Differential excitatory control of 2 parallel basket cell networks in amygdala microcircuits

Abstract

Information processing in neural networks depends on the connectivity among excitatory and inhibitory neurons. The presence of parallel, distinctly controlled local circuits within a cortical network may ensure an effective and dynamic regulation of microcircuit function. By applying a combination of optogenetics, electrophysiological recordings, and high resolution microscopic techniques, we uncovered the organizing principles of synaptic communication between principal neurons and basket cells in the basal nucleus of the amygdala. In this cortical structure, known to be critical for emotional memory formation, we revealed the presence of 2 parallel basket cell networks expressing either parvalbumin or cholecystokinin. While the 2 basket cell types are mutually interconnected within their own category via synapses and gap junctions, they avoid innervating each other, but form synaptic contacts with axo-axonic cells. Importantly, both basket cell types have the similar potency to control principal neuron spiking, but they receive excitatory input from principal neurons with entirely diverse features. This distinct feedback synaptic excitation enables a markedly different recruitment of the 2 basket cell types upon the activation of local principal neurons. Our data suggest fundamentally different functions for the 2 parallel basket cell networks in circuit operations in the amygdala.

Fig 1. Cholecystokinin-expressing basket cells (CCKBCs) and parvalbumin-containing basket cells (PVBCs) form 2 parallel inhibitory circuits, but both innervate axo-axonic cells (AACs).

(A) Horizontal section taken from the amygdala region of a transgenic mouse expressing red fluorescent protein under the control of the Cck promoter (CCK-DsRed). BLA, basolateral amygdala complex; EC, entorhinal cortex; Hip, hippocampus; DG, dentate gyrus; Me, medial amygdala; Ce, central amygdala; Pir, piriform cortex. Scale: 500 μm. (B) Red fluorescent protein (DsRed) expression and cholinergic fibers visualized by immunostaining against the vesicular acetylcholine transporter (VAChT) parcel out the basolateral amygdala complex into 3 parts: the anterior part of the basal amygdala (BAa) contains a pronounced DsRed signal and strong VAChT immunoreactivity, the posterior part of the BA (BAp) is characterized by low and high levels of DsRed and VAChT expression, respectively, and the lateral amygdala (LA) has low levels of both signals. Scale: 200 μm. (C) Maximum z intensity projection image of a PVBC and a CCKBC recorded in the BAa of a PV-eGFP x CCK-DsRed double transgenic mouse. The soma of the PVBC showed eGFP, but not DsRed signal (solid arrowhead), and its boutons were immunostained for calbindin (Calb, bottom left panels) that typifies PVBCs. The axon terminals of the CCKBC, which soma contained DsRed but not eGFP signal (open arrowhead), were immunoreactive for cannabinoid receptor type 1 (CB1) (bottom right panels). (D) Maximum z intensity projection image of an AAC and a CCKBC recorded simultaneously in the BAa. The axon collaterals of the AAC, which soma contained eGFP but not DsRed signal (solid arrowhead), formed close apposition with the axon initial segments visualized by immunostaining against ankyrin G (AnkG) and lacked Calb (bottom left panels), typical for AACs. The boutons of the CCKBC (open arrowhead), which soma showed DsRed but not eGFP signal, expressed CB1 (bottom right panels), characteristic for CCKBCs. Scale: 50 μm, insets 2 μm. (E) Representative traces from whole-cell paired recordings from post hoc identified different interneuron→interneuron pairs. Ten superimposed consecutive traces are in gray, averages are in bold. Scale: y: 10 mV/10 pA, x: 20 ms. (F) Connection probability matrix of interneuron→interneuron synaptic coupling was obtained by paired recordings. (G) Comparison of the amplitude of inhibitory postsynaptic currents (IPSCs) (18.8 ± 5.5 pA, n = 11 CCKBC→CCKBC pairs; 30.8 ± 18.3 pA, n = 9 CCKBC→AAC pairs; 32.2 ± 9.0 pA, n = 6 PVBC→PVBC pairs; 21.1 ± 4.5 pA, n = 6 PVBC→AAC pairs, Kruskal-Wallis ANOVA p = 0.2) (S1 Data). Each data point on the plot is an average obtained from a pair recording, and lines represent mean. (H) Comparison of the short-term plasticity of the different types of interneuron→interneuron connections (IPSC₁₀/IPSC₁: 5.72 ± 1.91, n = 7 CCKBC→CCKBC pairs; 2.87 ± 0.52, n = 7 CCKBC→AAC pairs; 0.48 ± 0.05, n = 6 PVBC→PVBC pairs; 0.53 ± 0.08, n = 4 PVBC→AAC pairs. Kruskal-Wallis ANOVA, p < 0.001) (S1 Data). Data are presented as mean ± SEM.

Fig 2. Cholecystokinin-expressing basket cells (CCKBCs) and parvalbumin-containing basket cells (PVBCs) avoid innervating each other but establish contacts on axo-axonic cells (AACs).

(A) Quantification of somatic cannabinoid receptor type 1 (CB1)- and parvalbumin (PV)-expressing inhibitory inputs (arrows) onto CCKBCs, PVBCs, and AACs using fluorescent immunostainings against gephyrin, CB1, PV, and vesicular gamma-aminobutyric acid (GABA) transporter (VGAT)/glutamate decarboxylase 65/67 (PanGAD). Axon terminals contacting the interneuron soma and other structures are labeled with arrows and arrowheads, respectively. Delineation of the cell bodies, terminals, and gephyrin puncta are shown for clarity. Scales: 5 μm and 1 μm. (B) Density of CB1- and PV-immunostained terminals contacting the different types of perisomatic region-targeting interneurons (CB1 on PVBC: 0.06 ± 0.03; CB1 on AAC: 0.81 ± 0.21; CB1 on CCKBC: 1.79 ± 0.16; Kruskal-Wallis ANOVA, p < 0.001; PVBC versus AAC p < 0.001, PVBC versus CCKBC p < 0.001, AAC versus CCKBC p = 0.002, Mann–Whitney U test; PV on PVBC: 1.25 ± 0.19; PV on AAC: 1.38 ± 0.33; PV on CCKBC: 0.08 ± 0.03, Kruskal-Wallis ANOVA, p < 0.001, PVBC versus AAC p = 0.36, PVBC versus CCKBC p < 0.001, AAC versus CCKBC p < 0.001, Mann–Whitney U test)(S2 Data), labels on columns show the number of examined somata in each group. Mean ± SEM.

Fig 3. Similar potent inhibition of principal neuron (PN) activity by cholecystokinin-expressing basket cells (CCKBCs) and parvalbumin-containing basket cells (PVBCs).

Maximum z intensity projection image of a CCKBC (A) and a PVBC (B). Scales: 50 μm. Upper insets in (A) and (B) indicate the positions of the recorded basket cells (BCs, white arrows) in horizontal slices prepared from mice expressing red fluorescent protein under the control of the Cck promoter (CCK-DsRed) and mice expressing enhanced green fluorescent protein under the control of the Pvalb promoter (PV-EGFP), respectively. Hip, hippocampus; BA, basal amygdala; Pir, piriform cortex. Scale for the insets: 500 μm. Lower insets show the firing patterns of the same BCs in response to a step current injection (+400 and −100 pA). Scales: 10 mV, 100 ms. (C, D) Representative inhibitory postsynaptic current (IPSC, middle traces) and potential (IPSP, lower traces) recordings in a CCKBC (blue)→PN (black) pair (C) or in a PVBC (orange)→PN pair (D) in response to 3 action potentials at 30 Hz (upper traces). Ten superimposed consecutive traces are in gray, average in black. Scales for (C) and (D): y: 20 mV/20 pA/0.5 mV, x: 20 ms. (E) CCKBCs and PVBCs give rise to IPSCs with similar amplitude (80.53 ± 18.36 pA, n = 13 CCKBC pairs and 150.38 ± 30.24 pA, n = 13 PVBC pairs, Mann–Whitney U test, p = 0.073) (S3 Data), and (F) 3 IPSPs with the similar area (121.83 ± 43.71 mV*ms, n = 13 CCKBC pairs and 119.78 ± 30.35 mV*ms, n = 11 PVBC pairs, Mann–Whitney U test, p = 0.45) (S3 Data) recorded in PNs. (G) Sinusoidal current trains with peak-to-peak amplitude of 30 pA (gray cycles) and 50 pA (black cycles) were injected into a PN to initiate firing, and 3 action potentials were evoked in the presynaptic CCKBC (blue) at the fourth cycle (red arrow). Schematic representation of the injection of 3 current pulses into the presynaptic interneuron (IN) at 30 Hz is shown in red. Voltage traces are offset for clarity. Scale: 10 mV, 200 ms. (H) Raw data of the experiments are shown in (G). Black and gray dots refer to the firing probability observed at the sinusoid current amplitudes of 50 pA and 30 pA, respectively. An open circle indicates the cycle when the presynaptic interneuron fired 3 action potentials. (I) Comparison of the inhibitory efficacy of CCKBCs and PVBCs (72.53 ± 8.80%, n = 13 CCKBC pairs and 78.70 ± 8.29%, n = 13 PVBC pairs, Mann–Whitney U test, p = 0.63) (S3 Data). Each data point on the plots represents an average obtained in a pair recording, and lines represent means.

Fig 4. Parvalbumin-containing basket cells (PVBCs) receive a higher number of excitatory inputs than cholecystokinin-expressing basket cells (CCKBCs).

VGluT1-immunopositive puncta opposing the dendrite of a CCKBC (A) and a PVBC (B) and containing bassoon staining were regarded as a contact (arrows). Open arrowhead shows a vesicular glutamate transporter 1 (VGluT1)-immunoreactive bouton in which bassoon didn't oppose the biocytin-labeled dendrite, presumably contacting a neighboring structure. Scale: 1 μm. (C) Each data point on the plot represents an average density obtained on a dendritic segment, lines represent mean. Biocytin-labeled dendritic segments of 5 CCKBCs and 6 PVBCs were investigated. (CCKBC: 0.32 ± 0.06, PVBC: 0.76 ± 0.06, Mann–Whitney U test, ***p < 0.001) (S4 Data).

Fig 5. Excitatory inputs of the 2 basket cell (BC) types have distinct properties.

(A) Representative traces depicting miniature excitatory postsynaptic currents (mEPSCs) in the presence of 0.5 μM tetrodotoxin (TTX) in a cholecystokinin-expressing basket cells (CCKBC) (left, blue) and in a parvalbumin-containing basket cells (PVBC) (right, orange) (holding potential, –65 mV). Scales: x: 0.5 s, y: 30 pA. (B) Cumulative probability distributions of mEPSC peak amplitudes (left) and interevent intervals (right) obtained in CCKBCs (blue, n = 5) and PVBCs (orange, n = 5). PVBCs receive mEPSCs with larger amplitude and at higher frequency than CCKBCs (Kolmogorov-Smirnov test, ***p < 0.001). Data points in the insets represent the median of the cumulative distributions obtained for each cell, and lines indicate means (peak amplitude: CCKBCs, 12.62 ± 0.43 pA, n = 5, PVBCs, 21.19 ± 3.01 pA, n = 5; interevent interval: CCKBCs, 86.97 ± 15.03 ms, n = 5; PVBCs, 34.87 ± 8.02 ms, n = 5; Mann–Whitney U test, *p < 0.05) (S5 Data). (C) Examples of deconvolved confocal images showing bassoon-apposing puncta immunostained for α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors in a dendritic portion of a PVBC (arrows, upper panels) and a CCKBC (arrows, lower panels). Scale bars: 0.4 μm. (D) Left, the same dendritic fragments shown in (C) with the superimposed stochastic optical reconstruction microscopy (STORM) images obtained for AMPA receptor staining (yellow), aligned and filtered using VividSTORM. Right, higher magnification of representative puncta depicting AMPA receptors along the interneuron dendrites, showing the deconvolved confocal image aligned with the STORM image. Each yellow sphere corresponds to a localization point (LP), while the white lines delineate the 2D convex hull area based on the AMPA LP clusters. Scales: left 0.5 μm, right 0.2 μm. (E) Cumulative probability distributions of the normalized number of LPs (left) and 2D convex hull area (right) obtained in CCKBCs (n = 30 puncta, blue) and PV-expressing interneurons (n = 53 puncta, orange). AMPA receptor clusters on CCKBC dendrites are smaller and contain lower number of LPs than those observed on parvalbumin (PV)-immunoreactive dendrites (normalized number of LPs: CCKBCs, 53.13 ± 8.58, PV-containing interneurons 69.28 ± 8.90; 2D convex hull area: CCKBCs, 0.04 ± 0.007 μm², PV-containing interneurons, 0.08 ± 0.006 μm²; Kolmogorov-Smirnov test, *p < 0.05, ***p < 0.001) (S5 Data). Each data point in insets represents a cluster, and lines indicate means (Mann–Whitney U test, *p < 0.05, ***p < 0.001).

Fig 6. Principle neurons (PNs) give rise to distinct excitatory synaptic inputs onto the 2 basket cell (BC) types.

(A) Panoramic images showing a cholecystokinin-expressing basket cell (CCKBC)→PN (left) and a parvalbumin-containing basket cell (PVBC)→PN (right) pair in the anterior part of the basal amygdala (BA). Interneurons were filled with Cascade Blue and PNs with biocytin. Insets show a higher magnification of the cell pairs recorded. Scales: 250 μm. (B) Representative traces of unitary excitatory postsynaptic currents (EPSCs) evoked by action potentials in a PN→CCKBC (upper traces) and a PN→PVBC pair (lower traces). Ten superimposed consecutive traces are in gray, and averages are in blue and orange. Scales: x: 2 ms; PN→CCKBC, y: 80 mV/8 pA; PN→PVBC, y: 90 mV/30 pA. (C) PVBCs receive unitary excitatory postsynaptic currents (uEPSCs) from local PNs with larger amplitude and lower failure rate than CCKBCs (potency, CCKBC: 21.5 ± 1.6 pA, PVBC: 94.25 ± 15.08 pA; failure rate, CCKBC: 0.58 ± 0.04, PVBC: 0.25 ± 0.04; Mann–Whitney U test, ***p < 0.001) (S6 Data). Each data point on the plots represents an average obtained in a pair recording, and lines represent means. (D) Schematic illustration of the experimental design to excite PNs using blue light. mCherry-containing construct was injected into the BA of transgenic mice (bottom drawing illustrates a horizontal slice showing the expression site in red). Among adeno-associated virus (AAV)-infected PNs (red circles), a CCKBC (blue) was recorded in whole-cell mode, while its excitatory input from neighboring PNs was tested by light illumination and/or whole-cell recording. BA, basal amygdala; Hip, hippocampus. (E) Connectivity map. Concentric circles indicate Δ100 μm distance from the interneuron in the center (S6 Data). Each dot represents a tested PN soma location. (F) PVBCs receive excitatory synaptic inputs from their neighboring PNs with high probability, while CCKBCs are innervated by PNs via their local collaterals with low probability, independently of the inter-somatic distance (S6 Data).

Fig 7. Principal neurons PNs innervate the 2 basket cell (BC) types differently.

Maximum z intensity projection image of a representative PN→parvalbumin-containing basket cell (PVBC) pair (A) and a PN→cholecystokinin-expressing basket cells (CCKBC) pair (F). Cascade Blue was used to label the BCs, while biocytin was introduced into the PNs to visualize them with streptavidin-conjugated Alexa647. Scales for (A and F): 50 μm. Neurolucida reconstructions (B and G) and the resulted dendrogram analysis (C and H) of the PVBC and CCKBC shown in (A) and in (F), respectively, marking the location of connections 1 and 2 (c1 and c2) established by the monosynaptically connected presynaptic PNs. Scales for (B, C, G, H): 50 μm. (D and E) High-power magnification 3D confocal images of putative contacts c1 and c2, showing close appositions between the dendrite-targeting boutons of the PN (yellow) and the PVBC dendrites (blue). Scale: 2 μm. (I) High power magnification 3D confocal images of a putative contact c1, showing close appositions between the bouton of the PN (yellow) and the CCKBC dendrites (blue). Scales: 2 μm. (J) PNs establish significantly more contacts on PVBCs than on CCKBCs (1.79 ± 0.26, n = 14 PN→CCKBC pairs, 3.23 ± 0.57, n = 13 PN→PVBC pairs, Mann–Whitney U test, *p < 0.05) (S7 Data). Each data point on the plot was obtained from a pair recording, and lines represent means. (K) 3D confocal image showing a close apposition between a biocytin-labeled PN terminal (yellow) and a Cascade Blue-labeled BC dendrite (blue). Scale: 2 μm. (L) Electron microscopic analysis of the contact shown in (K) confirmed the presence of the synaptic junction (arrow). Scale: 500 nm. (M) A strong relationship was found between the number of contact sites and the unitary excitatory postsynaptic current (uEPSC) potency in the case of PN→CCKBC pairs, while no correlation could be observed in the case of PN→PVBC pairs (S7 Data).

Fig 8. Spiking of parvalbumin-containing basket cells (PVBCs) requires a lower activity level of principal neuron (PN) population than cholecystokinin-expressing basket cells (CCKBCs).

(A) Experimental design. mCherry-containing construct was injected into the basal amygdala (BA) of triple transgenic mice (bottom drawing illustrates a horizontal slice showing the expression site in red; Hip, hippocampus; BA, basal amygdala). Among adeno-associated virus (AAV)-infected PNs (red circles), a CCKBC (blue) and a PVBC (orange) were simultaneously recorded first in loose-patch mode, followed by recordings in whole-cell mode, while neighboring PNs were excited by light illumination. (B) Representative traces of loose-patch and whole-cell recordings show different spiking thresholds and, concomitantly, larger light-evoked responses in a PVBC (lower traces) compared to a CCKBC (upper traces) in response to PN stimulation at different light intensities. Averages calculated from 3 consecutive events are shown in blue and orange for the CCKBC and the PVBC, respectively. Scales: y: 25 pA, loose-patch; y: 200 pA whole-cell; x: 10 ms. (C) A summary graph of the experiment shown in (B); the firing probability (solid circles, left axis) and the integral of the light-evoked responses (open circles, right axis) detected in basket cells (BCs) upon gradually elevated stimulation intensities (S8 Data). (D) The activation threshold of PVBCs is significantly lower than that recorded simultaneously in CCKBCs. Firing threshold values are normalized to PVBC firing threshold (CCKBC: 8.67 ± 1.91, n = 9 dual recordings; Paired Sample Wilcoxon Signed Rank Test, p < 0.01) (S8 Data). (E) Maximum z intensity projection image of the in vitro biocytin-filled interneurons from which data are shown in (B) and (C). Right, images illustrate the presence of calbindin (Calb+) immunopositivity in the PVBC (blue), while epifluorescent images show the expression of enhanced green fluorescent protein (eGFP) in the PVBC and red fluorescent protein (DsRed) in the CCKBC together with the corresponding differential interference contrast images. Scales: 50 μm, insets: 25 μm.

Fig 9. The wiring diagram of the 3 perisomatic region-targeting interneurons and principal neurons (PNs) in the basal amygdala (BA).

(A) Connectivity among the distinct interneuron types and from interneurons to the PNs. Both synaptic connections and gap junctions are indicated. Arrows show the direction of action potential spread along the axons. (B) Connectivity from the PNs to interneurons.