Parvalbumin+ interneurons obey unique connectivity rules and establish a powerful lateral-inhibition microcircuit in dentate gyrus

Abstract

Parvalbumin-positive (PV⁺) GABAergic interneurons in hippocampal microcircuits are thought to play a key role in several higher network functions, such as feedforward and feedback inhibition, network oscillations, and pattern separation. Fast lateral inhibition mediated by GABAergic interneurons may implement a winner-takes-all mechanism in the hippocampal input layer. However, it is not clear whether the functional connectivity rules of granule cells (GCs) and interneurons in the dentate gyrus are consistent with such a mechanism. Using simultaneous patch-clamp recordings from up to seven GCs and up to four PV⁺ interneurons in the dentate gyrus, we find that connectivity is structured in space, synapse-specific, and enriched in specific disynaptic motifs. In contrast to the neocortex, lateral inhibition in the dentate gyrus (in which a GC inhibits neighboring GCs via a PV⁺ interneuron) is ~ 10-times more abundant than recurrent inhibition (in which a GC inhibits itself). Thus, unique connectivity rules may enable the dentate gyrus to perform specific higher-order computations.

Fig. 1

Octuple recording from GCs and PV⁺ interneurons in the dentate gyrus. a Octuple recording from five GCs and two PV⁺ interneurons (seven cells successfully recorded). Infrared differential interference contrast video micrograph of the dentate gyrus in a 300-µm slice preparation, with eight recording pipettes. Shaded areas represent the 2D projections of cell bodies (blue, GCs; red and yellow, PV⁺ interneurons). Blue dashed lines, boundaries of GC layer. b Partial reconstruction of one GC and two PV⁺ interneurons in the same recording as shown in (a). Cells were filled with biocytin during recording and visualized using 3,3′-diaminobenzidine as chromogen. For clarity, only the somatodendritic domains were drawn for the PV⁺ interneurons. Insets, biocytin-labeled putative synaptic contacts, corresponding to boxes in main figure. c Connectivity matrix of an octuple recording (all eight cells successfully recorded). Subpanels on the diagonal (AP traces) represent the presynaptic cells, subpanels outside the diagonal (EPSC or IPSC traces) indicate postsynaptic cells. In this example, 56 connections were tested; 7 excitatory GC–PV⁺ interneuron connections, 7 inhibitory PV⁺ interneuron–GC connections, and 42 connections between GCs. Brief transients in a subset of traces represent capacitive coupling artifacts, as shown in previous publications^, . d Expanded view of presynaptic APs and postsynaptic currents, corresponding to the boxed areas in (c). In this octuple recording, an inhibitory synaptic connection was identified between the PV⁺ interneuron (red) and GC 5 (blue) and an excitatory synaptic connection was found between GC 1 (blue) and the PV⁺ interneuron (red). The presence of a unidirectional excitatory GC–PV⁺ interneuron connection and a unidirectional inhibitory PV⁺ interneuron–GC connection documents the existence of lateral inhibition in this recording. e Coexistence of different synapses in an octuple recording. In this recording, an excitatory GC–PV⁺ interneuron connection, an inhibitory PV⁺ interneuron–GC connection, a chemical inhibitory connection between the PV⁺ interneurons, and an electrical connection between the PV⁺ interneurons were found (from left to right). Same recording as in (a) and (b)

Fig. 2

Differential connectivity of PV⁺, CCK⁺, and SST⁺ interneurons in the dentate gyrus. a Light micrograph of a SST⁺ interneuron filled with biocytin during recording, and visualized using 3,3′-diaminobenzidine as chromogen. Cells were identified by genetic labeling in SST-Cre mice. Axon branches in the molecular layer (red arrows) suggest that the cell was a HIPP or TML interneuron^, . GCL, granule cell layer. b Light micrograph of a CCK⁺ interneuron filled with biocytin. Cells were identified by genetic labeling in CCK-Cre;DLX 5/6-Flp mice. Axon branches in the inner molecular layer (red arrows) suggest that the cell was a HICAP interneuron^–. c, d Excitatory and inhibitory connectivity of SST⁺ interneurons. GC–SST⁺ interneuron unitary EPSCs are shown in (c), SST⁺ interneuron–GC IPSCs are illustrated in (d). Individual synaptic responses (gray) and average trace (magenta or blue, 15 traces) are shown overlaid. Note the facilitation of EPSCs during train stimulation in (c). e, f Excitatory and inhibitory connectivity of CCK⁺ interneurons. GC–CCK⁺ interneuron EPSCs are shown in (e), CCK⁺ interneuron–GC IPSCs are illustrated in (f). Note the asynchronous release during and after train stimulation in (f), which is highly characteristic of CCK⁺ interneuron output synapses. g Comparison of average connection probability for pairs with an intersomatic distance of ≤ 100 µm. Whereas PV⁺ interneurons were highly connected, SST⁺ and CCK⁺ interneurons showed a markedly lower excitatory and inhibitory connectivity (number of tested connections 767, 71, and 165). Error bars represent 95%-confidence intervals estimated from a binomial distribution

Fig. 3

Rules of excitatory and inhibitory connectivity in GC–PV⁺ interneuron networks. a Unitary EPSCs, with individual synaptic responses (gray) and average trace (red, 15 traces) in a representative GC–PV⁺ interneuron pair. b GC–PV⁺ interneuron connection probability plotted versus intersomatic distance. Connection probability was determined as the ratio of the number of found connections over that of all possible connections in a given distance range. Error bars represent 95%-confidence intervals estimated from a binomial distribution. Data points were fit with a sigmoidal function; shaded area indicates the distance range in which connection probability decayed to half-maximal value (space constant). Red dashed line, maximal connection probability. Maximal connection probability (c_max) was 11.3%, and space constant (d_half) was 144 µm. c Peak amplitude of unitary EPSCs at GC–PV⁺ interneuron synapses, plotted against intersomatic distance. Data points were fit by linear regression; dashed lines indicate 95%-confidence intervals. d–f Similar plots as shown in (a–c), but for IPSCs generated at inhibitory PV⁺ interneuron–GC synapses. Maximal connection probability was 28.9%, and space constant was 215 µm. g Bootstrap analysis of maximal connection probability and space constant. Histograms indicate distributions of c_max (left) and d_half (right) for 10,000 bootstrap replications of the inhibitory PV⁺ interneuron–GC connections. Red arrows indicate experimental mean values for GC–PV⁺ interneuron synapses. h Number of reciprocally coupled GC–PV⁺ interneuron pairs (excitatory and inhibitory synapse; “recurrent inhibition motif”) and unidirectionally coupled PV⁺ interneuron–GC pairs (inhibitory synapse only; “lateral inhibition motif”). Note that the number of lateral inhibition motifs was almost 10-times higher than that of recurrent inhibition motifs, demonstrating the high abundance of lateral inhibition in the dentate gyrus microcircuit

Fig. 4

Rules of chemical and electrical connectivity between PV⁺ interneurons. a Left, light micrograph of a biocytin-labeled PV⁺ interneuron–PV⁺ interneuron pair. Right, unitary IPSCs, with individual synaptic responses (gray) and average trace (red, 15 traces) in the same pair. GCL, granule cell layer; IML, inner molecular layer. b PV⁺ interneuron–PV⁺ interneuron chemical connection probability (left) and IPSC peak amplitude (right) plotted versus intersomatic distance. Connection probability data points were fit with a sigmoidal function, IPSC amplitude data were analyzed by linear regression. Maximal connection probability was 58.1%, and space constant was 141 μm. c Electrical coupling between two PV⁺ interneurons. Voltage changes in the pre- and postsynaptic cell caused by the injections of long polarizing current pulses (left, + 200 pA; right, −200 pA; 200 ms) in one of the coupled cells. d PV⁺ interneuron–PV⁺ interneuron electrical connection probability (left) and coupling coefficient (right) plotted versus intersomatic distance. Maximal connection probability was 77.3%, and space constant was 146 μm. The coupling coefficient (CC) was calculated as the mean ratio of steady-state voltages (V₂/V₁, V₁/V₂) during application of current pulses in one of the cells (cell 1 and cell 2, respectively)

Fig. 5

Overabundance of disynaptic connectivity motifs in GC–PV⁺ interneuron networks and different functional properties of synapses embedded in motifs. a Graph analysis of disynaptic connectivity motifs. In total, there are five possible disynaptic connectivity motifs with two cells and 20 disynaptic motifs involving three cells. Arrows with open triangles indicate excitatory synapses, arrows with filled circles represent inhibitory synapses, and arrows with zigzag lines indicate gap junctions. Number indicates motif index. b Analysis of the number of motifs in 10,000 simulated data sets. Connection probability for the simulated data set was specified according to the experimentally determined spatial rules. Left, absolute motif number in experimental (black) and simulated data set (red, median; gray, 90%-confidence interval). Center, bar plot of relative abundance of various motifs (number of motifs in experimental data set over mean number in simulated data set). Error bars were taken from bootstrap analysis. Right, bar plot of z score of the different motifs. Light red area indicates z score in the interval [−1, 1]. Motifs 2, 3, 7, and 9 were significantly enriched above the chance level (P = 0.03145, 0.0085, 0.0272, and 0.0068 after multiple comparison correction). In contrast, motifs 6, 8, 10, 12, and 16 were slightly, but not significantly underrepresented (P = 0.15 for motif 6). Note that motifs 5, 17, 19–21, and 23–25 were not encountered in the present data set, because of the lack of connectivity between GCs. c Comparison of EPSC peak amplitude (left) and IPSC peak amplitude (right) in bidirectionally versus unidirectionally coupled GC–PV⁺ interneuron pairs. Peak amplitudes were not significantly different (P = 0.33 and 0.58, respectively). d Comparison of IPSC peak amplitude in PV⁺ interneuron–PV⁺ interneuron pairs connected by different chemical or electrical synapse motifs. IPSC peak amplitude was significantly larger in pairs with bidirectional inhibitory connections than with unidirectional connections (P = 0.016) and slightly higher in connections with than without gap junctions (P = 0.057). Asterisk indicates P < 0.05