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) isβ~β10-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.
Similar content being viewed by others
Introduction
Throughout the brain, fast-spiking, parvalbumin-expressing (PV+) GABAergic interneurons play a key role in several higher functions, such as feedforward and feedback inhibition, high-frequency network oscillations, and pattern separation1. Understanding how PV+ interneurons contribute to these complex computations requires a detailed and quantitative analysis of their synaptic connectivity. While early studies suggested that connectivity of PV+ interneurons is random2, more recent work highlighted several specific connectivity rules3,4,5,6,7 (Supplementary TableΒ 1). Analysis of principal neuron (PN)βinterneuron (IN) connectivity in the neocortex revealed that reciprocally connected pairs occurred much more frequently than expected in a random network3,4,5,6,7. Moreover, synaptic strength appeared to be higher in these reciprocally connected motifs4,6. Whether these connectivity rules also apply in other microcircuits, such as the hippocampus, has not been determined yet.
Pattern separation is a fundamental network computation in which PV+ interneurons are likely to be involved. Pattern separation is thought to be particularly important in the dentate gyrus, where conversion of overlapping synaptic input patterns into non-overlapping action potential (AP) output patterns8,9,10,11,12 may facilitate reliable storage of information in the downstream CA3 network9,13,14. Previous studies suggested a model of pattern separation based on a winner-takes-all mechanism mediated by feedback inhibition15,16,17,18,19. Such a model has received experimental support in the olfactory system20,21,22. While some studies suggested that similar mechanisms may operate in the dentate gyrus23,24, it is not clear whether the rules of PNβIN connectivity are adequate to support such a model. Specifically, two forms of feedback inhibition need to be distinguished: recurrent inhibition, in which an active PN inhibits itself via reciprocal PNβIN connections, and lateral inhibition, in which an active PN inhibits neighboring PNs but not itself25,26. A winner-takes-all mechanism likelyΒ requires lateral inhibition; recurrent inhibition may be counter-productive, because it could suppress potential winners17,26,27. However, in both neocortex and brain areas directly connected to the hippocampus, recurrent inhibition and lateral inhibition are equally abundant3,4,5,6,7 (Supplementary TableΒ 1). Such a circuit design would seem incompatible with efficient pattern separation.
To resolve this apparent contradiction, we examined the functional connectivity rules in PNβIN networks in the dentate gyrus, using simultaneous recordings from up to seven granule cells (GCs) and up to four GABAergic interneurons. Our experiments reveal a uniquely high abundance of lateral inhibition mediated by PV+ interneurons.
Results
Octuple recordings from neurons in the dentate gyrus
To determine the functional connectivity rules between PNs and INs in the dentate gyrus, we performed simultaneous whole-cell recordings from up to eight neurons (up to seven GCs and up to four INs) in vitro (Fig.Β 1a, b). PV+ interneurons, somatostatin-positive (SST+), and cholecystokinin-positive (CCK+) interneurons were identified in genetically modified mice, obtained by crossing Cre or Flp recombinase-expressing lines with tdTomato or EGFP reporter lines. PV+ interneurons showed the characteristic fast-spiking AP phenotype during sustained current injection, whereas both SST+ and CCK+ interneurons generated APs with lower frequency, corroborating the reliability of the genetic labeling (Supplementary FigureΒ 1).
To probe synaptic connectivity, we stimulated presynaptic neurons under current-clamp conditions, and recorded excitatory postsynaptic currents (EPSCs) or inhibitory postsynaptic currents (IPSCs) in postsynaptic neurons in the voltage-clamp configuration (Fig.Β 1cβe, Fig.Β 2). In total, we tested 9098 possible connections in 50 octuples, 72 septuples, 68 sextuples, 48 quintuples, 17 quadruples, 10 triples, and 5 pairs in 270 slices. Interestingly, PV+ interneurons showed a much higher connectivity than both SST+ and CCK+ interneurons. For GCβPV+ interneuron pairs with intersomatic distanceββ€β100βΒ΅m, the mean connection probability was 11.0% for excitatory GCβPV+ interneuron and 28.8% for inhibitory PV+ interneuronβGC connectivity (Fig.Β 2g). In contrast, for both SST+ interneurons and CCK+ interneurons, the mean connection probability was substantially lower (1.4 and 2.8% for SST+ interneurons, 1.2 and 12.1% for CCK+ interneurons; Fig.Β 2g). Excitatory interactions between GCs were completely absent, and disynaptic inhibitory interactions between GCs28,29 were extremely sparse (0.124%). These results indicate that in the dentate gyrus PV+ interneurons show a markedly higher connectivity than SST+ and CCK+ interneurons, extending previous observations in the neocortex30.
Connectivity rules for excitatory input of PV+ interneurons
As PV+ interneurons showed the highest input and output connectivity, we focused our functional connectivity analysis on this interneuron subtype. We first examined the rules of excitatory synaptic connectivity between GCs and PV+ interneurons by measuring EPSCs (Fig.Β 3aβc). We found that PV+ interneurons were highly and locally connected to GCs. The connection probability showed a peak of 11.3%, and steeply declined as a function of intersomatic distance, with a space constant of 144βΒ΅m (Fig.Β 3b). In contrast, the EPSC peak amplitude showed no significant distance dependence (Fig.Β 3c). To determine the efficacy of unitary GCβPV+ interneuron connections, we measured unitary excitatory postsynaptic potentials (EPSPs). Unitary EPSPs had a mean peak amplitude of 1.79βΒ±β0.36βmV (range: 0.30β7.16βmV; Supplementary FigureΒ 2a, b)28,31,32. To assess the efficacy of these events in triggering spikes in the presence of ongoing synaptic activity from multiple sources, we performed in vivo whole-cell recordings from fast-spiking interneurons in the dentate gyrus in awake mice running on a linear treadmill (Supplementary FigureΒ 2cβg). Under in vivo conditions, the difference between baseline membrane potential and threshold was 10.3βΒ±β1.8βmV (three in vivo recordings from fast-spiking interneurons in dentate gyrus). Thus, although the largest unitary EPSPs were close to the threshold of AP initiation, they were insufficient to trigger a spike. However, the high focal GCβPV+ interneuron connectivity (Fig.Β 3b) may enable activation of PV+ interneurons by spatial summation.
Connectivity rules for inhibitory output of PV+ interneurons
Next, we examined the rules of inhibitory synaptic connectivity between GCs and PV+ interneurons by measuring IPSCs (Fig.Β 3dβf). Similar to excitatory GCβPV+ interneuron connectivity, inhibitory PV+ interneuronβGC connectivity was distance-dependent (Fig.Β 3e). However, maximal connection probability was higher (28.9%) and the range of connectivity was wider (215βΒ΅m) than that of excitation. Bootstrap analysis revealed that both maximal connectivity and space constant were significantly shorter for excitatory GCβPV+ interneuron synapses than for inhibitory PV+ interneuronβGC synapses (Pβ<β0.0001 and Pβ=β0.0042, respectively; Fig.Β 3g). Thus, different connectivity rules apply for excitatory and inhibitoryΒ GC βPV+ interneuron connections (focal excitation versus broad inhibition).
To compare the connectivity rules in the dentate gyrus with those in other brain regions, we quantified the ratio of excitatory to inhibitory connection probability. We found that inhibition was much more abundant than excitation, with a connection probability ratio of 3.83, substantially higher than in other brain areas (Supplementary TableΒ 1). Furthermore, we quantified the abundance of lateral and recurrent motifs in pairs of neurons. In our total sample of 1301 GCβPV+ interneuron pairs, we found 296 unidirectional inhibitory connections, but only 32 bidirectional connections (Fig.Β 3h). Thus, the ratio of lateral inhibition to recurrent inhibition was 9.25, substantially higher than in other circuits (Supplementary TableΒ 1). These results indicate that connectivity rules of PV+ interneurons in the dentate gyrus are uniqueΒ in comparison to other previously examined circuits.
Connectivity rules for mutual inhibition of PV+ interneurons
Finally, we analyzed the functional connectivity rules for synapses between interneurons (Fig.Β 4). Chemical inhibitory synapses between PV+ interneurons showed a connectivity pattern that was more focal than that of inhibitory PV+ interneuronβGC synapses (Fig.Β 4a, b). Likewise, electrical synapses between PV+ interneurons33,34,35 showed a focal connectivity pattern (Fig.Β 4c, d). Bootstrap analysis revealed that the maximal connectivity was significantly higher, while the space constant was significantly shorter for inhibitory PV+βPV+ interneuron synapses than for PV+ interneuronβGC synapses (Pβ=β0.0001 and Pβ=β0.0036, respectively). Furthermore, recordings from GCs and multiple PV+ interneurons provided direct evidence for the suggestion33 that EPSPs propagate through gap junctions, although the peak amplitude is markedly attenuated (Supplementary FigureΒ 3). Taken together, these results indicate that connectivity rules in PNβIN microcircuits are synapse-specific. Different connectivity rules apply to excitatory and inhibitory synapses between PNs and INs (GCβPV+ versus PV+βGC), and to inhibitory synapses terminating on different postsynaptic target cells (PV+βGC versus PV+βPV+ synapses).
Disynaptic connectivity motifs
Previous studies demonstrated that recurrent PNβPV+ interneuron connectivity motifs are enriched above the chance level expected for a random network in several cortical microcircuits3,4,5,6,7,36. To test this hypothesis, we analyzed the abundance of all 25 possible disynaptic connectivity motifs in our sample (Fig.Β 5)37. To probe whether connectivity was random38 or nonrandom14,39,40,41,42, we compared motif numbers in our experimental data to a simulated data set assuming random connectivity with experimentally determined distance-dependent connection probabilities (Fig.Β 5a, b).
Among the 25 possible disynaptic motifs, four types of motifs were significantly enriched above the chance level: (1) Gap junction connections between PV+ interneurons, (2) mutual inhibition motifs (PV+ interneuronβPV+ interneuron connections) combined with gap junction connections43, (3) convergence motifs (connections of multiple GCs on a single PV+ interneuron), and (4) divergence motifs (connections of one PV+ interneuron onto multiple GCs; Fig.Β 5b; Pβ<β0.05 after correction for multiple comparisons). Surprisingly, reciprocal GCβPV+ interneuron motifs were not significantly enriched.
Previous studies further demonstrated that the amplitude of unitary IPSCs is higher in bidirectionally than in unidirectionally connected PNβIN pairs4,6. In contrast, in the dentate gyrus neither the amplitude of EPSCs nor that of IPSCs was significantly different between bidirectionally and unidirectionally connected GCβPV+ interneuron pairs (Fig.Β 5c). However, the amplitude of IPSCs was significantly larger in PV+ interneuronβPV+ interneuron pairs coupled by reciprocal inhibitory synapses (Fig.Β 5d). Taken together, these results indicate that in the dentate gyrus, like in other cortical areas, synaptic connectivity of PV+ interneurons is nonrandom. However, both the types of enriched motifs and the rules setting synaptic strength differ from those in other circuits3,4.
Discussion
Our results demonstrate that the rules of functional connectivity in the PNβIN network of the dentate gyrus fundamentally differ from those in other cortical circuits. In the dentate gyrus, unidirectionally inhibitory connections are ~10-times more frequent than reciprocal connections, demonstrating a massive prevalence of lateral inhibition in this circuit (Supplementary TableΒ 1). In contrast, in neocortex, entorhinal cortex, and presubiculum, reciprocal connections are equally or more abundant than unidirectional connections, implying powerful recurrent inhibition3,4,5,6,7 (Supplementary TableΒ 1). Furthermore, in the dentate gyrus mutual inhibition motifs, convergence motifs, and divergence motifs are statistically overrepresented. In contrast, in the neocortex, interneuron connectivity has been suggested to be largely random2. Collectively, these results suggest that the dentate gyrus network obeys unique connectivity rules.
The specific connectivity rules of the dentate circuit raise the intriguing possibility that these rules represent an adaptation to specific network functions implemented in this brain region. A major function of the dentate gyrus is pattern separation8,9,10,11,12, thought to be generated by a βwinner-takes-allβ mechanism15,16,17,18,19. In an ideal pattern separation circuit, a small population of activated βwinner cellsβ must be able to efficiently and rapidly inhibit a large population of βnon-winner cellsβ. The dentate gyrus connectivity rules are well suited for these functions. First, powerful lateral inhibition efficiently suppresses non-winners, whereas winners remain unaffected. Second, the combination of local connectivity and rapid axonal signaling mechanisms of PV+ interneurons1,44 implements a high-speed suppression mechanism, as required for efficient pattern separation. Previous modeling work suggested that scale-free network organization and the presence of hub neurons may enhance the robustness of network computations45,46. Our results may support this view, since the high abundance convergence and divergence motifs are consistent with scale-free architectural properties.
Furthermore, the connectivity rules of the PNβIN network may be important for the generation of network oscillations in dentate gyrus47. In particular, the high chemical and electrical INβIN connectivity establishes an efficient gamma oscillator circuit. The dense and focal electricalβchemical connectivity may explain the high power and frequency of gamma oscillations in the dentate gyrus47,48,49. Previous modeling work suggested that the small-world interneuron network architecture will support the emergence of coherent gamma oscillations50,51. Our results support this notion, since the high abundance of electricalβchemical INβIN motifs would be consistent with small-world architectural properties52. The establishment of a robust gamma oscillation circuit may, conversely, be important for the pattern separation process. Proposed models of pattern separation imply that the separation of patterns takes place in the time period during the recovery from a preceding gamma cycle17. Whether and how the pattern separation computation and the generation of gamma oscillations can coexist in the same circuit remains to be determined.
Our results suggest the possibility that the uniquely high abundance of lateral inhibition in dentate gyrus may contribute to pattern separation (Supplementary TableΒ 1). What then is the function of recurrent inhibition in all other brain areas, such as the neocortical circuits? In the neocortex, PN activity is high, which requires a mechanism to establish excitationβinhibition balance; reciprocal PNβIN connectivity seems well suited for this purpose7,20. In contrast, in the dentate gyrus PN activity is low, and such a balancing function may not be required53,54,55,56,57. Additionally, reciprocal PNβIN connectivity could contribute to the generation of slower network oscillations in these brain regions, for example in the lower gamma or beta frequency range, which are characteristic for the neocortex.
Our results are consistent with the idea that local connectivity rules can shape diverse network computations across multiple circuits. In the dentate gyrus, the unique PNβIN connectivity rules may determine the properties of pattern separation, grid-to-place code conversion, or processing of context information17,58. In the neocortex, PNβIN connectivity may determine network stability and excitationβinhibition balance7,20. In the hippocampal CA3 network, functional PNβPN connectivity rules shape pattern completion14, whereas in the neocortex functional PNβPN connectivity may shape response properties such as orientation selectivity41. Thus, the present results contribute to the emerging view that local connectivity rules are major determinants of higher computations in neuronal networks. Future work will be needed to test this hypothesis in both network models and behavioral experiments.
Methods
Hippocampal slice preparation
Experiments on genetically modified mice were performed in strict accordance with institutional, national, and European guidelines for animal experimentation and were approved by the Bundesministerium fΓΌr Wissenschaft, Forschung und Wirtschaft of Austria (A. Haslinger, Vienna; BMWFW-66.018/0007-WF/II/3b/2014; BMWF-66.018/0010-WF/V/3b/2015; BMWFW-66.018/0020-WF/V/3b/2016).
To genetically label PV+ interneurons, C57BL/6βJ PV-Cre knockin mice (http://jaxmice.jax.org/strain/008069) crossed with Ai14 loxP-flanked red fluorescent protein tdTomato reporter mice (https://www.jax.org/strain/007914) were used. To identify SST+ interneurons, somatostatin-ires-Cre mice (C-SSTtm1Npa, kindly provided by H. van der Putten; Novartis Pharma; MTD37295, Basel, Switzerland) were crossed with Ai14 tdTomato reporter mice. Finally, to label CCK+ interneurons, CCK-ires-Cre;DLX 5/6-Flp mice (https://www.jax.org/strain/012706 and https://www.jax.org/strain/010815)Β were crossed with dual reporter mice expressing either EGFP or tdTomato (RCEβ=βR26R CAG boosted EGFP mice, https://www.jax.org/strain/010812; Ai65, https://www.jax.org/strain/021875)59. Mice (20- to 44-days-old; mostly postnatal day 20β25) of either sex were lightly anesthetized with isoflurane (Forane, AbbVie, Vienna). For animals up to postnatal day 30, mice were sacrificed by decapitation. For animals older than 30 days, transcardial perfusion was performed with ice-cold sucrose-artificial cerebrospinal fluid (sucrose-ACSF) solution. Animals were deeply anesthetized with isoflurane followed by the intraperitoneal injection of a mixture of xylazine (0.5βml, 2%), ketamine (1βml, 10%), acepromazine (0.3βml, 1.4%), and physiological NaCl solution (1.5βml, 0.9%). Anesthetics were applied at a dose of 0.033βml/10βg body weight. The depth of the anesthesia was verified by the absence of toe pinch reflexes.
For preparing slices, the brain was rapidly removed and immersed in ice-cold sucrose-ACSF solution during dissection. A block of tissue containing the hippocampus was transferred to a vibratome (VT 1200, Leica) and transverse slices of 300-Β΅m thickness were cut with blade oscillation amplitude of 1.25βmm and blade forward movement velocity of 0.03βmmβsβ160. Finally, slices were incubated at ~35βΒ°C in standard artificial cerebrospinal fluid (ACSF) for 30βminutes and subsequently maintained at ~22βΒ°C for maximally 5βh before transfer into the recording chamber.
Solutions and chemicals
The ACSF used for in vitroΒ recordings contained 125βmM NaCl, 25βmM NaHCO3, 25βmM glucose, 2.5βmM KCl, 1.25βmM NaH2PO4, 2βmM CaCl2, and 1βmM MgCl2. The sucrose-ACSF used for dissection contained 64βmM NaCl, 25βmM NaHCO3, 10βmM glucose, 120βmM sucrose, 2.5βmM KCl, 1.25βmM NaH2PO4, 0.5βmM CaCl2, and 7βmM MgCl2. The osmolarity of the solutions was 290β315βmOsm and the pH was maintained at ~7.3 when equilibrated with a 95% O2/5% CO2 gas mixture. The intracelluar solution for in vitro recordings contained 120 mM K-gluconate, 40 mM KCl, 2 mM MgCl2, 2 mM Na2ATP,Β 10 mM HEPES, 0.1 mM EGTA, and 0.3% biocytin, pH adjusted to 7.28 with KOH.Β Chemicals were purchased from Merck or Sigma-Aldrich.
Multi-cell recordings
Glass micropipettes were fabricated from thick-walled borosilicate tubing (2βmm outer diameter, 1βmm inner diameter) and had open-tip resistances of 3β8 MΞ©. They were manually positioned with eight LN mini 25 micromanipulators (Luigs and Neumann) under visual control14 provided by a modified Olympus BX51 microscope equipped with a 60x water-immersion objective (LUMPlan FI/IR, NAβ=β0.90, Olympus,Β 2.05 βmm working distance) and infrared differential interference contrast video microscopy and epifluorescence. To preserve connectivity, cell bodies ~30β120 ΞΌm below the surface of the slice were targeted for recording. Interneurons were identified on the basis of tdTomato or EGFP fluorescence in epifluorescence illumination and the AP phenotype upon 1-s current pulses (>50βHz in a series of pulses of 100β1,200 pA for PV+ interneurons). Mature GCs were identified on the basis of morphological appearance in the infrared image and on the basis of passive and active membrane properties. Cells with input resistanceβ>β500βMΞ©, potentially representing newborn GCs61, were not included in the analysis. Cells with resting potentials more positive than β55βmV were immediately discarded. In total, the number of successfully recorded cells per recording varied between eight and two. Recording temperature was ~22βΒ°C (range: 20β24βΒ°C, room temperature).
Electrical signals were acquired with four two-channel Multiclamp 700B amplifiers (Molecular Devices), low-pass filtered at 6β10βkHz, and digitized at 20βkHz with a Cambridge Electronic Design 1401 mkII AD/DA converter using custom-made stimulation-acquisition scripts running under Signal 6.0 software (CED). For current-clamp recordings, pipette capacitance was ~80% compensated and series resistance was balanced by the bridge circuit of the amplifier; settings were readjusted throughout the experiment when necessary. For voltage-clamp recordings, series resistance was not compensated, but repeatedly monitored using 2-mV hyperpolarizing pulses.
To test for synaptic connections, a presynaptic neuron was stimulated with a train of five or ten current pulses (2βms, 1β2βnA) at frequencies of 20 or 50βHz, while all other neurons were voltage-clamped at β70βmV (Fig.Β 1c). A connection was defined as monosynaptic if synaptic currents had latenciesββ€β4.0βms and peak amplitudes were larger than 2.5 times the standard deviation of the baseline of the average trace (computed from 15β30 individual traces). Events with latenciesβ>β4.0βms were considered polysynaptic.Β For distal SST+βGC synapses,Β connectivity may be underestimated, because ofΒ substantial attenuation of synaptic signals by cable filtering.Β
Data analysis
Recordings were analyzed using Stimfit and Python-based scripts62. Synaptic latency was measured from the peak of the presynaptic AP to the onset of the postsynaptic potential or current. Kinetic analysis of EPSCs or IPSCs was performed in pairs with series resistance ofβ<β15 MΞ©. Distance was measured from soma center to soma center. Analysis of the axonal arbor of PV+ interneurons and GCs revealed that the axonal length was 2.21βΒ±β0.20 and 1.59βΒ±β0.07 times larger than the corresponding intersomatic distance (Supplementary FigureΒ 4). Connection probability was calculated as number of connected pairs over total number of tested pairs in each 50-Β΅m distance interval. 95%-confidence intervals were obtained according to binomial distributions. Distance dependence of connectivity was fit with a sigmoidal function f(x)β=βA [1β+βExp[(x β B)/C]β1, where x is absolute distance, and A, B, and C are fitted parameters. Throughout the text, the maximal connection probability (cmax) was determined as f(0), and the space constant (dhalf) was determined as the xβ value that specified the condition f(xβ)/f(0)β=β0.5. To test whether connectivity differed between synapses, 10,000 bootstrap replications of the inhibitory PV+ interneuronβGC data set were obtained, and the mean values of the GCβPV+ interneuron and PV+ interneuronβPV+ interneuron experimental data sets were compared against the simulated distribution63. Values are given as meanβΒ±βstandard error of the mean (S.E.M.). Box plots show lower quartile (Q1), median (horizontal line), and upper quartile (Q3). The interquartile range (IQRβ=βQ3βQ1) is represented as the height of the box. Whiskers extend to the most extreme data point that is no more than 1.5 x IQR from the edge of the box (Tukey style). Statistical comparisons were done either with a non-parametric MannβWhitney U two-sided test or by linear regression, testing whether the slope was significantly different from 0.
To test whether disynaptic motifs64 occurred significantly more frequently than expected by chance, we simulated the entire set of recording configurations including PV+ interneurons (41 octuples, 62 septuples, 54 sextuples, 37 quintuples, 14 quadruples, 7 triples, and 3 pairs in 218 slices) 10,000 times, assuming random connectivity14,38,64. The connection probabilities were set to the experimentally determined distance-dependent values. For each simulated data set, we counted the number of all 25 possible disynaptic motifs (Fig.Β 5a). From the 10,000 bootstrap replications, mean, median, and confidence intervals for these counts were determined. P values were calculated as the number of replications in which the motif number was equal to or larger than the empirical number, divided by the number of replications. If a motif was never encountered in the 10,000 replications, P was assumed asβ<β0.0001. For assessing statistical significance, correction for multiple testing was performed using a BenjaminiβHochberg method that controls the false discovery rate65. P values for m comparisons were sorted in increasing order (P1ββ€βP2ββ€ββ¦ββ€βPm), the first Pi value that satisfied the condition Piββ€βi / m 0.05 was identified (starting with Pm), and the motifs corresponding to Pj values with 1ββ€βjββ€βi were considered significant. For illustration purposes, P values were converted into z scores, using the quantiles of a standard normal distribution.
Morphological analysis
Neurons that were filled with biocytin (0.3%) for >1βh were processed for morphological analysis. After withdrawal of the pipettes, resulting in the formation of outside-out patches at the pipette tips, slices were fixed for 12β24βh at 4βΒ°C in a 0.1βM phosphate buffer (PB) solution containing 2.5% paraformaldehyde, 1.25% glutaraldehyde, and 15% (v/v) saturated picric acid solution. After fixation, slices were treated with hydrogen peroxide (1%, 10βmin) to block endogenous peroxidases, and rinsed in PB several times. Membranes were permeabilized with 1% Triton X100 in PB for 1βh. Slices were then transferred to a PB solution containing 1% avidin-biotinylated horseradish peroxidase complex (ABC, Vector Laboratories) and 1% Triton X100 forβ~β12βh. Excess ABC was removed by several rinses in PB and the slices were developed with 0.05% 3,3β²-diaminobenzidine tetrahydrochloride (DAB) and subsequently hydrogen peroxide. Finally, slices were embedded in Mowiol (Sigma-Aldrich).
In vivo recordings from dentate gyrus PV+ interneurons
Whole-cell patch-clamp recordings in vivo were performed in male 35- to 63-day-old mice as described previously53. Animals were in the head-fixed, fully awake configuration, and were running on a linear belt treadmill66,67. The head-bar implantation and craniotomy were performed under anesthesia by intraperitoneal injection of 80βmg/kg ketamine (Intervet) and 8βmg/kg xylazine (Graeub), followed by local anesthesia with lidocaine. A custom-made steel head-bar was attached to the skull using superglue and dental cement. The day before recording, two small (~0.5βmm in diameter) craniotomies, one for the patch electrode and one for a local field potential (LFP) electrode, were drilled at the following coordinates: 2.0βmm caudal, 1.2βmm lateral for whole-cell recording; 2.5βmm caudal, 1.2βmm lateral for the LFP recording. The dura was left intact, and craniotomies were covered with silicone elastomer (Kwik-Cast, World Precision Instruments). Pipettes were fabricated from borosilicate glass capillaries (1.75βmm outer diameter, 1.25βmm inner diameter). Long-taper whole-cell patch electrodes (9β12 MΞ©) were filled with a solution containing: 130 mM K-gluconate, 2βmM KCl, 2βmM MgCl2, 2βmM Na2ATP, 0.3βmM NaGTP, 10βmM HEPES, 18βmM sucrose, 10 or 0.1Β mM EGTA, and 0.3% biocytin, pH adjusted to 7.28 with KOH. Whole-cell patch electrodes were advanced through the cortex with 500β600βmbar of pressure to prevent the electrode tip from clogging. After passing the hippocampus CA1 subfield, the pressure was reduced to 20βmbar. After the blind whole-cell recording was obtained, series resistance was calculated by applying a test pulse (β+β50βmV and β10βmV) under voltage-clamp conditions. Recordings were immediately discarded if series resistance exceeded 100βMΞ©. After the bridge balance was compensated, step currents from β100 pA to 400 pA were injected to calculate input resistance and maximal firing frequency of the recorded cells. All the recordings were done in current-clamp experiment configuration without holding current injection using a Heka EPC double amplifier. Signals were low-pass filtered at 10βkHz (Bessel) and sampled at 25βkHz with Heka Patchmaster acquisition software. After recording, the patch pipettes were slowly withdrawn to form an outside-out patch, verifying the integrity of the seal. Data included were obtained from three fast-spiking cells in the dentate gyrus, which generated APs during sustained current injection at a frequency of >100βHz. To determine the relative AP threshold, spontaneous action potentials (sAPs) were detected, using either a single sAP or the first AP in a burst. The membrane potential preceding the sAP was measured in a 10β20βms time window before the sAP. sAP absolute threshold was determined from a dV/dtβV phase plot; the rising phase was fit with an exponential function including a shift factor, and the intersection of the fit curve with the baseline was defined as threshold.
Data availability
Original data, analysis programs, and computer code will be provided by the corresponding author (P.J.) upon request.
References
Hu, H., Gan, J. & Jonas, P. Interneurons. Fast-spiking, parvalbumin+ GABAergic interneurons: from cellular design to microcircuit function. Science 345, 1255263 (2014).
Packer, A. M. & Yuste, R. Dense, unspecific connectivity of neocortical parvalbumin-positive interneurons: a canonical microcircuit for inhibition? J. Neurosci. 31, 13260β13271 (2011).
Holmgren, C., Harkany, T., Svennenfors, B. & Zilberter, Y. Pyramidal cell communication within local networks in layer 2/3 of rat neocortex. J. Physiol. 551, 139β153 (2003).
Yoshimura, Y. & Callaway, E. M. Fine-scale specificity of cortical networks depends on inhibitory cell type and connectivity. Nat. Neurosci. 8, 1552β1559 (2005).
Couey, J. J. et al. Recurrent inhibitory circuitry as a mechanism for grid formation. Nat. Neurosci. 16, 318β324 (2013).
Peng, Y., Barreda TomΓ‘s, F. J., Klisch, C., Vida, I. & Geiger, J. R. P. Layer-specific organization of local excitatory and inhibitory synaptic connectivity in the rat presubiculum. Cereb. Cortex 27, 2435β2452 (2017).
Znamenskiy, P., et al Functional selectivity and specific connectivity of inhibitory neurons in primary visual cortex. //www.biorxiv.org/, https://doi.org/10.1101/294835 (2018).
Leutgeb, J. K., Leutgeb, S., Moser, M. B. & Moser, E. I. Pattern separation in the dentate gyrus and CA3 of the hippocampus. Science 315, 961β966 (2007).
Rolls, E. T. Pattern separation, completion, and categorisation in the hippocampus and neocortex. Neurobiol. Learn. Mem. 129, 4β28 (2016).
Cayco-Gajic, N. A., Clopath, C. & Silver, R. A. Sparse synaptic connectivity is required for decorrelation and pattern separation in feedforward networks. Nat. Commun. 8, 1116 (2017).
Chavlis, S. & Poirazi, P. Pattern separation in the hippocampus through the eyes of computational modeling. Synapse 71, e21972 (2017).
Leal, S. L. & Yassa, M. A. Integrating new findings and examining clinical applications of pattern separation. Nat. Neurosci. 21, 163β173 (2018).
Bischofberger, J., Engel, D., Frotscher, M. & Jonas, P. Timing and efficacy of transmitter release at mossy fiber synapses in the hippocampal network. PflΓΌgers Arch. 453, 361β372 (2006).
Guzman, S. J., SchlΓΆgl, A., Frotscher, M. & Jonas, P. Synaptic mechanisms of pattern completion in the hippocampal CA3 network. Science 353, 1117β1123 (2016).
Majani, E., Erlarson, R. & Abu-Mostafa, Y. On the k-winners takes-all network. Adv. Neural Inf. Process Syst. 1, 634β642 (1988).
Maass, W. On the computational power of winner-take-all. Neural Comput. 12, 2519β2535 (2000).
de Almeida, L., Idiart, M. & Lisman, J. E. A second function of gamma frequency oscillations: an E%-max winner-take-all mechanism selects which cells fire. J. Neurosci. 29, 7497β7503 (2009).
Tetzlaff, T., Helias., M., Einevoll, G. T. & Diesmann, M. Decorrelation of neural-network activity by inhibitory feedback. PLoS. Comput. Biol. 8, e1002596 (2012).
Faghihi, F. & Moustafa, A. A. A computational model of pattern separation efficiency in the dentate gyrus with implications in schizophrenia. Front. Syst. Neurosci. 9, 42 (2015).
Wiechert, M. T., Judkewitz, B., Riecke, H. & Friedrich, R. W. Mechanisms of pattern decorrelation by recurrent neuronal circuits. Nat. Neurosci. 13, 1003β1010 (2010).
Lin, A. C., Bygrave, A. M., de Calignon, A., Lee, T. & MiesenbΓΆck, G. Sparse, decorrelated odor coding in the mushroom body enhances learned odor discrimination. Nat. Neurosci. 17, 559β568 (2014).
Gschwend, O. et al. Neuronal pattern separation in the olfactory bulb improves odor discrimination learning. Nat. Neurosci. 18, 1474β1482 (2015).
Engin, E. et al. Tonic inhibitory control of dentate gyrus granule cells by Ξ±5-containing GABAA receptors reduces memory interference. J. Neurosci. 35, 13698β13712 (2015).
Temprana, S. G. et al. Delayed coupling to feedback inhibition during a critical period for the integration of adult-born granule cells. Neuron 85, 116β130 (2015).
Andersen, P., Eccles, J. C. & LΓΈyning, Y. Recurrent inhibition in the hippocampus with identification of the inhibitory cell and its synapses. Nature 198, 540β542 (1963).
Jonas, P. & BuzsΓ‘ki, G. Neural inhibition. Scholarpedia 2, 3286 (2007).
Chow, S. F., Wick, S. D. & Riecke, H. Neurogenesis drives stimulus decorrelation in a model of the olfactory bulb. PLoS. Comput. Biol. 8, e1002398 (2012).
Miles, R. Synaptic excitation of inhibitory cells by single CA3 hippocampal pyramidal cells of the guinea-pig in vitro. J. Physiol. 428, 61β77 (1990).
Jouhanneau, J. S., Kremkow, J. & Poulet, J. F. A. Single synaptic inputs drive high-precision action potentials in parvalbumin expressing GABA-ergic cortical neurons in vivo. Nat. Commun. 9, 1540 (2018).
Pala, A. & Petersen, C. C. H. In vivo measurement of cell-type-specific synaptic connectivity and synaptic transmission in layer 2/3 mouse barrel cortex. Neuron 85, 68β75 (2015).
Scharfman, H. E., Kunkel, D. D. & Schwartzkroin, P. A. Synaptic connections of dentate granule cells and hilar neurons: results of paired intracellular recordings and intracellular horseradish peroxidase injections. Neuroscience 37, 693β707 (1990).
Geiger, J. R. P., LΓΌbke, J., Roth, A., Frotscher, M. & Jonas, P. Submillisecond AMPA receptor-mediated signaling at a principal neuron-interneuron synapse. Neuron 18, 1009β1023 (1997).
Galarreta, M. & Hestrin, S. A network of fast-spiking cells in the neocortex connected by electrical synapses. Nature 402, 72β75 (1999).
Bartos, M., Vida, I., Frotscher, M., Geiger, J. R. P. & Jonas, P. Rapid signaling at inhibitory synapses in a dentate gyrus interneuron network. J. Neurosci. 21, 2687β2698 (2001).
Galarreta, M. & Hestrin, S. Spike transmission and synchrony detection in networks of GABAergic interneurons. Science 292, 2295β2299 (2001).
Larimer, P. & Strowbridge, B. W. Nonrandom local circuits in the dentate gyrus. J. Neurosci. 28, 12212β12223 (2008).
SchrΓΆter, M., Paulsen, O. & Bullmore, E. T. Micro-connectomics: probing the organization of neuronal networks at the cellular scale. Nat. Rev. Neurosci. 18, 131β146 (2017).
ErdΓΆs, P. & RΓ©nyi, A. On random graphs. I. Publ. Math. 6, 290β297 (1959).
Song, S., SjΓΆstrΓΆm, P. J., Reigl, M., Nelson, S. & Chklovskii, D. B. Highly nonrandom features of synaptic connectivity in local cortical circuits. PLoS Biol. 3, e68 (2005).
Perin, R., Berger, T. K. & Markram, H. A synaptic organizing principle for cortical neuronal groups. Proc. Natl Acad. Sci. USA 108, 5419β5424 (2011).
Cossell, L. et al. Functional organization of excitatory synaptic strength in primary visual cortex. Nature 518, 399β403 (2015).
Jouhanneau, J. S., Kremkow, J., Dorrn, A. L. & Poulet, J. F. In vivo monosynaptic excitatory transmission between layer 2 cortical pyramidal neurons. Cell Rep. 13, 2098β2106 (2015).
Rieubland, S., Roth, A. & HΓ€usser, M. Structured connectivity in cerebellar inhibitory networks. Neuron 81, 913β929 (2014).
Hu, H. & Jonas, P. A supercritical density of Na+ channels ensures fast signaling in GABAergic interneuron axons. Nat. Neurosci. 17, 686β693 (2014).
Barabasi, A. L. & Albert, R. Emergence of scaling in random networks. Science 286, 509β512 (1999).
Albert, R., Jeong, H. & Barabasi, A. L. Error and attack tolerance of complex networks. Nature 406, 378β382 (2000).
Bartos, M., Vida, I. & Jonas, P. Synaptic mechanisms of synchronized gamma oscillations in inhibitory interneuron networks. Nat. Rev. Neurosci. 8, 45β56 (2007).
Bragin, A. et al. Gamma (40β100 Hz) oscillation in the hippocampus of the behaving rat. J. Neurosci. 15, 47β60 (1995).
StrΓΌber, M., Sauer, J. F., Jonas, P. & Bartos, M. Distance-dependent inhibition facilitates focality of gamma oscillations in the dentate gyrus. Nat. Commun. 8, 758 (2017).
BuzsΓ‘ki, G., Geisler, C., Henze, D. A. & Wang, X. J. Interneuron diversity series: circuit complexity and axon wiring economy of cortical interneurons. Trends Neurosci. 27, 186β193 (2004).
Watts, D. J. & Strogatz, S. H. Collective dynamics of βsmall-worldβ networks. Nature 393, 440β442 (1998).
Song, H. F. & Wang, X.-J. Simple, distance-dependent formulation of the Watts-Strogatz model for directed and undirected small-world networks. Phys. Rev. E 90, 062801 (2014).
PernΓa-Andrade, A. J. & Jonas, P. Theta-gamma-modulated synaptic currents in hippocampal granule cells in vivo define a mechanism for network oscillations. Neuron 81, 140β152 (2014).
Pilz, G. A. et al. Functional imaging of dentate granule cells in the adult mouse hippocampus. J. Neurosci. 36, 7407β7414 (2016).
Danielson, N. B. et al. In vivo imaging of dentate gyrus mossy cells in behaving mice. Neuron 93, 552β559 (2017).
GoodSmith, D. et al. Spatial representations of granule cells and mossy cells of the dentate gyrus. Neuron 93, 677β690 (2017).
Senzai, Y. & BuzsΓ‘ki, G. Physiological properties and behavioral correlates of hippocampal granule cells and mossy cells. Neuron 93, 691β704 (2017).
Kesner, R. P. An analysis of dentate gyrus function (an update). Behav. Brain Res. 354, 84β91 (2018).
Taniguchi, H. et al. A resource of Cre driver lines for genetic targeting of GABAergic neurons in cerebral cortex. Neuron 71, 995β1013 (2011).
Bischofberger, J., Engel, D., Li, L., Geiger, J. R. P. & Jonas, P. Patch-clamp recording from mossy fiber terminals in hippocampal slices. Nat. Protoc. 1, 2075β2081 (2006).
Schmidt-Hieber, C., Jonas, P. & Bischofberger, J. Enhanced synaptic plasticity in newly generated granule cells of the adult hippocampus. Nature 429, 184β187 (2004).
Guzman, S. J., SchlΓΆgl, A. & Schmidt-Hieber, C. Stimfit: quantifying electrophysiological data with Python. Front. Neuroinform. 8, 16 (2014).
Efron, B. & Tibshirani, R. J. An introduction to the bootstrap. (Chapman & Hall/CRC, London, 1998).
Zhao, L., Beverlin, B. 2nd, Netoff, T. & Nykamp, D. Q. Synchronization from second order network connectivity statistics. Front. Comput. Neurosci. 5, 28 (2011).
Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. B 57, 289β300 (1995).
Royer, S. et al. Control of timing, rate and bursts of hippocampal place cells by dendritic and somatic inhibition. Nat. Neurosci. 15, 769β775 (2012).
Bittner, K. C. et al. Conjunctive input processing drives feature selectivity in hippocampal CA1 neurons. Nat. Neurosci. 18, 1133β1142 (2015).
Han, Z. S., Buhl, E. H., LΓΆrinczi, Z. & Somogyi, P. A high degree of spatial selectivity in the axonal and dendritic domains of physiologically identified local-circuit neurons in the dentate gyrus of the rat hippocampus. Eur. J. Neurosci. 5, 395β410 (1993).
Hosp, J. A. et al. Morpho-physiological criteria divide dentate gyrus interneurons into classes. Hippocampus 24, 189β203 (2014).
Hefft, S. & Jonas, P. Asynchronous GABA release generates long-lasting inhibition at a hippocampal interneuronβprincipal neuron synapse. Nat. Neurosci. 8, 1319β1328 (2005).
Acknowledgements
We thank Drs. Ryuichi Shigemoto and Federico Stella for critically reading an earlier version of the manuscript, and Nancy Kopell for useful suggestions. We are grateful to Florian Marr for excellent technical assistance and biocytin labeling, Alois SchlΓΆgl for programming, Christina Altmutter for technical help, Eleftheria Kralli-Beller for manuscript editing, and the Scientific Service Units of IST Austria for efficient support. Finally, we thank Rainer Friedrich for helpful discussions and Caroline Uhler for advice on statistics. This project received funding from the European Research Council (ERC) under the European Unionβs Horizon 2020 research and innovation programme (grant agreement No 692692) and the Fond zur FΓΆrderung der Wissenschaftlichen Forschung (Z 312-B27, Wittgenstein award), both to P.J..
Author information
Authors and Affiliations
Contributions
S.J.G. and P.J. designed the experiments, C.E. performed the in vitro experiments, X.Z. performed the in vivo experiments, C.E., S.J.G., and P.J. analyzed the data, and P.J. wrote the paper. All authors jointly revised the paper.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisherβs note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the articleβs Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the articleβs Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Espinoza, C., Guzman, S.J., Zhang, X. et al. Parvalbumin+ interneurons obey unique connectivity rules and establish a powerful lateral-inhibition microcircuit in dentate gyrus. Nat Commun 9, 4605 (2018). https://doi.org/10.1038/s41467-018-06899-3
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-018-06899-3
This article is cited by
-
The relationship between gamma-band neural oscillations and language skills in youth with Autism Spectrum Disorder and their first-degree relatives
Molecular Autism (2024)
-
Dynamic and selective engrams emerge with memory consolidation
Nature Neuroscience (2024)
-
Extracellular matrix abnormalities in the hippocampus of subjects with substance use disorder
Translational Psychiatry (2024)
-
Functional Neuroligin-2-MDGA1 interactions differentially regulate synaptic GABAARs and cytosolic gephyrin aggregation
Communications Biology (2024)
-
Modification of temporal pattern sensitivity for inputs from medial entorhinal cortex by lateral inputs in hippocampal granule cells
Cognitive Neurodynamics (2024)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.