions of recording and stimulation electrodes (modified from Paxinos and Franklin 2001). TBS Induced LTP in Cortico-LA Synapses Notably, a requirement for BDNF in LTP induction can depend on the specific LTP paradigm in use (Kang et al. 1997; Zakharenko et al. 2003; Abidin et al. 2006; Edelmann et al. 2015). Thus, to determine whether other types of LTP at cortico-LA synapses were affected in BDNF+/− mice, we performed an additional series of LTP experiments using TBS. The TBS protocol consisted of 2 trains (separated by 20 s) of 4 stimuli at 100 Hz, repeated 10 times at 5 Hz. The magnitude of LTP induced by this protocol in 2-month-old animals was indistinguishable between WT and BDNF+/− mice (P = 0.1810; WT: 113.6 ± 3.8%, n = 9 slices/N = 7 mice; BDNF+/−: 123.9 ± 9.4%, n = 6 slices/N = 5 mice). However, at older ages (≥3 months), this paradigm did not induce LTP irrespective of genotype (WT: 107.6 ± 5.6%, n = 9/N = 8 mice, P = 0.1289; BDNF+/−: 107.4 ± 9.7%, n = 9/N = 7 mice, P = 0.5703, Supplementary Fig. 2). Thus, in mice older than 3 months, LTP at cortico-LA synapses was induced more efficiently by trains of 100 Hz stimulation than by TBS. HFS Induced LTP in Cortico-LA Afferents in the Presence of Gabazine We observed unimpaired LTP at glutamatergic cortico-LA synapses in BDNF+/− mice when GABAergic inhibition was intact (Fig. 1A). However, chronic reduction of BDNF in BDNF+/− mice is known to reduce GABAergic inhibition in different brain areas (Kohara et al. 2007; Abidin et al. 2008; Laudes et al. 2012). Therefore, a potential impairment of LTP at glutamatergic cortical afferents to the LA could be compensated by reduced GABAergic inhibition in BDNF+/− mice. To test this hypothesis, recordings were performed in the presence of the specific GABAA receptor antagonist gabazine. As blockade of GABAergic inhibition was described to strongly enhance excitability in amygdala slices (Gean and Shinnick-Gallagher 1987; Isoardi et al. 2004; Huang and Kandel 2007), we applied gabazine at a low nonsaturating concentration (0.1 µM), which effectively prevented epileptiform activity. As assessed by patch-clamp recordings, this concentration reduced evoked inhibitory postsynaptic current amplitudes to 51.1 ± 3.1% of the control value before drug addition (n = 6, see Supplementary Fig. 3). In the presence of gabazine, LTP elicited by HFS amounted to 133.6 ± 12.7% (n = 8/N = 5 mice; P = 0.0078; Fig. 2A) in WT slices, and to 124.0 ± 10.4% (n = 11/N = 6 mice, P = 0.0137; Fig. 2A), in slices from BDNF+/− mice. The magnitude of LTP was not significantly different between WT and BDNF+/− mice, respectively (P = 0.3860). Therefore, a parallel decline of GABAergic inhibition due to chronic BDNF reduction cannot explain the unaffected LTP at cortico-LA synapses in BDNF+/− mice. Figure 2. View largeDownload slide Intact LTP at cortico-LA synapses in adult BDNF+/− mice in the presence of 0.1 µM gabazine. (A) LTP was induced by 3 trains of 100 stimuli at 100 Hz separated by 30 s or (B) by 4 trains of 100 stimuli at 100 Hz separated by 300 s. (A, B) Time course of averaged evoked field potential amplitudes in response to stimulation of cortical afferents in all slices recorded from WT and BDNF+/− mice, respectively. Note that LTP was induced independently of genotype. Insets at the right depict averaged field potential recordings 5 min before LTP induction and during the last 5 min of recordings for WT and BDNF+/− mice, respectively. (A) WT: n = 8 slices from N = 5 animals, BDNF+/−: n = 11 slices from N = 6 animals, ns, not significant. (B) WT: n = 11 slices from N = 9 animals, BDNF+/−: n = 9 slices from N = 6 animals, ns: not significantly different. Figure 2. View largeDownload slide Intact LTP at cortico-LA synapses in adult BDNF+/− mice in the presence of 0.1 µM gabazine. (A) LTP was induced by 3 trains of 100 stimuli at 100 Hz separated by 30 s or (B) by 4 trains of 100 stimuli at 100 Hz separated by 300 s. (A, B) Time course of averaged evoked field potential amplitudes in response to stimulation of cortical afferents in all slices recorded from WT and BDNF+/− mice, respectively. Note that LTP was induced independently of genotype. Insets at the right depict averaged field potential recordings 5 min before LTP induction and during the last 5 min of recordings for WT and BDNF+/− mice, respectively. (A) WT: n = 8 slices from N = 5 animals, BDNF+/−: n = 11 slices from N = 6 animals, ns, not significant. (B) WT: n = 11 slices from N = 9 animals, BDNF+/−: n = 9 slices from N = 6 animals, ns: not significantly different. It could be argued that the LTP protocol we used so far was not suitable to detect differences between genotypes. Therefore, we next tested a paradigm similar to the one used by Huang et al. (2000), which is known to induce protein synthesis-dependent late-LTP in the LA. This LTP paradigm consisted of 4 trains of 100 stimuli at 100 Hz separated by 5 min, and led to significant LTP as assessed 2 h after tetanization (WT: 123.6 ± 5.9%, n = 11 slices/N = 9 mice, P = 0.001; BDNF+/−: 121.5 ± 7.4%, n = 9 slices/N = 6 mice, P = 0.0273; Fig. 2B). The 2 genotypes did not show a difference in the magnitude of LTP (P = 0.8197). These results revealed intact LTP in response to distinct types of HFS at cortico-LA inputs in adult BDNF+/− mice (≥3 months), yet their fear learning was impaired. Thus, LTP at cortico-LA afferents does not enable formation of fear memory in BDNF+/− mice. This suggested that synaptic plasticity at intra-amygdala synapses might be impaired in BDNF+/− mice, and thereby contribute to reduced fear memory learning. Therefore, we tested LTP at these intra-amygdala synapses (see below). HFS Induced LTP at Intra-Amygdala Synapses We first focused on the basal amygdala (BL) as an intra-amygdala target of the LA (Pitkänen et al. 1997) which is implicated in fear learning (Amano et al. 2011 and references therein). Recordings were again performed in the presence of 0.1 µM gabazine. The stimulation electrode was positioned in the LA. Interestingly, we found similar magnitudes of LTP at these glutamatergic synapses in the BL (WT: 153.1 ± 15.3%, n = 9 slices/N = 8 mice, P = 0.0039; BDNF+/−: 153.1 ± 27.6%, n = 5 slices/N = 5 mice, P = 0.0079) in WT and BDNF+/− mice (P = 1.0; Fig. 3A). Figure 3. View largeDownload slide Intact HFS-LTP in distinct intra-amygdala circuits in adult BDNF+/− mice. (A, B) LTP was induced by 4 trains of 100 stimuli at 100 Hz separated by 300 s. Time course of averaged evoked field potential amplitudes in response to stimulation of intrabasal (A) or medial central amygdala afferents (B) in all slices recorded from WT and BDNF+/− mice. Note that LTP was successfully induced independent of genotype. Insets at the right depict averaged field potential traces 5 min before LTP induction and during the last 5 min of recordings for WT and BDNF+/− mice. (A) WT: n = 9 slices from N = 8 animals, BDNF+/−: n = 5 slices from N = 5 animals, ns, not significant. (B) WT: n = 4 slices from N = 3 animals, BDNF+/−: n = 5 slices from N = 3 animals, ns: not significantly different. Figure 3. View largeDownload slide Intact HFS-LTP in distinct intra-amygdala circuits in adult BDNF+/− mice. (A, B) LTP was induced by 4 trains of 100 stimuli at 100 Hz separated by 300 s. Time course of averaged evoked field potential amplitudes in response to stimulation of intrabasal (A) or medial central amygdala afferents (B) in all slices recorded from WT and BDNF+/− mice. Note that LTP was successfully induced independent of genotype. Insets at the right depict averaged field potential traces 5 min before LTP induction and during the last 5 min of recordings for WT and BDNF+/− mice. (A) WT: n = 9 slices from N = 8 animals, BDNF+/−: n = 5 slices from N = 5 animals, ns, not significant. (B) WT: n = 4 slices from N = 3 animals, BDNF+/−: n = 5 slices from N = 3 animals, ns: not significantly different. Consequently, we addressed LTP at subsequent glutamatergic synapses in the amygdala network. Besides the LA, the central nucleus of the amygdala (CeA) may be involved in acquisition and consolidation of fear memories (Wilensky et al. 2006). It is known that the medial part of the central nucleus is the main output of the amygdala and activates brainstem projections leading to the expression of fear responses (Pape and Pare 2010). Therefore, we analyzed synaptic plasticity in this structure. However, LTP in the medial central amygdala that was elicited by HFS of the BL was not affected in BDNF+/− animals (WT 165.5 ± 13.1%, n = 4 slices/N = 3 mice, BDNF+/− 153.8 ± 13.0%, n = 5 slices/N = 3 mice). Both genotypes showed significant potentiation compared with pretetanus values (WT P = 0.0286, BDNF+/−P = 0.0079) but no group differences (P = 0.7302, Fig. 3B). Field potential amplitudes were not significantly different between WT and BDNF+/− at any time point after HFS-induced LTP. Taken together, these results indicate intact LTP at intra-amygdala glutamatergic synapses. Altered synaptic plasticity at these synapses can therefore not account for the impaired fear learning in adult BDNF+/− mice. Occlusion of Fear Learning Induced Synaptic Changes and LTP at Cortico-LA Afferents Taking the unaltered LTP in adult BDNF+/− mice into account (≥3 months old), the question arises whether our LTP paradigm at cortico-LA afferents was actually relevant for fear learning. Therefore, LTP was tested ex vivo in slices of cued fear-conditioned WT mice. As control, we established a pseudoconditioning paradigm in which the animals received the same number of tone and foot shock presentations, but in an unpaired manner. To verify that this pseudoconditioning does not lead to any cued fear learning, we first compared this paradigm with our established fear conditioning paradigm (Fig. 4). During the conditioning, we observed a significant increase in freezing behavior in both groups (Fig. 4A; F20,356 = 8.8, P < 0.0001). This similar increase in freezing in the pseudoconditioned animals was expected, as they received the same number of foot shocks during the training session as the fear-conditioned animals. No differences in either the magnitude or the time course of freezing were observed between the 2 groups (factor training type [pseudo vs. trained]: F1356 = 0.0, P = 1; interaction of the factors training type × time: F20,356 = 1.0, P = 0.49). One day later, we tested the fear memory of the animals by presenting the CS in a novel context (Fig. 4B). Here, we observed a clear difference between the 2 groups: while the trained animals showed an obvious increase in freezing upon CS presentation, the pseudoconditioned animals did not. The statistical analysis revealed strong effects for the factors training type (F1,59 = 21.8, P < 0.0001) and phase (F2,59 = 50.6, P < 0.0001) as well as for the interaction of these 2 factors (F2,59 = 19.7, P < 0.0001). Post hoc Tukey comparisons revealed a significant increase in freezing only in the trained animals during the CS presentation, while in the pseudoconditioned mice there was no difference between the freezing behavior expressed in the different phases. This experiment demonstrates that our paradigm reliably induces fear behavior. In contrast, pseudoconditioning does not induce any cued fear learning, although freezing during training is similar between the 2 groups. Figure 4. View largeDownload slide Comparison of fear conditioning with the pseudoconditioning paradigm. To establish an appropriate control group for fear-conditioned animals in the ex vivo occlusion experiments, we designed a pseudoconditioning paradigm in which the animals received the same number of tone and foot shock presentations as the fear conditioning group, but in an explicitly unpaired manner. During the fear conditioning training (A), both groups showed a similar increase in freezing, most probably due to the foot shock presentations. In the fear retrieval test (B), which was performed 24 h later in a novel context, only animals that underwent the fear conditioning protocol (“trained”) showed an increased freezing behavior upon CS presentation. Pseudoconditioned animals showed a very low level of freezing throughout the experiment that remained in the range of the freezing during the habituation period. (*) indicates significant differences between trained and pseudoconditioned animals, as well as between CS and pre-CS and habituation. Figure 4. View largeDownload slide Comparison of fear conditioning with the pseudoconditioning paradigm. To establish an appropriate control group for fear-conditioned animals in the ex vivo occlusion experiments, we designed a pseudoconditioning paradigm in which the animals received the same number of tone and foot shock presentations as the fear conditioning group, but in an explicitly unpaired manner. During the fear conditioning training (A), both groups showed a similar increase in freezing, most probably due to the foot shock presentations. In the fear retrieval test (B), which was performed 24 h later in a novel context, only animals that underwent the fear conditioning protocol (“trained”) showed an increased freezing behavior upon CS presentation. Pseudoconditioned animals showed a very low level of freezing throughout the experiment that remained in the range of the freezing during the habituation period. (*) indicates significant differences between trained and pseudoconditioned animals, as well as between CS and pre-CS and habituation. After verifying that our 2 conditioning paradigms resulted in cued fear learning to the tone in fear-conditioned animals but absence of this cued fear learning in pseudoconditioned animals, we trained animals of both genotypes with the respective protocols and tested for occlusion of LTP at cortico-LA synapses. Of note, every CS presentation in the absence of the US results in extinction learning (Rescorla and Wagner 1972). Therefore, LTP was recorded in animals without a preceding fear retrieval test to avoid effects of synaptic changes linked to fear extinction learning. Similar to the previous experiment (compare Fig. 4A), we observed a general increase in freezing (Fig. 5A, B) upon the foot shock presentations in all groups (factor “time”: F20,1112 = 22,3, P < 0.0001). Importantly, there were no differences between genotypes (F1,1112 = 0.02, P = 0.9) or training paradigms (F1,1112 = 0.02, P = 0.89) as well as no interaction of these 3 factors (time × genotype × training paradigm: F20,1112 = 0.56, P = 0.94). These behavioral data verified similar conditions for both groups in response to the training session that preceded the subsequent electrophysiological recordings. Figure 5. View largeDownload slide Occlusion experiments in WT and BDNF+/−mice at 24 h (C, D) and 2 h (E, F) after fear conditioning. (A, B) Animals of both genotypes were either fear conditioned (“trained”) or served as a pseudoconditioned control (“pseudo”). During the conditioning training, all groups showed a comparable increase in freezing behavior (WT pseudoconditioned n = 23, trained n = 24; BDNF+/− pseudoconditioned n = 14, trained n = 24). (C, D) Slice recordings were performed 24 h after training. LTP was induced by 3 trains of 100 stimuli at 100 Hz separated by 30 s. Time course of averaged evoked field potentials in response to stimulation of cortico-LA afferents in all slices recorded from WT (C) and BDNF+/− mice (D). Insets at the right depict averaged field recordings 5 min before LTP induction and during the last 5 min of recordings for mice with unpaired or paired training. In WT animals, LTP was significantly reduced in fear-conditioned mice (trained) as opposed to mice subjected to unpaired training (pseudoconditioning). In BDNF+/− mice, LTP was unchanged by paired or nonpaired CS–US presentations. (E, F) Slice recordings were performed at least 4 h after training. In WT mice (E), LTP was significantly reduced in fear-conditioned mice (trained) as opposed to WT mice subjected to unpaired training (pseudoconditioning). In BDNF+/− mice (F), LTP was unchanged by paired or nonpaired CS–US presentations. (C) WT mice: Paired training, n = 9 slices from N = 6 animals, pseudoconditioning, n = 10 slices from N = 6 animals, *: significant, P = 0.0435. (D) BDNF+/− mice: Paired training, n = 12 slices from N = 5 animals, pseudoconditioning, n = 7 slices from N = 3 animals, ns: not significant, P = 0.4726. (E) WT mice: Paired training, n = 15 slices from N = 10 animals, pseudoconditioning, n = 13 slices from N = 10 animals, *: significant, P = 0.0304. (F) BDNF+/− mice: Paired training, n = 12 slices from N = 9 animals, pseudoconditioning, n = 7 slices from N = 6 animals, ns: not significant, P = 0.5828. Figure 5. View largeDownload slide Occlusion experiments in WT and BDNF+/−mice at 24 h (C, D) and 2 h (E, F) after fear conditioning. (A, B) Animals of both genotypes were either fear conditioned (“trained”) or served as a pseudoconditioned control (“pseudo”). During the conditioning training, all groups showed a comparable increase in freezing behavior (WT pseudoconditioned n = 23, trained n = 24; BDNF+/− pseudoconditioned n = 14, trained n = 24). (C, D) Slice recordings were performed 24 h after training. LTP was induced by 3 trains of 100 stimuli at 100 Hz separated by 30 s. Time course of averaged evoked field potentials in response to stimulation of cortico-LA afferents in all slices recorded from WT (C) and BDNF+/− mice (D). Insets at the right depict averaged field recordings 5 min before LTP induction and during the last 5 min of recordings for mice with unpaired or paired training. In WT animals, LTP was significantly reduced in fear-conditioned mice (trained) as opposed to mice subjected to unpaired training (pseudoconditioning). In BDNF+/− mice, LTP was unchanged by paired or nonpaired CS–US presentations. (E, F) Slice recordings were performed at least 4 h after training. In WT mice (E), LTP was significantly reduced in fear-conditioned mice (trained) as opposed to WT mice subjected to unpaired training (pseudoconditioning). In BDNF+/− mice (F), LTP was unchanged by paired or nonpaired CS–US presentations. (C) WT mice: Paired training, n = 9 slices from N = 6 animals, pseudoconditioning, n = 10 slices from N = 6 animals, *: significant, P = 0.0435. (D) BDNF+/− mice: Paired training, n = 12 slices from N = 5 animals, pseudoconditioning, n = 7 slices from N = 3 animals, ns: not significant, P = 0.4726. (E) WT mice: Paired training, n = 15 slices from N = 10 animals, pseudoconditioning, n = 13 slices from N = 10 animals, *: significant, P = 0.0304. (F) BDNF+/− mice: Paired training, n = 12 slices from N = 9 animals, pseudoconditioning, n = 7 slices from N = 6 animals, ns: not significant, P = 0.5828. Ex vivo LTP recordings in these animals were performed 1 day after training (Fig. 5C–F, see inset). LTP in cortico-LA afferents (pseudoconditioned control: 140.0 ± 9.5%, n = 10 slices/N = 6 mice, P = 0.002; trained: 111.6 ± 4.8%, n = 9/N = 6 mice, P = 0.0273) was significantly occluded in fear-conditioned WT animals compared with pseudoconditioned control mice (P = 0.0435; Fig. 5C). This occlusion is consistent with the hypothesis that cued fear learning and LTP at cortico-LA afferents induced by our protocol involve common synaptic mechanisms. In contrast, LTP in BDNF+/− mice (pseudoconditioned control: 116 ± 4.4%, n = 7/N = 3 mice, P = 0.0156; trained: 112.4 ± 5.8%, n = 12/N = 5 mice, P = 0.0476) was not occluded in fear-conditioned animals compared with pseudoconditioned control mice (P = 0.4726; Fig. 5D). Thus, fear learning seems to induce long-term changes at cortico-LA synapses in WT mice as indicated by the occlusion of LTP 24 h after training. These learning-induced long-term changes are absent in BDNF+/− mice, which accordingly do not show fear memory 24 h after training. To get further insights into the dynamics of these long-term synaptic changes, we tested occlusion of cortico-LA LTP at earlier time points after conditioning training. Of note, behavioral testing by itself, for example, due to increased stress by additional handling or the application of foot shocks, may affect subsequent measurements of synaptic plasticity. Therefore, we first tested suitable delays between end of pseudoconditioning and preparation of brain slices that allowed induction of LTP in WT mice. LTP recordings started always at least 2 h after slice preparation. Under these conditions neither field recordings with 20 min (109.2 ± 9.4%, n = 8/N = 3 mice, P = 0.7422) nor 60 min delay between pseudoconditioning and preparation of slices (113.9 ± 8.2%, n = 7/N = 3 mice, P = 0.0781) showed LTP. In contrast, stable LTP was expressed when mice were sacrificed 2 h after pseudoconditioning and recordings started at the earliest 4 h after the training procedure. When using this time window, LTP was successfully induced in pseudoconditioned WT mice (control: 122.3 ± 6.4%, n = 13/N = 10 mice, P = 0.0012; trained: 104.5 ± 2.8%, n = 15/N = 10 mice, P = 0.0413, Fig. 5E) as well as BDNF+/− mice (control: 116.7 ± 6.5%, n = 7/N = 6 mice, P = 0.0313; trained: 111.1 ± 4.5%, n = 12/N = 9 mice, P = 0.0122; Fig. 5F). Notably, LTP was occluded in WT mice (P = 0.030 Fig. 5E), but not in BDNF+/− mice (P = 0.5828, Fig. 5F). These data suggest that synaptic plasticity at cortico-LA synapses is relevant for early fear memory consolidation in WT mice, that is, ≤4 h after training. However, these synaptic changes are absent in BDNF+/− mice probably contributing to the declining fear memory in BDNF+/− mice, that initially (i.e., 30 min after training) show intact fear memory acquisition. Assessment of Early Fear Memories To get more detailed insights into the characteristics of the fear learning deficit in BDNF+/− mice, we tested the early fear memory in different cohorts at distinct time points between 30 min and 24 h after fear conditioning (Fig. 6B, C). During the fear conditioning training (Fig. 6A), both genotypes showed an increase in freezing behavior (time: F20,1574 = 45.1, P < 0.0001). Overall, BDNF+/− mice showed slightly more freezing than their WT littermates, but this was not statistically significant (genotype: F1,1574 = 0.0, P = 0.98; genotype × time: F20,1574 = 1.56, P = 0.06). Thus, there was no obvious difference between the 2 genotypes during the fear conditioning training, confirming our previous results (Endres and Lessmann 2012). In the fear memory tests, WT animals showed very stable and comparable fear expression at all tested time points, whereas the fear memory in BDNF+/− mice seemed to decline over time. This decline is also evident from elevated pre-CS freezing. Since the pre-CS freezing reflects the average of all 5 pre-CS periods for a given test interval, this elevated pre-CS freezing probably reflects unspecific freezing. As BDNF+/− mice show this elevation in freezing behavior only after the first CS presentation, it seems like this unspecific freezing is triggered by the first CS presentation. Overall, these data suggest that the memory of the temporal relation between CS and US vanishes in BDNF+/− mice. Twenty-four hours after fear conditioning training, the BDNF+/− animals showed an additional significantly reduced freezing to the CS. These observations are supported by the results of an ANOVA revealing significant effects for the factors “genotype” (BDNF+/− vs. WT), “phase” (Habituation vs. pre-CS vs. CS) and “time point of testing” (genotype: F1,312 = 4.12, P = 0.042; phase: F2,312 = 126.9, P < 0.0001; time point: F1,312 = 8.58, P = 0.004). In addition, the ANOVA revealed a strong interaction between the factors “phase” and “genotype” (F2,312 = 12.81, P < 0.0001), further supporting a generally different fear memory expression between the 2 genotypes. However, there was no significant interaction of the factors “phase” × “genotype” × “time point of testing” as well as for “phase” × “time point of testing” and “genotype” × “time point of testing” (F’s ≤ 2.03, P’s ≥ 0.15). As the ANOVA revealed significant effects for all 3 single factors, we also performed a post hoc Tukey test, which revealed significant differences between the freezing behavior expressed during the habituation period compared with the CS periods for both genotypes at all tested time points. Overall, these findings suggest successful fear memory retrieval for both genotypes at all tested time points, with the exception that at 24 h after fear conditioning, BDNF+/− mice exhibited significantly less freezing than their WT littermates. This corroborates our previous finding of impaired fear learning in these animals (Endres and Lessmann 2012). Importantly, WT animals showed significantly more freezing during the CS compared with the pre-CS periods at all tested time points, while the BDNF+/− mice exhibited similar freezing levels during both phases (CS and pre-CS), especially at later time points of testing. This observation could be interpreted as an indication for generalized fear in BDNF+/− mice. However, since the freezing levels during the habituation periods were similar between BDNF+/− and WT animals, this observation speaks in favor of reduced fear memory accuracy rather than a generalized fear behavior. To better account for these differences in pre-CS to CS freezing, we analyzed the Δfreezing scores (i.e., freezing during CS minus freezing during pre-CS periods, Fig. 6B) of the animals. In order to see whether there was a significant increase in freezing, we tested whether these Δ-scores differ from zero (i.e., no increase in freezing) by single sided t-test comparisons. For BDNF+/− mice, this analysis revealed a significant Δfreezing only at 2 h and a tendency of increased Δfreezing (P = 0.09) 30 min after fear conditioning. In contrast, their WT littermates exhibited a significant Δfreezing at all tested time points. This analysis further suggests the idea of a continuous loss in fear memory precision with ongoing time after training. Figure 6. View largeDownload slide Early fear memory retrieval in BDNF+/− and WT mice. (A) Freezing during the conditioning training. No significant differences between the 2 groups were observed (WT n = 51; BDNF+/−n = 57). (B) Comparison of Δfreezing scores at different time points after fear conditioning. The Δ-scores were calculated for all 5 CS presentations per animal and then averaged. Significance of differences of Δ-scores from zero was determined (*: P < 0.05, ~:P < 0.1). (C) Fear memory retrieval at different time points after fear conditioning, ranging from 0.5 to 24 h. The freezing during the pre-CS and CS represents the average over pre-CS and CS periods. Significant differences compared with habituation and pre-CS periods are indicated by (#), while (+) represents only a significant difference to the habituation, and ($) represents a significant difference between WT and BDNF+/−. (Number of observations for WT/BDNF+/− 30 min: 8/8; 2 h: 13/13; 4 h: 11/14; 6 h: 11/14; 24 h: 8/8). (D) Discriminative early fear memory. In young animals (2 months old, left panel), a clear discrimination of CS− and CS+ was observed 4 h after conditioning (n = 16 per time point). Five- to six-month-old BDNF+/− mice (right panel) failed to discriminate between the 2 stimuli at this time point, while WT littermates did discriminate (WT n = 12; BDNF+/−n = 13; *: P < 0.05). Figure 6. View largeDownload slide Early fear memory retrieval in BDNF+/− and WT mice. (A) Freezing during the conditioning training. No significant differences between the 2 groups were observed (WT n = 51; BDNF+/−n = 57). (B) Comparison of Δfreezing scores at different time points after fear conditioning. The Δ-scores were calculated for all 5 CS presentations per animal and then averaged. Significance of differences of Δ-scores from zero was determined (*: P < 0.05, ~:P < 0.1). (C) Fear memory retrieval at different time points after fear conditioning, ranging from 0.5 to 24 h. The freezing during the pre-CS and CS represents the average over pre-CS and CS periods. Significant differences compared with habituation and pre-CS periods are indicated by (#), while (+) represents only a significant difference to the habituation, and ($) represents a significant difference between WT and BDNF+/−. (Number of observations for WT/BDNF+/− 30 min: 8/8; 2 h: 13/13; 4 h: 11/14; 6 h: 11/14; 24 h: 8/8). (D) Discriminative early fear memory. In young animals (2 months old, left panel), a clear discrimination of CS− and CS+ was observed 4 h after conditioning (n = 16 per time point). Five- to six-month-old BDNF+/− mice (right panel) failed to discriminate between the 2 stimuli at this time point, while WT littermates did discriminate (WT n = 12; BDNF+/−n = 13; *: P < 0.05). To further test for the precision of early fear memory, we performed a discriminative fear conditioning paradigm. Here, we trained animals with 2 randomly appearing tone stimuli. One tone (8 kHz, CS+) was always followed by a foot shock, while the other tone (2 kHz CS−) was not. When we tested these animals 3 h after fear conditioning (Fig. 6D), even 2- to 3-month-old WT mice did not differentiate between CS+ and CS−, that is, they responded with a comparable fear response to both stimuli. An ANOVA revealed significant main effects for the factors “time point of testing” (F1,150 = 14.2, P = 0.0002) and “phase” (F4,150 = 33.4, P < 0.0001), as well as a tendency for the interaction of these 2 factors (F4,150 = 2.0, P = 0.10). Post hoc Tukey comparisons revealed significant differences between pre-CS and CS periods for both CS+ and CS−, if animals were tested 3 h after training. However, when animals were tested 4 h after conditioning, a significant difference between pre-CS and CS existed only in case of CS+ presentations. Hence, the ability to discriminate between CS+ and CS− in our fear conditioning paradigm seems to start around 4 h after training in adult (2–3 months old) WT mice. Therefore, we decided to test adult BDNF+/− mice and their WT littermates 4 h after fear conditioning (Fig. 6D, right panel). Here, we observed a similar pattern as in the previous experiment for WT mice, while BDNF+/− mice displayed no discrimination between CS+ and CS−. An ANOVA revealed a significant main effect for the factor “phase” (F4,115 = 15.6, P < 0.0001), no significant effect for the factor “genotype” (F1,115 = 0.3, P = 0.6) but a tendency for the interaction of these 2 factors (F4,115 = 2.3, P = 0.09). Post hoc Tukey comparisons revealed that only in WT mice a selective significant increase in freezing between pre-CS+ and CS+ could be observed, thus indicating that WT but not BDNF+/− mice could discriminate between CS+ and CS−. In conclusion, our data demonstrate that 24 h after fear conditioning, BDNF+/− mice exhibited an impaired fear memory expression. In addition, our data suggest a declining fear memory precision starting in the very early stages of fear memory consolidation, that is, at 4 h after fear conditioning. Discussion In the present study, we demonstrate that, although LTP in the lateral, basal, and central medial nuclei of the amygdala is intact in adult BDNF+/− mice (≥3 months old), these animals show impaired fear memory consolidation. Nevertheless, early fear memory in adult BDNF+/− mice is still functional. Thus, the unaltered synaptic plasticity in these amygdala circuits is in line with successful acquisition of fear memories in BDNF+/− mice. Moreover, our data provide evidence that fear learning in WT mice results in long-lasting changes in synaptic plasticity at cortico-LA synapses, as indicated by occlusion of LTP 24 h after fear conditioning. Thus, fear learning in WT mice occludes LTP, similar to what has been reported previously for rats (Tsvetkov et al. 2002; Schroeder and Shinnick-Gallagher 2004, 2005). In line with the impaired long-term fear memory in adult BDNF+/− mice, we did not observe occlusion of LTP at cortico-LA synapses 24 h after fear conditioning. Thus, our findings support the notion that in WT mice the successful consolidation of cued fear memories relies, at least in part, on long-term synaptic changes at cortico-LA synapses. In addition, our results demonstrate that such learning-related processes resulting in these long-term modifications are not functional in adult BDNF+/− mice, probably leading to the observed deficit in fear memory consolidation. Interestingly, in BDNF+/− mice LTP is also not occluded when assessed 4–6 h after fear conditioning, at a time when these animals still show successful but unprecise retrieval of early fear memory. These observations suggest that the learning-related changes in synaptic plasticity at cortico-LA synapses observed in WT mice are not required for the expression of early fear memories in BDNF+/− mice. However, as we observed a reduction in fear memory precision (i.e., increase in pre-CS freezing and lack of CS+/CS− discrimination) in BDNF+/− mice from 4 h onward, changes in synaptic plasticity at these synapses might be required to form a precise fear memory trace, in WT as well as in in BDNF+/− mice. Since we found intact LTP in amygdala circuits of adult BDNF+/− mice while fear consolidation was impaired, a decline in synaptic plasticity at other synapses seems to contribute to the lack of long-term memory in BDNF+/− mice. Likely candidates are the prelimbic or the perirhinal cortex (PRhC), since for both areas a crucial role of BDNF-TrkB signaling for the consolidation of cued fear memories has been reported (Choi et al. 2010; Schulz-Klaus et al. 2013). Age-Dependent Learning Deficit and LTP in the Amygdala The present study was undertaken to analyze the cellular mechanisms underlying the age-dependent learning deficit that we recently described for adult (≥3 months old) BDNF+/− mice (Endres and Lessmann 2012). In a previous study, we had shown a BDNF-dependent impairment of LTP in thalamic afferents to the LA already at the age of 1 month (Meis et al. 2012). We therefore assumed that fear learning in these younger animals might preferentially be mediated by stable cortico-LA LTP (Meis et al. 2012). Indeed, it has been shown previously that disruption of auditory fear conditioning is only achieved by combined lesions of the thalamic and the cortical sensory pathways to the LA (Romanski and LeDoux 1992). Thus, we hypothesized that an age-dependent decline in LTP at cortico-LA synapses might account for the observed learning deficit in adult BDNF+/− mice. Since BDNF requirement for LTP has been shown to critically depend on the induction paradigm (Kang et al. 1997; Zakharenko et al. 2003; Abidin et al. 2006; Edelmann et al. 2015) different LTP protocols were tested. Additionally, experiments were performed in the presence of the GABAA receptor antagonist gabazine to account for any compensatory changes in the inhibitory system by chronic BDNF reduction (Gottmann et al. 2009), which could mask LTP deficits. Interestingly, successful LTP induction at cortico-LA synapses appeared to be independent of age and chronic BDNF reduction across different LTP induction paradigms and irrespective of intact GABAergic inhibition. Overall, these data suggest that intact cortico-LA LTP alone does not enable fear memory consolidation in adult BDNF+/− mice. Therefore, we additionally analyzed synaptic plasticity in other subnuclei of the amygdala network that may contribute to the observed learning deficit in BDNF+/− mice. First, we focused on the BL, which was previously reported to undergo LTP upon repeated LA stimulation (Rammes et al. 2000). Moreover, a decrease in fear learning was associated with a reduction of LTP in this nucleus in several knockout mouse models (Brambilla et al. 1997; Humeau et al. 2007; Huynh et al. 2009; Bourgognon et al. 2012) or after pharmacological interference (Sinai et al. 2010). However, we did not observe any decrease in LTP in the basal nucleus upon intra-BL HFS in BDNF+/− as compared with WT mice. Likewise, LTP remained unchanged at glutamatergic input synapses to the medial central amygdala in BDNF+/− mice, representing the main output structure of the amygdala (Pape and Paré 2010), which is also involved in fear acquisition and consolidation (Wilensky et al. 2006). Interestingly, the paraventricular nucleus of the thalamus (PVT) was recently shown to modulate fear learning by activation of TrkB receptors in the lateral central amygdala (Penzo et al. 2015). Nonetheless, our present results do not speak in favor of altered synaptic plasticity at glutamatergic inputs to the lateral CeA in BDNF+/− mice, since LTP at BL-medial CeA synapses was intact in these animals. Overall, we observed unaltered early LTP in BDNF+/− mice at all synapses investigated (cortico-LA, intra-BL, and BL-medial CeA) which parallels the stable early fear memory retrieval (4 h after training) but not the deficit in long-term fear memory (at 24 h). Thus, additional synaptic plasticity processes not analyzed in our study seem to be required for the consolidation of long-term fear memories. Age-Dependent Changes in Synaptic Plasticity and Fear Learning The age-related learning deficit that we observed in BDNF+/− mice could result from qualitative changes in the formation of synaptic circuits during ontogenesis that are involved in the coding of fear memory. Only a few previous studies analyzed fear learning as well as the corresponding cellular mechanisms between adolescence and adulthood. For example, it was shown that, both, synaptic networks in the amygdala and the connectivity between amygdala and prefrontal cortex are altered roughly 1–3 months after birth (Cunningham et al. 2002, 2008; Pan et al. 2009b). In addition, experience-driven maturation of synaptic transmission at cortico-LA synapses was reported to occur most likely during early adulthood (Gambino et al. 2010). It remains to be determined by future studies whether the altered fear memory in BDNF+/− animals can be related to one of these changes. Early Fear Memory in BDNF+/− Mice The preserved LTP at cortico-LA and intra-amygdala synapses in our study points to stable amygdala function in BDNF+/− mice. Interestingly, early fear memory in age-matched BDNF+/− mice was comparable to that of WT littermates when tested 0.5–6 h after cued fear conditioning, as is evident from similar increments in freezing between habituation and CS presentations (see Fig. 6C). This indicates intact acquisition of fear memories in BDNF+/− mice and suggests that synaptic networks required to generate early fear memory are functional in BDNF+/− mice. Nevertheless, the 2 genotypes differ in pre-CS freezing duration with longer testing intervals between conditioning and retrieval. In WT mice, pre-CS freezing duration within the session declines to a similar level as freezing during the habituation, whereas in BDNF+/− mice pre-CS freezing reaches a level in the range of CS freezing. This finding is mirrored by the analysis of the Δfreezing scores, which revealed significant increments in freezing behavior in WT mice at any tested time point. In contrast, BDNF+/− mice exhibited no significant increment in freezing when tested 4 h or later after fear conditioning training. In conclusion, these observations suggest that despite a successful retrieval of early fear memories in BDNF+/− mice, the precision of fear memory to the actual CS presentation gets poorer during the early consolidation processes. This might be taken as a first indication of the fear memory deficit observed 24 h after fear conditioning. To gain a more detailed insight into the precision of the early fear memory, we introduced a discriminative fear learning task. In 2-month-old WT mice, we observed successful discrimination between CS+ and CS− when testing 4 h after fear conditioning but not at an interval of 3 h (see Fig. 6D). Importantly, these responses were elicited specifically by the tone presentations, as indicated by low pre-CS freezing values during the retrieval session. Thus, it seems like this early fear memory is not fine-tuned to the CS. Consequently, the animals responded to both stimuli with a high but specific fear response. With ongoing consolidation processes, the precision of the fear memory increased. This is evident from the much lower levels of freezing to the CS− with increasing time interval between training and retrieval session. Alternatively to this interpretation, fear conditioning might just induce a longer lasting state of arousal (Rodrigues et al. 2009), which caused the animals to respond with a generalized fear behavior to all unexpected stimuli that appear during testing. Since in WT animals the discrimination between CS+ and CS− begins around 4 h after training, we tested the discriminative memory abilities of aged BDNF+/− mice only at this time point. Here, we observed that 5- to 6-month-old BDNF+/− mice exhibited comparable levels of fear to CS+ and CS− presentations. In addition, they showed similar elevated pre-CS freezing levels as observed in the nondiscriminative fear conditioning experiments (compare Fig. 5). In contrast, their WT littermates showed a specific fear response to the CS+, even though the response to the CS− was more pronounced than in younger WT animals. This suggests that early fear discrimination might decline with aging. We previously observed that BDNF+/− mice do not exhibit altered anxiety in the elevated plus maze and open field at any age (Endres and Lessmann 2012). Therefore, it seems unlikely that the elevated fear responses during the pre-CS and CS− periods are due to an altered anxiety level in BDNF+/− mice. In conclusion, these results support the notion that fear memory in BDNF+/− mice, even though still retrievable early after training, undergoes a constant decline in memory precision, which is due to a lack of required—maybe BDNF-dependent—consolidation processes. Occlusion of LTP by Previous Fear Learning Several lines of experimental data suggest that LTP in the amygdala might be the cellular mechanism underlying cued fear learning (for review, see Sigurdsson et al. 2007). In this respect, fear conditioning was shown previously to induce long-lasting changes at different amygdala synapses, as reflected by facilitated synaptic transmission as well as a lack of subsequent electrical LTP induction (McKernan and Shinnick-Gallagher 1997; Tsvetkov et al. 2002; Schroeder and Shinnick-Gallagher 2004, 2005; Hong et al. 2011, 2012). These ex vivo experiments were performed exclusively in previously fear-conditioned rats but not in mice. We therefore investigated LTP occlusion in age-matched mice to verify that LTP induced by our specific paradigm was indeed affected by preceding fear learning. In fact, LTP was significantly reduced by pretraining of mice compared with pseudoconditioned animals (see Fig. 5C, E). These findings indicate that LTP induction at cortico-LA synapses by our HFS paradigm and fear learning share identical cellular processes in WT mice. However, we did not observe such an occlusion of cortico-LA LTP in BDNF+/− mice, indicating that in these animals no learning-dependent long-term changes in cortico-LA afferents occurred. As expected, this lack of occlusion at the cellular level parallels the behavioral deficit in fear consolidation in BDNF+/− mice when assessed 24 h after training (present study; Endres and Lessmann 2012). We also tried to perform occlusion experiments shortly after fear training to gain insights into the synaptic processes involved in acquisition of fear memory. Successful LTP induction was achieved in slices of pseudoconditioned mice, which were sacrificed 2 h after training, when field recordings started 4–8 h after training. This LTP was again occluded in WT mice, but not in BDNF+/− mice, as already observed for recordings obtained 1 day after training. This suggests that changes in synaptic transmission occur at these synapses in WT mice, which however, are lacking in BDNF+/− mice. Interestingly, we observed in the same time range (i.e., 4–8 h) an ongoing decline in fear memory precision in BDNF+/− mice. This was evident from the reduced Δfreezing scores that were due to an increased pre-CS freezing. Overall, this might indicate that the lack of fear memory-related changes at the cortico-LA afferents contributes to the reduced fear memory precision observed at the early stages (i.e., 4–8 h) after fear conditioning training. Altered Synaptic Plasticity in Other Fear Learning-Related Brain Areas We focused exclusively on synaptic plasticity within the amygdala network in this study, as the amygdala is widely accepted as key brain area for cued fear learning (Sigurdsson et al. 2007; Ehrlich et al. 2009; Pape and Paré 2010). Besides the amygdala, synaptic plasticity in several other brain areas contributes to successful formation of cued fear memory, for example, the prelimbic medioprefrontal cortex (PL, Burgos-Robles et al. 2009; Sotres-Bayon and Quirk 2010; Sierra-Mercado et al. 2011) or the PRhC (Schulz et al. 2004; Kealy and Commins 2011; Kent and Brown 2012). Interestingly, interfering with BDNF signaling in these brain areas by regional BDNF-knockout (Choi et al. 2010) or acute pharmacological inhibition of TrkB signaling (Schulz-Klaus et al. 2013) resulted in a lack of fear memory consolidation. Since the expression of BDNF protein changes with aging (Croll et al. 1998; Katoh-Semba et al. 1998; von Bohlen und Halbach 2010; Boger et al. 2011) and these changes are differentially regulated in different brain areas (Silhol et al. 2005; Psotta et al. 2013), BDNF- and age-dependent synaptic plasticity might vary depending on the specific brain area under study. Thus, an age-dependent change affecting fear consolidation may take place in the PL and/or PRhC in BDNF+/− mice. These changes might in turn alter synaptic structures or synaptic transmission in distinct brain areas (including the amygdala) that are essential for fear memory consolidation. In this respect, it is an interesting question whether we stimulated fibers originating from the above mentioned cortical regions in our LTP recordings as well. We placed our stimulation electrode at the external capsule which carries mainly fibers originating from higher sensory cortices such as for example, the auditory cortex (de Olmos et al. 1985), which is required for tone discrimination (LeDoux 1995). Since afferents from the PL or PRhC reach the amygdala either more ventrally (PL, Vertes, 2004) or more horizontally (PRhC, von Bohlen und Halbach and Albrecht 2002), it seems unlikely that these fibers are stimulated in our LTP recordings. Nevertheless, future studies addressing the mechanisms of synaptic plasticity in the PL and/or the PRhC in relation to BDNF availability and age will be required to test their contribution to the age-dependent fear learning deficit in BDNF+/− mice. Possible Sources of BDNF BDNF mRNA and protein was detected in the rodent amygdala at moderate to high levels (Conner et al. 1997; Yan et al. 1997; Krause et al. 2008), and fear conditioning induces a selective increase of BDNF levels in the BLA (Rattiner et al. 2005; Ou and Gean 2006). In addition, the temporal association cortex, sending afferents to the LA, shows substantial BDNF expression (Ernfors et al. 1990; Castren et al. 1995; Conner et al. 1997). In contrast, the LA is poorly innervated by the paraventricular thalamic nucleus, a structure recently identified as a major source of BDNF for the lateral nucleus of the central amygdala (Penzo et al. 2015). Accordingly, postsynaptic as well as presynaptic BDNF may contribute to LTP induction/expression (Edelmann et al. 2014) at cortico-LA afferents. Conclusion Overall, we demonstrate that fear learning induces long-lasting changes in synaptic plasticity at cortico-LA synapses in WT mice, which are absent in adult BDNF+/− mice. Moreover, we show that precision of early fear memory declines with time after training in BDNF+/− mice already at 4 h after training. Therefore, consolidation-relevant processes seem to occur at cortico-LA synapses in WT mice, which are absent in BDNF+/− mice. Thus, intact LTP at cortico-LA synapses parallels the formation of the initial CS–US association in BDNF+/− mice. However, due to a lack of consolidation-relevant synaptic plasticity, which might take place outside the amygdala circuitry, cortico-LA synapses are unable to undergo long-term changes in BDNF+/− mice that seem to be required to form a stable fear memory. In this respect, future studies should aim at elucidating how BDNF-dependent synaptic plasticity inside and outside the amygdala orchestrates the formation of fear memories in adult animals. Supplementary Material Supplementary material is available at Cerebral Cortex online. Funding Deutsche Forschungsgemeinschaft (SFB 779, TP B06). Notes We thank Kathrin Friese, Colette Obst, Angela Jahn, Regina Ziegler, and Evelyn Friedl for excellent technical assistance, and Gloria Deisy Quiceno for linguistic corrections of the manuscript. Conflict of Interest: None declared. References Abidin I, Eysel UT, Lessmann V, Mittmann T. 2008. Impaired GABAergic inhibition in the visual cortex of brain-derived neurotrophic factor heterozygous knockout mice. J Physiol . 586: 1885– 1901. Google Scholar CrossRef Search ADS PubMed  Abidin I, Köhler T, Weiler E, Zoidl G, Eysel UT, Lessmann V, Mittmann T. 2006. Reduced presynaptic efficiency of excitatory synaptic transmission impairs LTP in the visual cortex of BDNF-heterozygous mice. Eur J Neurosci . 24: 3519– 3531. Google Scholar CrossRef Search ADS PubMed  Amano T, Duvarci S, Popa D, Paré D. 2011. The fear circuit revisited: contributions of the basal amygdala nuclei to conditioned fear. J Neurosci . 31: 15481– 15489. Google Scholar CrossRef Search ADS PubMed  Bauer EP, Schafe GE, LeDoux JE. 2002. NMDA receptors and L-type voltage-gated calcium channels contribute to long-term potentiation and different components of fear memory formation in the lateral amygdala. J Neurosci . 22: 5239– 5249. Google Scholar PubMed  Boger HA, Mannangatti P, Samuvel DJ, Saylor AJ, Bender TS, McGinty JF, Fortress AM, Zaman V, Huang P, Middaugh LD, et al.  . 2011. Effects of brain-derived neurotrophic factor on dopaminergic function and motor behavior during aging. Genes Brain Behav . 10: 186– 198. Google Scholar CrossRef Search ADS PubMed  Bourgognon JM, Schiavon E, Salah-Uddin H, Skrzypiec AE, Attwood BK, Shah RS, Patel SG, Mucha M, John Challiss RA, Forsythe ID, et al.  . 2012. Regulation of neuronal plasticity and fear by a dynamic change in PAR1-G protein coupling in the amygdala. Mol Psychiatry . doi:10.1038/mp.2012.133. Brambilla R, Gnesutta N, Minichiello L, White G, Roylance AJ, Herron CE, Ramsey M, Wolfer DP, Cestari V, Rossi-Arnaud C, et al.  . 1997. A role for the Ras signalling pathway in synaptic transmission and long-term memory. Nature . 390: 281– 286. Google Scholar CrossRef Search ADS PubMed  Burgos-Robles A, Vidal-Gonzalez I, Quirk GJ. 2009. Sustained conditioned responses in prelimbic prefrontal neurons are correlated with fear expression and extinction failure. J Neurosci . 29: 8474– 8482. Google Scholar CrossRef Search ADS PubMed  Castren E, Thoenen H, Lindholm D. 1995. Brain-derived neurotrophic factor messenger RNA is expressed in the septum, hypothalamus and in adrenergic brain stem nuclei of adult rat brain and is increased by osmotic stimulation in the paraventricular nucleus. Neuroscience . 64: 71– 80. Google Scholar CrossRef Search ADS PubMed  Choi DC, Maguschak KA, Ye K, Jang SW, Myers KM, Ressler KJ. 2010. Prelimbic cortical BDNF is required for memory of learned fear but not extinction or innate fear. Proc Natl Acad Sci USA . 107: 2675– 2680. Google Scholar CrossRef Search ADS PubMed  Conner JM, Lauterborn JC, Yan Q, Gall CM, Varon S. 1997. Distribution of brain-derived neurotrophic factor (BDNF) protein and mRNA in the normal adult rat CNS: evidence for anterograde axonal transport. J Neurosci . 17: 2295– 2313. Google Scholar PubMed  Cowansage KK, LeDoux JE, Monfils MH. 2010. Brain-derived neurotrophic factor: a dynamic gatekeeper of neural plasticity. Curr Mol Pharmacol . 3: 12– 29. Google Scholar CrossRef Search ADS PubMed  Croll SD, Ip NY, Lindsay RM, Wiegand SJ. 1998. Expression of BDNF and trkB as a function of age and cognitive performance. Brain Res . 812: 200– 208. Google Scholar CrossRef Search ADS PubMed  Cunningham MG, Bhattacharyya S, Benes FM. 2002. Amygdalo-cortical sprouting continues into early adulthood: implications for the development of normal and abnormal function during adolescence. J Comp Neurol . 453: 116– 130. Google Scholar CrossRef Search ADS PubMed  Cunningham MG, Bhattacharyya S, Benes FM. 2008. Increasing Interaction of amygdalar afferents with GABAergic interneurons between birth and adulthood. Cereb Cortex . 18: 1529– 1535. Google Scholar CrossRef Search ADS PubMed  Daftary SS, Calderon G, Rios M. 2012. Essential role of brain-derived neurotrophic factor in the regulation of serotonin transmission in the basolateral amygdala. Neuroscience . 224: 125– 134. Google Scholar CrossRef Search ADS PubMed  de Olmos J, Alheid GF, Beltramino CA. 1985. Amygdala. In: Paxinos G, editor. The rat nervous system . Sydney: Academic Press. p. 223– 334. Drephal C, Schubert M, Albrecht D. 2006. Input-specific long-term potentiation in the rat lateral amygdala of horizontal slices. Neurobiol Learn Mem . 85: 272– 282. Google Scholar CrossRef Search ADS PubMed  Doyère V, Schafe GE, Sigurdsson T, LeDoux JE. 2003. Long-term potentiation in freely moving rats reveals asymmetries in thalamic and cortical inputs to the lateral amygdala. Eur J Neurosci . 17: 2703– 2715. Google Scholar CrossRef Search ADS PubMed  Edelmann E, Cepeda-Prado E, Franck M, Lichtenecker P, Brigadski T, Lessmann V. 2015. Theta burst firing recruits BDNF release and signaling in postsynaptic CA1 neurons in spike timing-dependent LTP. Neuron . 86: 1041– 1054. Google Scholar CrossRef Search ADS PubMed  Edelmann E, Lessmann V, Brigadski T. 2014. Pre- and postsynaptic twists in BDNF secretion and action in synaptic plasticity. Neuropharmacology . 76: 610– 627. Google Scholar CrossRef Search ADS PubMed  Ehrlich I, Humeau Y, Grenier F, Ciocchi S, Herry C, Lüthi A. 2009. Amygdala inhibitory circuits and the control of fear memory. Neuron . 62: 757– 771. Google Scholar CrossRef Search ADS PubMed  Endres T, Lessmann V. 2012. Age-dependent deficits in fear learning in heterozygous BDNF knock-out mice. Learn Mem . 19: 561– 570. Google Scholar CrossRef Search ADS PubMed  Ernfors P, Wetmore C, Olson L, Persson H. 1990. Identification of cells in rat brain and peripheral tissues expressing mRNA for members of the nerve growth factor family. Neuron . 5: 511– 526. Google Scholar CrossRef Search ADS PubMed  Gambino F, Khelfaoui M, Poulain B, Bienvenu T, Chelly J, Humeau Y. 2010. Synaptic maturation at cortical projections to the lateral amygdala in a mouse model of Rett syndrome. PLoS One . 5: e11399. Google Scholar CrossRef Search ADS PubMed  Gean PW, Shinnick-Gallagher P. 1987. Picrotoxin induced epileptiform activity in amygdaloid neurons. Neurosci Lett . 73: 149– 154. Google Scholar CrossRef Search ADS PubMed  Gottmann K, Mittmann T, Lessmann V. 2009. BDNF signaling in the formation, maturation and plasticity of glutamatergic and GABAergic synapses. Exp Brain Res . 199: 203– 234. Google Scholar CrossRef Search ADS PubMed  Hong I, Kim J, Lee J, Park S, Song B, Kim J, An B, Park K, Lee HW, Lee S, et al.  . 2011. Reversible plasticity of fear memory-encoding amygdala synaptic circuits even after fear memory consolidation. PLoS One . 6: e24260. Google Scholar CrossRef Search ADS PubMed  Hong I, Kim J, Song B, Park K, Shin K, Eom KD, Han PL, Lee S, Choi S. 2012. Fear conditioning occludes late-phase long-term potentiation at thalamic input synapses onto the lateral amygdala in rat brain slices. Neurosci Lett . 506: 121– 125. Google Scholar CrossRef Search ADS PubMed  Huang YY, Kandel ER. 2007. 5-Hydroxytryptamine induces a protein kinase A/mitogen-activated protein kinase-mediated and macromolecular synthesis-dependent late phase of long-term potentiation in the amygdala. J Neurosci . 27: 3111– 3119. Google Scholar CrossRef Search ADS PubMed  Huang YY, Martin KC, Kandel ER. 2000. Both protein kinase A and mitogen-activated protein kinase are required in the amygdala for the macromolecular synthesis-dependent late phase of long-term potentiation. J Neurosci . 20: 6317– 6325. Google Scholar PubMed  Humeau Y, Reisel D, Johnson AW, Borchardt T, Jensen V, Gebhardt C, Bosch V, Gass P, Bannerman DM, Good MA, et al.  . 2007. A pathway-specific function for different AMPA receptor subunits in amygdala long-term potentiation and fear conditioning. J Neurosci . 27: 10947– 10956. Google Scholar CrossRef Search ADS PubMed  Huynh DP, Maalouf M, Silva AJ, Schweizer FE, Pulst SM. 2009. Dissociated fear and spatial learning in mice with deficiency of ataxin-2. PLoS One . 4: e6235. Google Scholar CrossRef Search ADS PubMed  Isoardi NA, Martijena ID, Carrer HF, Molina VA. 2004. Increased fear learning coincides with neuronal dysinhibition and facilitated LTP in the basolateral amygdala following benzodiazepine withdrawal in rats. Neuropsychopharmacology . 29: 1852– 1864. Google Scholar CrossRef Search ADS PubMed  Kang H, Welcher AA, Shelton D, Schuman EM. 1997. Neurotrophins and time: different roles for TrkB signaling in hippocampal longterm potentiation. Neuron . 19: 653– 664. Google Scholar CrossRef Search ADS PubMed  Katoh-Semba R, Semba R, Takeuchi IK, Kato K. 1998. Age-related changes in levels of brain-derived neurotrophic factor in selected brain regions of rats, normal mice and senescence-accelerated mice: a comparison to those of nerve growth factor and neurotrophin-3. Neurosci Res . 31: 227– 234. Google Scholar CrossRef Search ADS PubMed  Kealy J, Commins S. 2011. The rat perirhinal cortex: a review of anatomy, physiology, plasticity, and function. Prog Neurobiol . 93: 522– 548. Google Scholar CrossRef Search ADS PubMed  Kent BA, Brown TH. 2012. Dual functions of perirhinal cortex in fear conditioning. Hippocampus . 22: 2068– 2079. Google Scholar CrossRef Search ADS PubMed  Kohara K, Yasuda H, Huang Y, Adachi N, Sohya K, Tsumoto T. 2007. A local reduction in cortical GABAergic synapses after a loss of endogenous brain-derived neurotrophic factor, as revealed by single-cell gene knock-out method. J Neurosci . 27: 7234– 7244. Google Scholar CrossRef Search ADS PubMed  Korte M, Carroll P, Wolf E, Brem G, Thoenen H, Bonhoeffer T. 1995. Hippocampal long-term potentiation is impaired in mice lacking brain-derived neurotrophic factor. Proc Natl Acad Sci USA . 92: 8856– 8860. Google Scholar CrossRef Search ADS PubMed  Krause S, Schindowski K, Zechel S, von Bohlen und Halbach O. 2008. Expression of trkB and trkC receptors and their ligands brain-derived neurotrophic factor and neurotrophin-3 in the murine amygdala. J Neurosci Res . 86: 411– 421. Google Scholar CrossRef Search ADS PubMed  Kulisch C, Eckers N, Albrecht D. 2011. Method of euthanasia affects amygdala plasticity in horizontal brain slices from mice. J Neurosci Methods . 201: 340– 345. Google Scholar CrossRef Search ADS PubMed  Laudes T, Meis S, Munsch T, Lessmann V. 2012. Impaired transmission at corticothalamic excitatory inputs and intrathalamic GABAergic synapses in the ventrobasal thalamus of heterozygous BDNF knockout mice. Neuroscience . 222: 215– 227. Google Scholar CrossRef Search ADS PubMed  LeDoux JE. 1995. Emotion: clues from the brain. Annu Rev Psychol . 46: 209– 235. Google Scholar CrossRef Search ADS PubMed  Li C, Dabrowska J, Hazra R, Rainnie DG. 2011. Synergistic activation of dopamine D1 and TrkB receptors mediate gain control of synaptic plasticity in the basolateral amygdala. PLoS One . 6: e26065. Google Scholar CrossRef Search ADS PubMed  Matthies H, Schulz S, Thiemann W, Siemer H, Schmidt H, Krug M, Höllt V. 1997. Design of a multiple slice interface chamber and application for resolving the temporal pattern of CREB phosphorylation in hippocampal long-term potentiation. J Neurosci Methods . 78: 173– 179. Google Scholar CrossRef Search ADS PubMed  McKernan MG, Shinnick-Gallagher P. 1997. Fear conditioning induces a lasting potentiation of synaptic currents in vitro. Nature . 390: 607– 611. Google Scholar CrossRef Search ADS PubMed  Meis S, Bergado-Acosta JR, Yanagawa Y, Obata K, Stork O, Munsch T. 2008. Identification of a neuropeptide S responsive circuitry shaping amygdala activity via the endopiriform nucleus. PLoS One . 3: e2695. Google Scholar CrossRef Search ADS PubMed  Meis S, Endres T, Lessmann V. 2012. Postsynaptic BDNF signalling regulates long-term potentiation at thalamo-amygdala afferents. J Physiol . 590: 193– 208. Google Scholar CrossRef Search ADS PubMed  Musumeci G, Sciarretta C, Rodríguez-Moreno A, Al Banchaabouchi M, Negrete-Díaz V, Costanzi M, Berno V, Egorov AV, von Bohlen Und Halbach O, Cestari V, et al.  . 2009. TrkB modulates fear learning and amygdalar synaptic plasticity by specific docking sites. J Neurosci . 29: 10131– 10143. Google Scholar CrossRef Search ADS PubMed  Ou LC, Gean PW. 2006. Regulation of amygdala-dependent learning by brain-derived neurotrophic factor is mediated by extracellular signal-regulated kinase and phosphatidylinositol-3-kinase. Neuropsychopharmacology . 31: 287– 296. Google Scholar CrossRef Search ADS PubMed  Ou LC, Yeh SH, Gean PW. 2010. Late expression of brain-derived neurotrophic factor in the amygdala is required for persistence of fear memory. Neurobiol Learn Mem . 93: 372– 382. Google Scholar CrossRef Search ADS PubMed  Pan BX, Dong Y, Ito W, Yanagawa Y, Shigemoto R, Morozov A. 2009a. Selective gating of glutamatergic inputs to excitatory neurons of amygdala by presynaptic GABAb receptor. Neuron . 61: 917– 929. Google Scholar CrossRef Search ADS PubMed  Pan BX, Ito W, Morozov A. 2009b. Divergence between thalamic and cortical inputs to lateral amygdala during juvenile-adult transition in mice. Biol Psychiatry . 66: 964– 971. Google Scholar CrossRef Search ADS PubMed  Pape HC, Pare D. 2010. Plastic synaptic networks of the amygdala for the acquisition, expression, and extinction of conditioned fear. Physiol Rev . 90: 419– 463. Google Scholar CrossRef Search ADS PubMed  Park H, Poo MM. 2013. Neurotrophin regulation of neural circuit development and function. Nat Rev Neurosci . 14: 7– 23. Google Scholar CrossRef Search ADS PubMed  Paxinos G, Franklin KBJ. 2001. The mouse brain in stereotaxic coordinates . San Diego: Academic Press. Penzo MA, Robert V, Tucciarone J, De Bundel D, Wang M, Van Aelst L, Darvas M, Parada LF, Palmiter RD, He M, et al.  . 2015. The paraventricular thalamus controls a central amygdala fear circuit. Nature . 519: 455– 459. Google Scholar CrossRef Search ADS PubMed  Pitkänen A, Savander V, LeDoux JE. 1997. Organization of intra-amygdaloid circuitries in the rat: an emerging framework for understanding functions of the amygdala. Trends Neurosci . 20: 517– 523. Google Scholar CrossRef Search ADS PubMed  Psotta L, Lessmann V, Endres T. 2013. Impaired fear extinction learning in adult heterozygous BDNF knock-out mice. Neurobiol Learn Mem . 103: 34– 38. Google Scholar CrossRef Search ADS PubMed  Rammes G, Steckler T, Kresse A, Schütz G, Zieglgänsberger W, Lutz B. 2000. Synaptic plasticity in the basolateral amygdala in transgenic mice expressing dominant-negative cAMP response element-binding protein (CREB) in forebrain. Eur J Neurosci . 12: 2534– 2546. Google Scholar CrossRef Search ADS PubMed  Rattiner LM, Davis M, Ressler KJ. 2005. Brain-derived neurotrophic factor in amygdala-dependent learning. Neuroscientist . 11: 323– 333. Google Scholar CrossRef Search ADS PubMed  Rescorla RA, Wagner AR. 1972. A theory of Pavlovian conditioning: variations in the effectiveness of reinforcement and non-reinforcement. In: Black AH, Prokasy WF, editors. Classical conditioning II: Theory and research . New York: Appleton-Century-Crofts. p. 64– 99. Rioult-Pedotti MS, Friedman D, Donoghue JP. 2000. Learning-induced LTP in neo-cortex. Science . 290: 533– 536. Google Scholar CrossRef Search ADS PubMed  Rodrigues SM, LeDoux JE, Sapolsky RM. 2009. The influence of stress hormones on fear circuitry. Annu Rev Neurosci . 32: 289– 313. Google Scholar CrossRef Search ADS PubMed  Rogan MT, Stäubli UV, LeDoux JE. 1997. Fear conditioning induces associative long-term potentiation in the amygdala. Nature . 390: 604– 607. Google Scholar CrossRef Search ADS PubMed  Romanski LM, LeDoux JE. 1992. Equipotentiality of thalamo-amygdala and thalamo-cortico-amygdala circuits in auditory fear conditioning. J Neurosci . 12: 4501– 4509. Google Scholar PubMed  Schroeder BW, Shinnick-Gallagher P. 2004. Fear memories induce a switch in stimulus response and signaling mechanisms for long-term potentiation in the lateral amygdala. Eur J Neurosci . 20: 549– 556. Google Scholar CrossRef Search ADS PubMed  Schroeder BW, Shinnick-Gallagher P. 2005. Fear learning induces persistent facilitation of amygdala synaptic transmission. Eur J Neurosci . 22: 1775– 1783. Google Scholar CrossRef Search ADS PubMed  Schulz B, Fendt M, Richardson R, Schnitzler HU. 2004. Temporary inactivation of the perirhinal cortex by muscimol injections block acquisition and expression of fear-potentiated startle. Eur J Neurosci . 19: 713– 720. Google Scholar CrossRef Search ADS PubMed  Schulz-Klaus B, Lessmann V, Endres T. 2013. BDNF-dependent consolidation of fear memories in the perirhinal cortex. Front Behav Neurosci . 7: 205. Google Scholar CrossRef Search ADS PubMed  Sierra-Mercado D, Padilla-Coreano N, Quirk GJ. 2011. Dissociable roles of prelimbic and infralimbic cortices, ventral hippocampus, and basolateral amygdala in the expression and extinction of conditioned fear. Neuropsychopharmacology . 36: 529– 538. Google Scholar CrossRef Search ADS PubMed  Sigurdsson T, Doyère V, Cain CK, LeDoux JE. 2007. Long-term potentiation in the amygdala: a cellular mechanism of fear learning and memory. Neuropharmacology . 52: 215– 227. Google Scholar CrossRef Search ADS PubMed  Silhol M, Bonnichon V, Rage F, Tapia-Arancibia L. 2005. Age-related changes in brain-derived neurotrophic factor and tyrosine kinase receptor isoforms in the hippocampus and hypothalamus in male rats. Neuroscience . 132: 613– 624. Google Scholar CrossRef Search ADS PubMed  Sinai L, Duffy S, Roder JC. 2010. Src inhibition reduces NR2B surface expression and synaptic plasticity in the amygdala. Learn Mem . 17: 364– 371. Google Scholar CrossRef Search ADS PubMed  Sotres-Bayon F, Quirk GJ. 2010. Prefrontal control of fear: more than just extinction. Curr Opin Neurobiol . 20: 231– 235. Google Scholar CrossRef Search ADS PubMed  Tsvetkov E, Carlezon WA, Benes FM, Kandel ER, Bolshakov VY. 2002. Fear conditioning occludes LTP-induced presynaptic enhancement of synaptic transmission in the cortical pathway to the lateral amygdala. Neuron . 34: 289– 300. Google Scholar CrossRef Search ADS PubMed  Vertes RP. 2004. Differential projections of the infralimbic and prelimbic cortex in the rat. Synapse . 51: 32– 58. Google Scholar CrossRef Search ADS PubMed  Von Bohlen und Halbach O, Albrecht D. 2002. Reciprocal connections of the hippocampal area CA1, the lateral nucleus of the amygdala and cortical areas in a combined horizontal slice preparation. Neurosci Res . 44: 91– 100. Google Scholar CrossRef Search ADS PubMed  Von Bohlen und Halbach O. 2010. Involvement of BDNF in age-dependent alterations in the hippocampus. Front Aging Neurosci . 2: 36. Google Scholar PubMed  Watanabe Y, Ikegaya Y, Saito H, Abe K. 1995. Roles of GABAA, NMDA and muscarinic receptors in induction of long-term potentiation in the medial and lateral amygdala in vitro. Neurosci Res . 21: 317– 322. Google Scholar CrossRef Search ADS PubMed  Whitlock JR, Heynen AJ, Shuler MG, Bear MF. 2006. Learning induces long-term potentiation in the hippocampus. Science . 313: 1093– 1097. Google Scholar CrossRef Search ADS PubMed  Wilensky AE, Schafe GE, Kristensen MP, LeDoux JE. 2006. Rethinking the fear circuit: the central nucleus of the amygdala is required for the acquisition, consolidation, and expression of Pavlovian fear conditioning. J Neurosci . 26: 12387– 12396. Google Scholar CrossRef Search ADS PubMed  Yan Q, Rosenfeld RD, Matheson CR, Hawkins N, Lopez OT, Bennett L, Welcher AA. 1997. Expression of brain-derived neurotrophic factor protein in the adult rat central nervous system. Neuroscience . 78: 431– 448. Google Scholar CrossRef Search ADS PubMed  Yoshii A, Constantine-Paton M. 2010. Postsynaptic BDNF-TrkB signaling in synapse maturation, plasticity, and disease. Dev Neurobiol . 70: 304– 322. Google Scholar PubMed  Zakharenko SS, Patterson SL, Dragatsis I, Zeitlin SO, Siegelbaum SA, Kandel ER, Morozov AP. 2003. Presynaptic BDNF required for a presynaptic but not postsynaptic component of LTP at hippocampal CA1-CA3 synapses. Neuron . 39: 975– 990. Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Cerebral Cortex Oxford University Press

The Relation Between Long-Term Synaptic Plasticity at Glutamatergic Synapses in the Amygdala and Fear Learning in Adult Heterozygous BDNF-Knockout Mice

Loading next page...
 
/lp/ou_press/the-relation-between-long-term-synaptic-plasticity-at-glutamatergic-K0LEklDZc0
Publisher
Oxford University Press
Copyright
© The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com
ISSN
1047-3211
eISSN
1460-2199
D.O.I.
10.1093/cercor/bhx032
Publisher site
See Article on Publisher Site

Abstract

Abstract Brain-derived neurotrophic factor (BDNF) heterozygous knockout mice (BDNF+/− mice) show fear learning deficits from 3 months of age onwards. Here, we addressed the question how this learning deficit correlates with altered long-term potentiation (LTP) in the cortical synaptic input to the lateral amygdala (LA) and at downstream intra-amygdala synapses in BDNF+/- mice. Our results reveal that the fear learning deficit in BDNF+/− mice was not paralleled by a loss of LTP, neither at cortical inputs to the LA nor at downstream intra-amygdala glutamatergic synapses. As we did observe early fear memory (30 min after training) in BDNF+/− mice while long-term memory (24 h post-training) was absent, the stable LTP in cortico-LA and downstream synapses is in line with the intact acquisition of fear memories. Ex vivo recordings in acute slices of fear-conditioned wildtype (WT) mice revealed that fear learning induces long-lasting changes at cortico-LA synapses that occluded generation of LTP 4 and 24 h after training. Overall, our data show that the intact LTP in the tested amygdala circuits is consistent with intact acquisition of fear memories in both WT and BDNF+/− mice. In addition, the lack of learning-induced long-term changes at cortico-LA synapses in BDNF+/− mice parallels the observed deficit in fear memory consolidation. amygdala, BDNF, fear conditioning, field recording, long-term potentiation Introduction The neurotrophin brain-derived neurotrophic factor (BDNF) and its corresponding receptor TrkB (tropomyosin-related kinase B) are known to be essential for learning and memory as well as for regulation of synaptic strength and synaptic plasticity (Gottmann et al. 2009; Cowansage et al. 2010; Yoshii and Constantine-Paton 2010; Park and Poo 2013; Edelmann et al. 2014). Several recent studies demonstrated a critical role for BDNF signaling in amygdala-dependent fear learning. Thus, overexpression of a nonfunctional, truncated TrkB receptor (tTrkB) in the amygdala led to impaired fear learning (Rattiner et al. 2005). Furthermore, acute inhibition of BDNF/TrkB signaling blocked the acquisition and prevented the formation of long-term fear memories, depending on the time of application of TrkB inhibitors (Ou and Gean 2006; Ou et al. 2010). Fear learning was also impaired by a point mutation in the Y816F or Y515F phosphorylation site of the TrkB receptor, respectively (Musumeci et al. 2009). Fear learning relies on synaptic plasticity at thalamic and cortical inputs to the dorsal part of the lateral amygdala (LA; Sigurdsson et al. 2007; Ehrlich et al. 2009; Pape and Paré 2010). In addition, recent data demonstrated that synaptic plasticity in the LA depends on intact BDNF/TrkB signaling (Li et al. 2011; Daftary et al. 2012; Meis et al. 2012). In this respect, long-term potentiation (LTP) at thalamic afferents to the LA is prevented by acute inhibition of BDNF/TrkB signaling (Meis et al. 2012). In addition, chronic BDNF deficiency in heterozygous BDNF-knockout mice (BDNF+/− mice) abolished LTP at thalamo-LA synapses already at an age of 1 month. In contrast, LTP at cortico-LA synapses was unaffected (Meis et al. 2012). Interestingly, we could recently demonstrate that these mice exhibit deficits in fear learning when they are 3 months of age or older (Endres and Lessmann 2012). As LTP at thalamo-LA afferents is already impaired in 1-month-old animals while fear conditioning is still functional up to 2 months, intact fear learning at 2 months might be enabled by synaptic plasticity at cortico-LA synapses. However, LTP at this synapse may become incapable to contribute to fear memory storage at older ages. This idea is supported by several studies demonstrating postpubertal changes in synaptic properties of thalamo-LA and cortico-LA synapses as well as in network morphology (e.g., Cunningham et al. 2002, 2008; Pan et al. 2009a; Gambino et al. 2010). To test our hypothesis, we performed in vitro field potential recordings in amygdala slices of adult (≥3 months) BDNF+/− mice and WT littermates. Surprisingly, LTP at cortico-LA synapses, as well as at different intra-amygdala glutamatergic synapses, was unaffected in BDNF+/− mice. Therefore, we analyzed in a second series of experiments whether cued fear learning alters plasticity of cortico-LA synapses. As it has been shown for several brain regions—including the amygdala—learning results in occlusion of LTP at synapses involved in memory formation (see e.g., Rioult-Pedotti et al. 2000; Tsvetkov et al. 2002; Whitlock et al. 2006). Therefore, we tested for occlusion of LTP in slices of previously fear-conditioned mice and indeed observed occlusion of LTP in WT mice 4 and 24 h after fear conditioning. This indicates fear learning-related long-term changes at these synapses. However, this learning-induced occlusion of LTP was absent in BDNF+/− mice at both time points (4 and 24 h). As we observed on the one hand intact LTP and on the other hand a lack of learning-related long-term changes at cortico-LA synapses in BDNF+/− mice, we performed a detailed analysis of the early fear memory (i.e., 0.5–6 h after fear conditioning) in BDNF+/− and WT mice. Here, we observed robust early fear memory also in BDNF+/− mice that gradually declined in precision over time. These results suggest that acquisition of fear memory is still intact in adult BDNF+/− mice, but due to a defective fear memory consolidation no stable long-term fear memory is established in these mice. Overall, our results suggest that cued fear learning results in long-lasting changes at cortico-LA synapses in WT mice. These learning-induced changes in plasticity are absent in BDNF+/− mice 4 h after fear conditioning, probably resulting in the observed deficit in fear memory consolidation. Materials and Methods Animals We used 3- to 5-month-old male BDNF+/− mice (Korte et al. 1995), which were outbred on a C57BL/6 J (Charles River) genetic background. In addition, 2-month-old animals were utilized in 2 sets of experiments as stated in the Results section. Wildtype (WT) littermates served as controls. The animals were housed in groups of 4 animals and had free access to food and water. All experiments were carried out in accordance with the European Committees Council Directive (86/609/EEC) and were approved by the local animal care committee (Landesverwaltungsamt Sachsen Anhalt, IPHY/G/01-872/08 & IPHY/G/01–1191/13). Slice Preparation Standard procedures were used to prepare coronal slices from male BDNF+/− mice or their WT littermates, respectively. Mice at postnatal days P86–P140 were deeply anesthetized by inhalation of 4% isofluran (Kulisch et al. 2011) and killed by decapitation. A block of tissue containing the amygdala was rapidly removed and placed in standard artificial cerebrospinal fluid (ACSF) containing (in mM): NaCl, 125; KCl, 2.5; NaH2PO4, 0.8; NaHCO3, 25; MgCl2, 1; CaCl2, 2; glucose, 10; bubbled with 95% O2/5% CO2. Coronal slices (400 μm thick) were prepared on a vibratome (Model 1000, The Vibratome Company), and were incubated in ACSF in an interface chamber for at least 2 h before recording. Field Potential Recordings Field potential recordings were performed in an interface chamber at 32 ± 1 °C (Matthies et al. 1997). Recording pipettes were pulled from borosilicate glass (GC150TF-10, Clark Electromedical Instruments), filled with ACSF (3–4 MΩ) and positioned in the LA (see insets in respective figures). A concentric bipolar electrode (FHC Inc.) was placed on the surface of the slice above the external capsule when stimulating cortico-LA synapses. For assessment of LTP at intra-amygdala afferents, the stimulation electrode was positioned within the basal amygdala, and the recording pipette was placed in the basal or central medial amygdala (CeA). For ex vivo recordings in fear-conditioned or pseudoconditioned mice (see below), animals were sacrificed 2 or 24 h after fear conditioning, and recordings started 4–8 h or 26–30 h after training, respectively. Field potentials were evoked by stimuli of 100 µs duration delivered by a stimulus isolator (Isoflex, AMPI) at 0.016 Hz. Stimulus intensity was adjusted to evoke responses of halfmaximal amplitude. Signals were amplified by a DAM-80 amplifier (WPI) and digitized with a CED 1401plus interface (Cambridge Electronic Design), controlled by a custom-made software (Reymann and Frey, LIN Magdeburg). Since in the amygdala the recorded field potentials consist of a summation of excitatory postsynaptic potentials (EPSPs) and synchronized action potentials, the analysis of the field potential amplitudes, instead of slopes, is more reliable. In addition, EPSP amplitudes are less sensitive to variability and noise than the slope (compare discussion in Drephal et al. 2006). Therefore, in keeping with many other LTP studies in the amygdala (see e.g., Watanabe et al. 1995; McKernan and Shinnick-Gallagher 1997; Rogan et al. 1997; Doyère et al. 2003) we also analyzed EPSP amplitudes in our recordings. Signal amplitude was measured as the sum of 1) the difference between onset and peak of the negative voltage deflection and 2) the difference of the peak of the negative voltage deflection and the succeeding positive peak, divided by 2 (Drephal et al. 2006; Kulisch et al. 2011). To induce LTP, we tested different types of stimulation patterns because, based on previous literature, specific induction paradigms may rely on distinct signaling pathways (Huang et al. 2000; Bauer et al. 2002). A high-frequency stimulation (HFS) pattern was composed of 3 trains of 100 stimuli at 100 Hz separated by 30 s. A second type of HFS consisted of 4 trains of 100 stimuli at 100 Hz separated by 5 min. Theta-burst stimulation (TBS) comprised 2 trains of 4 stimuli at 100 Hz, repeated 10 times at 5 Hz, separated by 20s. Stimulation protocols were executed at time point zero. For comparison of LTP between genotypes, LTP was quantified by normalizing and averaging field potentials during the last 5 min of experiments (i.e., 55–60 min or 115–120 min after LTP induction) relative to 30 min baseline. For statistical analysis of successful LTP induction, we compared averaged field potential amplitudes 5 min before LTP induction with the respective amplitudes during the last 5 min of recordings. Patch-Clamp Recordings Whole-cell patch-clamp recordings of IPSCs were done as described previously (Meis et al. 2008). Briefly, single slices were transferred to a submerged chamber. Recordings were made using a patch-clamp amplifier (EPC-9, Heka). Patch pipettes were pulled from borosilicate glass (GC150TF-10, Clark Electromedical Instruments) to resistances of 2–3 MΩ, and filled with (in mM): Csgluconate, 107; CsCl, 13; MgCl2, 1; CaCl2, 0.07; EGTA, 11; HEPES, 10; MgATP, 3, NaGTP, 0.5 (pH 7.2 with KOH). A liquid junction potential of 10 mV of the pipette solution was corrected for. After obtaining the whole-cell configuration, neurons were held at 0 mV. Drugs All chemicals were obtained from Sigma, except for K252a (Alomone). Assessment of Fear Learning For fear conditioning, we used an automated fear conditioning setup (TSE-Systems). The animals were placed in a cubic box (23 × 23 cm2), located in a sound attenuating chamber. The floor of the test box consisted of a grid floor, by which the unconditioned stimulus (US, 1 s, 0.7 mA, scrambled foot shock) was delivered. The conditioned stimulus (CS, 30 s, 8 kHz sine tone, 70 dB sound pressure level [SPL]) was presented by a loudspeaker located at the ceiling of the test box. An array of infrared light beams around the arena enabled tracking the activity of the animals. To provide different contexts between fear conditioning and fear memory retrieval tests, the color of the test boxes (black or transparent) as well as the cleaning agents (70% ethanol or Deskosept; Dr Schumacher GmbH) were randomly changed to compose distinct contextual environments. For fear conditioning training, the animals were permitted to explore the test chamber for the first 2 min to become habituated. Then, the CS and US were presented 3 times with random interstimulus intervals (90–240 s) in a paired manner, that is, the CS (30 s) coterminated with the US (1 s). This protocol is identical to the one used in our previous studies, resulting in impaired fear learning in BDNF+/− at 3 months of age and beyond (Endres and Lessmann 2012). As a control for the occlusion experiments, we presented the CS and US in a random unpaired manner. To assess the cued fear memory of the animals, we exposed the animals 5 times to the CS in a different context (as described above). Besides analyzing freezing behavior during the CS periods (5× 30 s), we also quantified the average freezing during the 30-s periods before each of the 5 CS presentations (pre-CS freezing). In order to get more detailed insights into the characteristics of the learning deficit, we performed the fear memory test in different groups at different time points, that is, 0.5, 2, 4, 6, or 24 h, after fear conditioning. To quantify the precision of early fear memory, we performed a tone frequency-dependent discriminative fear learning task. To this aim, one tone (CS−, 30 s, 2 kHz sine tone, 75 dB SPL) was presented in the absence of a coterminating foot shock, while another tone (CS+, 30 s, 8 kHz sine tone, 75 dB SPL) was followed by a foot shock (1 s, 0.7 mA). Both CS+ and CS− were presented 3 times at random sequence during the conditioning session with random interstimulus intervals. Three to four hours later, we tested the fear memory by presenting CS+ and CS− in a random sequence in a novel context (identical to the classical fear conditioning procedure). Data Analysis Nonparametric data were analyzed by Wilcoxon signed rank test, or Mann–Whitney U-test, by using Graph Pad Prism software. Normally distributed data were analyzed by performing an analysis of variance (ANOVA, JMP 8, SAS Institute), followed by post hoc Tukey comparisons. All electrophysiological recordings were normalized to average baseline amplitudes and analyzed with Origin 8.0. (OriginLab Corporation). To statistically analyze the fear conditioning experiments, repeated measure ANOVAs were performed using “phase of the experiment” (habituation, pre-CS, CS, or pre-CS+, pre-CS−, CS+, CS−) as within-subject factor and “genotype” or “time of testing” as between-subject factors. All data are presented as mean ± standard error of the mean (SEM). Differences were considered statistically significant at P < 0.05. For electrophysiological recordings, the lowercase n (“n”) indicates the number of slices tested, while the capital n (“N”) refers to the number of animals from which these slices were obtained from. Results HFS Induced LTP in Cortico-LA Afferents To correlate LTP at LA inputs with fear learning, we focused in this study on LTP at cortico-LA synapses in adult mice (≥3 months). LTP was induced by an HFS pattern composed of 3 trains of 100 stimuli at 100 Hz, separated by 30 s. In slices from adult WT mice, the average field potential amplitude was significantly increased 60 min after the induction when compared with pretetanus values (125.5 ± 9.5%, n = 12 slices from N = 7 mice; P = 0.009; Fig. 1A). A similar significant increase in field potential amplitudes in response to LTP induction was observed in BDNF+/− mice of analogous age (114.9 ± 7.6%, n = 14 slices/N = 10 mice, P = 0.02). This potentiation in BDNF+/− mice was not significantly different from WT slices (P = 0.2472, Fig. 1A). Thus, HFS-induced LTP at this input structure is not hampered in adult BDNF+/− mice. To verify that the LTP paradigm we used was nevertheless dependent on BDNF signaling, we performed a control experiment in WT mice (≥ 3 months) in the continuous presence of the tyrosine kinase inhibitor K252a (100 nM, preincubation time ≥2 h). In WT slices exposed to the same concentration of DMSO as used to dissolve K252a (0.1%), the average field potential amplitude was significantly increased 60 min after LTP induction (160.4 ± 9.7%, n = 8 slices/N = 5 mice; P = 0.0078; Fig. 1B). In contrast, LTP was not induced in the presence of K252a (97.1 ± 2.9%, n = 9 slices/N = 6 mice; P = 0.2031). The difference between the 2 data sets was highly significant (P < 0.0001, Fig. 1B). Likewise, LTP was also prevented in the presence of K252a in BDNF+/− mice (P = 0.0043, Supplementary Fig. 1). These results suggest that the LTP induced by our tetanic stimulation is dependent on BDNF/TrkB signaling. However, BDNF levels in BDNF+/− mice do not seem to be reduced to a level that prevents LTP expression. Figure 1. View largeDownload slide BDNF dependency of HFS-LTP at cortico-LA synapses in adult BDNF+/− mice. LTP was induced by 3 trains of 100 stimuli at 100 Hz separated by 30 s intervals. (A) Time course of averaged evoked field potentials in response to stimulation of cortical afferents in all slices recorded from WT and BDNF+/− mice. In both genotypes, LTP could be reliably induced. Insets at the right depict averaged field potential recording traces 5 min before LTP induction and during the last 5 min of recordings for WT and BDNF+/− mice, for 2 representative individual slices. WT: n = 12 slices from N = 7 animals, BDNF+/−: n = 14 slices from N = 10 animals, ns: not significantly different. (B) Time course of averaged evoked field potentials in response to stimulation of cortico-LA afferents in slices obtained from WT animals with and without application of the tyrosine kinase inhibitor K252a, labeled as white and black data points, respectively. LTP was abolished by application of 100 nM K252a (preincubation ≥2 h). Insets at the right depict averaged field potential recording traces 5 min before LTP induction and during the last 5 min of recordings, for 2 representative individual slices. Control, 0.1% DMSO: n = 8 slices from N = 5 animals, K252a: n = 9 slices from N = 6 animals, ***: significant, P < 0.0001. Insets at the left depict positions of recording and stimulation electrodes (modified from Paxinos and Franklin 2001). Figure 1. View largeDownload slide BDNF dependency of HFS-LTP at cortico-LA synapses in adult BDNF+/− mice. LTP was induced by 3 trains of 100 stimuli at 100 Hz separated by 30 s intervals. (A) Time course of averaged evoked field potentials in response to stimulation of cortical afferents in all slices recorded from WT and BDNF+/− mice. In both genotypes, LTP could be reliably induced. Insets at the right depict averaged field potential recording traces 5 min before LTP induction and during the last 5 min of recordings for WT and BDNF+/− mice, for 2 representative individual slices. WT: n = 12 slices from N = 7 animals, BDNF+/−: n = 14 slices from N = 10 animals, ns: not significantly different. (B) Time course of averaged evoked field potentials in response to stimulation of cortico-LA afferents in slices obtained from WT animals with and without application of the tyrosine kinase inhibitor K252a, labeled as white and black data points, respectively. LTP was abolished by application of 100 nM K252a (preincubation ≥2 h). Insets at the right depict averaged field potential recording traces 5 min before LTP induction and during the last 5 min of recordings, for 2 representative individual slices. Control, 0.1% DMSO: n = 8 slices from N = 5 animals, K252a: n = 9 slices from N = 6 animals, ***: significant, P < 0.0001. Insets at the left depict positions of recording and stimulation electrodes (modified from Paxinos and Franklin 2001). TBS Induced LTP in Cortico-LA Synapses Notably, a requirement for BDNF in LTP induction can depend on the specific LTP paradigm in use (Kang et al. 1997; Zakharenko et al. 2003; Abidin et al. 2006; Edelmann et al. 2015). Thus, to determine whether other types of LTP at cortico-LA synapses were affected in BDNF+/− mice, we performed an additional series of LTP experiments using TBS. The TBS protocol consisted of 2 trains (separated by 20 s) of 4 stimuli at 100 Hz, repeated 10 times at 5 Hz. The magnitude of LTP induced by this protocol in 2-month-old animals was indistinguishable between WT and BDNF+/− mice (P = 0.1810; WT: 113.6 ± 3.8%, n = 9 slices/N = 7 mice; BDNF+/−: 123.9 ± 9.4%, n = 6 slices/N = 5 mice). However, at older ages (≥3 months), this paradigm did not induce LTP irrespective of genotype (WT: 107.6 ± 5.6%, n = 9/N = 8 mice, P = 0.1289; BDNF+/−: 107.4 ± 9.7%, n = 9/N = 7 mice, P = 0.5703, Supplementary Fig. 2). Thus, in mice older than 3 months, LTP at cortico-LA synapses was induced more efficiently by trains of 100 Hz stimulation than by TBS. HFS Induced LTP in Cortico-LA Afferents in the Presence of Gabazine We observed unimpaired LTP at glutamatergic cortico-LA synapses in BDNF+/− mice when GABAergic inhibition was intact (Fig. 1A). However, chronic reduction of BDNF in BDNF+/− mice is known to reduce GABAergic inhibition in different brain areas (Kohara et al. 2007; Abidin et al. 2008; Laudes et al. 2012). Therefore, a potential impairment of LTP at glutamatergic cortical afferents to the LA could be compensated by reduced GABAergic inhibition in BDNF+/− mice. To test this hypothesis, recordings were performed in the presence of the specific GABAA receptor antagonist gabazine. As blockade of GABAergic inhibition was described to strongly enhance excitability in amygdala slices (Gean and Shinnick-Gallagher 1987; Isoardi et al. 2004; Huang and Kandel 2007), we applied gabazine at a low nonsaturating concentration (0.1 µM), which effectively prevented epileptiform activity. As assessed by patch-clamp recordings, this concentration reduced evoked inhibitory postsynaptic current amplitudes to 51.1 ± 3.1% of the control value before drug addition (n = 6, see Supplementary Fig. 3). In the presence of gabazine, LTP elicited by HFS amounted to 133.6 ± 12.7% (n = 8/N = 5 mice; P = 0.0078; Fig. 2A) in WT slices, and to 124.0 ± 10.4% (n = 11/N = 6 mice, P = 0.0137; Fig. 2A), in slices from BDNF+/− mice. The magnitude of LTP was not significantly different between WT and BDNF+/− mice, respectively (P = 0.3860). Therefore, a parallel decline of GABAergic inhibition due to chronic BDNF reduction cannot explain the unaffected LTP at cortico-LA synapses in BDNF+/− mice. Figure 2. View largeDownload slide Intact LTP at cortico-LA synapses in adult BDNF+/− mice in the presence of 0.1 µM gabazine. (A) LTP was induced by 3 trains of 100 stimuli at 100 Hz separated by 30 s or (B) by 4 trains of 100 stimuli at 100 Hz separated by 300 s. (A, B) Time course of averaged evoked field potential amplitudes in response to stimulation of cortical afferents in all slices recorded from WT and BDNF+/− mice, respectively. Note that LTP was induced independently of genotype. Insets at the right depict averaged field potential recordings 5 min before LTP induction and during the last 5 min of recordings for WT and BDNF+/− mice, respectively. (A) WT: n = 8 slices from N = 5 animals, BDNF+/−: n = 11 slices from N = 6 animals, ns, not significant. (B) WT: n = 11 slices from N = 9 animals, BDNF+/−: n = 9 slices from N = 6 animals, ns: not significantly different. Figure 2. View largeDownload slide Intact LTP at cortico-LA synapses in adult BDNF+/− mice in the presence of 0.1 µM gabazine. (A) LTP was induced by 3 trains of 100 stimuli at 100 Hz separated by 30 s or (B) by 4 trains of 100 stimuli at 100 Hz separated by 300 s. (A, B) Time course of averaged evoked field potential amplitudes in response to stimulation of cortical afferents in all slices recorded from WT and BDNF+/− mice, respectively. Note that LTP was induced independently of genotype. Insets at the right depict averaged field potential recordings 5 min before LTP induction and during the last 5 min of recordings for WT and BDNF+/− mice, respectively. (A) WT: n = 8 slices from N = 5 animals, BDNF+/−: n = 11 slices from N = 6 animals, ns, not significant. (B) WT: n = 11 slices from N = 9 animals, BDNF+/−: n = 9 slices from N = 6 animals, ns: not significantly different. It could be argued that the LTP protocol we used so far was not suitable to detect differences between genotypes. Therefore, we next tested a paradigm similar to the one used by Huang et al. (2000), which is known to induce protein synthesis-dependent late-LTP in the LA. This LTP paradigm consisted of 4 trains of 100 stimuli at 100 Hz separated by 5 min, and led to significant LTP as assessed 2 h after tetanization (WT: 123.6 ± 5.9%, n = 11 slices/N = 9 mice, P = 0.001; BDNF+/−: 121.5 ± 7.4%, n = 9 slices/N = 6 mice, P = 0.0273; Fig. 2B). The 2 genotypes did not show a difference in the magnitude of LTP (P = 0.8197). These results revealed intact LTP in response to distinct types of HFS at cortico-LA inputs in adult BDNF+/− mice (≥3 months), yet their fear learning was impaired. Thus, LTP at cortico-LA afferents does not enable formation of fear memory in BDNF+/− mice. This suggested that synaptic plasticity at intra-amygdala synapses might be impaired in BDNF+/− mice, and thereby contribute to reduced fear memory learning. Therefore, we tested LTP at these intra-amygdala synapses (see below). HFS Induced LTP at Intra-Amygdala Synapses We first focused on the basal amygdala (BL) as an intra-amygdala target of the LA (Pitkänen et al. 1997) which is implicated in fear learning (Amano et al. 2011 and references therein). Recordings were again performed in the presence of 0.1 µM gabazine. The stimulation electrode was positioned in the LA. Interestingly, we found similar magnitudes of LTP at these glutamatergic synapses in the BL (WT: 153.1 ± 15.3%, n = 9 slices/N = 8 mice, P = 0.0039; BDNF+/−: 153.1 ± 27.6%, n = 5 slices/N = 5 mice, P = 0.0079) in WT and BDNF+/− mice (P = 1.0; Fig. 3A). Figure 3. View largeDownload slide Intact HFS-LTP in distinct intra-amygdala circuits in adult BDNF+/− mice. (A, B) LTP was induced by 4 trains of 100 stimuli at 100 Hz separated by 300 s. Time course of averaged evoked field potential amplitudes in response to stimulation of intrabasal (A) or medial central amygdala afferents (B) in all slices recorded from WT and BDNF+/− mice. Note that LTP was successfully induced independent of genotype. Insets at the right depict averaged field potential traces 5 min before LTP induction and during the last 5 min of recordings for WT and BDNF+/− mice. (A) WT: n = 9 slices from N = 8 animals, BDNF+/−: n = 5 slices from N = 5 animals, ns, not significant. (B) WT: n = 4 slices from N = 3 animals, BDNF+/−: n = 5 slices from N = 3 animals, ns: not significantly different. Figure 3. View largeDownload slide Intact HFS-LTP in distinct intra-amygdala circuits in adult BDNF+/− mice. (A, B) LTP was induced by 4 trains of 100 stimuli at 100 Hz separated by 300 s. Time course of averaged evoked field potential amplitudes in response to stimulation of intrabasal (A) or medial central amygdala afferents (B) in all slices recorded from WT and BDNF+/− mice. Note that LTP was successfully induced independent of genotype. Insets at the right depict averaged field potential traces 5 min before LTP induction and during the last 5 min of recordings for WT and BDNF+/− mice. (A) WT: n = 9 slices from N = 8 animals, BDNF+/−: n = 5 slices from N = 5 animals, ns, not significant. (B) WT: n = 4 slices from N = 3 animals, BDNF+/−: n = 5 slices from N = 3 animals, ns: not significantly different. Consequently, we addressed LTP at subsequent glutamatergic synapses in the amygdala network. Besides the LA, the central nucleus of the amygdala (CeA) may be involved in acquisition and consolidation of fear memories (Wilensky et al. 2006). It is known that the medial part of the central nucleus is the main output of the amygdala and activates brainstem projections leading to the expression of fear responses (Pape and Pare 2010). Therefore, we analyzed synaptic plasticity in this structure. However, LTP in the medial central amygdala that was elicited by HFS of the BL was not affected in BDNF+/− animals (WT 165.5 ± 13.1%, n = 4 slices/N = 3 mice, BDNF+/− 153.8 ± 13.0%, n = 5 slices/N = 3 mice). Both genotypes showed significant potentiation compared with pretetanus values (WT P = 0.0286, BDNF+/−P = 0.0079) but no group differences (P = 0.7302, Fig. 3B). Field potential amplitudes were not significantly different between WT and BDNF+/− at any time point after HFS-induced LTP. Taken together, these results indicate intact LTP at intra-amygdala glutamatergic synapses. Altered synaptic plasticity at these synapses can therefore not account for the impaired fear learning in adult BDNF+/− mice. Occlusion of Fear Learning Induced Synaptic Changes and LTP at Cortico-LA Afferents Taking the unaltered LTP in adult BDNF+/− mice into account (≥3 months old), the question arises whether our LTP paradigm at cortico-LA afferents was actually relevant for fear learning. Therefore, LTP was tested ex vivo in slices of cued fear-conditioned WT mice. As control, we established a pseudoconditioning paradigm in which the animals received the same number of tone and foot shock presentations, but in an unpaired manner. To verify that this pseudoconditioning does not lead to any cued fear learning, we first compared this paradigm with our established fear conditioning paradigm (Fig. 4). During the conditioning, we observed a significant increase in freezing behavior in both groups (Fig. 4A; F20,356 = 8.8, P < 0.0001). This similar increase in freezing in the pseudoconditioned animals was expected, as they received the same number of foot shocks during the training session as the fear-conditioned animals. No differences in either the magnitude or the time course of freezing were observed between the 2 groups (factor training type [pseudo vs. trained]: F1356 = 0.0, P = 1; interaction of the factors training type × time: F20,356 = 1.0, P = 0.49). One day later, we tested the fear memory of the animals by presenting the CS in a novel context (Fig. 4B). Here, we observed a clear difference between the 2 groups: while the trained animals showed an obvious increase in freezing upon CS presentation, the pseudoconditioned animals did not. The statistical analysis revealed strong effects for the factors training type (F1,59 = 21.8, P < 0.0001) and phase (F2,59 = 50.6, P < 0.0001) as well as for the interaction of these 2 factors (F2,59 = 19.7, P < 0.0001). Post hoc Tukey comparisons revealed a significant increase in freezing only in the trained animals during the CS presentation, while in the pseudoconditioned mice there was no difference between the freezing behavior expressed in the different phases. This experiment demonstrates that our paradigm reliably induces fear behavior. In contrast, pseudoconditioning does not induce any cued fear learning, although freezing during training is similar between the 2 groups. Figure 4. View largeDownload slide Comparison of fear conditioning with the pseudoconditioning paradigm. To establish an appropriate control group for fear-conditioned animals in the ex vivo occlusion experiments, we designed a pseudoconditioning paradigm in which the animals received the same number of tone and foot shock presentations as the fear conditioning group, but in an explicitly unpaired manner. During the fear conditioning training (A), both groups showed a similar increase in freezing, most probably due to the foot shock presentations. In the fear retrieval test (B), which was performed 24 h later in a novel context, only animals that underwent the fear conditioning protocol (“trained”) showed an increased freezing behavior upon CS presentation. Pseudoconditioned animals showed a very low level of freezing throughout the experiment that remained in the range of the freezing during the habituation period. (*) indicates significant differences between trained and pseudoconditioned animals, as well as between CS and pre-CS and habituation. Figure 4. View largeDownload slide Comparison of fear conditioning with the pseudoconditioning paradigm. To establish an appropriate control group for fear-conditioned animals in the ex vivo occlusion experiments, we designed a pseudoconditioning paradigm in which the animals received the same number of tone and foot shock presentations as the fear conditioning group, but in an explicitly unpaired manner. During the fear conditioning training (A), both groups showed a similar increase in freezing, most probably due to the foot shock presentations. In the fear retrieval test (B), which was performed 24 h later in a novel context, only animals that underwent the fear conditioning protocol (“trained”) showed an increased freezing behavior upon CS presentation. Pseudoconditioned animals showed a very low level of freezing throughout the experiment that remained in the range of the freezing during the habituation period. (*) indicates significant differences between trained and pseudoconditioned animals, as well as between CS and pre-CS and habituation. After verifying that our 2 conditioning paradigms resulted in cued fear learning to the tone in fear-conditioned animals but absence of this cued fear learning in pseudoconditioned animals, we trained animals of both genotypes with the respective protocols and tested for occlusion of LTP at cortico-LA synapses. Of note, every CS presentation in the absence of the US results in extinction learning (Rescorla and Wagner 1972). Therefore, LTP was recorded in animals without a preceding fear retrieval test to avoid effects of synaptic changes linked to fear extinction learning. Similar to the previous experiment (compare Fig. 4A), we observed a general increase in freezing (Fig. 5A, B) upon the foot shock presentations in all groups (factor “time”: F20,1112 = 22,3, P < 0.0001). Importantly, there were no differences between genotypes (F1,1112 = 0.02, P = 0.9) or training paradigms (F1,1112 = 0.02, P = 0.89) as well as no interaction of these 3 factors (time × genotype × training paradigm: F20,1112 = 0.56, P = 0.94). These behavioral data verified similar conditions for both groups in response to the training session that preceded the subsequent electrophysiological recordings. Figure 5. View largeDownload slide Occlusion experiments in WT and BDNF+/−mice at 24 h (C, D) and 2 h (E, F) after fear conditioning. (A, B) Animals of both genotypes were either fear conditioned (“trained”) or served as a pseudoconditioned control (“pseudo”). During the conditioning training, all groups showed a comparable increase in freezing behavior (WT pseudoconditioned n = 23, trained n = 24; BDNF+/− pseudoconditioned n = 14, trained n = 24). (C, D) Slice recordings were performed 24 h after training. LTP was induced by 3 trains of 100 stimuli at 100 Hz separated by 30 s. Time course of averaged evoked field potentials in response to stimulation of cortico-LA afferents in all slices recorded from WT (C) and BDNF+/− mice (D). Insets at the right depict averaged field recordings 5 min before LTP induction and during the last 5 min of recordings for mice with unpaired or paired training. In WT animals, LTP was significantly reduced in fear-conditioned mice (trained) as opposed to mice subjected to unpaired training (pseudoconditioning). In BDNF+/− mice, LTP was unchanged by paired or nonpaired CS–US presentations. (E, F) Slice recordings were performed at least 4 h after training. In WT mice (E), LTP was significantly reduced in fear-conditioned mice (trained) as opposed to WT mice subjected to unpaired training (pseudoconditioning). In BDNF+/− mice (F), LTP was unchanged by paired or nonpaired CS–US presentations. (C) WT mice: Paired training, n = 9 slices from N = 6 animals, pseudoconditioning, n = 10 slices from N = 6 animals, *: significant, P = 0.0435. (D) BDNF+/− mice: Paired training, n = 12 slices from N = 5 animals, pseudoconditioning, n = 7 slices from N = 3 animals, ns: not significant, P = 0.4726. (E) WT mice: Paired training, n = 15 slices from N = 10 animals, pseudoconditioning, n = 13 slices from N = 10 animals, *: significant, P = 0.0304. (F) BDNF+/− mice: Paired training, n = 12 slices from N = 9 animals, pseudoconditioning, n = 7 slices from N = 6 animals, ns: not significant, P = 0.5828. Figure 5. View largeDownload slide Occlusion experiments in WT and BDNF+/−mice at 24 h (C, D) and 2 h (E, F) after fear conditioning. (A, B) Animals of both genotypes were either fear conditioned (“trained”) or served as a pseudoconditioned control (“pseudo”). During the conditioning training, all groups showed a comparable increase in freezing behavior (WT pseudoconditioned n = 23, trained n = 24; BDNF+/− pseudoconditioned n = 14, trained n = 24). (C, D) Slice recordings were performed 24 h after training. LTP was induced by 3 trains of 100 stimuli at 100 Hz separated by 30 s. Time course of averaged evoked field potentials in response to stimulation of cortico-LA afferents in all slices recorded from WT (C) and BDNF+/− mice (D). Insets at the right depict averaged field recordings 5 min before LTP induction and during the last 5 min of recordings for mice with unpaired or paired training. In WT animals, LTP was significantly reduced in fear-conditioned mice (trained) as opposed to mice subjected to unpaired training (pseudoconditioning). In BDNF+/− mice, LTP was unchanged by paired or nonpaired CS–US presentations. (E, F) Slice recordings were performed at least 4 h after training. In WT mice (E), LTP was significantly reduced in fear-conditioned mice (trained) as opposed to WT mice subjected to unpaired training (pseudoconditioning). In BDNF+/− mice (F), LTP was unchanged by paired or nonpaired CS–US presentations. (C) WT mice: Paired training, n = 9 slices from N = 6 animals, pseudoconditioning, n = 10 slices from N = 6 animals, *: significant, P = 0.0435. (D) BDNF+/− mice: Paired training, n = 12 slices from N = 5 animals, pseudoconditioning, n = 7 slices from N = 3 animals, ns: not significant, P = 0.4726. (E) WT mice: Paired training, n = 15 slices from N = 10 animals, pseudoconditioning, n = 13 slices from N = 10 animals, *: significant, P = 0.0304. (F) BDNF+/− mice: Paired training, n = 12 slices from N = 9 animals, pseudoconditioning, n = 7 slices from N = 6 animals, ns: not significant, P = 0.5828. Ex vivo LTP recordings in these animals were performed 1 day after training (Fig. 5C–F, see inset). LTP in cortico-LA afferents (pseudoconditioned control: 140.0 ± 9.5%, n = 10 slices/N = 6 mice, P = 0.002; trained: 111.6 ± 4.8%, n = 9/N = 6 mice, P = 0.0273) was significantly occluded in fear-conditioned WT animals compared with pseudoconditioned control mice (P = 0.0435; Fig. 5C). This occlusion is consistent with the hypothesis that cued fear learning and LTP at cortico-LA afferents induced by our protocol involve common synaptic mechanisms. In contrast, LTP in BDNF+/− mice (pseudoconditioned control: 116 ± 4.4%, n = 7/N = 3 mice, P = 0.0156; trained: 112.4 ± 5.8%, n = 12/N = 5 mice, P = 0.0476) was not occluded in fear-conditioned animals compared with pseudoconditioned control mice (P = 0.4726; Fig. 5D). Thus, fear learning seems to induce long-term changes at cortico-LA synapses in WT mice as indicated by the occlusion of LTP 24 h after training. These learning-induced long-term changes are absent in BDNF+/− mice, which accordingly do not show fear memory 24 h after training. To get further insights into the dynamics of these long-term synaptic changes, we tested occlusion of cortico-LA LTP at earlier time points after conditioning training. Of note, behavioral testing by itself, for example, due to increased stress by additional handling or the application of foot shocks, may affect subsequent measurements of synaptic plasticity. Therefore, we first tested suitable delays between end of pseudoconditioning and preparation of brain slices that allowed induction of LTP in WT mice. LTP recordings started always at least 2 h after slice preparation. Under these conditions neither field recordings with 20 min (109.2 ± 9.4%, n = 8/N = 3 mice, P = 0.7422) nor 60 min delay between pseudoconditioning and preparation of slices (113.9 ± 8.2%, n = 7/N = 3 mice, P = 0.0781) showed LTP. In contrast, stable LTP was expressed when mice were sacrificed 2 h after pseudoconditioning and recordings started at the earliest 4 h after the training procedure. When using this time window, LTP was successfully induced in pseudoconditioned WT mice (control: 122.3 ± 6.4%, n = 13/N = 10 mice, P = 0.0012; trained: 104.5 ± 2.8%, n = 15/N = 10 mice, P = 0.0413, Fig. 5E) as well as BDNF+/− mice (control: 116.7 ± 6.5%, n = 7/N = 6 mice, P = 0.0313; trained: 111.1 ± 4.5%, n = 12/N = 9 mice, P = 0.0122; Fig. 5F). Notably, LTP was occluded in WT mice (P = 0.030 Fig. 5E), but not in BDNF+/− mice (P = 0.5828, Fig. 5F). These data suggest that synaptic plasticity at cortico-LA synapses is relevant for early fear memory consolidation in WT mice, that is, ≤4 h after training. However, these synaptic changes are absent in BDNF+/− mice probably contributing to the declining fear memory in BDNF+/− mice, that initially (i.e., 30 min after training) show intact fear memory acquisition. Assessment of Early Fear Memories To get more detailed insights into the characteristics of the fear learning deficit in BDNF+/− mice, we tested the early fear memory in different cohorts at distinct time points between 30 min and 24 h after fear conditioning (Fig. 6B, C). During the fear conditioning training (Fig. 6A), both genotypes showed an increase in freezing behavior (time: F20,1574 = 45.1, P < 0.0001). Overall, BDNF+/− mice showed slightly more freezing than their WT littermates, but this was not statistically significant (genotype: F1,1574 = 0.0, P = 0.98; genotype × time: F20,1574 = 1.56, P = 0.06). Thus, there was no obvious difference between the 2 genotypes during the fear conditioning training, confirming our previous results (Endres and Lessmann 2012). In the fear memory tests, WT animals showed very stable and comparable fear expression at all tested time points, whereas the fear memory in BDNF+/− mice seemed to decline over time. This decline is also evident from elevated pre-CS freezing. Since the pre-CS freezing reflects the average of all 5 pre-CS periods for a given test interval, this elevated pre-CS freezing probably reflects unspecific freezing. As BDNF+/− mice show this elevation in freezing behavior only after the first CS presentation, it seems like this unspecific freezing is triggered by the first CS presentation. Overall, these data suggest that the memory of the temporal relation between CS and US vanishes in BDNF+/− mice. Twenty-four hours after fear conditioning training, the BDNF+/− animals showed an additional significantly reduced freezing to the CS. These observations are supported by the results of an ANOVA revealing significant effects for the factors “genotype” (BDNF+/− vs. WT), “phase” (Habituation vs. pre-CS vs. CS) and “time point of testing” (genotype: F1,312 = 4.12, P = 0.042; phase: F2,312 = 126.9, P < 0.0001; time point: F1,312 = 8.58, P = 0.004). In addition, the ANOVA revealed a strong interaction between the factors “phase” and “genotype” (F2,312 = 12.81, P < 0.0001), further supporting a generally different fear memory expression between the 2 genotypes. However, there was no significant interaction of the factors “phase” × “genotype” × “time point of testing” as well as for “phase” × “time point of testing” and “genotype” × “time point of testing” (F’s ≤ 2.03, P’s ≥ 0.15). As the ANOVA revealed significant effects for all 3 single factors, we also performed a post hoc Tukey test, which revealed significant differences between the freezing behavior expressed during the habituation period compared with the CS periods for both genotypes at all tested time points. Overall, these findings suggest successful fear memory retrieval for both genotypes at all tested time points, with the exception that at 24 h after fear conditioning, BDNF+/− mice exhibited significantly less freezing than their WT littermates. This corroborates our previous finding of impaired fear learning in these animals (Endres and Lessmann 2012). Importantly, WT animals showed significantly more freezing during the CS compared with the pre-CS periods at all tested time points, while the BDNF+/− mice exhibited similar freezing levels during both phases (CS and pre-CS), especially at later time points of testing. This observation could be interpreted as an indication for generalized fear in BDNF+/− mice. However, since the freezing levels during the habituation periods were similar between BDNF+/− and WT animals, this observation speaks in favor of reduced fear memory accuracy rather than a generalized fear behavior. To better account for these differences in pre-CS to CS freezing, we analyzed the Δfreezing scores (i.e., freezing during CS minus freezing during pre-CS periods, Fig. 6B) of the animals. In order to see whether there was a significant increase in freezing, we tested whether these Δ-scores differ from zero (i.e., no increase in freezing) by single sided t-test comparisons. For BDNF+/− mice, this analysis revealed a significant Δfreezing only at 2 h and a tendency of increased Δfreezing (P = 0.09) 30 min after fear conditioning. In contrast, their WT littermates exhibited a significant Δfreezing at all tested time points. This analysis further suggests the idea of a continuous loss in fear memory precision with ongoing time after training. Figure 6. View largeDownload slide Early fear memory retrieval in BDNF+/− and WT mice. (A) Freezing during the conditioning training. No significant differences between the 2 groups were observed (WT n = 51; BDNF+/−n = 57). (B) Comparison of Δfreezing scores at different time points after fear conditioning. The Δ-scores were calculated for all 5 CS presentations per animal and then averaged. Significance of differences of Δ-scores from zero was determined (*: P < 0.05, ~:P < 0.1). (C) Fear memory retrieval at different time points after fear conditioning, ranging from 0.5 to 24 h. The freezing during the pre-CS and CS represents the average over pre-CS and CS periods. Significant differences compared with habituation and pre-CS periods are indicated by (#), while (+) represents only a significant difference to the habituation, and ($) represents a significant difference between WT and BDNF+/−. (Number of observations for WT/BDNF+/− 30 min: 8/8; 2 h: 13/13; 4 h: 11/14; 6 h: 11/14; 24 h: 8/8). (D) Discriminative early fear memory. In young animals (2 months old, left panel), a clear discrimination of CS− and CS+ was observed 4 h after conditioning (n = 16 per time point). Five- to six-month-old BDNF+/− mice (right panel) failed to discriminate between the 2 stimuli at this time point, while WT littermates did discriminate (WT n = 12; BDNF+/−n = 13; *: P < 0.05). Figure 6. View largeDownload slide Early fear memory retrieval in BDNF+/− and WT mice. (A) Freezing during the conditioning training. No significant differences between the 2 groups were observed (WT n = 51; BDNF+/−n = 57). (B) Comparison of Δfreezing scores at different time points after fear conditioning. The Δ-scores were calculated for all 5 CS presentations per animal and then averaged. Significance of differences of Δ-scores from zero was determined (*: P < 0.05, ~:P < 0.1). (C) Fear memory retrieval at different time points after fear conditioning, ranging from 0.5 to 24 h. The freezing during the pre-CS and CS represents the average over pre-CS and CS periods. Significant differences compared with habituation and pre-CS periods are indicated by (#), while (+) represents only a significant difference to the habituation, and ($) represents a significant difference between WT and BDNF+/−. (Number of observations for WT/BDNF+/− 30 min: 8/8; 2 h: 13/13; 4 h: 11/14; 6 h: 11/14; 24 h: 8/8). (D) Discriminative early fear memory. In young animals (2 months old, left panel), a clear discrimination of CS− and CS+ was observed 4 h after conditioning (n = 16 per time point). Five- to six-month-old BDNF+/− mice (right panel) failed to discriminate between the 2 stimuli at this time point, while WT littermates did discriminate (WT n = 12; BDNF+/−n = 13; *: P < 0.05). To further test for the precision of early fear memory, we performed a discriminative fear conditioning paradigm. Here, we trained animals with 2 randomly appearing tone stimuli. One tone (8 kHz, CS+) was always followed by a foot shock, while the other tone (2 kHz CS−) was not. When we tested these animals 3 h after fear conditioning (Fig. 6D), even 2- to 3-month-old WT mice did not differentiate between CS+ and CS−, that is, they responded with a comparable fear response to both stimuli. An ANOVA revealed significant main effects for the factors “time point of testing” (F1,150 = 14.2, P = 0.0002) and “phase” (F4,150 = 33.4, P < 0.0001), as well as a tendency for the interaction of these 2 factors (F4,150 = 2.0, P = 0.10). Post hoc Tukey comparisons revealed significant differences between pre-CS and CS periods for both CS+ and CS−, if animals were tested 3 h after training. However, when animals were tested 4 h after conditioning, a significant difference between pre-CS and CS existed only in case of CS+ presentations. Hence, the ability to discriminate between CS+ and CS− in our fear conditioning paradigm seems to start around 4 h after training in adult (2–3 months old) WT mice. Therefore, we decided to test adult BDNF+/− mice and their WT littermates 4 h after fear conditioning (Fig. 6D, right panel). Here, we observed a similar pattern as in the previous experiment for WT mice, while BDNF+/− mice displayed no discrimination between CS+ and CS−. An ANOVA revealed a significant main effect for the factor “phase” (F4,115 = 15.6, P < 0.0001), no significant effect for the factor “genotype” (F1,115 = 0.3, P = 0.6) but a tendency for the interaction of these 2 factors (F4,115 = 2.3, P = 0.09). Post hoc Tukey comparisons revealed that only in WT mice a selective significant increase in freezing between pre-CS+ and CS+ could be observed, thus indicating that WT but not BDNF+/− mice could discriminate between CS+ and CS−. In conclusion, our data demonstrate that 24 h after fear conditioning, BDNF+/− mice exhibited an impaired fear memory expression. In addition, our data suggest a declining fear memory precision starting in the very early stages of fear memory consolidation, that is, at 4 h after fear conditioning. Discussion In the present study, we demonstrate that, although LTP in the lateral, basal, and central medial nuclei of the amygdala is intact in adult BDNF+/− mice (≥3 months old), these animals show impaired fear memory consolidation. Nevertheless, early fear memory in adult BDNF+/− mice is still functional. Thus, the unaltered synaptic plasticity in these amygdala circuits is in line with successful acquisition of fear memories in BDNF+/− mice. Moreover, our data provide evidence that fear learning in WT mice results in long-lasting changes in synaptic plasticity at cortico-LA synapses, as indicated by occlusion of LTP 24 h after fear conditioning. Thus, fear learning in WT mice occludes LTP, similar to what has been reported previously for rats (Tsvetkov et al. 2002; Schroeder and Shinnick-Gallagher 2004, 2005). In line with the impaired long-term fear memory in adult BDNF+/− mice, we did not observe occlusion of LTP at cortico-LA synapses 24 h after fear conditioning. Thus, our findings support the notion that in WT mice the successful consolidation of cued fear memories relies, at least in part, on long-term synaptic changes at cortico-LA synapses. In addition, our results demonstrate that such learning-related processes resulting in these long-term modifications are not functional in adult BDNF+/− mice, probably leading to the observed deficit in fear memory consolidation. Interestingly, in BDNF+/− mice LTP is also not occluded when assessed 4–6 h after fear conditioning, at a time when these animals still show successful but unprecise retrieval of early fear memory. These observations suggest that the learning-related changes in synaptic plasticity at cortico-LA synapses observed in WT mice are not required for the expression of early fear memories in BDNF+/− mice. However, as we observed a reduction in fear memory precision (i.e., increase in pre-CS freezing and lack of CS+/CS− discrimination) in BDNF+/− mice from 4 h onward, changes in synaptic plasticity at these synapses might be required to form a precise fear memory trace, in WT as well as in in BDNF+/− mice. Since we found intact LTP in amygdala circuits of adult BDNF+/− mice while fear consolidation was impaired, a decline in synaptic plasticity at other synapses seems to contribute to the lack of long-term memory in BDNF+/− mice. Likely candidates are the prelimbic or the perirhinal cortex (PRhC), since for both areas a crucial role of BDNF-TrkB signaling for the consolidation of cued fear memories has been reported (Choi et al. 2010; Schulz-Klaus et al. 2013). Age-Dependent Learning Deficit and LTP in the Amygdala The present study was undertaken to analyze the cellular mechanisms underlying the age-dependent learning deficit that we recently described for adult (≥3 months old) BDNF+/− mice (Endres and Lessmann 2012). In a previous study, we had shown a BDNF-dependent impairment of LTP in thalamic afferents to the LA already at the age of 1 month (Meis et al. 2012). We therefore assumed that fear learning in these younger animals might preferentially be mediated by stable cortico-LA LTP (Meis et al. 2012). Indeed, it has been shown previously that disruption of auditory fear conditioning is only achieved by combined lesions of the thalamic and the cortical sensory pathways to the LA (Romanski and LeDoux 1992). Thus, we hypothesized that an age-dependent decline in LTP at cortico-LA synapses might account for the observed learning deficit in adult BDNF+/− mice. Since BDNF requirement for LTP has been shown to critically depend on the induction paradigm (Kang et al. 1997; Zakharenko et al. 2003; Abidin et al. 2006; Edelmann et al. 2015) different LTP protocols were tested. Additionally, experiments were performed in the presence of the GABAA receptor antagonist gabazine to account for any compensatory changes in the inhibitory system by chronic BDNF reduction (Gottmann et al. 2009), which could mask LTP deficits. Interestingly, successful LTP induction at cortico-LA synapses appeared to be independent of age and chronic BDNF reduction across different LTP induction paradigms and irrespective of intact GABAergic inhibition. Overall, these data suggest that intact cortico-LA LTP alone does not enable fear memory consolidation in adult BDNF+/− mice. Therefore, we additionally analyzed synaptic plasticity in other subnuclei of the amygdala network that may contribute to the observed learning deficit in BDNF+/− mice. First, we focused on the BL, which was previously reported to undergo LTP upon repeated LA stimulation (Rammes et al. 2000). Moreover, a decrease in fear learning was associated with a reduction of LTP in this nucleus in several knockout mouse models (Brambilla et al. 1997; Humeau et al. 2007; Huynh et al. 2009; Bourgognon et al. 2012) or after pharmacological interference (Sinai et al. 2010). However, we did not observe any decrease in LTP in the basal nucleus upon intra-BL HFS in BDNF+/− as compared with WT mice. Likewise, LTP remained unchanged at glutamatergic input synapses to the medial central amygdala in BDNF+/− mice, representing the main output structure of the amygdala (Pape and Paré 2010), which is also involved in fear acquisition and consolidation (Wilensky et al. 2006). Interestingly, the paraventricular nucleus of the thalamus (PVT) was recently shown to modulate fear learning by activation of TrkB receptors in the lateral central amygdala (Penzo et al. 2015). Nonetheless, our present results do not speak in favor of altered synaptic plasticity at glutamatergic inputs to the lateral CeA in BDNF+/− mice, since LTP at BL-medial CeA synapses was intact in these animals. Overall, we observed unaltered early LTP in BDNF+/− mice at all synapses investigated (cortico-LA, intra-BL, and BL-medial CeA) which parallels the stable early fear memory retrieval (4 h after training) but not the deficit in long-term fear memory (at 24 h). Thus, additional synaptic plasticity processes not analyzed in our study seem to be required for the consolidation of long-term fear memories. Age-Dependent Changes in Synaptic Plasticity and Fear Learning The age-related learning deficit that we observed in BDNF+/− mice could result from qualitative changes in the formation of synaptic circuits during ontogenesis that are involved in the coding of fear memory. Only a few previous studies analyzed fear learning as well as the corresponding cellular mechanisms between adolescence and adulthood. For example, it was shown that, both, synaptic networks in the amygdala and the connectivity between amygdala and prefrontal cortex are altered roughly 1–3 months after birth (Cunningham et al. 2002, 2008; Pan et al. 2009b). In addition, experience-driven maturation of synaptic transmission at cortico-LA synapses was reported to occur most likely during early adulthood (Gambino et al. 2010). It remains to be determined by future studies whether the altered fear memory in BDNF+/− animals can be related to one of these changes. Early Fear Memory in BDNF+/− Mice The preserved LTP at cortico-LA and intra-amygdala synapses in our study points to stable amygdala function in BDNF+/− mice. Interestingly, early fear memory in age-matched BDNF+/− mice was comparable to that of WT littermates when tested 0.5–6 h after cued fear conditioning, as is evident from similar increments in freezing between habituation and CS presentations (see Fig. 6C). This indicates intact acquisition of fear memories in BDNF+/− mice and suggests that synaptic networks required to generate early fear memory are functional in BDNF+/− mice. Nevertheless, the 2 genotypes differ in pre-CS freezing duration with longer testing intervals between conditioning and retrieval. In WT mice, pre-CS freezing duration within the session declines to a similar level as freezing during the habituation, whereas in BDNF+/− mice pre-CS freezing reaches a level in the range of CS freezing. This finding is mirrored by the analysis of the Δfreezing scores, which revealed significant increments in freezing behavior in WT mice at any tested time point. In contrast, BDNF+/− mice exhibited no significant increment in freezing when tested 4 h or later after fear conditioning training. In conclusion, these observations suggest that despite a successful retrieval of early fear memories in BDNF+/− mice, the precision of fear memory to the actual CS presentation gets poorer during the early consolidation processes. This might be taken as a first indication of the fear memory deficit observed 24 h after fear conditioning. To gain a more detailed insight into the precision of the early fear memory, we introduced a discriminative fear learning task. In 2-month-old WT mice, we observed successful discrimination between CS+ and CS− when testing 4 h after fear conditioning but not at an interval of 3 h (see Fig. 6D). Importantly, these responses were elicited specifically by the tone presentations, as indicated by low pre-CS freezing values during the retrieval session. Thus, it seems like this early fear memory is not fine-tuned to the CS. Consequently, the animals responded to both stimuli with a high but specific fear response. With ongoing consolidation processes, the precision of the fear memory increased. This is evident from the much lower levels of freezing to the CS− with increasing time interval between training and retrieval session. Alternatively to this interpretation, fear conditioning might just induce a longer lasting state of arousal (Rodrigues et al. 2009), which caused the animals to respond with a generalized fear behavior to all unexpected stimuli that appear during testing. Since in WT animals the discrimination between CS+ and CS− begins around 4 h after training, we tested the discriminative memory abilities of aged BDNF+/− mice only at this time point. Here, we observed that 5- to 6-month-old BDNF+/− mice exhibited comparable levels of fear to CS+ and CS− presentations. In addition, they showed similar elevated pre-CS freezing levels as observed in the nondiscriminative fear conditioning experiments (compare Fig. 5). In contrast, their WT littermates showed a specific fear response to the CS+, even though the response to the CS− was more pronounced than in younger WT animals. This suggests that early fear discrimination might decline with aging. We previously observed that BDNF+/− mice do not exhibit altered anxiety in the elevated plus maze and open field at any age (Endres and Lessmann 2012). Therefore, it seems unlikely that the elevated fear responses during the pre-CS and CS− periods are due to an altered anxiety level in BDNF+/− mice. In conclusion, these results support the notion that fear memory in BDNF+/− mice, even though still retrievable early after training, undergoes a constant decline in memory precision, which is due to a lack of required—maybe BDNF-dependent—consolidation processes. Occlusion of LTP by Previous Fear Learning Several lines of experimental data suggest that LTP in the amygdala might be the cellular mechanism underlying cued fear learning (for review, see Sigurdsson et al. 2007). In this respect, fear conditioning was shown previously to induce long-lasting changes at different amygdala synapses, as reflected by facilitated synaptic transmission as well as a lack of subsequent electrical LTP induction (McKernan and Shinnick-Gallagher 1997; Tsvetkov et al. 2002; Schroeder and Shinnick-Gallagher 2004, 2005; Hong et al. 2011, 2012). These ex vivo experiments were performed exclusively in previously fear-conditioned rats but not in mice. We therefore investigated LTP occlusion in age-matched mice to verify that LTP induced by our specific paradigm was indeed affected by preceding fear learning. In fact, LTP was significantly reduced by pretraining of mice compared with pseudoconditioned animals (see Fig. 5C, E). These findings indicate that LTP induction at cortico-LA synapses by our HFS paradigm and fear learning share identical cellular processes in WT mice. However, we did not observe such an occlusion of cortico-LA LTP in BDNF+/− mice, indicating that in these animals no learning-dependent long-term changes in cortico-LA afferents occurred. As expected, this lack of occlusion at the cellular level parallels the behavioral deficit in fear consolidation in BDNF+/− mice when assessed 24 h after training (present study; Endres and Lessmann 2012). We also tried to perform occlusion experiments shortly after fear training to gain insights into the synaptic processes involved in acquisition of fear memory. Successful LTP induction was achieved in slices of pseudoconditioned mice, which were sacrificed 2 h after training, when field recordings started 4–8 h after training. This LTP was again occluded in WT mice, but not in BDNF+/− mice, as already observed for recordings obtained 1 day after training. This suggests that changes in synaptic transmission occur at these synapses in WT mice, which however, are lacking in BDNF+/− mice. Interestingly, we observed in the same time range (i.e., 4–8 h) an ongoing decline in fear memory precision in BDNF+/− mice. This was evident from the reduced Δfreezing scores that were due to an increased pre-CS freezing. Overall, this might indicate that the lack of fear memory-related changes at the cortico-LA afferents contributes to the reduced fear memory precision observed at the early stages (i.e., 4–8 h) after fear conditioning training. Altered Synaptic Plasticity in Other Fear Learning-Related Brain Areas We focused exclusively on synaptic plasticity within the amygdala network in this study, as the amygdala is widely accepted as key brain area for cued fear learning (Sigurdsson et al. 2007; Ehrlich et al. 2009; Pape and Paré 2010). Besides the amygdala, synaptic plasticity in several other brain areas contributes to successful formation of cued fear memory, for example, the prelimbic medioprefrontal cortex (PL, Burgos-Robles et al. 2009; Sotres-Bayon and Quirk 2010; Sierra-Mercado et al. 2011) or the PRhC (Schulz et al. 2004; Kealy and Commins 2011; Kent and Brown 2012). Interestingly, interfering with BDNF signaling in these brain areas by regional BDNF-knockout (Choi et al. 2010) or acute pharmacological inhibition of TrkB signaling (Schulz-Klaus et al. 2013) resulted in a lack of fear memory consolidation. Since the expression of BDNF protein changes with aging (Croll et al. 1998; Katoh-Semba et al. 1998; von Bohlen und Halbach 2010; Boger et al. 2011) and these changes are differentially regulated in different brain areas (Silhol et al. 2005; Psotta et al. 2013), BDNF- and age-dependent synaptic plasticity might vary depending on the specific brain area under study. Thus, an age-dependent change affecting fear consolidation may take place in the PL and/or PRhC in BDNF+/− mice. These changes might in turn alter synaptic structures or synaptic transmission in distinct brain areas (including the amygdala) that are essential for fear memory consolidation. In this respect, it is an interesting question whether we stimulated fibers originating from the above mentioned cortical regions in our LTP recordings as well. We placed our stimulation electrode at the external capsule which carries mainly fibers originating from higher sensory cortices such as for example, the auditory cortex (de Olmos et al. 1985), which is required for tone discrimination (LeDoux 1995). Since afferents from the PL or PRhC reach the amygdala either more ventrally (PL, Vertes, 2004) or more horizontally (PRhC, von Bohlen und Halbach and Albrecht 2002), it seems unlikely that these fibers are stimulated in our LTP recordings. Nevertheless, future studies addressing the mechanisms of synaptic plasticity in the PL and/or the PRhC in relation to BDNF availability and age will be required to test their contribution to the age-dependent fear learning deficit in BDNF+/− mice. Possible Sources of BDNF BDNF mRNA and protein was detected in the rodent amygdala at moderate to high levels (Conner et al. 1997; Yan et al. 1997; Krause et al. 2008), and fear conditioning induces a selective increase of BDNF levels in the BLA (Rattiner et al. 2005; Ou and Gean 2006). In addition, the temporal association cortex, sending afferents to the LA, shows substantial BDNF expression (Ernfors et al. 1990; Castren et al. 1995; Conner et al. 1997). In contrast, the LA is poorly innervated by the paraventricular thalamic nucleus, a structure recently identified as a major source of BDNF for the lateral nucleus of the central amygdala (Penzo et al. 2015). Accordingly, postsynaptic as well as presynaptic BDNF may contribute to LTP induction/expression (Edelmann et al. 2014) at cortico-LA afferents. Conclusion Overall, we demonstrate that fear learning induces long-lasting changes in synaptic plasticity at cortico-LA synapses in WT mice, which are absent in adult BDNF+/− mice. Moreover, we show that precision of early fear memory declines with time after training in BDNF+/− mice already at 4 h after training. Therefore, consolidation-relevant processes seem to occur at cortico-LA synapses in WT mice, which are absent in BDNF+/− mice. Thus, intact LTP at cortico-LA synapses parallels the formation of the initial CS–US association in BDNF+/− mice. However, due to a lack of consolidation-relevant synaptic plasticity, which might take place outside the amygdala circuitry, cortico-LA synapses are unable to undergo long-term changes in BDNF+/− mice that seem to be required to form a stable fear memory. In this respect, future studies should aim at elucidating how BDNF-dependent synaptic plasticity inside and outside the amygdala orchestrates the formation of fear memories in adult animals. Supplementary Material Supplementary material is available at Cerebral Cortex online. Funding Deutsche Forschungsgemeinschaft (SFB 779, TP B06). Notes We thank Kathrin Friese, Colette Obst, Angela Jahn, Regina Ziegler, and Evelyn Friedl for excellent technical assistance, and Gloria Deisy Quiceno for linguistic corrections of the manuscript. Conflict of Interest: None declared. References Abidin I, Eysel UT, Lessmann V, Mittmann T. 2008. Impaired GABAergic inhibition in the visual cortex of brain-derived neurotrophic factor heterozygous knockout mice. J Physiol . 586: 1885– 1901. Google Scholar CrossRef Search ADS PubMed  Abidin I, Köhler T, Weiler E, Zoidl G, Eysel UT, Lessmann V, Mittmann T. 2006. Reduced presynaptic efficiency of excitatory synaptic transmission impairs LTP in the visual cortex of BDNF-heterozygous mice. Eur J Neurosci . 24: 3519– 3531. Google Scholar CrossRef Search ADS PubMed  Amano T, Duvarci S, Popa D, Paré D. 2011. The fear circuit revisited: contributions of the basal amygdala nuclei to conditioned fear. J Neurosci . 31: 15481– 15489. Google Scholar CrossRef Search ADS PubMed  Bauer EP, Schafe GE, LeDoux JE. 2002. NMDA receptors and L-type voltage-gated calcium channels contribute to long-term potentiation and different components of fear memory formation in the lateral amygdala. J Neurosci . 22: 5239– 5249. Google Scholar PubMed  Boger HA, Mannangatti P, Samuvel DJ, Saylor AJ, Bender TS, McGinty JF, Fortress AM, Zaman V, Huang P, Middaugh LD, et al.  . 2011. Effects of brain-derived neurotrophic factor on dopaminergic function and motor behavior during aging. Genes Brain Behav . 10: 186– 198. Google Scholar CrossRef Search ADS PubMed  Bourgognon JM, Schiavon E, Salah-Uddin H, Skrzypiec AE, Attwood BK, Shah RS, Patel SG, Mucha M, John Challiss RA, Forsythe ID, et al.  . 2012. Regulation of neuronal plasticity and fear by a dynamic change in PAR1-G protein coupling in the amygdala. Mol Psychiatry . doi:10.1038/mp.2012.133. Brambilla R, Gnesutta N, Minichiello L, White G, Roylance AJ, Herron CE, Ramsey M, Wolfer DP, Cestari V, Rossi-Arnaud C, et al.  . 1997. A role for the Ras signalling pathway in synaptic transmission and long-term memory. Nature . 390: 281– 286. Google Scholar CrossRef Search ADS PubMed  Burgos-Robles A, Vidal-Gonzalez I, Quirk GJ. 2009. Sustained conditioned responses in prelimbic prefrontal neurons are correlated with fear expression and extinction failure. J Neurosci . 29: 8474– 8482. Google Scholar CrossRef Search ADS PubMed  Castren E, Thoenen H, Lindholm D. 1995. Brain-derived neurotrophic factor messenger RNA is expressed in the septum, hypothalamus and in adrenergic brain stem nuclei of adult rat brain and is increased by osmotic stimulation in the paraventricular nucleus. Neuroscience . 64: 71– 80. Google Scholar CrossRef Search ADS PubMed  Choi DC, Maguschak KA, Ye K, Jang SW, Myers KM, Ressler KJ. 2010. Prelimbic cortical BDNF is required for memory of learned fear but not extinction or innate fear. Proc Natl Acad Sci USA . 107: 2675– 2680. Google Scholar CrossRef Search ADS PubMed  Conner JM, Lauterborn JC, Yan Q, Gall CM, Varon S. 1997. Distribution of brain-derived neurotrophic factor (BDNF) protein and mRNA in the normal adult rat CNS: evidence for anterograde axonal transport. J Neurosci . 17: 2295– 2313. Google Scholar PubMed  Cowansage KK, LeDoux JE, Monfils MH. 2010. Brain-derived neurotrophic factor: a dynamic gatekeeper of neural plasticity. Curr Mol Pharmacol . 3: 12– 29. Google Scholar CrossRef Search ADS PubMed  Croll SD, Ip NY, Lindsay RM, Wiegand SJ. 1998. Expression of BDNF and trkB as a function of age and cognitive performance. Brain Res . 812: 200– 208. Google Scholar CrossRef Search ADS PubMed  Cunningham MG, Bhattacharyya S, Benes FM. 2002. Amygdalo-cortical sprouting continues into early adulthood: implications for the development of normal and abnormal function during adolescence. J Comp Neurol . 453: 116– 130. Google Scholar CrossRef Search ADS PubMed  Cunningham MG, Bhattacharyya S, Benes FM. 2008. Increasing Interaction of amygdalar afferents with GABAergic interneurons between birth and adulthood. Cereb Cortex . 18: 1529– 1535. Google Scholar CrossRef Search ADS PubMed  Daftary SS, Calderon G, Rios M. 2012. Essential role of brain-derived neurotrophic factor in the regulation of serotonin transmission in the basolateral amygdala. Neuroscience . 224: 125– 134. Google Scholar CrossRef Search ADS PubMed  de Olmos J, Alheid GF, Beltramino CA. 1985. Amygdala. In: Paxinos G, editor. The rat nervous system . Sydney: Academic Press. p. 223– 334. Drephal C, Schubert M, Albrecht D. 2006. Input-specific long-term potentiation in the rat lateral amygdala of horizontal slices. Neurobiol Learn Mem . 85: 272– 282. Google Scholar CrossRef Search ADS PubMed  Doyère V, Schafe GE, Sigurdsson T, LeDoux JE. 2003. Long-term potentiation in freely moving rats reveals asymmetries in thalamic and cortical inputs to the lateral amygdala. Eur J Neurosci . 17: 2703– 2715. Google Scholar CrossRef Search ADS PubMed  Edelmann E, Cepeda-Prado E, Franck M, Lichtenecker P, Brigadski T, Lessmann V. 2015. Theta burst firing recruits BDNF release and signaling in postsynaptic CA1 neurons in spike timing-dependent LTP. Neuron . 86: 1041– 1054. Google Scholar CrossRef Search ADS PubMed  Edelmann E, Lessmann V, Brigadski T. 2014. Pre- and postsynaptic twists in BDNF secretion and action in synaptic plasticity. Neuropharmacology . 76: 610– 627. Google Scholar CrossRef Search ADS PubMed  Ehrlich I, Humeau Y, Grenier F, Ciocchi S, Herry C, Lüthi A. 2009. Amygdala inhibitory circuits and the control of fear memory. Neuron . 62: 757– 771. Google Scholar CrossRef Search ADS PubMed  Endres T, Lessmann V. 2012. Age-dependent deficits in fear learning in heterozygous BDNF knock-out mice. Learn Mem . 19: 561– 570. Google Scholar CrossRef Search ADS PubMed  Ernfors P, Wetmore C, Olson L, Persson H. 1990. Identification of cells in rat brain and peripheral tissues expressing mRNA for members of the nerve growth factor family. Neuron . 5: 511– 526. Google Scholar CrossRef Search ADS PubMed  Gambino F, Khelfaoui M, Poulain B, Bienvenu T, Chelly J, Humeau Y. 2010. Synaptic maturation at cortical projections to the lateral amygdala in a mouse model of Rett syndrome. PLoS One . 5: e11399. Google Scholar CrossRef Search ADS PubMed  Gean PW, Shinnick-Gallagher P. 1987. Picrotoxin induced epileptiform activity in amygdaloid neurons. Neurosci Lett . 73: 149– 154. Google Scholar CrossRef Search ADS PubMed  Gottmann K, Mittmann T, Lessmann V. 2009. BDNF signaling in the formation, maturation and plasticity of glutamatergic and GABAergic synapses. Exp Brain Res . 199: 203– 234. Google Scholar CrossRef Search ADS PubMed  Hong I, Kim J, Lee J, Park S, Song B, Kim J, An B, Park K, Lee HW, Lee S, et al.  . 2011. Reversible plasticity of fear memory-encoding amygdala synaptic circuits even after fear memory consolidation. PLoS One . 6: e24260. Google Scholar CrossRef Search ADS PubMed  Hong I, Kim J, Song B, Park K, Shin K, Eom KD, Han PL, Lee S, Choi S. 2012. Fear conditioning occludes late-phase long-term potentiation at thalamic input synapses onto the lateral amygdala in rat brain slices. Neurosci Lett . 506: 121– 125. Google Scholar CrossRef Search ADS PubMed  Huang YY, Kandel ER. 2007. 5-Hydroxytryptamine induces a protein kinase A/mitogen-activated protein kinase-mediated and macromolecular synthesis-dependent late phase of long-term potentiation in the amygdala. J Neurosci . 27: 3111– 3119. Google Scholar CrossRef Search ADS PubMed  Huang YY, Martin KC, Kandel ER. 2000. Both protein kinase A and mitogen-activated protein kinase are required in the amygdala for the macromolecular synthesis-dependent late phase of long-term potentiation. J Neurosci . 20: 6317– 6325. Google Scholar PubMed  Humeau Y, Reisel D, Johnson AW, Borchardt T, Jensen V, Gebhardt C, Bosch V, Gass P, Bannerman DM, Good MA, et al.  . 2007. A pathway-specific function for different AMPA receptor subunits in amygdala long-term potentiation and fear conditioning. J Neurosci . 27: 10947– 10956. Google Scholar CrossRef Search ADS PubMed  Huynh DP, Maalouf M, Silva AJ, Schweizer FE, Pulst SM. 2009. Dissociated fear and spatial learning in mice with deficiency of ataxin-2. PLoS One . 4: e6235. Google Scholar CrossRef Search ADS PubMed  Isoardi NA, Martijena ID, Carrer HF, Molina VA. 2004. Increased fear learning coincides with neuronal dysinhibition and facilitated LTP in the basolateral amygdala following benzodiazepine withdrawal in rats. Neuropsychopharmacology . 29: 1852– 1864. Google Scholar CrossRef Search ADS PubMed  Kang H, Welcher AA, Shelton D, Schuman EM. 1997. Neurotrophins and time: different roles for TrkB signaling in hippocampal longterm potentiation. Neuron . 19: 653– 664. Google Scholar CrossRef Search ADS PubMed  Katoh-Semba R, Semba R, Takeuchi IK, Kato K. 1998. Age-related changes in levels of brain-derived neurotrophic factor in selected brain regions of rats, normal mice and senescence-accelerated mice: a comparison to those of nerve growth factor and neurotrophin-3. Neurosci Res . 31: 227– 234. Google Scholar CrossRef Search ADS PubMed  Kealy J, Commins S. 2011. The rat perirhinal cortex: a review of anatomy, physiology, plasticity, and function. Prog Neurobiol . 93: 522– 548. Google Scholar CrossRef Search ADS PubMed  Kent BA, Brown TH. 2012. Dual functions of perirhinal cortex in fear conditioning. Hippocampus . 22: 2068– 2079. Google Scholar CrossRef Search ADS PubMed  Kohara K, Yasuda H, Huang Y, Adachi N, Sohya K, Tsumoto T. 2007. A local reduction in cortical GABAergic synapses after a loss of endogenous brain-derived neurotrophic factor, as revealed by single-cell gene knock-out method. J Neurosci . 27: 7234– 7244. Google Scholar CrossRef Search ADS PubMed  Korte M, Carroll P, Wolf E, Brem G, Thoenen H, Bonhoeffer T. 1995. Hippocampal long-term potentiation is impaired in mice lacking brain-derived neurotrophic factor. Proc Natl Acad Sci USA . 92: 8856– 8860. Google Scholar CrossRef Search ADS PubMed  Krause S, Schindowski K, Zechel S, von Bohlen und Halbach O. 2008. Expression of trkB and trkC receptors and their ligands brain-derived neurotrophic factor and neurotrophin-3 in the murine amygdala. J Neurosci Res . 86: 411– 421. Google Scholar CrossRef Search ADS PubMed  Kulisch C, Eckers N, Albrecht D. 2011. Method of euthanasia affects amygdala plasticity in horizontal brain slices from mice. J Neurosci Methods . 201: 340– 345. Google Scholar CrossRef Search ADS PubMed  Laudes T, Meis S, Munsch T, Lessmann V. 2012. Impaired transmission at corticothalamic excitatory inputs and intrathalamic GABAergic synapses in the ventrobasal thalamus of heterozygous BDNF knockout mice. Neuroscience . 222: 215– 227. Google Scholar CrossRef Search ADS PubMed  LeDoux JE. 1995. Emotion: clues from the brain. Annu Rev Psychol . 46: 209– 235. Google Scholar CrossRef Search ADS PubMed  Li C, Dabrowska J, Hazra R, Rainnie DG. 2011. Synergistic activation of dopamine D1 and TrkB receptors mediate gain control of synaptic plasticity in the basolateral amygdala. PLoS One . 6: e26065. Google Scholar CrossRef Search ADS PubMed  Matthies H, Schulz S, Thiemann W, Siemer H, Schmidt H, Krug M, Höllt V. 1997. Design of a multiple slice interface chamber and application for resolving the temporal pattern of CREB phosphorylation in hippocampal long-term potentiation. J Neurosci Methods . 78: 173– 179. Google Scholar CrossRef Search ADS PubMed  McKernan MG, Shinnick-Gallagher P. 1997. Fear conditioning induces a lasting potentiation of synaptic currents in vitro. Nature . 390: 607– 611. Google Scholar CrossRef Search ADS PubMed  Meis S, Bergado-Acosta JR, Yanagawa Y, Obata K, Stork O, Munsch T. 2008. Identification of a neuropeptide S responsive circuitry shaping amygdala activity via the endopiriform nucleus. PLoS One . 3: e2695. Google Scholar CrossRef Search ADS PubMed  Meis S, Endres T, Lessmann V. 2012. Postsynaptic BDNF signalling regulates long-term potentiation at thalamo-amygdala afferents. J Physiol . 590: 193– 208. Google Scholar CrossRef Search ADS PubMed  Musumeci G, Sciarretta C, Rodríguez-Moreno A, Al Banchaabouchi M, Negrete-Díaz V, Costanzi M, Berno V, Egorov AV, von Bohlen Und Halbach O, Cestari V, et al.  . 2009. TrkB modulates fear learning and amygdalar synaptic plasticity by specific docking sites. J Neurosci . 29: 10131– 10143. Google Scholar CrossRef Search ADS PubMed  Ou LC, Gean PW. 2006. Regulation of amygdala-dependent learning by brain-derived neurotrophic factor is mediated by extracellular signal-regulated kinase and phosphatidylinositol-3-kinase. Neuropsychopharmacology . 31: 287– 296. Google Scholar CrossRef Search ADS PubMed  Ou LC, Yeh SH, Gean PW. 2010. Late expression of brain-derived neurotrophic factor in the amygdala is required for persistence of fear memory. Neurobiol Learn Mem . 93: 372– 382. Google Scholar CrossRef Search ADS PubMed  Pan BX, Dong Y, Ito W, Yanagawa Y, Shigemoto R, Morozov A. 2009a. Selective gating of glutamatergic inputs to excitatory neurons of amygdala by presynaptic GABAb receptor. Neuron . 61: 917– 929. Google Scholar CrossRef Search ADS PubMed  Pan BX, Ito W, Morozov A. 2009b. Divergence between thalamic and cortical inputs to lateral amygdala during juvenile-adult transition in mice. Biol Psychiatry . 66: 964– 971. Google Scholar CrossRef Search ADS PubMed  Pape HC, Pare D. 2010. Plastic synaptic networks of the amygdala for the acquisition, expression, and extinction of conditioned fear. Physiol Rev . 90: 419– 463. Google Scholar CrossRef Search ADS PubMed  Park H, Poo MM. 2013. Neurotrophin regulation of neural circuit development and function. Nat Rev Neurosci . 14: 7– 23. Google Scholar CrossRef Search ADS PubMed  Paxinos G, Franklin KBJ. 2001. The mouse brain in stereotaxic coordinates . San Diego: Academic Press. Penzo MA, Robert V, Tucciarone J, De Bundel D, Wang M, Van Aelst L, Darvas M, Parada LF, Palmiter RD, He M, et al.  . 2015. The paraventricular thalamus controls a central amygdala fear circuit. Nature . 519: 455– 459. Google Scholar CrossRef Search ADS PubMed  Pitkänen A, Savander V, LeDoux JE. 1997. Organization of intra-amygdaloid circuitries in the rat: an emerging framework for understanding functions of the amygdala. Trends Neurosci . 20: 517– 523. Google Scholar CrossRef Search ADS PubMed  Psotta L, Lessmann V, Endres T. 2013. Impaired fear extinction learning in adult heterozygous BDNF knock-out mice. Neurobiol Learn Mem . 103: 34– 38. Google Scholar CrossRef Search ADS PubMed  Rammes G, Steckler T, Kresse A, Schütz G, Zieglgänsberger W, Lutz B. 2000. Synaptic plasticity in the basolateral amygdala in transgenic mice expressing dominant-negative cAMP response element-binding protein (CREB) in forebrain. Eur J Neurosci . 12: 2534– 2546. Google Scholar CrossRef Search ADS PubMed  Rattiner LM, Davis M, Ressler KJ. 2005. Brain-derived neurotrophic factor in amygdala-dependent learning. Neuroscientist . 11: 323– 333. Google Scholar CrossRef Search ADS PubMed  Rescorla RA, Wagner AR. 1972. A theory of Pavlovian conditioning: variations in the effectiveness of reinforcement and non-reinforcement. In: Black AH, Prokasy WF, editors. Classical conditioning II: Theory and research . New York: Appleton-Century-Crofts. p. 64– 99. Rioult-Pedotti MS, Friedman D, Donoghue JP. 2000. Learning-induced LTP in neo-cortex. Science . 290: 533– 536. Google Scholar CrossRef Search ADS PubMed  Rodrigues SM, LeDoux JE, Sapolsky RM. 2009. The influence of stress hormones on fear circuitry. Annu Rev Neurosci . 32: 289– 313. Google Scholar CrossRef Search ADS PubMed  Rogan MT, Stäubli UV, LeDoux JE. 1997. Fear conditioning induces associative long-term potentiation in the amygdala. Nature . 390: 604– 607. Google Scholar CrossRef Search ADS PubMed  Romanski LM, LeDoux JE. 1992. Equipotentiality of thalamo-amygdala and thalamo-cortico-amygdala circuits in auditory fear conditioning. J Neurosci . 12: 4501– 4509. Google Scholar PubMed  Schroeder BW, Shinnick-Gallagher P. 2004. Fear memories induce a switch in stimulus response and signaling mechanisms for long-term potentiation in the lateral amygdala. Eur J Neurosci . 20: 549– 556. Google Scholar CrossRef Search ADS PubMed  Schroeder BW, Shinnick-Gallagher P. 2005. Fear learning induces persistent facilitation of amygdala synaptic transmission. Eur J Neurosci . 22: 1775– 1783. Google Scholar CrossRef Search ADS PubMed  Schulz B, Fendt M, Richardson R, Schnitzler HU. 2004. Temporary inactivation of the perirhinal cortex by muscimol injections block acquisition and expression of fear-potentiated startle. Eur J Neurosci . 19: 713– 720. Google Scholar CrossRef Search ADS PubMed  Schulz-Klaus B, Lessmann V, Endres T. 2013. BDNF-dependent consolidation of fear memories in the perirhinal cortex. Front Behav Neurosci . 7: 205. Google Scholar CrossRef Search ADS PubMed  Sierra-Mercado D, Padilla-Coreano N, Quirk GJ. 2011. Dissociable roles of prelimbic and infralimbic cortices, ventral hippocampus, and basolateral amygdala in the expression and extinction of conditioned fear. Neuropsychopharmacology . 36: 529– 538. Google Scholar CrossRef Search ADS PubMed  Sigurdsson T, Doyère V, Cain CK, LeDoux JE. 2007. Long-term potentiation in the amygdala: a cellular mechanism of fear learning and memory. Neuropharmacology . 52: 215– 227. Google Scholar CrossRef Search ADS PubMed  Silhol M, Bonnichon V, Rage F, Tapia-Arancibia L. 2005. Age-related changes in brain-derived neurotrophic factor and tyrosine kinase receptor isoforms in the hippocampus and hypothalamus in male rats. Neuroscience . 132: 613– 624. Google Scholar CrossRef Search ADS PubMed  Sinai L, Duffy S, Roder JC. 2010. Src inhibition reduces NR2B surface expression and synaptic plasticity in the amygdala. Learn Mem . 17: 364– 371. Google Scholar CrossRef Search ADS PubMed  Sotres-Bayon F, Quirk GJ. 2010. Prefrontal control of fear: more than just extinction. Curr Opin Neurobiol . 20: 231– 235. Google Scholar CrossRef Search ADS PubMed  Tsvetkov E, Carlezon WA, Benes FM, Kandel ER, Bolshakov VY. 2002. Fear conditioning occludes LTP-induced presynaptic enhancement of synaptic transmission in the cortical pathway to the lateral amygdala. Neuron . 34: 289– 300. Google Scholar CrossRef Search ADS PubMed  Vertes RP. 2004. Differential projections of the infralimbic and prelimbic cortex in the rat. Synapse . 51: 32– 58. Google Scholar CrossRef Search ADS PubMed  Von Bohlen und Halbach O, Albrecht D. 2002. Reciprocal connections of the hippocampal area CA1, the lateral nucleus of the amygdala and cortical areas in a combined horizontal slice preparation. Neurosci Res . 44: 91– 100. Google Scholar CrossRef Search ADS PubMed  Von Bohlen und Halbach O. 2010. Involvement of BDNF in age-dependent alterations in the hippocampus. Front Aging Neurosci . 2: 36. Google Scholar PubMed  Watanabe Y, Ikegaya Y, Saito H, Abe K. 1995. Roles of GABAA, NMDA and muscarinic receptors in induction of long-term potentiation in the medial and lateral amygdala in vitro. Neurosci Res . 21: 317– 322. Google Scholar CrossRef Search ADS PubMed  Whitlock JR, Heynen AJ, Shuler MG, Bear MF. 2006. Learning induces long-term potentiation in the hippocampus. Science . 313: 1093– 1097. Google Scholar CrossRef Search ADS PubMed  Wilensky AE, Schafe GE, Kristensen MP, LeDoux JE. 2006. Rethinking the fear circuit: the central nucleus of the amygdala is required for the acquisition, consolidation, and expression of Pavlovian fear conditioning. J Neurosci . 26: 12387– 12396. Google Scholar CrossRef Search ADS PubMed  Yan Q, Rosenfeld RD, Matheson CR, Hawkins N, Lopez OT, Bennett L, Welcher AA. 1997. Expression of brain-derived neurotrophic factor protein in the adult rat central nervous system. Neuroscience . 78: 431– 448. Google Scholar CrossRef Search ADS PubMed  Yoshii A, Constantine-Paton M. 2010. Postsynaptic BDNF-TrkB signaling in synapse maturation, plasticity, and disease. Dev Neurobiol . 70: 304– 322. Google Scholar PubMed  Zakharenko SS, Patterson SL, Dragatsis I, Zeitlin SO, Siegelbaum SA, Kandel ER, Morozov AP. 2003. Presynaptic BDNF required for a presynaptic but not postsynaptic component of LTP at hippocampal CA1-CA3 synapses. Neuron . 39: 975– 990. Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Cerebral CortexOxford University Press

Published: Apr 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

Monthly Plan

  • Read unlimited articles
  • Personalized recommendations
  • No expiration
  • Print 20 pages per month
  • 20% off on PDF purchases
  • Organize your research
  • Get updates on your journals and topic searches

$49/month

Start Free Trial

14-day Free Trial

Best Deal — 39% off

Annual Plan

  • All the features of the Professional Plan, but for 39% off!
  • Billed annually
  • No expiration
  • For the normal price of 10 articles elsewhere, you get one full year of unlimited access to articles.

$588

$360/year

billed annually
Start Free Trial

14-day Free Trial