Fear conditioning induces distinct patterns of gene expression in lateral amygdala

Authors


R. Lamprecht, Department of Neurobiology and Ethology, University of Haifa, Haifa 31905, Israel. E-mail:rlamp@research.haifa.ac.il

Abstract

The lateral nucleus of the amygdala (LA) has been implicated in the formation of long-term associative memory (LTM) of stimuli associated with danger through fear conditioning. The current study aims to detect genes that are expressed in LA following associative fear conditioning. Using oligonucleotide microarrays, we monitored gene expression in rats subjected to paired training where a tone co-terminates with a footshock, or unpaired training where the tone and footshock are presented in a non-overlapping manner. The paired protocol consistently leads to auditory fear conditioning memory formation, whereas the unpaired protocol does not. When the paired group was compared with the unpaired group 5 h after training, the expression of genes coding for the limbic system-associated membrane protein (Lsamp), kinesin heavy chain member 2 (Kif2), N-ethylmaleimide-sensitive fusion protein (NSF) and Hippocalcin-like 4 protein (Hpcal4) was higher in the paired group. These genes encode proteins that regulate neuronal axonal morphology (Lsamp, Kif2), presynaptic vesicle cycling and release (Hpcal4 and NSF), and AMPA receptor maintenance in synapses (NSF). Quantitative real-time PCR (qPCR) showed that Kif2 and Lsamp are expressed hours following fear conditioning but minutes after unpaired training. Hpcal4 is induced by paired stimulation only 5 h after the training. These results show that fear conditioning induces a unique temporal activation of molecular pathways involved in regulating synaptic transmission and axonal morphology in LA, which is different from non-associative stimulation.

Long-term memory (LTM) formation is believed to involve alterations of synaptic efficacy produced by modifications in neural transmission because of physiochemical and/or structural modifications of synaptic communication within neuronal networks (Bliss & Collingridge 1993; Hebb 1949; Konorski 1948; Lamprecht & LeDoux 2004; Martin et al. 2000; Tsien 2000). These modifications in turn are made possible by gene expression, which leads to synthesis of RNA and proteins (Bailey et al. 1996; Davis & Squire 1984; Dudai 1989; Goelet et al. 1986; Kandel 2001). The aim of the current study was to identify genes that are expressed during LTM formation after Pavlovian fear conditioning, a leading model for the study of cellular and molecular mechanisms of memory storage (Blair et al. 2001; Davis & Whalen 2001; Dityatev & Bolshakov 2005; Fanselow & LeDoux 1999; Fanselow & Poulos 2005; LeDoux 2000; Maren 2005; Sah et al. 2003). In fear conditioning, an association is formed between a neutral conditioned stimulus (CS), such as a tone, and an aversive unconditioned stimulus (US), typically a mild footshock. This behavioral paradigm is especially useful as a tool for studying the molecular basis of LTM because the putative site of memory, the lateral nucleus of the amygdala (LA), has been identified (Fanselow & LeDoux 1999; Fanselow & Poulos 2005; Maren 2005; Rodrigues et al. 2004; Schafe et al. 2001). Fear conditioning LTM formation requires gene expression and protein synthesis. Indeed, blockade of RNA or protein synthesis in the LA and nearby areas prevents the formation of fear conditioning LTM without disrupting short-term memory (STM) (Bailey et al. 1999; Duvarci et al. 2008; Maren et al. 2003; Schafe & LeDoux 2000). This observation indicates that RNA and protein syntheses are essential for the formation of fear conditioning memory in the amygdala.

To detect gene expression in the LA following fear conditioning, we used the Affymetrix microarray technology, which allows the simultaneous detection of changes in the expression of thousands of genes. Other studies have investigated gene expression in amygdala after fear conditioning (Keeley et al. 2001; Mei et al. 2005; Pollak et al. 2008; Stork et al. 2001). Here we focused specifically on the LA. To dissect the LA from adjacent amygdala areas, we used the laser microdissection (LMD) technique. To specifically test memory that reflects the CS–US association, as opposed to non-associative factors, such as generalized stress that can also induce gene expression, we compared the expression of genes in LA in animals that received associative training (paired CS and US presentations where the US occurs during the CS) vs. animals that received the same number of CS and US presentations but in a non-overlapping manner. We initially focused on gene expression in animals killed 5 h after training, because genes involved in key neuronal functions and synaptic plasticity are expressed around this time after synaptic stimulation (Hong et al. 2004). The differential expression between the paired and unpaired groups at 5 h was verified using real-time PCR (qPCR). In addition, qPCR was performed on the same genes but in a different group of animals killed 30 min after training to identify possible temporal changes in the expression of these genes following the training procedure.

Methods

Animals

All studies involved male Sprague–Dawley rats (Hilltop Labs, Scottdale, PA, USA and Harlan Laboratories, Jerusalem, Israel) weighing 250–300 g. The animals were housed separately in plastic Nalgene cages and placed on a 12-h light/dark cycle with ad libitum food and water. All procedures were in accordance with the National Institutes of Health guide and were approved by the New York University and University of Haifa Animal Care and Use Committees.

Fear conditioning

Fear conditioning took place in a Plexiglas rodent conditioning chamber with a metal grid floor dimly illuminated by a single house light and enclosed within a sound-attenuating chamber (Coulbourn Instruments, Lehigh Valley, PA, USA). Rats were habituated to the training chamber for 15 min for four consecutive days. On the fifth day, the rats were divided into three groups: (1) naï ve group exposed to the chamber for 15 min with no CS or US; (2) paired group was acclimated for 90 seconds in the conditioning chamber and presented with five pairings of tone for 20 seconds (CS; 5 kHz, 75 dB) that co-terminated with a foot shock (US; 0.5 second, 1.3 mA). The inter-trial interval (ITI) was of 120 seconds in average; and (3) unpaired group was acclimated for 90 seconds in the conditioning chamber and presented with five unpaired stimulation of shock and tone where a 0.5-second footshock (1.3 mA) was followed by ITI of 60 seconds in average, then a 20-second tone (5 kHz, 75 dB), and another ITI of 120 seconds in average.

Testing of conditioned fear memory

Rats were tested 24 h (LTM) after conditioning for tone memory in a different chamber with black walls and a plastic floor to minimize the effect of the context. Animals were subjected to five tones (20 seconds; 5 kHz, 75 dB) with an average ITI of 120 seconds. Animal behavior was recorded and the video images were transferred to a computer equipped with an analysis program. The percentage of changed pixels between two adjacent 1-second images was used as a measure of activity. Time period spent freezing was measured, results of individual five test tones were summed and percentage freezing from the total (100 seconds) was calculated for each animal.

Laser microdissection

Rats were killed at 30 min or 5 h after paired or unpaired stimulation or 30 min after non-stimulation training (naï ve group), and their brains were rapidly removed, frozen on dry ice and kept at −80°C until further use. The brains were subsequently cut at 20 μm serial frozen sections on a standard cryostat, and sections at the level of the LA were mounted on special polyethylene naphthalate (PEN) slides (Leica, Bannockburn, IL, USA) (sections were collected from the anterior part of LA through the posterior end of LA). Slides were immediately frozen on dry ice and kept at −80°C. At the day of LMD, sections were removed from the freezer, immediately fixed in 100% ethanol for 1 min and then rehydrated in a graded series of ethanol/water (95%, 75%, 50%; each 1 min). Sections were then stained for 20 seconds in 1% Thionin in 0.05m NaCH3COO buffer, pH 4.5, dehydrated (50%, 75%, 95%, 100% ethanol, 1 min each), dried for 2 min, and immediately used for microdissection. All solutions were prepared with diethylpyrocarbonate (DEPC)-treated water. The LA was identified on a computer screen, and the entire region of interest was laser-microdissected from sections throughout the LA (AS LMD Leica; Fig. 1) into a microcentrifuge tube containing RNA extraction solution (PicoPure RNA Isolation Kit; Molecular Devices, Sunnyvale, CA, USA).

Figure 1.

Behavioral outcome of training and representative photograph depicting the laser microdissection of the lateral amygdala. (a) Animals were subjected to the naï ve, paired or unpaired protocols (n = 4 each) and tested for freezing responses when subjected again to the CS tones 24 h later. *P < 0.001. Values are mean ± SEM. (b) A thionine-stained brain before (b1) and after (b2) laser microdissection. The dissections were targeted for the dorsal LA (b3).

RNA isolation, probe preparation, microarray hybridization and image analysis

Total RNA was extracted from the LMD samples using PicoPure RNA Isolation Kit according to the manufacturer's protocol. Quality of RNA was ensured before labeling by analyzing it using the RNA 6000 NanoAssay and a Bioanalyzer 2100 (Agilent, Santa Clara, CA, USA). Samples with a 28S/18S ribosomal peak ratio of 1.8–2.0 were considered suitable for labeling. Total RNA was used for complementary DNA (cDNA) synthesis using an oligo(dT)-T7 primer and the SuperScript Double-Stranded cDNA Synthesis Kit (Invitrogen, Carlsbad, CA, USA). Synthesis, linear amplification and labeling of cRNA were accomplished by in vitro transcription using the MessageAmp aRNA Kit (Ambion, Austin, TX, USA) and biotinylated nucleotides (Enzo Diagnostics, Farmingdale, NY, USA). Ten micrograms of labeled and fragmented cRNA were then hybridized to the rat array 230A (Affymetrix, Santa Clara, CA, USA) at 45°C for 16 h. Posthybridization staining and washing were processed according to the manufacturer's instruction (Affymetrix). Finally, chips were scanned with a high-numerical Aperture and flying objective lens (FOL) in the GS3000 scanner (Affymetrix). The image was quantified using MAS 5.1 (MicroArray Suite, Affymetrix) with the default parameters for the statistical algorithm and all probe set scaling with a target intensity of 500. Quality control of hybridization was assessed: no significant differences (P > 0.05; exact Mann–Whitney test) in 3′ and 5′ ratios of GAPDH, Hexokinase and beta-actin RNAs in microarrays were found between the paired and unpaired groups. Furthermore, no significant changes in level of housekeeping genes beta-2-microglobulin (B2M) and cyclophilin A (PPIA) between paired and unpaired groups were detected.

Data analysis

The abundance of each gene transcript was scored as an average difference value by comparing the intensity of hybridization to 20 sets of perfect match 25-mer oligonucleotides relative to 20 sets of mismatched oligonucleotides using Affymetrix Genechip® Operating software (version 1.4). Results of individual genechips from paired animals were compared with those from unpaired animals using all possible pair-wise comparisons. Genes were not included in the analysis unless they were given a score of ‘I’ (increased) or ‘D’ (decreased) and ‘P’ (present) by the Affymetrix Genechip® software in all comparisons. Genes that were considered differentially expressed were those that met criteria for fold change (average > 1.5-fold increase) and signal strength (average > 100-twice the background).

Real-time PCR

The RNA of genes, which were found to be differentially expressed between paired and unpaired groups 5 h after training by the microarray experiments, were further subjected to qPCR performed using an ABI Prism 7900 Sequence Detector (Applied Biosystems, Foster City, CA, USA) and gene-specific TaqMan FAM/MGB assays (Applied Biosystems; Assays ID: Kif2-Rn01497645_m1; Lsamp-Rn00568353_m1; Hpcal4-Rn00569340_g1; Nsf-Rn01645680_g1; HSP-90β-Rn01511686_g1; glyceraldehyde-3-phosphate dehydrogenase (GAPDH)-Rn99999916_s1; beta-2-microglobulin (B2M)- Rn00560865_m1; cyclophilin A (PPIA)-Rn00690933_m1) mostly as previously described (Dracheva et al. 2008) Each 10 μl PCR reaction consisted of 2.5 μl of the relevant cDNA (diluted 25 times in H2O), a specific TaqMan assay, and 5 μl of the 2 × PCR Gold Master Mix (Applied Biosystems), which contained ROX, Aplitag Gold DNA polymerase, AmpErase UNG, dATP, dCTP, dGTP, dUTP and MgCl2. For all target and endogenous control genes (ECGs), the standard thermal cycling program was applied (Dracheva et al. 2008). All assays were run in triplicate for each sample, and only one cDNA was amplified in each reaction (monoplex). To account for the differences in the amount of the input material between the samples, the expression level of target genes was normalized to the expression levels of three different ECGs (GAPDH, B2M, PPIA) (Vandesompele et al. 2002). The relative expression level of the target transcripts was determined using Relative Standard Curve Method (RSCM; see Guide to Performing Relative Quantitation of Gene Expression Using Real-Time Quantitative PCR, Applied Biosystems). This approach provides accurate quantitative results as it accounts for differences in the efficiencies between target and control amplifications. Standard curves were generated for each target assay and for each ECG assay by the association between the threshold cycle (Ct) values and different quantities of a ‘calibrator’ cDNA. The ‘calibrator’ was prepared by mixing small quantities of all experimental samples. Using the linear equations of standard curves, the relative amounts of each target and each ECG mRNA were calculated in each sample. The relative expression level of the target mRNA was computed as the ratio between the target mRNA and the geometric mean of the three above-mentioned ECGs.

Statistics

Comparisons between RNA levels for main effect of groups were carried out using one-way analysis of variance (anova) followed by Tukey's post hoc test with an α level of 0.05. Statistical significance for differences between two groups was assessed with Mann–Whitney test.

Results

Fear conditioning induces the expression of genes encoding proteins regulating neuronal morphology and transmission in LA

Fear learning is known to induce the expression of immediate early genes encoding transcription factors in LA (Ressler et al. 2002). We were, therefore, interested in examining the expression of late response genes in LA following fear conditioning. Thus, we sampled the tissue 5 h after completion of conditioning. Rats were trained with one of the two conditions: paired training, five pairings of a tone (CS) that co-terminated with a footshock (US) and unpaired training, in which five presentations of the CS and US occurred in a non-overlapping manner (Lamprecht et al. 2002). Two groups were then analyzed; the first for behavioral outcome of training and the second for gene expression in LA. Figure 1a shows that when the CS tones were presented 24 h after training, freezing is significantly higher in the paired group (anovaF2,9 = 32.26, P < 0.001) when compared with unpaired (P < 0.001) or naï ve (no CS and US presentation; P < 0.001) groups. No significant differences were detected between the naï ve and unpaired animals (P = 0.82). Brains from the second group (from animals trained with paired or unpaired protocols) were removed 5 h after training, and the LA was dissected from rats using the laser microdissection technique (Fig. 1b). RNA was extracted, amplified and subjected to gene expression analysis using Affymetrix microarray. RNA from each rat was analyzed by a separate microarray (n = 3, each group). Genes that were differentially expressed in all possible comparisons between the paired and unpaired animals were subjected for additional analysis, in which only those genes with average signal level > 100 and with fold change > 1.50 were filtered. Figure 2 shows the genes that were expressed at higher level in the paired animals when compared with the unpaired rats 5 h after fear conditioning in LA. Fold changes between paired and unpaired groups of these genes were significantly different (P < 0.001 for all genes that were compared) from fold changes of the control genes B2M (fold change = 0.011) or cyclophilin A (PPIA; fold change = −0.011). The majority of the differentially expressed genes detected in the study encode proteins involved in the regulation of axonal growth and synaptic transmission.

Figure 2.

Fear conditioning induces gene expression 5 h after training. The average of fold changes in LA obtained for all possible comparisons between paired and unpaired animals 5 h after training (n = 3 each) detected by microarray studies. Only genes with an average fold change above 1.5 are presented. Hpcal4, hippocalcin-like 4; Hsp90β, heat shock 90 kDa protein 1, beta; Kif2, kinesin heavy chain member 2; Lsamp, limbic system-associated membrane protein; NSF, N-ethylmaleimide-sensitive fusion protein. Values are mean ± SEM. Fold changes detected for these genes between paired and unpaired groups were significantly different (P < 0.001) from those detected for control genes beta-2-microglobulin (B2M; fold change = 0.011) or cyclophilin A (PPIA; fold change = −0.011).

Temporal expression of genes in LA depends on whether the CS and US are presented in a paired or unpaired manner

To verify the findings obtained by the microarray method, the expression of the identified genes was also assessed by qPCR. In addition to the sister aliquots of the RNA samples used for the microarray analysis from animals killed 5 h after conditioning, a different set of samples was obtained from animals killed 30 min after training. We chose 30 min because it has been shown that neuronal stimulation and learning induce gene expression at that time-point (Keeley et al. 2006; Ressler et al. 2002). We explored whether the genes found to be differentially expressed between paired and unpaired groups after 5 h were also expressed 30 min following the training.

Table 1 shows that qPCR confirmed most of the changes in gene expression between the paired and unpaired groups 5 h after training in LA that were detected by Affymetrix microarray. We also found that genes were expressed in different temporal patterns (Fig. 3). At the 30-min time-point, Kif 2 (anovaF4,9 = 45.72, P < 0.001) and Lsamp (anovaF4,9 = 10.957, P < 0.003) were induced by the unpaired stimulation (unpaired vs. naï ve: Kif2- P < 0.001; Lsamp- P < 0.02) but not by the paired stimulus. Furthermore, the level of their RNAs in unpaired group was significantly higher than in the paired group (unpaired vs. paired: Kif2- P < 0.001; Lsamp- P < 0.008). At 5 h after training, the expression of Kif2, Lsamp and Hpcal4 (anovaF4,9 = 6.45, P < 0.01) was induced by the paired stimulation (paired vs. naï ve: Kif2- P < 0.001; Hpcal4- P < 0.03; Lsamp = P < 0.02) but not by the unpaired protocol. The level of the RNAs in paired group was also significantly higher than that in the unpaired group (paired vs. unpaired: Kif2- P < 0.001; Hpcal4- P < 0.02; Lsamp- P < 0.04). The Nsf (anovaF4,9 = 7.83, P < 0.005) gene transcript was significantly higher in the paired group when compared with the unpaired group 5 h after training (P < 0.007). Both groups were not significantly different from the naï ve group. The above results show that gene expression can be responsive specifically to one of the training protocols (e.g. paired only for Hpcal4) or to both paired and unpaired training where each protocol lead to different temporal pattern of gene expression (Kif2, Lsamp).

Table 1.  Fold change in RNA levels in LA between paired and unpaired groups 5 h after training
GeneAffymetrix numberFold change (Affymetrix)Fold change (qPCR)
  1. ND, not determined

Hippocalcin-like 4 (Hpcal4)1387639_at2.142.84
Hippocalcin-like 4 (Hpcal4)1375765_at2.5ND
N-ethylmaleimide-sensitive factor (Nsf)1369689_at1.761.73
N-ethylmaleimide-sensitive factor (Nsf)1369690_at1.59ND
Kinesin heavy chain (Kif2)1375508_at2.312.53
Limbic system-associated membrane protein (Lsamp)1370550_at1.561.5
Heat shock 90 kDa protein 1, beta (Hsp90β)1375335_at1.741.22
Fetal Alzheimer antigen1374283_at1.8ND
Figure 3.

Different temporal patterns of gene expression in LA in paired and unpaired groups. The results of the qPCR analysis for the genes that were differentially expressed between the paired, unpaired and naive groups (n = 3 paired or unpaired, n = 2 naive) show different temporal responses to paired or unpaired stimulation. The expression of a gene in LA depends on whether the CS and US stimuli are coupled or separated as well as on the time after training. Mean ± SEM of values detected by qPCR analysis are shown. *P < 0.05 between stimulus and naï ve and between paired and unpaired groups. In NSF, *P < 0.05 only between paired and unpaired groups.

To show that Kif2, Lsamp, Hpcal4 and Nsf expressions are induced in specific brain area, we monitored the level of these genes, using qPCR, in paired and unpaired groups in olfactory bulb, which serves as a control area. No differences were observed in gene expression between paired and unpaired groups 30 min (HpCal4-P = 0.4; Kif2-P = 0.2: Lsamp-P = 0.1; Nsf-P = 0.1) or 5 h (HpCal4-P = 0.4; Kif2-P = 1.0: Lsamp-P = 1.0; Nsf- P = 0.7) after training.

Discussion

In the current study, we examined whether fear conditioning leads to expression of specific genes in LA. Toward that end, we used the Affymetrix microarray to compare the level of gene expression in LA 5 h after training in animals that underwent associative fear conditioning (paired CS–US) as compared with non-associative training (same sensory simulation but in a non-overlapping pattern). We detected differential expression of genes encoding proteins that are involved in neuronal morphogenesis and synaptic transmission. We extended these findings by monitoring the temporal pattern of the expression of these genes at 30 min and 5 h after training using qPCR. We discovered that the patterns of expression depended on the training protocol and time after training, inasmuch as a given gene could be expressed late following fear conditioning but early following unpaired training.

We detected seven genes that were differentially expressed between paired and unpaired groups out of all genes represented in microarray. The relatively low number of differentially expressed genes between the paired and unpaired groups detected 5 h after training may be a result of several factors: First, the tone or shock per se are known to induce gene expression in amygdala (Campeau et al. 1991). These genes will not be detected in the type of comparison performed in our study. Second, we have used a stringent analysis procedure, where only genes that were detected to be differentially expressed after all possible comparisons of the paired and unpaired animals were considered. This approach reduces false positives. Third, although we have dissected the LA specifically, this brain area contains many cells that do not participate in fear conditioning. These cells may express basal level of the same genes that are induced by fear conditioning and thereby could mask their detection. Dissection of specific cells that participate in fear memory in LA may increase signal to noise ratio and unveil new genes participating in fear conditioning.

Two genes involved in axonal morphology were found to be differentially expressed: the Kinesin heavy chain member 2 (Kif2) and limbic system-associated membrane protein (Lsamp). Kif2 is a member of the kinesin family of motor proteins involved in neuronal morphogenesis (Hirokawa & Takemura 2004). It has been found to regulate microtubule depolymerization in neurons and is suggested to control the length of the axons (e.g. Kif2a−/− mice show aberrantly elongated axonal branches (Homma et al. 2003)). Kif2 has also been shown to translocate proteins, such as the growth-cone-specific IGF-1 receptor, to growth cone as well as to play an important role in expansion of the nerve growth cone (Morfini et al. 1997; Pfenninger et al. 2003). Kif2 gene was shown to be expressed following nerve growth factor (NGF) or brain-derived neurotrophic factor (BDNF) application in tissue culture (Morfini et al. 1997; Pfenninger et al. 2003). Lsamp also encodes a protein that regulates axonal growth. Lsamp has been found on the surface of neurons and their growing axons (Horton & Levitt 1988) and is expressed mainly in the limbic areas including the perirhinal, hippocampus, cigulate, amygdala and limbic thalamic neurons (Pimenta et al. 1995). Lsamp has homophilic binding property and is suggested to mediate interactions between neuronal populations of the limbic system. It mediates neuronal growth (e.g. expression of Lsamp in neurons facilitates neurite outgrowth) and is involved in proper targeting of neuronal pathways (e.g. inhibition of its activity in vivo results in abnormal neuronal connectivity) (Keller et al. 1989; Pimenta et al. 1995). Thus, Lsamp, which is involved in formation of homophilic binding, could be important for proper targeting of the growing axons in LA (as observed in neuronal circuits of the developing hippocampus), whereas Kif2, which mediates axonal morphology, may possibly support alteration in axonal patterning in LA.

The role of presynaptic axons in fear conditioning is further supported by our finding of an increase in the expression of Hippocalcin like-4 (Hpcal4). Hpcal4 [also known as visinin-like protein 2 (VILIP-2) or neural visinin-like protein 2 (NVP-2)] is a calcium-binding protein that belongs to the neuronal Ca 2+ sensor superfamily (Burgoyne 2007). Hpcal4 (VILIP-2 ) has been shown to slow the rate of inactivation of Cav2.1 calcium channels (Few et al. 2005; Lautermilch et al. 2005). Those channels conduct P/Q type calcium currents that initiate exocytosis of neurotransmitters (Regehr & Mintz 1994; Takahashi & Momiyama 1993). By regulating presynaptic Cav2.1 channels, Hpcal4 may be directly involved in shaping presynaptic calcium transients, thus influencing the time–course and the amount of neurotransmitter release. Our finding that another gene with 89% similarity to mouse Hpcal4 (1375765_at) is also differentially expressed between the paired and unpaired groups suggests that regulation of calcium transients in LA after fear conditioning may require several calcium sensor proteins. The increase of Hpcal4 could represent cellular processes that accompany changes in axonal morphogenesis namely the need for an increase in presynaptic proteins or processes regulating vesicle release in preexisting presynapses.

Another gene that was differentially expressed between the paired and unpaired groups and also encodes a protein that is intimately involved in synaptic transmission is the N-ethylmaleimide-sensitive fusion (NSF) protein (Zhao et al. 2007). In presynapses, NSF is involved in vesicle fusion cycle by forming complexes with soluble NSF attachment protein (SNAP) to catalyze the disassembly of cis–soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complexes for reactivation (Brunger 2001). In postsynpases, NSF is involved in insertion of AMPA receptors into the synapse: Inhibition of NSF-GluR2 interactions leads to the decrease in synaptic currents (Lee et al. 2002). Furthermore, NSF regulates GABAA receptor levels in synapse and GABAB receptors desensitization (Goto et al. 2005; Pontier et al. 2006). Thus, changes in NSF expression may modulate both pre and postsynaptic transmission in LA by regulating vesicle cycle, excitatory (AMPA) and/or inhibitory (GABA) transmission. This, in turn, may lead to alteration of synaptic efficacy after learning.

Our qPCR experiments confirmed the results obtained by the Affymetrix microarray analysis and showed different temporal patterns of gene expression in LA following the training. Interestingly, the level of Kif2 and Lsamp RNAs in paired and unpaired groups depends on the time after training. Kif2 and Lsamp were found at higher level in LA of paired animals when compared with unpaired groups 5 h after training. In contrast, the levels of Kif2 and Lsamp were higher at the unpaired group 30 min after training. This may represent a cellular mechanism where genes are induced shortly after stressful events (the US in unpaired), but if the stressful event (US) is coupled to a meaningful sensory stimulus (CS), these genes are expressed in a different pattern engaging more robustly with the later consolidation processes of fear memory. Consistently, exposure of rats to stressful situation (cat odor) for 30 min has previously been shown to induce rapid expression of Lsamp in the rat amygdala (including basolateral, central and medial nuclei) detected at the end of the trial (Kõks et al. 2004). In addition, rats with low exploratory activity in the elevated plus-maze (anxious rats) have shown an increased level of Lsamp gene in amygdala compared with high exploratory rats (Nelovkov et al. 2006). Furthermore, the basic neuroanatomical organization and sensory and motor development are normal in the Lsamp KO mice (Catania et al. 2008); however, these mice exhibit heightened response to novelty in several behavioral tests, which may reflect an altered response to environmental stressors. Lsamp may, therefore, mediate changes in the fine neuronal connectivity that subserve these behaviors. Collectively, these results suggest that expression of the Lsamp gene in amygdala mediates emotional stress associated with the foot shock (US) in the unpaired stimulation 30 min after training but not 5 h afterward, as at this time the level of Lsamp in unpaired group is not different from controls. At that time-point (5 h after training), Lsamp is induced only after paired (learning) stimulation.

Alternatively, changes in Kif2 and Lsamp may represent cellular mechanisms responding differently to stimuli that are close in time (paired CS–US) compared with stimuli that are dispersed in time (unpaired). These stimuli patterns lead to different behavioral responses. The paired CS–US stimulation leads to fear learning, whereas the unpaired may lead to learning of safety (Rogan et al. 2005). In the later case, the tone represents a safety situation when it is not coupled to a footshock. Thus, the differential expression of the genes in response to paired or unpaired stimuli may underlie cellular mechanisms mediating fear or safety learning.

Another possibility is that genes expressed in the unpaired group could mediate contextual fear memory formation in LA. It has been shown that tone-shock paired training induces fear memories to the tone, whereas unpaired training do not (Lamprecht et al. 2002; Fig. 1a). However, unpaired training induces significantly increased contextual fear memory when compared with the paired group (Trifilieff et al. 2007). This study also shows that unpaired training induces an activation of ERK1/2 in LA. Thus, although we have habituated the rats to the context for 4 days before training, gene expression in LA in the unpaired group may result from context–shock association.

It would be interesting to examine the upstream molecular mechanisms whereby neurons use to translate the temporal relationships between the CS and US into differential gene expression. The difference in the patterns of Kif2 and Lsamp induction in response to the paired and unpaired stimulation might indicate that these training protocols engage different intracellular signaling pathways in amygdala neurons. Unpaired stimulation may activate transcription factors rapidly by posttranslational modifications (e.g. phosphorylation) that can, in turn, facilitate their binding to the Kif2 and Lsamp promoters leading to fast transcription of these genes. Paired stimulation, however, may induce more elaborated pathways leading first to the expression and translation of transcription factors that then bind to the genes' promoters leading to the late induction of Kif2 and Lsamp. These cellular pathways have been shown to be used for the induction of transcription factors followed by expression of late response genes (Ginty et al. 1992). Differential activation of signaling molecules at the synapse may engage different nuclear signaling pathways and, in turn, induces distinct patterns of gene expression (West et al. 2002). For example, stimuli leading to increased calcium influx at the synapse may result in different activation of transcription factors than stimuli leading to elevated cAMP levels at the synapse (Kornhauser et al. 2002). Future studies will attempt to understand how the function of these genes at different time-points in LA leads to neuronal and behavioral responses underlying fear conditioning.

Acknowledgments

This research was supported by grants to J. E. L. (National Institute of Health Grants P50 MH058911, R01 MH046516 and National Science Foundation Grant IBN-0131433) and to S. D. (VISN3 Mental Illness Research and Education Clinical Center). We thank the Memorial Sloan-Kettering Cancer Center Genomics Core laboratory for microarray assay and analysis.

Ancillary