Synaptic properties of SOM- and CCK-expressing cells in dentate gyrus interneuron networks

Abstract

Hippocampal GABAergic cells are highly heterogeneous, but the functional significance of this diversity is not fully understood. By using paired recordings of synaptically connected interneurons in slice preparations of the rat and mouse dentate gyrus (DG), we show that morphologically identified interneurons form complex neuronal networks. Synaptic inhibitory interactions exist between cholecystokinin (CCK)-expressing hilar commissural associational path (HICAP) cells and among somatostatin (SOM)-containing hilar perforant path-associated (HIPP) interneurons. Moreover, both interneuron types inhibit parvalbumin (PV)-expressing perisomatic inhibitory basket cells (BCs), whereas BCs and HICAPs rarely target HIPP cells. HICAP and HIPP cells produce slow, weak, and unreliable inhibition onto postsynaptic interneurons. The time course of inhibitory signaling is defined by the identity of the presynaptic and postsynaptic cell. It is the slowest for HIPP-HIPP, intermediately slow for HICAP-HICAP, but fast for BC-BC synapses. GABA release at interneuron-interneuron synapses also shows cell type-specific short-term dynamics, ranging from multiple-pulse facilitation at HICAP-HICAP, biphasic modulation at HIPP-HIPP to depression at BC-BC synapses. Although dendritic inhibition at HICAP-BC and HIPP-BC synapses appears weak and slow, channelrhodopsin 2-mediated excitation of SOM terminals demonstrates that they effectively control the activity of target interneurons. They markedly reduce the discharge probability but sharpen the temporal precision of action potential generation. Thus, dendritic inhibition seems to play an important role in determining the activity pattern of GABAergic interneuron populations and thereby the flow of information through the DG circuitry.

Keywords: GABA; cholecystokinin; dentate gyrus; interneuron; somatostatin; synaptic transmission.

Figure 1.

Morphological, physiological, and synaptic characteristics of perisomatic and dendritic inhibitory interneurons in rodent dentate gyrus. A, Left, Confocal image stacks of pairs of synaptically connected GABAergic interneurons intracellularly labeled with biocytin and visualized with streptavidin conjugated with AlexaFluor-647 (white labeling). A₁, Perisomatic inhibitory interneurons with axon arbours located in the granule cell layer (gcl). Right, Both perisomatic inhibitory BCs are PV-positive as revealed by antibody labeling. From top to bottom, top green, biocytin labeling, arrows point to both somata; middle red, PV expression; bottom, colocalization of both markers. A₂, HICAPs with axon collaterals mainly located in the inner molecular layer (iml). Right, A single intracellularly labeled HICAP cell was identified as CCK-positive. Biocytin was visualized by streptavidin conjugated to AlexaFluor-750. Inset, Characteristic discharge pattern of the HICAP cell (0.7 nA, 1 s). Calibration: 200 ms, 50 mV. A₃, HIPPs with axon arbors located predominantly in the outer molecular layer (oml). Right, Both neurons coexpress somatostatin (SOM; arrows). B, Passive and active membrane properties of identified BCs (B₁), HICAPs (B₂), and HIPPs (B₃). B1–B3, Top traces, Voltage trajectories of cell pairs shown in A during 1-s-long current injections (−100, −50, 300–800 pA). B₁–B₃, Summary graphs show left the input resistance (R_in) of the recorded cell types and right the half-duration of single action potentials. Each circle represents a single data point; colored circles with lines represent mean values ± SEM. C, uIPSCs recorded at pairs shown in A. A presynaptic action potential (top) evokes uIPSCs in the postsynaptic cell. Single uIPSCs (6 traces) are shown superimposed (middle), and the average uIPSC (30 traces) at −70 mV is shown (bottom). Schematic illustration on top represents the recoded neuron types; orange represents BC–BC; green represents HICAP–HICAP; blue represents HIPP–HIPP pairs. C₃, Bottom, Average uIPSCs shown in C1–C3 were peak normalized and superimposed. ***p ≤ 0.001. **p ≤ 0.01. *p ≤ 0.05. ¹Significantly different from HICAP. ²Significantly different from HIPP.

Figure 2.

Weak, unreliable, and slow inhibition at homologous interneuron output synapses. A, Summary graph of synaptic latency, 20%–80% rise time, and decay time constant (τ, amplitude weighted decay time defined with a biexponential fit) of uIPSCs at BC–BC (6 pairs), HICAP–HICAP (15 pairs), and HIPP–HIPP (6 pairs) synapses. All cells were morphologically identified. Values are averages of >30 traces, including failures. **p ≤ 0.01. *p ≤ 0.05. B, Left, Superposition of five traces shows differences in the timing of single uIPSCs at BC–BC and HICAP–HICAP synapses. Middle, Summary histogram of synaptic latencies from representative BC–BC (top) and HICAP–HICAP (bottom) pairs. Right, Summary bar graph of the CV of synaptic latency. Note the larger jitter in the timing of uIPSCs at HICAP–HICAP synapses. C, Summary bar graph of peak amplitudes of average uIPSCs, including failures. D, Probability of possible unidirectional synaptic connections among homologous interneuron types (10.7% BC–BC, 6 pairs of 56 simultaneous dual BC recordings; 25.9% HICAP–HICAP, 15 pairs of 58 simultaneous dual HICAP cell-like recordings; and 5.2% HIPP–HIPP, 6 pairs of 115 dual SOM-GFP cell recordings). Error bars indicate mean ± SEM. Circles represent single data points. ***p ≤ 0.001. **p ≤ 0.01. *p ≤ 0.05.

Figure 3.

Reconstructions of homologous pairs of synaptically connected HICAPs and HIPPs. Soma and dendrites of the presynaptic neuron are depicted in green, and the axon is shown in red. Soma and dendrites of the postsynaptic neuron are drawn in black and the axon was left out for clarity. A, Neurolucida reconstruction of a DAB-labeled HICAP–HICAP pair. Arrowheads point to synaptic contact sites validated by electron microscopy. Inset, One of five identified synaptic sites. Number relates to the location of the depicted synapse. B, Simple neurite tracer reconstruction of a HIPP–HIPP pair (same as in Fig. 1A₃). Insets, Deconvolved confocal image stacks of 2 visually identified putative contact sites. Blue circles and numbers represent areas in which the putative synapses are located. They were identified as close appositions between the presynaptic axon and the postsynaptic dendrite. Gray lines indicate borders between layers. oml, outer molecular layer; gcl, granule cell layer; iml, inner molecular layer.

Figure 4.

Dynamic properties of GABA release at homologous BC- versus HICAP- and HIPP-mediated inhibitory output synapses. A Left, Multiple-pulse depression at BC–BC synapses. Top, Ten action potentials were evoked at 50 Hz in the presynaptic BC. Bottom, Superposition of 6 traces of uIPSCs. Average trace of uIPSCs from 30 individual traces (failures included). Middle, Multiple-pulse facilitation at HICAP–HICAP synapses. Same experiment as at BC–BC synapses. Note the asynchronously evoked uIPSCs during the course of the train. Right, HIPP–HIPP synapses show a biphasic response upon repetitive activation of the presynaptic cell. The initial strong multiple-pulse facilitation is followed by a decline in facilitation. Dashed lines indicate that uIPSC amplitudes were measured from the preceding baseline. B, Summary plots represent amplitude ratios between IPSC_n to IPSC₁ plotted as a function of the subsequently evoked uIPSC in the train. Lines represent fit results to the data (left and middle, exponential function; right, parabolic function). C, CV analysis of multiple-pulse depression (left), multiple-pulse facilitation (middle), and biphasic response (right). The inverse of the square of the CV⁻² of the amplitude of the second (open circles) and fourth uIPSC (filled circle) was plotted against the mean peak amplitude (A_n); data were normalized to the CV⁻² and mean amplitude of the first uIPSC (CV₁⁻², A₁). Dashed line indicates the identity line (6 BC–BC, 10 HICAP–HICAP and 5 HIPP–HIPP pairs).

Figure 5.

Recovery from multiple-pulse depression at BC–BC and facilitation at HICAP–HICAP and HIPP–HIPP synapses. A–C, A train of 10 action potentials at 50 Hz (left) was applied to the presynaptic interneuron to evoke multiple-pulse modulation of uIPSCs. Recovery from multiple-pulse modulation was determined by evoking single action potentials in the presynaptic cell with increasing time intervals after the train. Values below arrows indicate length of the time interval after the train (6 BC–BC, 7 HICAP–HICAP and 5 HIPP–HIPP pairs). Right, Time course of recovery from multiple-pulse modulation. Amplitude ratio of evoked IPSC after the train and the first IPSC in the train (IPSC_n/IPSC₁) are plotted against the time interval. Lines represent monoexponential fit to the data with a time constant (τ) of 0.65 s for BC–BC synapses (6 pairs), 1.95 s for HICAP–HICAP (7 pairs), and 0.36 s for HIPP–HIPP (5 pairs) synapses. Averages from >10 single traces are shown. Filled circles represent the mean amplitude of the last uIPSC in the 50 Hz train. Data points are mean ± SEM. D, Top, Confocal image stack of a BC–BC pair. Arrowheads indicate location of axon collaterals in the granule cell layer (gcl). Middle, Micrograph of a DAB-labeled HICAP–HICAP pair. Arrowheads indicate axon collaterals in the inner molecular layer (iml). Bottom, Confocal image stack of an intracellularly labeled HIPP–HIPP pair with biocytin (left) that expresses somatostatin (SOM, right). Stars represent cell bodies. Gray dotted lines indicate borders between the gcl and hilus, the gcl and the iml, and between the outer molecular layer (oml) and hippocampal fissure (ml, molecular layer).

Figure 6.

Kinetic and dynamic properties of uIPSCs at heterologous interneuron–BC synapses. A, Summary graphs of synaptic latency, 20%–80% rise time, decay time constant (τ), and CV in synaptic latency of uIPSCs at BC–BC (6 pairs), HICAP–BC (8 pairs), and HIPP–BC (6 pairs) synapses. B, Top, Confocal image stack of a HIPP–BC pair. Arrows indicate axon collaterals in the outer and middle molecular layer (oml). Areas surrounded by blue circles are shown at high magnification on the right. Bottom, Antibody labeling against PV (left); superposition of PV and biocytin (bio) labeling (right). Star represents the BC; + indicates the HIPP soma located behind the PV⁺ cell body. Right, Deconvolved image stack of 3 visually identified putative synapses. C, Probability of possible unidirectional connections between interneuron types and postsynaptic BCs (HICAP–BCs, 15 pairs of 92 simultaneous dual HICAP-like and BC-like recordings; HIPP–BCs 12 pairs of 94 simultaneous dual SOM-GFP cell and BC recordings). D, Multiple-pulse dynamics at interneuron–BC synapses. Top, Trains of 10 action potentials were applied at 50 Hz to the presynaptic HIPP cell (top trace), and uIPSCs were recorded at postsynaptic BCs (lower trace; average of 30 single traces). IPSC_n/IPSC₁ is plotted as a function of the subsequently evoked IPSCs during the train. Note the constant inhibitory signaling at HICAP–BC-like (green circles) and HIPP–BC synapses (blue circles) as well as multiple-pulse depression at BC–BC synapses (orange circles, same data as in Fig. 4B, left). Line indicates exponential fit. Error bars indicate mean ± SEM. A, Circles represent single data points. D, Circles represent mean ± SEM. **p ≤ 0.01. *p ≤ 0.05.

Figure 7.

SOM cell-mediated dendritic inhibition reduces frequency and improves precision in action potential generation in target BCs. A, Left, Confocal image stack shows selective expression of ChR2-tdT in SOM-expressing interneurons upon stereotaxic injection of rAAV-ChR2-tdT (Material and Methods). Arrows indicate cell bodies colocalizing ChR2 and SOM. Red square outlines the area in which intensity in ChR2-tdT labeling was quantified using ImageJ software and plotted against the distance across the DG. Middle, Intensity of tdT-expression quantified as arbitrary units (AU). Note the higher expression intensity in the hilus and outer molecular layer (oml). Right, Percentage of SOM⁺ cells coexpressing ChR2-tdT⁺ and ChR2-tdT⁺ colocalizing SOM⁺. B, Illustration of the experimental condition. A confocal image stack of ChR2-tdT was superimposed with a labeled BC (white). Whole-cell recordings were performed from a BC during extracellular stimulation (3 pulses, 20 Hz) of the PP with a pipette positioned in the outer molecular layer (oml). SOM-ChR2-tdT⁺ cells were recruited by applying a blue light pulse (473 nm, 2 ms) to the oml. Circle represents the location of the light spot (diameter ∼40 μm) close to distal apical dendrites of the target cell. Gray lightning represents electrical PP stimulation; blue lightning represents light pulse application. Inset, Single action potential generation in ChR2-tdT-SOM⁺ cells upon blue light illumination (16 cells tested). C, Left, inset, The experimental configuration. Top, Voltage response to PP stimulation. Middle, Average IPSCs (blue) recorded in the same BC evoked by 3 blue light pulses at 20 Hz. Bottom, Superimposed average EPSCs evoked before (black) and in the presence (red) of ChR2 activation. Right, Superposition of single action potentials induced by the second PP stimulus during control conditions (upper traces) and in the presence of blue light (lower traces). Gray bars represent the jitter in spike generation. Blue bar represents time of light application. Red dotted lines and red arrows indicate a shift in the mean latency between PP stimulation and spike threshold. D, Left, Bar graph represents probability of action potential generation in BCs during the first and second spike of a train before (white bars) and after (gray bars) light-induced dendritic inhibition. Right, Bar graphs represent the SD in spike timing (jitter; 6 BCs) and mean latency of spike generation in controls and during SOM-mediated dendritic inhibition. Symbols connected by lines represent single experiments. Error bars indicate mean ± SEM. ***p ≤ 0.0001. **p ≤ 0.01. *p ≤ 0.05. *1p = 0.07.