Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2010 Mar 26.
Published in final edited form as: Neuron. 2009 Mar 26;61(6):917–929. doi: 10.1016/j.neuron.2009.01.029

Selective gating of glutamatergic inputs to excitatory neurons of amygdala by presynaptic GABAb receptor

Bing-Xing Pan 1,*, Yulin Dong 2, Wataru Ito 1, Yuchio Yanagawa 3, Ryuichi Shigemoto 2, Alexei Morozov 1,*
PMCID: PMC2673456  NIHMSID: NIHMS99973  PMID: 19324000

Abstract

GABAb receptor (GABAbR)-mediated suppression of glutamate release is critical for limiting glutamatergic transmission across the central nervous system. Here we show that, upon tetanic stimulation of afferents to lateral amygdala, presynaptic GABAbR-mediated inhibition only occurs in glutamatergic inputs to principle neurons (PNs), but not to interneurons (INs), despite the presence of GABAbR in terminals to both types of neurons. The selectivity is caused by differential local GABA accumulation; it requires GABA reuptake, and parallels distinct spatial distributions of presynaptic GABAbR in terminals to PNs and INs. Moreover, GABAbR-mediated suppression of theta-burst induced long-term potentiation (LTP) occurs only in the inputs to PNs, but not to INs. Thus, target cell-specific control of glutamate release by presynaptic GABAbR orchestrates the inhibitory dominance inside amygdala and may contribute to prevention of non-adaptive defensive behaviors.

Introduction

The amygdala is responsible for formation and storage of fear memories (Davis, 2000; Muller et al., 1997). It differs from other brain regions by the low firing due to a strong inhibitory tone (Bordi et al., 1993; Pare and Collins, 2000) assumed to be essential for an organism to respond appropriately to sensory signals. When a signal indicates a threat, the “silence” is broken, allowing excitatory neurons to fire and activate the downstream defensive circuits (Quirk et al., 1995; Repa et al., 2001). Conversely, when a signal does not predict danger, it is suppressed by the extensive interneuronal network. In pathological states, this inhibition may become impaired, and even neutral signals can cause non-adaptive fear and anxiety (Quirk and Gehlert, 2003; Rodriguez Manzanares et al., 2005).

The lateral amygdala nucleus (LA) serves as the major amygdala entrance for sensory information from cortical and subcortical areas (LeDoux, 2007). Upon activation of LA afferents, INs release GABA, which hyperpolarizes PNs through GABA receptor and inhibits their firing (Lang and Pare, 1998). Besides acting postsynaptically, GABA can diffuse out of the synaptic cleft and suppress neighboring glutamatergic afferents via presynaptic GABAbR. Such suppression has been found in many areas of the central nervous system including the hippocampus, (Isaacson et al., 1993), cerebellum (Dittman and Regehr, 1997) and the amygdala (Yamada et al., 1999), and mainly on glutamatergic projections to PNs. Yet, very little is known about the GABAbR gating of inputs to INs.

The finding that pharmacological activation of presynaptic GABAbR similarly suppresses glutamatergic inputs to INs and PNs (Lei and McBain, 2003; Porter and Nieves, 2004) raises a possibility that physiological stimuli may be equally effective in recruiting this receptor on inputs to INs and PNs. Indeed, in rat neocortex, repetitive firing of bitufted interneurons elicits presynaptic GABAbR-dependent inhibition of their glutamatergic afferents (Zilberter et al., 1999). If the same inhibitory mechanism was present in LA, it would depress glutamatergic inputs to INs and compromise the strong inhibitory tone which is needed for suppression of inappropriate defensive behaviors. This rationale implies that, in LA, GABAbR-mediated presynaptic inhibition of inputs to INs should differ from those to PNs.

To investigate regulation of glutamatergic release in terminals targeting PNs and INs, we recorded from both types of cells, which were readily distinguished in transgenic mice expressing GFP selectively in INs. Physiological activation of inhibitory network selectively inhibited glutamatergic terminals to PNs, but not to INs. This selectivity arose from differences in GABA diffusion in the vicinity of PNs and INs, and was related to distinct distributions of presynaptic GABAbR between the two types of terminals. This target cell-specific control of glutamatergic release may underlie the prevalence of inhibition inside LA and may help in preventing inappropriate defensive response to sensory signals.

Results

Target specific, GABAbR-mediated presynaptic suppression of glutamatergic inputs to LA

To examine the modulation of glutamatergic transmission in the LA afferents by synaptically recruited interneuronal network, we applied a brief tetanus (priming) to either thalamic or cortical input and recorded responses from either PNs or INs inside LA to test-pulse given in the parallel non-primed input 100 ms after the tetanus (Fig. 1A). As PNs and INs receive inputs from cortical and subcortical afferents, we performed priming of a heterologous pathway to avoid homosynaptic plasticity induced by repetitive stimuli (Shaban et al., 2006). 100 μM picrotoxin was routinely added in ACSF to block GABAa receptor (GABAaR). We first confirmed that the cortical and thalamic inputs were stimulated in isolation from each other. If the inputs were activated independently, EPSC evoked by their simultaneous stimulation (recorded sum EPSC) should be equal to the sum of EPSCs evoked by stimulation of each pathway (predicted sum EPSC) (Tsvetkov et al., 2004). We found no difference between the predicted and recorded sum EPSCs in both cell types (ratios between recorded and predicted sum EPSCs, PN: 1.01±0.03, n=7 cells/3 mice, p=0.952; IN: 0.99±0.04, n=8 cells/4 mice, p=0.851 when compared to 1) (Supplementary Fig. 1), confirming that the two pathways to either LA PNs or INs are activated independently.

Fig. 1. Target-specific and activity-dependent suppression of glutamatergic transmission to LA by presynaptic GABAbR.

Fig. 1

Open in a new tab

(A) Upper: electrode placement. Stimulation electrodes (black) on cortical and thalamic inputs, and recording electrode (gray). Lower: stimulation patterns. Single pulses in either cortical or thalamic input alone, or preceded (100 ms delay) by priming (10 pulses at 200 Hz) of the parallel input. (B) Left: inhibition of cortically evoked EPSCs in PNs and INs by thalamic priming. Right: inhibition of thalamically evoked EPSCs by cortical priming. Insets show averaged responses of representative neurons to test stimulation alone (solid), or preceded by tetanus in the parallel input (dashed). Stimulus artifacts were truncated for clarity. (C) EPSCs evoked in a PN (upper) and IN (bottom) by the minimal stimulation of cortical input alone (left), or preceded by thalamic priming (right). (D) Summary plots showing changes in the failure rates (left) and potencies (right) of the cortically-evoked minimal responses in PNs and INs following thalamic priming. (E) Effects of 10 μM CGP 52432 in the bath and 100 μM GDP-βS in the pipette on heterosynaptic inhibition in PNs. Stimulus artifacts were truncated for clarity. (F) Positive correlation between changes in 1/CV2 and amplitudes of cortically evoked EPSCs following thalamic priming. (G) and (H): Dependence of heterosynaptic inhibition of cortical EPSCs in PNs and INs on the intensity (G) and frequency (H) of the thalamic tetanus. *p<0.05, **p<0.01.

The priming of the thalamic input suppressed cortically-evoked EPSCs in PNs by 34.7±4.6% (p<0.001, n=18 cells/8 mice) but had negligible effect on EPSCs in INs (% suppression, 4.6±3.7, n=13 cells/6 mice, p=0.282) (Fig. 1B). Similarly, the priming of the cortical input suppressed thalamically-evoked EPSCs in PNs but not in INs (% of inhibition, PN: 37.4±6.2, n=10 cells/5 mice, p<0.001; IN: 5.6±2.7, n=10 cells/5 mice, p=0.071) (Fig. 1B). Thus, heterosynaptic inhibition of EPSCs was induced by priming of either pathway and occurred only in PNs and not in INs.

Since bulk electrical stimulation may engage polysynaptic component contaminating glutamatergic EPSCs (Jungling et al., 2008), we examined whether the inhibition directly involved the inputs to PNs and tested effects of the thalamic priming on EPSCs evoked by “minimal stimulation” of the cortical input, which presumably activates only a single presynaptic fiber (Stevens and Wang, 1994). Thalamic priming substantially increased the failure rate of minimal EPSCs in PNs but not in INs (% of increase, PN: 40.1±10.0, n=8 cells/4 mice; IN: 1.3±8.3, n=8 cells/5 mice, p=0.014 in comparison with that in PNs) without affecting amplitude of successful EPSC in either cell type (% of decrease, PN: −5.6±7.6%, p=0.311; IN: 0.78±5.6%, p=0.756) (Fig. 1C, D). Thus, the target-specific heterosynaptic inhibition occurs in inputs to LA PNs and is expressed presynaptically.

Since sustained interneuronal firing increases levels of diffusible GABA, which can depress glutamatergic transmission through activation of presynaptic GABAbR (Isaacson et al., 1993; Vogt and Nicoll, 1999), we next examined the role of GABAbR in the priming-induced inhibition of glutamatergic inputs to PNs. In agreement with the findings in other brain areas, perfusion of amygdala slices with GABAbR antagonist CGP 52432 (10 μM) abolished the inhibition (% of inhibition:−3.4±1.4%, n=7 cells/3 mice, p<0.001, compared with the % inhibition in absence of CGP 52432) (Fig. 1E). By contrast, blocking the downstream signaling of postsynaptic GABAbR by including in the pipette 100 μM GDP-βS, a nonhydrolyzable G-protein inhibitor (Supplementary Fig. 2), did not affect the inhibition (% inhibition: 36.1±10.5, n=7 cells/3 mice; p=0.882 when compared with no GDP-βS control) (Fig. 1E), excluding the involvement of postsynaptic GABAbR. Moreover, the relative changes of the mean EPSC amplitudes following the priming positively correlated with the changes of (1/CV)2 (CV: coefficient variation) (r=0.808) (Fig. 1F), confirming that the priming-induced suppression of cortical EPSCs in PNs is mediated by activation of presynaptic GABAbR.

What is the origin of GABA which activates the presynaptic GABAbRs? It may come from axo-axonic inhibitory synapses formed on the presynaptic terminals. However, anatomical studies indicate that most axo-axonic synapses in the central nervous system are formed on axonal initial segments (Khirug et al., 2008; Somogyi et al., 1998), whereas the synapses on presynaptic terminals are rare (Barbaresi, 2005; Wang et al., 1997). As no such synapses have been reported in amygdala, the likely origin of GABA was its spillover (Isaacson et al., 1993). Since the amounts of diffusible GABA should depend on the activation of interneuronal network by the priming, we investigated how modulation of priming would affect the suppression of cortical EPSCs. In PNs, the suppression increased progressively with increasing intensity and frequency of priming, whereas in INs, only high intensity priming caused moderate suppression. Moreover, the suppression in INs did not show clear dependence on the frequency (Fig. 1G and H). The overall dependence of EPSC suppression on the intensity and frequency of priming was higher in PNs than in INs (repeated ANOVA, intensity: F[2, 4]=4.9, p=0.003, data from 6 PNs and 5 INs; frequency: F[2, 4]=3.7, p=0.012, data from 6 PNs and 5 INs).

Different distribution of GABAbR in excitatory terminals to PNs and INs

In LA, glutamatergic afferents project to PNs and INs. How can GABA spillover selectively inhibit terminals to PNs? One explanation would be the absence of GABAbR in the terminals targeting INs. To test this possibility, we examined the expression of presynaptic GABAbR by electron microscopy (EM) using the pre-embedding immunogold method. Unexpectedly, the receptors were found not only in terminals to PNs, but also in those to INs (Fig. 2A–D). The analysis of 57 randomly selected GABAbR-positive glutamatergic terminals showed that the average number of immunoparticles in terminals to PNs was ~20% higher than that in terminals to INs (PN: 4.28±0.28/terminal; IN: 3.56±0.22/terminal, p=0.045). In both types of terminals, most GABAbRs were located along extrasynaptic plasma membrane. The proportion of extrasynaptic GABAbR in terminals to INs (92.5 %) was moderately higher than in those to PNs (72.2 %). Analysis of distances between individual gold particles and the closest edge of presynaptic active zone revealed that GABAbR in terminals to INs was located further away from synaptic active zone than those in terminals to PNs (p=0.021, the Kolmogorov-Smirnov test) (Fig. 2E, F).

Fig. 2. Expression of GABAbR in glutamatergic terminals to PNs and INs.

Fig. 2

Open in a new tab

(A–D) Electron micrographs of LA showing immunoreactivity for GFP and GABAb1 subunit. Arrows show immunogold particles corresponding to presynaptically localized GABAbR, arrowheads point to DAB staining revealing GFP expressed in INs. Distribution of GABAb1 subunit in terminals which make asymmetrical synapses on PN (A, B) and IN (C, D). Scale bar: 200 nm. T: terminal; PNd: dendritic shaft of principle neuron; PNs: spine of principle neuron; IN: interneuron. (E) Percentage of particles in 50 nm bins as a function of distance from the nearest edge of synapses. (F) Cumulative histograms of the distributions in (E) (p=0.021, the Kolmogorov-Smirnov test).

Given the lack of GABAbR-mediated inhibition of glutamatergic terminals to INs, but the presence of the receptor in these terminals, we examined whether those receptors were functional. We compared the effects of bath applied GABAbR agonist baclofen on cortically-evoked EPSCs in PNs and INs. The postsynaptic GABAbR-mediated currents were blocked by either replacing K+ for Cs+, or including GDP-βS in the pipette solution. Baclofen (10μM) inhibited transmission in cortical inputs to both PNs and INs (Fig. 3A, B). In parallel with the EM results showing moderately higher expression of GABAbR in terminals to PNs, the inhibition was moderately stronger in cortico-PN than in cortico-IN synapses (% inhibition, PN: 75.4±3.0, n=12 cells/5 mice; IN: 64.1±4.3, n=9 cells/4 mice, p=0.038) (Fig. 3C). Although the differences in the receptor expression and functionality were significant, they appeared too small to account for the high selectivity of presynaptic inhibition. An additional mechanism for the selectivity could be differential accumulation of the receptor ligand GABA around PNs and INs.

Fig. 3. Stronger inhibition by baclofen of cortico-PN than cortico-IN transmission.

Fig. 3

Open in a new tab

(A) Suppression of cortically evoked EPSCs by baclofen and its reversal by CGP 52432 in a PN (filled circle) and IN (open circle). EPSC amplitudes normalized to the mean value of baseline. (B) Effects of baclofen and CGP 52432 on basal EPSCs in 12 PNs (left) and 9 INs (right). (C) Summary of baclofen-induced inhibition in PNs and INs. *p<0.05, **p<0.01.

Spillover results in higher levels of GABA around PNs than INs

To compare GABA spillover around PNs and INs in response to a short-lasting tetanus in LA afferents, we used an outside-out membrane patch from a randomly selected PN as a sensor of spilled GABA (Isaacson et al., 1993). As PNs and INs are intermingled inside LA, and only their somatic areas can be distinguished under microscope, we measured GABA concentration near the soma. PN-IN sets in close proximity (less than 10 μm, an estimated distance between the nearest edges of two somas) were selected, and the pipette with the patch held at 0 mV was slowly reinserted into the slice and approached to the surface of each cell in a random order (Fig. 4A). The tetanus readily evoked brief channel activity, which was blocked by picrotoxin (100 μM), confirming the GABAaR origin of the evoked currents (Supplementary Fig. 3). For every PN-IN set tested, the evoked GABAaR activity approached to PN always exceeded that near IN (Fig. 4B), as reflected in a higher mean charge (PNs: 680.9±278.9 fC; INs: 205.4±112.8 fC; n=8 sets/4 mice, p=0.036) and amplitude (PNs: 9.1±2.7 pA; INs: 2.7±0.7 pA; n=8 sets/4 mice, p=0.021) of the currents (Fig. 4C, D). These results suggested that upon tetanic stimulation, more GABA was accumulated around PNs than INs in the vicinity of soma. However, since the heterosynaptic suppression occurs at terminals and most of them are located around dendritic tree, it was necessary to determine how priming affected levels of GABA in the vicinity of dendrites.

Fig. 4. Tetanic stimulation of amygdala afferents generates more diffusible GABA around PNs than around INs.

Fig. 4

Open in a new tab

(A) Example of a PN-IN set. Upper left: a set of PN and IN visualized under infrared DIC optics; upper right: green fluorescence emitted by IN. A pipette containing an outside-out patch positioned at the surface of PN (lower left) and IN (lower right). (B) Consecutive traces of GABAaR current across an outside-out patch in response to short lasting tetani (5 stimuli at 200 Hz) when the same patch was positioned at the surfaces of PN (left) or IN (right). Averaged traces are shown on the bottom. Stimulus artifacts were truncated for clarity. (C, D) Summary plots of charge (C) and amplitude (D) of GABAaR currents recorded near PN-IN sets. (E) Subcellular distribution of GABAbRs in PNs and INs revealed by EM. (F) Left: averaged traces of GABAbRs currents in a PN-IN set following short-term tetanus. Right: Changes of holding current in the same PN-IN set following perfusion of baclofen. (G) Ratios between tetanus- and baclofen-evoked GABAbRs currents in PN-IN sets. *p<0.05, ***p<0.001.

One strategy to investigate accumulation of GABA around dendrites is to measure postsynaptic GABAbR-mediated current which reflects levels of extrasynaptic GABA. It has been reported that in hippocampus, GABAbR was mainly expressed at neuronal periphery rather than soma (Kulik, A, et al 2003). The analysis of the subcellular distribution of postsynaptic GABAbR by EM in LA, revealed that in both PNs and INs the majority of GABAbRs were located in dendrites (numbers of golden particles per μm, PN, soma, 0.15±0.08, n=10; spine: 1.80±0.11; n=95; p<0.001 when compared to soma; dendritic shaft: 1.22±0.11, n=139, p<0.001; IN, soma, 0.00±0.0, n=5, dendrite: 0.41±0.08, n=90, p<0.001) (Fig. 4E). Thus, we reasoned that the GABAbR-mediated current evoked during the tetanus would mainly result from diffused GABA at neuronal periphery and could be used as a sensor of extrasynaptic GABA in the vicinity of dendritic tree.

To compare the GABAbR currents in PNs and INs, we performed simultaneous recording from sets of PN and IN located in close proximity. Since the amplitudes of these currents are also determined by the number of GABAbR, we used the ratio between the amplitude of GABAbR current evoked by tetanus and GABAbR current evoked in the same neuron during subsequent perfusion of the slice with baclofen (10μM) as an index reflecting the amount of GABA spilled at the neuronal periphery. For every PN-IN set tested, this ratio was always higher in PNs than in INs (PN: 0.57±0.09, IN: 0.31±0.05, n=7 sets/5 mice, p<0.001) (Fig. 4F, G), suggesting that the tetanus resulted in higher GABA accumulation near the periphery of PNs.

However, somatic recoding of the dendritically-generated GABAbR current is also influenced by dendritic cable filtering. If synapses on dendrites were closer to soma, the weaker filtering could result in larger GABAbR current. To compare distribution of GABAergic synapses along dendritic tree in PNs and INs, we examined kinetic properties of mIPSC, which are mediated by GABAa receptors. The relationship between rise time (10–90 %) and amplitude of mIPSC had a characteristic ‘triangle’ shape (Soltesz et al., 1995) in both PNs and INs (Supplementary Fig. 4A, B). This shape indicates dendritic filtering in both types of cell, because the slower IPSCs have a tendency to have smaller amplitudes. Consistent with previous studies, this relationship was not linear, likely due to high variability in amplitude and kinetics of the underlying synaptic currents (Bekkers and Stevens, 1996; Soltesz et al., 1995). The comparison of the rise time revealed a larger proportion of mIPSCs with a longer rise time in PNs (Supplementary Fig. 4C). As such IPSC typically originate from more distal GABAergic synapses (Maccaferri et al., 2000), this result indicates that GABAergic synapses on dendrites of PNs are further from soma than the synapses on dendrites of INs. Thus, our measurements of GABAbR currents may even underestimate the differences in local GABA accumulation around periphery of PNs and INs.

To determine the mechanisms underlying the differential accumulation of GABA, we investigated two factors which could affect levels of GABA: 1) the strength of GABAergic input to the target postsynaptic neurons; 2) the efficacy of GABA clearance.

In LA, inhibitory inputs to PNs are stronger than those to INs

To compare the strength of GABAergic inputs in PNs and INs, we evoked IPSCs by stimuli of increasing intensity. To minimize effects of factors that may influence amplitude of evoked EPSC, such as position and properties of the stimulation electrode, and the quality of individual slices, we performed simultaneous recording from seven sets of PN and IN that were in close proximity to each other (<10 μM). In all PN-IN sets the input-output slopes of PN IPSCs exceeded those of IN IPSCs (PNs: 21.5±4.4/106; INs: 6.4±2.0/106, n=7 sets/4 mice; p=0.011) (Fig. 5A), indicating higher synaptic efficacy of inhibitory transmission to PNs than INs. Meanwhile, miniature IPSCs recorded in the presence of tetrodotoxin (TTX, 1 μM) were more rare in INs (Supplementary Fig. 5), but did not differ in their amplitudes between PNs and INs (PN: 19.3±1.0 pA, n=8 cells/3 mice; IN: 18.1±0.9 pA, n=9 cells/3 mice, p=0.398) (Fig. 5B). Hence, the greater efficacy of inhibitory transmission in PNs was caused by their stronger GABAergic innervation and not by higher postsynaptic quantal responses.

Fig. 5. Synaptic efficacy of inhibitory transmission to PNs exceeds that to INs.

Fig. 5

Open in a new tab

(A)Input-output curves of IPSCs, cortically evoked and simultaneously recorded in sets of PNs and INs (left). The curve slope comparison within PN-IN sets reveals higher slopes in PN (right). (B) Representative traces of mIPSCs in PN and IN (left) and summary plots of mIPSC amplitude (right). (C) Input-output curves of EPSCs in sets of PNs and INs (left). The curve slopes do not differ within PN-IN sets (right). (D) Representative traces of mEPSCs in PN and IN (left) and summary plots of mEPSCs amplitude (right). *p<0.05.

In contrast to GABAergic inputs, no difference was found between synaptic efficacy of glutamatergic inputs (input-output curve slope, PNs: 5.0±1.5/106; INs: 5.4±1.7/106, n=7 sets/4 mice; p=0.878). The mean amplitudes of miniature EPSCs were also statistically indistinguishable (PN: 17.7±0.7 pA, n=8 cells/3 mice; IN: 19.7±2.3 pA, n=9 cells/3 mice, p=0.435) (Fig. 5C, D). These results indicate similar glutamatergic innervations of PNs and INs.

Role of GABA reuptake in sustaining target specificity of presynaptic inhibition in LA

The concentration of extrasynaptic GABA is determined by GABA release and reuptake. To determine whether the reuptake contributes to the target-specificity of presynaptic inhibition, we examined the effects of GABA reuptake inhibitor SKF 89976A (30μM). Since transmitter reuptake is a temperature-dependent process, the experiment was performed at 361. At this temperature thalamic priming also suppressed cortical EPSCs in PNs but not in INs (% inhibition, PN inputs: 37.5±7.0, n=9 cells/5 mice, p<0.001; IN inputs: 5.8±3.2; n=8 cells/4 mice, p=0.115). In the presence of SKF 89976A, the inhibition occurred in both cell types (% inhibition, PN: 49.1±6.9%, n=9 cells/5 mice, p<0.001; IN: 32.2±4.5%, n=8 cells/4 mice, p<0.001) indicating that GABA reuptake was required for the target-specificity of presynaptic inhibition. Meanwhile, the SKF 89976A-dependent component of the inhibition in INs was higher than that in PNs (inhibition increase by SKF 89976A, IN: 26.4±4.3%; PN: 11.5±4.2%, p=0.025) (Fig. 6A, B), indicating a possibility that GABA clearance was more efficient at INs, and its inhibition resulted in greater accumulation of GABA in the IN vicinity. However, considering that 1) in the above experiments, much more GABA accumulated around PNs than INs; 2) at higher GABA concentration, GABAbR is closer to saturation by the ligand (Kaupmann et al., 1998), it was also possible that GABA reuptake was similar around PNs and INs, but the same GABA rise following reuptake block suppressed EPSCs in INs more strongly than in PNs.

Fig. 6. Effects of GABA reuptake blocker on GABAbR-mediated presynaptic inhibition of glutamatergic inputs to PNs and INs.

Fig. 6

Open in a new tab

(A) Bath application of SKF 89976A augments the inhibition of cortically evoked EPSCs in both PNs and INs by thalamic tetanus of regular or weak intensity. (B) SKF 89976A-mediated inhibition in PNs and INs following thalamic tetani of different intensities. Plots show increases in the inhibition by the drug. (C) Average traces of GABAbR currents from a PN and IN before (gray) and after application of SKF 89976A (black). Stimulus artifacts were truncated for clarity. (D, E) Summary plots of increases in amplitudes (D) and decay constants (E) of GABAbR currents in PNs and INs following SKF 89976A application. *p<0.05, **p<0.01, ***p<0.001.

To differentiate between these possibilities, we reduced the intensity of the priming to decrease the amount of GABA accumulated around PNs (see methods) and reexamined the effect of SKF 89976A on the tetanus-induced inhibition of EPSCs in PNs. The inhibition by the weak priming became negligible in PNs, similar to that in INs when the regular intensity priming was applied (% inhibition: 7.0± 2.8, n=8 cells/3 mice, p=0.777, as compared with inhibition in INs) (Fig. 6A). Subsequent application of SKF 89976A markedly augmented the inhibition, and SKF 89976A-dependent component of the inhibition was similar to that obtained in INs (p=0.918), but stronger than that in PNs in the experiments with the regular intensity tetanus (p=0.048) (Fig. 6B).

Thus, GABA reuptake was similar around PNs and INs and the different effects of SKF 89976A on cortical inputs to these two types of neurons mostly arise from the different accumulation of GABA in their vicinity. This conclusion was confirmed by testing the effect of SKF 89976A on evoked GABAbR-mediated IPSCs (Fig. 4F). When we adjusted stimulus intensity to evoke currents of similar basal amplitudes (~20 pA) in both neuronal types, SKF 89976A augmented IPSCs similarly in both PNs and INs, as reflected in identical increases in current amplitudes (% increase: PN: 131.2±51.3, n=8 cells/4 mice; IN: 149.1±19.0, N=8 cells/4 mice; p=0.741) and decay constants (% increase: PN: 116.5±20.7; IN: 139.4±36.0; p=0.592) (Fig. 6C–E).

Presynaptic GABAbR suppresses LTP in cortical input to PNs but not INs

Synaptic plasticity in the sensory afferents to LA is believed to be a mechanism for fear learning (Dityatev and Bolshakov, 2005; LeDoux, 2000), and has been found in inputs not only to LA PNs, but also to INs (Szinyei et al., 2007). Recently, presynaptic GABAbR was shown to suppress non-associative plasticity in PNs and prevent fear generalization (Shaban et al., 2006). Given the difference in presynaptic inhibition of glutamatergic inputs to PNs and INs, we tested whether plasticity in these afferents was also regulated differently by the presynaptic GABAbR. To induce LTP, we selected theta stimulation, which mimics neuronal firing during oscillatory activity in LA (Seidenbecher et al., 2003).

The stimulation of cortical inputs in the presence of 100 μM picrotoxin evoked moderate LTP in INs (% of baseline: 128.8±7.9, n=11 cells/4 mice; p=0.005), but not in PNs (% of baseline: 109.0±7.7, n=10 cells/4 mice, p=0.269) (Fig. 7A, B). However, in the presence of GABAbR blocker CGP 52432 (10μM), LTP was readily evoked in PNs (% of baseline: 167.2±13.4, n=8 cells/3 mice, p=0.001); in contrast, LTP in INs was not affected (CGP 52432: 133.5±17.2, n=8 cells/3 mice, p=0.783, compared with LTP without CGP 52432) (Fig. 7A, B). Since CGP 52432 did not influence basal transmission in either neuronal type (data not shown), these results suggested that activation of GABAbRs during LTP induction suppressed LTP in inputs to PNs but not INs. To distinguish between the contributions of pre- versus post-synaptic GABAbR in this suppression, we blocked postsynaptic effects of the receptor by including 100 μM GDP-βS in the pipette solution (Supplementary Fig. 2). Under these conditions, theta stimulation still failed to evoke LTP in PNs (% of baseline: 112.1±6.6, n=6 cells/2 mice, p=0.124), but did evoke it when CGP 52432 was added in the bath (% of baseline: 144.2±10.7, n=5 cells/2 mice, p=0.015) (Fig. 7C) suggesting that in PNs, theta-induced LTP was suppressed by activation of pre- rather than postsynaptic GABAbRs.

Fig. 7. Presynaptic GABAbR suppresses LTP in cortical inputs to PNs but not to INs.

Fig. 7

Open in a new tab

(A–B) LTP in INs (A) and in PNs (B) in the absence (open circles) or presence (filled circles) of CGP 52432. (C) LTP in PNs with GDP-βS in the pipette (open circles), or combined with CGP 52432 in the bath (filled circles). (D) LTP in INs in the presence of SKF 89976A alone (open circles), or combined with CGP 53432 (filled circles).

To examine whether presynaptic GABAbR was indeed recruited during LTP induction and, if so, whether this recruitment occurred only in the inputs to PNs, we first analyzed how CGP 52432 affected progressive membrane depolarization in PNs and INs during LTP induction. Since action potentials often evoked during the first train of 4-stimuli interfered with the analysis, we examined membrane depolarization during the second 4-stimulus train. Inclusion of CGP 52432 resulted in a stronger increase in depolarization in PNs but not in INs (repeated ANOVA, PNs: F(2, 3)=16.9, p<0.001, 6 cells with CGP 52432 and 7 cells without CGP 52432; INs: F(2, 3)=0.27, p=0.849, 6 cells with CGP 52432 and 7 cells without CGP) (Fig. 8A, B). Second, we investigated the effect of CGP 52432 on EPSCs responses to the paired stimulation (50 ms interval) of the cortical inputs 100 ms after brief homosynaptic theta-stimuli. While theta-stimulation alone decreased paired pulse ratio (PPR) in both types of neurons (PN, control: 1.29±0.04; theta:1.04±0.03, n=9 cells/5 mice, p<0.001; IN, control: 1.30±0.08, theta: 0.96±0.07, n=5 cells/3 mice, p<0.001), the subsequent perfusion with CGP 52432 further reduced post-tetanic PPR in PNs but not in INs (PPR, PN: 0.90±0.02, p<0.001; IN: 0.95±0.05, p=0.752, compared to post-tetanic PPR without CGP 52432) (Fig. 8C and D). Since CGP 52432 had no effect on the basal PPR in either PNs or INs (data not shown), these data altogether indicated that cortical theta stimulation recruited presynaptic GABAbRs in cortical inputs to PNs but not INs.

Fig. 8. GABAbR is recruited by LTP-induction procedure.

Fig. 8

Open in a new tab

(A) EPSP induced in PNs during the second 4-pulse tetanus of the LTP induction procedure in the presence and absence of CGP 52432. Traces from representative neurons (left), summary plot showing changes in membrane potential expressed as % of EPSP amplitude in response to the first pulse in the 4-pulse tetanus. (B) Same as in (A), but for INs. (C) Averaged traces of EPSCs in a PN (upper) and IN (bottom) in response to paired pulse stimulation under control condition, or when preceded by thalamic tetanus in the absence and presence of CGP 52432. (D) Summary plots showing PPR in PNs and INs under conditions described in (C). ***p<0.001.

If the failure to recruit GABAbRs in the terminals to INs resulted from insufficient amounts of GABA, blocking the reuptake should enable suppression of LTP in INs. As expected, in the presence of SKF 89976A, INs did not show LTP (% of baseline: 102.7±4.3, n=4 cells/3 mice, p=0.579); however, when SKF 89976A was combined with CGP 52432, theta stimulation evoked LTP (% of baseline: 131.2±9.8, n=4 cells/2 mice, p=0.047), suggesting that the LTP suppression by SKF 89976A was mediated by GABAbR (Fig. 7D).

To investigate mechanisms of LTP evoked by theta stimulation in the presence of CGP 52432, we tested its Ca2+-dependence. BAPTA (10mM) in the pipette solution suppressed LTP in both PNs and INs (PN: 107.0±11.7% of baseline, n=6 cells/3 mice, p=0.007 compared with LTP without BAPTA; IN: 92.5±2.0% of baseline, n=4 cells/3 mice, p=0.031) (Supplementary Fig. 6A), indicating that the LTP requires postsynaptic Ca2+. A non-competitive NMDAR antagonist MK-801 (1mM) in the pipette solution blocked LTP in PNs but not in INs (% of baseline, PN: 111.4±13.0, n=5 cells/3 mice, p=0.016 compared with LTP without MK 801, IN: 135.2±4.4, n=4 cells/3 mice, p=0.951) (Supplementary Fig. 6B) indicating the LTP in PNs, but not in INs, requires postsynaptic NMDA receptor.

Discussion

Presynaptic GABAbR-mediated inhibition of glutamate release in the afferents to LA is target cell-specific. When the LA interneuronal network is activated by repetitive stimuli, the inhibition takes place only in synapses that target excitatory (PNs), but not inhibitory neurons (INs). The selectivity of the inhibition is achieved by higher increases in local GABA concentration around PNs than INs; it involves GABA reuptake, and correlates with differential distribution of GABAbR in terminals to PNs and INs. One consequence of this phenomenon is suppression of synaptic plasticity in cortical inputs to the excitatory, but not inhibitory neurons.

Target cell specific control of synaptic transmission in LA

Target-specific modulation of presynaptic release has been found in many areas across the CNS, in glutamatergic and GABAergic terminals (Ferraguti et al., 2005; Koester and Johnston, 2005; Marowsky et al., 2005; Pelkey and McBain, 2007; Reyes et al., 1998; Scanziani et al., 1998). Several presynaptic mechanisms are responsible for this phenomenon, including differential regulation of Ca2+ dynamics (Koester and Johnston, 2005), different Ca2+-diffusion distances between Ca2+ channels and their associated neurotransmitter release sites (Rozov et al., 2001), and selective expression of certain molecules in presynaptic terminals (Engel and Jonas, 2005; Shigemoto et al., 1996).

We found that the selective suppression of glutamatergic inputs to PNs in LA correlated with different expression and functionality of presynaptic GABAbR in glutamatergic inputs to PNs and INs. The EM analysis revealed that expression of GABAbR was slightly higher in terminals to PNs. Moreover, the proportion of the receptor expressed intrasynaptically in PN terminals was also higher. Since our EM analysis did not identify specific terminals formed by thalamic and cortical projections, there remains a possibility that GABAbRs distribution in these terminals may differ from the distribution revealed by our analysis. However, this possibility appear less likely given that moderately stronger inhibition of synaptic inputs to PNs by baclofen matched with the moderately higher expression of the GABAbR in those inputs. Yet, the differences found were too small to explain the highly selective presynaptic suppression of glutamatergic terminals to PNs.

On the other hand, the differences in accumulation of GABA in the vicinity of the terminals appear to be a major factor contributing to this phenomenon. First, measurement of GABA-evoked currents in a sniffer pipette with a GABAaR-containing patch detected higher levels of extrasynaptic GABA near the soma of PNs than INs. Second, tetanization of LA afferents activated a larger proportion of GABAbR in PNs than in INs, as revealed by a higher ratio of the tetanus-evoked to baclofen-evoked whole cell GABAbR-mediated currents. Given the EM finding of predominantly dendritic GABAbR expression, the postsynaptic GABAbR currents should mainly reflect the concentration of GABA near the dendrites. Therefore, our results indicate that a greater accumulation of GABA around the dendritic tree of PNs in response to the tetanic stimulation of LA inputs causes a stronger suppression of glutamatergic terminals projecting to PNs than to INs.

The likely reason for the differences in accumulation of GABA around PNs and INs is their different GABAergic innervation. While the synaptic efficacy of glutamatergic inputs to these two neuronal types was comparable, the efficacy of the GABAergic inputs was higher in PNs. Consistently, an earlier morphological study (Smith et al., 1998) suggests that more functional GABAergic terminals innervate PNs than INs.

The activity of the GABA transporter is also required for the selective inhibition. In the presence of the transporter blocker SKF 89976A, inhibition was no longer restricted to the inputs to PNs, but spread on inputs to INs. Notably, relative increase in inhibition by the reuptake blocker was larger in terminals to INs than in those to PNs, raising a possibility that the reuptake efficacy might be higher near PNs than near INs. Previous studies have demonstrated such differences in the hippocampus, where the reuptake was higher around pyramidal cells, because of the laminar distribution of GABA transporters (Engel et al., 1998). The possibility of differential reuptake efficacy, however, was ruled out, because SKF 89976A similarly potentiated tetanus-evoked postsynaptic GABAbR currents in PNs and INs when stimulation intensity was adjusted to equalize the basal current in PNs and INs. Why then did blocking GABA reuptake had different effects on presynaptic inhibition in PNs and INs? One possible reason is that saturation levels of GABAbR are different, depending on local GABA concentration. When the basal GABA concentration is low, GABAbR is far from being saturated by the ligand and the effect of GABA rise on the receptor activity following reuptake block may be stronger. Consistently, when we reduced the amount of diffusible GABA around PNs by decreasing priming intensity, the reuptake block produced more inhibition.

GABA diffusion in sustaining high inhibitory tone in amygdala

In amygdala, INs fire frequently, whereas PNs don’t. This inhibitory dominance has been mainly attributed to stronger postsynaptic inhibition of PNs. Several factors may account for this: first, INs receive fewer GABAergic projections than PNs (Smith et al., 1998), second, the reversal potential of their GABAa receptor current is more depolarized (Martina et al., 2001), third, their GABAbR and Ca2+-dependent potassium conductances are smaller (Lang and Pare, 1997; Lang and Pare, 1998). The present study reveals that, in addition to these differences, the selective suppression of inputs to PNs by extrasynaptic GABA also contributes to the predominance of inhibition in amygdala.

How then do the pre- and postsynaptic inhibition mechanisms cooperate? In contrast to GABAaR, which can be activated even by a single quantum of GABA, the GABAbR activation requires firing of a population of INs to produce enough GABA to overcome diffusion and uptake (Scanziani, 2000). When a weak stimulus arrives in amygdala, the amount of spilled GABA is not sufficient to activate GABAbR and the postsynaptic GABAaR may be mainly responsible for inhibition. With a stronger sensory stimulus, synchronous firing of INs will result in pooling of GABA from multiple inhibitory synapses, generating enough GABA to activate presynaptic GABAbR and inhibit glutamatergic inputs to PNs. At the same time, glutamatergic inputs to INs are not affected by spillover and thus the inhibitory tone is sustained.

One physiological consequence of the target cell-specific presynaptic inhibition appears to be selective suppression of LTP in sensory inputs to LA PNs. While theta stimulation of the cortical input readily evoked LTP in INs regardless of the presence or absence of CGP 52432, it evoked LTP in PNs only when GABAbR was blocked by CGP 52432. This action of CGP 52432 appeared to be presynaptic, because suppression of postsynaptic GABAbR-mediated currents by GDP-βS in the pipette did not affect LTP either in the presence or absence of CGP 52432. Consistent with the selective role of GABAbR in suppression of LTP, theta stimulation of the cortical input recruited this receptor only in terminals targeting PNs, as supported by two pieces of evidence; first, CGP 52432 augmented membrane depolarization during LTP induction in PNs, but not in INs, second, it decreased PPR only in PNs after theta-stimulation while having no effect on PPR prior to it. Thus, theta stimulation recruits GABAbRs to antagonize the release of glutamate only in the terminals synapsing onto PNs.

The failure of theta stimulation to recruit presynaptic GABAbRs in cortico-IN synapses arises from insufficient accumulation of extrasynaptic GABA. Indeed, when GABA reuptake was blocked with SKF 89976A, LTP in cortico-IN synapses became suppressed, and this suppression was reversed by co-application of CGP 52432.

While the parsimonious explanation of the dependency of theta-induced LTP in our experiments on GABAbR block is a direct suppression of glutamatergic release by GABAbR in the tested synapses, indirect effects of GABAbR from neighboring synapses are also possible. For example, inhibition of GABAbR by CGP 52432 may increase glutamate spillover and augment LTP induction by single Poisson train via postsynaptic Ca2+/NMDA receptor-independent mechanisms (Humeau et al., 2003; Shaban et al., 2006). However, our findings that LTP induced by theta stimulation in the presence of CGP 52432 in PNs did require postsynaptic Ca2+ and NMDA receptors indicate that even if it involved heterosynaptic mechanisms, most likely they were different from those recruited during the Poisson train stimulation. While postsynaptic Ca2+ was required for LTP in both PNs and INs, the LTP in INs was independent of postsynaptic NMDARs. Unlike PNs, LA INs express high levels of Ca2+-permeable AMPAR, which may mediate Ca2+-dependent synaptic plasticity in these cells (Mahanty and Sah, 1998).

Stimulation of LA amygdala afferents recruits at least two inhibitory mechanisms, the feedforward inhibition from INs triggered by LA afferents and the feedback inhibition from the INs activated by the firing of LA PNs. The presynaptic GABAbR may thus have a dual effect on PNs. By suppressing their glutamatergic afferents it may inhibit plasticity. On the other hand, by decreasing their firing it may weaken the feedback inhibition and, in theory, facilitate plasticity. However, our finding that theta stimulation failed to evoke LTP in the absence of CGP 52432 argues that it is the GABAbR-mediated suppression of glutamatergic inputs to PNs that plays the major role in gating LTP induction.

Previous studies have shown that LTP in LA can be induced by pairing and spike-timing dependent plasticity protocols without suppressing GABAbR (Pan et al., 2008; Tsvetkov et al., 2002). It raises a question why some forms of plasticity require suppression of GABAbR whereas others do not. We find that high frequency stimulation of LA afferents effectively recruits GABAbR whereas low frequency stimulation does not. This might be the reason why LTP induced by the pairing and spike-timing dependent plasticity protocols using low frequency stimulation does not depend on GABAbR.

While LTP in PNs is considered to be a synaptic mechanism for fear learning (Dityatev and Bolshakov, 2005), the tight control of its induction by presynaptic GABAbR suggests that transient removal of GABAbR-mediated inhibition may be required for fear learning. On the other hand, the easily induced plasticity in INs can serve as a mechanism for scaling up the activity of the inhibitory network to allow effective suppression of strong and repetitive sensory stimuli which do not predict danger. Despite the presence of GABAbR in glutamatergic terminals targeting INs, these receptors appear physiologically silent. One plausible possibility is that they may be recruited for the suppression of amygdala inhibitory network during fear learning or pathological states of fear. Understanding when and how presynaptic GABAb receptors on terminals to INs become activated may help to elucidate synaptic mechanisms of fear learning and non-adaptive fear.

Material and methods

Mice

Heterozygous GAD67-GFP(Δneo) mice with GFP in interneurons (Tamamaki et al., 2003) were used. All experiments were approved by NIMH Animal Care and Use Committee.

Immunohistochemistry for electron microscopy

Animals were anaesthetized by intraperitoneal injection of pentobarbital (50 mg/kg), and perfused through the aorta with 25 mM phosphate-buffered saline (PBS, pH 7.4) for 1 min, following by an ice-cold fixative containing 4% paraformaldehyde, 0.05% glutaraldehyde, 15% saturated picric acid, 0.1 M phosphate buffer (PB, pH 7.4) for 12 mins. The brains were immediately removed and the tissue blocks containing LA were cut on a vibratome (Leica VT1000S, Wetzlar, Germany) into 50 μm-thick sections and collected in 0.1 M PB.

Immunohistochemistry was carried out as described (Kulik et al., 2002). Free-floating sections were blocked in 10% normal goat serum (NGS) diluted in 50 mM Tris HCl pH 7.4, 0.9% NaCl (TBS, pH 7.4) for 1 h at room temperature followed by sequential incubation with (1) a mixture of a rabbit anti-GABAb1 antibody (B17, 1–2 μg/ml) (Kulik et al., 2002; Kulik et al., 2003) and a mouse antibody to GFP (gift from Dr. Shohei Mitani, 0.5 μg/ml) in TBS containing 1% NGS for 48 h at room temperature and (2) a mixture of goat anti-rabbit IgG coupled to 1.4 nm gold (Nanoprobe Inc., Stony Brook, NY, USA) and biotinylated goat anti-mouse antibody (1:100; Vector Labs, Burlingame, CA) for 4 h in the same solution. After several washes in PBS, the sections were postfixed in 1% glutaraldehyde diluted in the same buffer for 10 min, washed in water, followed by silver enhancement of the gold particles with HQ silver kit (Nanoprobes Inc.). After washing with PBS and TBS, the sections were incubated with ABC complex (1:100, ABC Elite Kit, Vector Labs) for 2 h and then DAB-H2O2 solution. After treatment with 1% osmium tetraoxide in 0.1 M PB, the sections were stained with uranyl acetate, dehydrated in graded series of ethanol, and flat-embedded on glass slides in Durcupan resin (Fluka, Buchs, Switzerland). Regions of interest were cut at 70 nm thickness on an ultramicrotome (Leica) and collected on grids. Following counterstaining with lead citrate, ultrastructural analyses were performed with Tecnai 12 transmission electron microscope (FEI, Eindhoven, the Netherlands).

To test the specificity of immunolabeling for GABAb1 receptor under electron microscopy, the primary antibody was either omitted or replaced with 5% (v/v) normal serum of the species of the primary antibody. Under this condition, no selective labeling was observed. To compare distribution of presynaptic GABAbR in glutamatergic terminals to PNs and INs, the Kolmogorov-Smirnov test was used.

Electrophysiology

Amygdala slices were prepared from 4–5 week old mice as described (Tsvetkov et al., 2002) with minor modifications. Mice were sacrificed by decapitation and brains were quickly removed to ice-cold oxygenated (95% O2/5% CO2) artificial cerebrospinal fluid (ACSF) containing (in mM): 119 NaCl, 2.5 KCl, 1 MgSO4, 2.5 CaCl2, 10 glucose, and 26 NaHCO3 (pH 7.30) and kept there for ~3 minutes. Slices containing LA were cut with a tissue slicer from Precisionary Instrument Inc. (Greenville, NC) and maintained at room-temperature for at least one hour before recording. Slices were transferred to a recording chamber superfused with ACSF at a rate of about 60 ml/h. Temperature was at 29±1 °C. Whole cell recordings were obtained from pyramidal cells and “local” interneurons (but not paracapsular intercalated cells) in the lateral amygdala with an EPC-10 amplifier and Pulse v8.76 software (HEKA Elektronik, Lambreht/Pfaltz, Germany). The PNs were visualized under DIC/infrared optics and the INs were detected with green fluorescence. Data were filtered at 1K Hz using the patch-clamp amplifier circuitry. Compound synaptic responses were evoked by field stimulation of the fibers entering either in the external capsule (cortical input) or internal capsule (thalamic input) by a 1–3 MΩ glass stimulation electrode filled with ACSF and positioned outside the amygdala border. In most experiments patch pipettes were filled with (in mM): 120 K-gluconate, 5 NaCl, 1 MgCl2, 10 HEPES, and 0.2 EGTA, 2 ATP-Mg, 0.1 GTP-Na. The pH was adjusted to 7.3 with KOH and osmolarity to 285 Osm with sucrose. In experiments investigating input-output curves of E(I)PSCs, miniature E(I)PSCs and presynaptic activation of GABAb receptor as well as local GABA concentration around PNs/INs, Cs2+ was used instead of K+ (the pH adjusted to 7.3 by CsOH). Patching pipette resistance was 4–6 MΩ when filled with a K+-containing solution and 3–5 MΩ when filled with a Cs+-containing solution. All membrane potentials were corrected by a junction potential of 12 mV and compound/miniature EPSCs were recorded at a holding potential of −70 mV while IPSCs were at 0 mV. Series resistance (Rs) was in the range of 10–20 MΩ and monitored throughout experiments. If Rs changed more than 20% during recording, the data were not included in analysis.

In experiments measuring inhibition of presynaptic release by priming, the stimulation intensity was adjusted to evoke similar amplitude EPSCs (~150 pA) by single pulses delivered in either pathway. “Minimal stimulation” was performed as described (Pan et al, 2008) except the stimulation intensity was adjusted to achieve EPSC failure rate of ~50%. In experiments with “reduced priming intensity”, the stimulation of thalamic pathway was decreased to generate EPSCs amplitudes equal to 1/3 of those evoked by stimulation of the cortical pathway. When testing effects of pharmacological activation of presynaptic GABAbR by baclofen on evoked EPSCs, postsynaptic GABAbR-mediated currents were blocked by either replacing internal K+ for Cs+, or by inclusion of GDP-βS in the pipette. The effectiveness of GDP-βS in blocking postsynaptic GABAbR currents in PNs and INs was verified by measuring changes of holding potential subsequent to baclofen application (Supplementary Fig 2). Since the effects of baclofen were similar in both conditions, data from experiment using Cs+ and GDP-βS were pooled together.

For assessing local GABA concentrations around the soma of PNs and INs following the activation of LA afferents, a short-lasting tetanus (5 stimuli at 200 Hz) was delivered to either cortical (5 PN-IN sets) or thalamic (3 PN-IN sets) pathway while the same pipette containing an outside-out patch from another PN was positioned at the surface of PNs and INs inside the slice tissue. The distance between selected PN and IN was measured by dividing the actual distance shown in the monitor by the microscope/camera magnification factor. The outside-out patch was obtained by gradually drawing the pipette after forming whole-cell patch until a GΩ-seal was reestablished. The intensity of tetanus was adjusted to evoke EPSC of ~150 pA in PN in response to single stimulus before forming outside-out patch. We routinely verified the stability of recording after the pipette movement between two cells by returning the pipette to the first neuron and repeating the recording (Supplementary Fig. 3). The amplitude of GABAaR current was obtained by averaging 5 consecutive traces and determining the difference between the mean current during the baseline and the mean current over a 2–3 ms window at the current peak of the average trace.

To compare levels of extrasynaptic GABA at the periphery of PNs and INs following activation of INs in LA we simultaneously recorded whole-cell GABAbR-mediated current from sets of PN and INs in response to the short-lasting tetanus applied through a stimulation electrode positioned inside LA close to either the external (3 PN-IN sets), or internal (4 PN-IN sets) capsule to randomize population of recruited interneurons. To isolate the current, recordings were performed in the presence of 20 mM CNQX, 20 μM APV and 100 mM picrotoxin while the cell was held at −70 mV. The stimulation intensity was adjusted to evoke ~30 pA GABAbRs currents in PNs.

For calculating E(I)PSC input-output curve slopes, we only included initial linear portions of the curves before responses began to saturate. Miniature postsynaptic currents were recorded in the presence of 1 μM TTX and additional 100 mM picrotoxin for mEPSCs, or 20 mM CNQX and 20 μM APV for mIPSCs.

In LTP experiments, stimulus intensity was adjusted to produce potency about 20–30% of the maximum synaptic response. To induce LTP in PNs and INs, the initial membrane potential was held at −80 mV and LTP was induced by 3 trains of theta stimuli (4 stimuli at 100 Hz repeated 25 times at intervals of 200 ms) separated by 30 seconds. Synaptic strength was measured as the initial slope (first 2–3 ms) of EPSP. LTP was quantified by normalizing the data collected in the last 5 minutes to the mean value of baseline EPSP, which was recorded at 0.066 Hz for at least 5 minutes before LTP induction.

In experiments examining effects of CGP 52432 on PPR following theta stimulation two trains each containing 4 stimuli of 100 Hz separated by 200 ms were delivered in the cortical inputs and followed (100 ms) by paired stimulation (50 ms interval) of the same pathway. 20 mM ATP was included in the pipette solution to minimize the possible contamination of PPR in INs by the effects of polyamine during postsynaptic depolarization (Lei and McBain, 2003). Data were expressed as Means ± SEM. Statistical significance was determined by using paired, unpaired or one group t-test as well as repeated ANOVA as appropriate.

Reagents

CNQX, APV, CGP 52432, SKF 89976A, Baclofen were from Tocris (Ellisville, MO). GDP-βS was from Sigma-Aldrich (St. Louis, MO).

Acknowledgments

This research was supported by NIMH Intramural Research Program, grant from SORST, Japan Science and Technology Agency, and Grant-in-Aids for Scientific Research from the MEXT, Japan. We thank Chris McBain and Vadim Bolshakov for comments on the manuscript.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Barbaresi P. GABA-immunoreactive neurons and terminals in the cat periaqueductal gray matter: a light and electron microscopic study. J Neurocytol. 2005;34:471–487. doi: 10.1007/s11068-006-9440-7. [DOI] [PubMed] [Google Scholar]
  2. Bekkers JM, Stevens CF. Cable properties of cultured hippocampal neurons determined from sucrose-evoked miniature EPSCs. J Neurophysiol. 1996;75:1250–1255. doi: 10.1152/jn.1996.75.3.1250. [DOI] [PubMed] [Google Scholar]
  3. Bordi F, LeDoux J, Clugnet MC, Pavlides C. Single-unit activity in the lateral nucleus of the amygdala and overlying areas of the striatum in freely behaving rats: rates, discharge patterns, and responses to acoustic stimuli. Behav Neurosci. 1993;107:757–769. doi: 10.1037/0735-7044.107.5.757. [DOI] [PubMed] [Google Scholar]
  4. Davis M. The role of the amygdala in conditionedf and unconditioned fear and anxiety. In: Aggleton JP, editor. The amygdala. New York: Oxford University Press; 2000. pp. 213–288. [Google Scholar]
  5. Dittman JS, Regehr WG. Mechanism and kinetics of heterosynaptic depression at a cerebellar synapse. J Neurosci. 1997;17:9048–9059. doi: 10.1523/JNEUROSCI.17-23-09048.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Dityatev AE, Bolshakov VY. Amygdala, long-term potentiation, and fear conditioning. Neuroscientist. 2005;11:75–88. doi: 10.1177/1073858404270857. [DOI] [PubMed] [Google Scholar]
  7. Engel D, Jonas P. Presynaptic action potential amplification by voltage-gated Na+ channels in hippocampal mossy fiber boutons. Neuron. 2005;45:405–417. doi: 10.1016/j.neuron.2004.12.048. [DOI] [PubMed] [Google Scholar]
  8. Engel D, Schmitz D, Gloveli T, Frahm C, Heinemann U, Draguhn A. Laminar difference in GABA uptake and GAT-1 expression in rat CA1. J Physiol. 1998;512(Pt 3):643–649. doi: 10.1111/j.1469-7793.1998.643bd.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Ferraguti F, Klausberger T, Cobden P, Baude A, Roberts JD, Szucs P, Kinoshita A, Shigemoto R, Somogyi P, Dalezios Y. Metabotropic glutamate receptor 8-expressing nerve terminals target subsets of GABAergic neurons in the hippocampus. J Neurosci. 2005;25:10520–10536. doi: 10.1523/JNEUROSCI.2547-05.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Humeau Y, Shaban H, Bissiere S, Luthi A. Presynaptic induction of heterosynaptic associative plasticity in the mammalian brain. Nature. 2003;426:841–845. doi: 10.1038/nature02194. [DOI] [PubMed] [Google Scholar]
  11. Isaacson JS, Solis JM, Nicoll RA. Local and diffuse synaptic actions of GABA in the hippocampus. Neuron. 1993;10:165–175. doi: 10.1016/0896-6273(93)90308-e. [DOI] [PubMed] [Google Scholar]
  12. Jungling K, Seidenbecher T, Sosulina L, Lesting J, Sangha S, Clark SD, Okamura N, Duangdao DM, Xu YL, Reinscheid RK, Pape HC. Neuropeptide S-mediated control of fear expression and extinction: role of intercalated GABAergic neurons in the amygdala. Neuron. 2008;59:298–310. doi: 10.1016/j.neuron.2008.07.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Kaupmann K, Malitschek B, Schuler V, Heid J, Froestl W, Beck P, Mosbacher J, Bischoff S, Kulik A, Shigemoto R, et al. GABA(B)-receptor subtypes assemble into functional heteromeric complexes. Nature. 1998;396:683–687. doi: 10.1038/25360. [DOI] [PubMed] [Google Scholar]
  14. Khirug S, Yamada J, Afzalov R, Voipio J, Khiroug L, Kaila K. GABAergic depolarization of the axon initial segment in cortical principal neurons is caused by the Na-K-2Cl cotransporter NKCC1. J Neurosci. 2008;28:4635–4639. doi: 10.1523/JNEUROSCI.0908-08.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Koester HJ, Johnston D. Target cell-dependent normalization of transmitter release at neocortical synapses. Science. 2005;308:863–866. doi: 10.1126/science.1100815. [DOI] [PubMed] [Google Scholar]
  16. Kulik A, Nakadate K, Nyiri G, Notomi T, Malitschek B, Bettler B, Shigemoto R. Distinct localization of GABA(B) receptors relative to synaptic sites in the rat cerebellum and ventrobasal thalamus. Eur J Neurosci. 2002;15:291–307. doi: 10.1046/j.0953-816x.2001.01855.x. [DOI] [PubMed] [Google Scholar]
  17. Kulik A, Vida I, Lujan R, Haas CA, Lopez-Bendito G, Shigemoto R, Frotscher M. Subcellular localization of metabotropic GABA(B) receptor subunits GABA(B1a/b) and GABA(B2) in the rat hippocampus. J Neurosci. 2003;23:11026–11035. doi: 10.1523/JNEUROSCI.23-35-11026.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lang EJ, Pare D. Synaptic and synaptically activated intrinsic conductances underlie inhibitory potentials in cat lateral amygdaloid projection neurons in vivo. J Neurophysiol. 1997;77:353–363. doi: 10.1152/jn.1997.77.1.353. [DOI] [PubMed] [Google Scholar]
  19. Lang EJ, Pare D. Synaptic responsiveness of interneurons of the cat lateral amygdaloid nucleus. Neuroscience. 1998;83:877–889. doi: 10.1016/s0306-4522(97)00420-x. [DOI] [PubMed] [Google Scholar]
  20. LeDoux J. The amygdala. Curr Biol. 2007;17:R868–874. doi: 10.1016/j.cub.2007.08.005. [DOI] [PubMed] [Google Scholar]
  21. LeDoux JE. Emotion circuits in the brain. Annu Rev Neurosci. 2000;23:155–184. doi: 10.1146/annurev.neuro.23.1.155. [DOI] [PubMed] [Google Scholar]
  22. Lei S, McBain CJ. GABA B receptor modulation of excitatory and inhibitory synaptic transmission onto rat CA3 hippocampal interneurons. J Physiol. 2003;546:439–453. doi: 10.1113/jphysiol.2002.034017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Maccaferri G, Roberts JD, Szucs P, Cottingham CA, Somogyi P. Cell surface domain specific postsynaptic currents evoked by identified GABAergic neurones in rat hippocampus in vitro. J Physiol . 2000;524(Pt 1):91–116. doi: 10.1111/j.1469-7793.2000.t01-3-00091.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Mahanty NK, Sah P. Calcium-permeable AMPA receptors mediate long- term potentiation in interneurons in the amygdala. Nature. 1998;394:683–687. doi: 10.1038/29312. [DOI] [PubMed] [Google Scholar]
  25. Marowsky A, Yanagawa Y, Obata K, Vogt KE. A specialized subclass of interneurons mediates dopaminergic facilitation of amygdala function. Neuron. 2005;48:1025–1037. doi: 10.1016/j.neuron.2005.10.029. [DOI] [PubMed] [Google Scholar]
  26. Martina M, Royer S, Pare D. Cell-type-specific GABA responses and chloride homeostasis in the cortex and amygdala. J Neurophysiol. 2001;86:2887–2895. doi: 10.1152/jn.2001.86.6.2887. [DOI] [PubMed] [Google Scholar]
  27. Muller J, Corodimas KP, Fridel Z, LeDoux JE. Functional inactivation of the lateral and basal nuclei of the amygdala by muscimol infusion prevents fear conditioning to an explicit conditioned stimulus and to contextual stimuli. Behav Neurosci. 1997;111:683–691. doi: 10.1037//0735-7044.111.4.683. [DOI] [PubMed] [Google Scholar]
  28. Pan BX, Vautier F, Ito W, Bolshakov VY, Morozov A. Enhanced cortico-amygdala efficacy and suppressed fear in absence of Rap1. J Neurosci. 2008;28:2089–2098. doi: 10.1523/JNEUROSCI.5156-07.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Pare D, Collins DR. Neuronal correlates of fear in the lateral amygdala: multiple extracellular recordings in conscious cats. J Neurosci. 2000;20:2701–2710. doi: 10.1523/JNEUROSCI.20-07-02701.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Pelkey KA, McBain CJ. Differential regulation at functionally divergent release sites along a common axon. Curr Opin Neurobiol. 2007;17:366–373. doi: 10.1016/j.conb.2007.04.008. [DOI] [PubMed] [Google Scholar]
  31. Porter JT, Nieves D. Presynaptic GABAB receptors modulate thalamic excitation of inhibitory and excitatory neurons in the mouse barrel cortex. J Neurophysiol. 2004;92:2762–2770. doi: 10.1152/jn.00196.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Quirk GJ, Gehlert DR. Inhibition of the amygdala: key to pathological states? Ann N Y Acad Sci. 2003;985:263–272. doi: 10.1111/j.1749-6632.2003.tb07087.x. [DOI] [PubMed] [Google Scholar]
  33. Quirk GJ, Repa C, LeDoux JE. Fear conditioning enhances short-latency auditory responses of lateral amygdala neurons: parallel recordings in the freely behaving rat. Neuron. 1995;15:1029–1039. doi: 10.1016/0896-6273(95)90092-6. [DOI] [PubMed] [Google Scholar]
  34. Repa JC, Muller J, Apergis J, Desrochers TM, Zhou Y, LeDoux JE. Two different lateral amygdala cell populations contribute to the initiation and storage of memory. Nat Neurosci. 2001;4:724–731. doi: 10.1038/89512. [DOI] [PubMed] [Google Scholar]
  35. Reyes A, Lujan R, Rozov A, Burnashev N, Somogyi P, Sakmann B. Target-cell-specific facilitation and depression in neocortical circuits. Nat Neurosci. 1998;1:279–285. doi: 10.1038/1092. [DOI] [PubMed] [Google Scholar]
  36. Rodriguez Manzanares PA, Isoardi NA, Carrer HF, Molina VA. Previous stress facilitates fear memory, attenuates GABAergic inhibition, and increases synaptic plasticity in the rat basolateral amygdala. J Neurosci. 2005;25:8725–8734. doi: 10.1523/JNEUROSCI.2260-05.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Rozov A, Burnashev N, Sakmann B, Neher E. Transmitter release modulation by intracellular Ca2+ buffers in facilitating and depressing nerve terminals of pyramidal cells in layer 2/3 of the rat neocortex indicates a target cell-specific difference in presynaptic calcium dynamics. J Physiol. 2001;531:807–826. doi: 10.1111/j.1469-7793.2001.0807h.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Scanziani M. GABA spillover activates postsynaptic GABA(B) receptors to control rhythmic hippocampal activity. Neuron. 2000;25:673–681. doi: 10.1016/s0896-6273(00)81069-7. [DOI] [PubMed] [Google Scholar]
  39. Scanziani M, Gahwiler BH, Charpak S. Target cell-specific modulation of transmitter release at terminals from a single axon. Proc Natl Acad Sci U S A. 1998;95:12004–12009. doi: 10.1073/pnas.95.20.12004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Seidenbecher T, Laxmi TR, Stork O, Pape HC. Amygdalar and hippocampal theta rhythm synchronization during fear memory retrieval. Science. 2003;301:846–850. doi: 10.1126/science.1085818. [DOI] [PubMed] [Google Scholar]
  41. Shaban H, Humeau Y, Herry C, Cassasus G, Shigemoto R, Ciocchi S, Barbieri S, van der Putten H, Kaupmann K, Bettler B, Luthi A. Generalization of amygdala LTP and conditioned fear in the absence of presynaptic inhibition. Nat Neurosci. 2006;9:1028–1035. doi: 10.1038/nn1732. [DOI] [PubMed] [Google Scholar]
  42. Shigemoto R, Kulik A, Roberts JD, Ohishi H, Nusser Z, Kaneko T, Somogyi P. Target-cell-specific concentration of a metabotropic glutamate receptor in the presynaptic active zone. Nature. 1996;381:523–525. doi: 10.1038/381523a0. [DOI] [PubMed] [Google Scholar]
  43. Smith Y, Pare JF, Pare D. Cat intraamygdaloid inhibitory network: ultrastructural organization of parvalbumin-immunoreactive elements. J Comp Neurol. 1998;391:164–179. doi: 10.1002/(sici)1096-9861(19980209)391:2<164::aid-cne2>3.0.co;2-0. [DOI] [PubMed] [Google Scholar]
  44. Soltesz I, Smetters DK, Mody I. Tonic inhibition originates from synapses close to the soma. Neuron. 1995;14:1273–1283. doi: 10.1016/0896-6273(95)90274-0. [DOI] [PubMed] [Google Scholar]
  45. Somogyi P, Tamas G, Lujan R, Buhl EH. Salient features of synaptic organisation in the cerebral cortex. Brain Res Brain Res Rev. 1998;26:113–135. doi: 10.1016/s0165-0173(97)00061-1. [DOI] [PubMed] [Google Scholar]
  46. Stevens CF, Wang Y. Changes in reliability of synaptic function as a mechanism for plasticity. Nature. 1994;371:704–707. doi: 10.1038/371704a0. [DOI] [PubMed] [Google Scholar]
  47. Szinyei C, Narayanan RT, Pape HC. Plasticity of inhibitory synaptic network interactions in the lateral amygdala upon fear conditioning in mice. Eur J Neurosci. 2007;25:1205–1211. doi: 10.1111/j.1460-9568.2007.05349.x. [DOI] [PubMed] [Google Scholar]
  48. Tamamaki N, Yanagawa Y, Tomioka R, Miyazaki J, Obata K, Kaneko T. Green fluorescent protein expression and colocalization with calretinin, parvalbumin, and somatostatin in the GAD67-GFP knock-in mouse. J Comp Neurol. 2003;467:60–79. doi: 10.1002/cne.10905. [DOI] [PubMed] [Google Scholar]
  49. Tsvetkov E, Carlezon WA, Benes FM, Kandel ER, Bolshakov VY. Fear conditioning occludes LTP-induced presynaptic enhancement of synaptic transmission in the cortical pathway to the lateral amygdala. Neuron. 2002;34:289–300. doi: 10.1016/s0896-6273(02)00645-1. [DOI] [PubMed] [Google Scholar]
  50. Vogt KE, Nicoll RA. Glutamate and gamma-aminobutyric acid mediate a heterosynaptic depression at mossy fiber synapses in the hippocampus. Proc Natl Acad Sci U S A. 1999;96:1118–1122. doi: 10.1073/pnas.96.3.1118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Wang QP, Guan JL, Nakai Y. Electron microscopic study of GABAergic synaptic innervation of nitric oxide synthase immunoreactive neurons in the dorsal raphe nucleus in the rat. Synapse. 1997;25:24–29. doi: 10.1002/(SICI)1098-2396(199701)25:1<24::AID-SYN3>3.0.CO;2-H. [DOI] [PubMed] [Google Scholar]
  52. Yamada J, Saitow F, Satake S, Kiyohara T, Konishi S. GABA(B) receptor-mediated presynaptic inhibition of glutamatergic and GABAergic transmission in the basolateral amygdala. Neuropharmacology. 1999;38:1743–1753. doi: 10.1016/s0028-3908(99)00126-4. [DOI] [PubMed] [Google Scholar]
  53. Zilberter Y, Kaiser KM, Sakmann B. Dendritic GABA release depresses excitatory transmission between layer 2/3 pyramidal and bitufted neurons in rat neocortex. Neuron. 1999;24:979–988. doi: 10.1016/s0896-6273(00)81044-2. [DOI] [PubMed] [Google Scholar]

RESOURCES