Abstract

Down syndrome is the most common genetic neurodevelopmental disorder associated with mild-to-moderate intellectual disability. A disturbed excitation-inhibition balance is thought to be a major cause for the intellectual deficits in DS. In this study, we used patch-clamp electrophysiology, optogenetic stimulation and immunohistochemistry to investigate synaptic inhibition from specific interneuron subpopulations onto granule cells of the dentate gyrus in Ts65Dn mice. Optogenetically evoked inhibitory postsynaptic currents (IPSCs) from somatostatin (SOM) interneurons onto dendrites of granule cells in the outer molecular layer (ML) did not differ between euploid (Eu) and Ts65Dn mice, indicating normal distal dendritic inhibition in Ts65Dn mice. In addition, optogenetically evoked IPSCs from parvalbumin (PV) interneurons were significantly reduced, indicating reduced functional somatic inhibition in Ts65Dn mice. In contrast, activation of cholecystokinin (CCK) positive interneurons by targeted electrical microstimulation of the inner molecular layer (iML) resulted in IPSCs of increased amplitude (Eu: 372.5 ± 51.97 pA, n = 10; Ts65Dn: 619.9 ± 74.68 pA, n = 9). GABAergic synaptic terminals of CCK interneurons express cannabinoid receptor 1 (CB1). Quantitative analysis of synapses double-labeled for CB1 and the vesicular GABA transporter (VGAT) revealed a significantly increased number of putative CCK interneuron terminals in the iML of Ts65Dn mice (Eu: 0.048 ± 0.014 puncta/μm3, Ts65Dn: 0.34 ± 0.12 puncta/μm3). In contrast, the density of PV-VGAT double-positive synapses within the granule cell layer did not differ between the two genotypes. Taken together, our results indicate that proximal dendritic inhibition from CCK interneurons is increased in the dentate gyrus of Ts65Dn mice, while PV interneuron-mediated somatic inhibition appears to be unchanged or functionally diminished.

Introduction

Down syndrome (DS) is the most common genetic disorder of intellectual disability (Desai, 1997; Antonarakis et al., 2020). In addition to the neurological system, DS affects the cardiovascular and musculoskeletal systems including common phenotypes of short stature, muscle hypotonia and congenital heart defects. Importantly, patients have a significantly higher risk of suffering from epilepsy and early-onset Alzheimer (Antonarakis et al., 2020). The underlying genetic mechanism is a triplication of the human chromosome 21 causing a perturbation in gene regulation and expression (Dierssen, 2012; Antonarakis et al., 2020). There are currently no approved pharmacological treatments to address cognitive deficits in DS patients. The most widely used model to investigate the pathophysiological mechanisms underlying impairments in DS is the Ts65Dn mouse in which a segmental trisomy for the distal murine chromosome 16 was produced by reciprocal translocation. This segment hosts genes, which are evolutionarily conserved from human chromosome 21, including the DS critical region (Reeves et al., 1995). Ts65Dn mice recapitulate most DS phenotypes. In particular, they show the typical neurological disturbances like deficits in short- and long-term memory, reduced neurogenesis and impaired performance on complex visual spatial learning tasks (Chakrabarti et al., 2007; Dierssen, 2012).

An imbalance of excitation and inhibition in the trisomic brain is thought to be the main cause for the deficits in declarative learning and memory (Kleschevnikov et al., 2004; Chakrabarti et al., 2010; Rueda et al., 2020). Long-term potentiation (LTP), thought to be the synaptic mechanisms underlying memory, is impaired in Ts65Dn mice (Malenka and Nicoll, 1999; Kleschevnikov et al., 2004). Calcium influx through NMDA receptors is necessary for most forms of LTP in the hippocampus (Malenka and Nicoll, 1999), and NMDA receptor activation has recently been shown to be impaired due to increased dendritic inhibition in the CA1 area (Schulz et al., 2019). Importantly, application of non-competitive GABAA receptor blockers (Pentylenetetrazol or Picrotoxin) in non-epileptic concentrations rescued the behavior in novel object recognition tasks and the induction of LTP (Kleschevnikov et al., 2004; Fernandez et al., 2007). Together, these results suggested that aberrantly increased GABAA receptor mediated inhibition leads to abnormalities in hippocampus-dependent learning and memory in DS.

The dentate gyrus (DG) of the hippocampus is essential for memory processing as it contributes to pattern separation (Leutgeb et al., 2007; McHugh et al., 2007; Myers and Scharfman, 2011; Madar et al., 2019). Different interneuron subpopulations control its activity and output along the trisynaptic path (Hu et al., 2014; Elgueta and Bartos, 2019). In the DG of Ts65Dn mice, Kleschevnikov and colleagues found an increased frequency of miniature IPSCs (mIPSCs) and strongly increased electrically evoked GABAA- and GABAB-mediated IPSCs (Kleschevnikov et al., 2012). Which specific interneurons mediated this increase in inhibition could not be determined using the available approaches. More recent studies using cell type or layer specific activation of GABAergic inputs have shown that enhanced dendritic inhibition likely mediated by somatostatin (SOM) expressing interneurons contributes to altered inhibition in the CA1 area and the prefrontal cortex (PFC) (Schulz et al., 2019; de Zorilla San Martin et al., 2020). Therefore, we used optogenetic and targeted electrical stimulation of specific inputs from GABAergic interneurons in combination with immunohistochemistry to investigate synaptic inhibition from specific interneuron subpopulations onto granule cells of the DG in Ts65Dn mice in the present study. The particular focus was on SOM-, cholecystokinin (CCK)-positive, and parvalbumin (PV-) interneurons mediating distal dendritic, proximal dendritic and perisomatic inhibition.

Materials and methodsAnimals

Electrical stimulation experiments were performed on Ts65Dn (B6EiC3Sn a/A-Ts (1716) 65Dn/J) mice obtained from The Jackson Laboratory. Ts65Dn females were backcrossed with B6C3F1/OlaHsd males (F1 hybrid males from C57BL/6JOlaHsd ♀ × C3H/HeNHsd ♂ obtained from Harlan). For optogenetic experiments, we used triple crosses of B6EiC3Sn a/A-Ts(1716)65Dn/J ♀ × F1 ♂ (B6.129P2-Pvalbtm1 (cre) Arbr/J × Ai32 (RCL-ChR2(H134R)/EYFP)) and B6EiC3Sn a/A-Ts(1716)65Dn/J ♀ × F1 ♂ (SST tm2.1(cre) Zjh/J × Ai32 (RCL-ChR2 (H134R)/EYFP)), referred to as Eu or Ts65Dn, in the remainder of the text. We identified the supernumerary chromosome in 1716 of Ts65Dn mice in the offspring by standard polymerase chain reaction (PCR) procedures (Reinholdt et al., 2011). Eu and Ts65Dn mice were littermates. In this study experiments were performed with 6-week-old and fully mature (8-week-old) mice of both sexes. Mice were housed in groups of up to 5 animals in standard ventilated cages with a 12 h light/dark cycle and had ad libitum access to food and water. All experiments were approved by the Basel Cantonal Committee on Animal Experimentation according to federal and cantonal regulations.

Perfusion and slice preparation for immunohistochemistry

Five Eu (3 male, 2 female) and four Ts65Dn (2 male, 2 female) fully mature mice were transcardially perfused under deep anesthesia with 0.9% saline followed by paraformaldehyde (PFA, 4% in 0.1 M phosphate buffered saline (PB), pH 7.4). The brains were post-fixed overnight in PFA at 4 °C. Afterwards they were cryoprotected in 30% sucrose and frozen in isopentane. Frontal sections with a thickness of 50 μm were cut on a cryostat and stored at 4 °C in 0.1 M PB for immunohistochemistry.

Slice preparation for patch clamp recordings

Mice (41 male, 22 female) were placed for approximately 10 min in an oxygen-enriched environment to increase cell survival. We anesthetized the animals with isoflurane (4% in O2, Vapor, Draeger). Decapitation was performed in accordance with the national and institutional guidelines as soon as the righting reflex was absent. Slices were cut as previously described (Geiger et al., 2002; Bischofberger et al., 2006): The brain was carefully removed in ice-cold sucrose-based solution that contained (in mM): 87 NaCl, 25 NaHCO3, 2.5 KCl, 1.25 NaH2PO4, 75 sucrose, 0.5 CaCl2, 7 MgCl2 and 10 glucose (equilibrated with 95% O2/5% CO2; Osmolarity: 325–328 mOsm). Parahorizontal slices of 350 μm thickness were cut with a VT1200 vibratome (Leica Microsystems) and collected in the same sucrose-based solution. Finally, they were incubated at 35 °C for 30 min and stored at room temperature until the experiments were performed.

Patch clamp recordings

We visually identified granule cells (GCs) by using infrared differential interference contrast video microscopy. The slices were continuously superfused with ACSF at room temperature, containing the following (in mM): 125 NaCl, 25 glucose, 25 NaHCO3, 2.5 KCl, 1.25 NaH2PO4, 2 CaCl2 and 1 MgCl2 (equilibrated with 95% O2/5% CO2; Osmolarity: 318–320 mOsm). Patch pipettes were pulled from borosilicate glass tubing (outer diameter: 2.0 mm; wall thickness: 0.5 mm; Hilgenberg) on a Flaming-Brown P-97 puller (Sutter Instruments).

GCs were voltage-clamped at −80 mV in the whole-cell configuration. The current signal was measured using a Multiclamp 700A amplifier (Molecular Devices), low-pass filtered with a cutoff frequency of 8 kHz and digitized at 20 kHz via a CED Power 1401 Interface (Cambridge Electronic Design). Patch pipettes had a resistance in a range of 2-4 MΩ. The internal solution depended on the type of experiment. In general, we used potassium-chloride-based (KCl) internal solutions for the recording of proximal synaptic inputs and cesium-chloride-based (CsCl) internal solutions for the recording of distal synaptic inputs from SOM interneurons. Specifically, we used the following KCl-based solution with addition of a potassium-gluconate for greater stability in the initial experiments employing electrical stimulation of GABAergic inputs in the iML of 6-week-old mice (in mM): 52 KGluc, 100 KCl, 10 HEPES, 0.1 EGTA, 2 MgCl2, 2 Na2ATP, 1 Phosphocreatine, 0.3 GTP, 5 QX314 (bromide or chloride) and 0.1% biocytin, adjusted to pH 7.28 with KOH. Subsequently, we employed KCl-based solutions for electrical stimulation (in mM): 120 KCl, 10 HEPES, 2 MgCl2, 20 BAPTA, 2 Na2ATP, 0.3 GTP, 5 QX314 (bromide or chloride); and optogenetic stimulation of PV interneurons (in mM): 140 KCl, 10 EGTA, 10 HEPES, 2 MgCl2, 2 Na2ATP, 0.3 GTP, 5 QX314 (bromide or chloride) and 0.1% biocytin. The pH was adjusted to 7.3 with KOH and the osmolarity was approximately 315 mOsm/L. For recording of distal dendritic inputs evoked by optogenetic stimulation of SOM-interneurons, pipettes were filled with a CsCl-based solution with addition of cesium-gluconate (CsGluc) containing the following in (mM): 40 CsGluc, 100 CsCl, 10 EGTA, 10 HEPES, 2 MgCl2, 2 Na2ATP, 2 TEA-Cl, 0.3 GTP and 5 QX314 (bromide or chloride) and 0.1% biocytin, adjusted to pH 7.3 with CsOH.

Criteria to include recordings in the analysis were the following: initial seal resistances were at least 1 GΩ; series resistance (Rs) changed less than 30% during the recording and did not exceed 20 MΩ. An online analysis was performed to monitor the Rs and input resistance (Rin) during the experiments, by applying a − 5 mV voltage step at the end of each sweep. Furthermore, we excluded experiments, if IPSCs were recorded more than 7 h after slice preparation. All experiments were performed at room temperature (20–22 °C). Data was collected using IGOR Pro 6.31 (WaveMetrics) and the CSF library support from Cambridge Electronic Design.

Extracellular synaptic stimulation

To stimulate synaptic inputs, pipettes with a resistance of 3–6 MΩ were filled with a HEPES-buffered Na2+-rich solution. The stimulating pipette was placed in the inner molecular layer (iML), close to the granule cell layer (GCL). Recordings in the electrical stimulation experiments were performed in presence of 3 mM Kynurenic acid (Kyn) to block glutamatergic receptors at glutamatergic synapses, which are also recruited by electrical stimulation to ensure that the recorded IPSCs were GABAergic. Two electrical stimuli of 0.2 ms length were applied at low intensities (10–30 μA) with an inter-stimulus-interval (ISI) of 200 ms.

Channelrhodopsin-2 (ChR2) mediated activation of GABAergic interneurons

A diode laser (DL-473, Rapp OptoElectronic GmbH) was used in this study. The laser was coupled to the epifluorescence port of the respective microscope (Zeiss Examiner, equipped with a 63x NA = 1.0 water immersion objective; Carl Zeiss Microscopy GmbH) via fiber optics and were controlled via TTL pulses. Whole field stimulation was performed using light with a wavelength of 473 nm. Two pulses were applied (1 ms duration each). Due to different decay time constants of the evoked IPSCs, the ISI was adjusted to trigger a second response after the first had almost returned back to baseline (stimulation of PV interneurons: ISI = 200 ms; stimulation of SOM interneurons: ISI = 500 ms). The field of illumination was targeted either to the GCL, when stimulating PV interneurons, or the border region of the ML to the stratum lacunosum moleculare to stimulate dendritic inputs mediated by SOM interneurons. Recordings of SOM inputs were done in presence of 3 mM Kyn to block glutamatergic receptors, because some Cre-driver lines may have unspecific expression in excitatory neurons (Müller-Komorowska et al., 2020).

Drugs and reagents

Picrotoxin (PTX, Sigma Aldrich) was dissolved at 50 mM in ethanol, Kyn (Tocris) at 300 mM in 0.6 N NaOH and gabazine (Tocris) at 10 mM in Millipore water. Drug aliquots were stored in the freezer at −20 °C. The dilutions of these drugs in ASCF were prepared at the day of the experiment at the following concentrations: 100 μM PTX, 3 mM Kyn and 10 μM gabazine.

Immunohistochemistry

Acute slices (350 μm thick) with biocytin-filled cells were fixed overnight in 4% PFA and underwent the following protocol. First, they were washed in 0.1 M PB three times (15 min each). Blocking was done for 1 h at room temperature with the following solution (per slice): 475 μL 0.3% Triton X-100, 25 μL normal donkey serum (NDS). Next, the slices were transferred in the solution of the primary antibodies for 72 h at 4 °C (per slice): 475 μL 0.3% Triton X-100, 25 μL NDS, chicken anti-green-fluorescent-protein (GFP, 1:1000, #ab13970, Abcam). After another four washing steps (15 min each) in 0.1 M PB slices were put in the secondary antibody solution (per slice): 475 μL 0.3% Triton X-100, 25 μL NDS, donkey anti-chicken Alexa 488 (1:500, #703–545-155, Jackson ImmunoResearch), Streptavidin Alexa Fluor 568 (1:1000, #s11226, MoBiTech) and DAPI (1:10000, Sigma Aldrich). We incubated slices overnight at 4 °C. The slices were washed in another three steps (10 min-2*15 min) in 0.1 M PB. Slices were placed on microscope slides, liquid was removed with filter paper and they were mounted directly with ProLong Gold Antifade (Thermo Fisher Scientific).

Another protocol was used for immunohistochemistry of 50 μm fixed cryostat slices. After washing in 0.1 M PB, slices were preincubated with 0.25% Triton X-100 and 10% normal horse serum (NHS) in 0.1 M PB for 30 min at room temperature. Followed by incubation with primary antibodies 3 h at room temperature and 4 °C overnight: guinea pig anti-vesicular GABA transporter (VGAT, 1:500, #131004, Synaptic Systems), rabbit anti-PV (1:1000, #PV-27, Swant) and rabbit anti-cannabinoid receptor type 1 (CB1, 1:500, #258003, Synaptic Systems). Slices were transferred into the secondary antibodies with the following dilutions: donkey anti-rabbit Alexa 488 (1:500, #A21206, Thermo Fisher Scientific), donkey anti-guinea pig Cy5 (1:200, #706175148, Jackson ImmunoResearch), and counterstained with DAPI (1:10000, Roche Diagnostics). After incubating 3 h at room temperature, slices were washed in 0.1 M PB, mounted and coverslipped with DAKO fluorescence mounting medium.

Imaging and 3D-reconstruction

Images of 350 μm thick slices were taken with a Zeiss LSM 700 confocal microscope. Zeiss ZEN software was used for the processing.

Cryostat slices (50 μm thick) were imaged with a Zeiss LSM 880 confocal microscope with AiryScan using a 40x water immersion (1.2 NA, additional 2x zoom factor) or a 63x oil immersion objective (1.4 NA, additional 2x zoom factor). Images were taken with a z-stack step size of 0.14 μm and had approximately a total thickness of 7 μm. We measured chromatic aberration with microscopic beads and could not find a distortion.

We imaged three slices per animal and four regions within each slice. In total 12 images per animal were analyzed. 3D-reconstruction was done in Imaris (Version 9.9.2 Bitplane). First, a baseline subtraction was performed on the VGAT channel. Next, puncta were automatically reconstructed with the spot analysis tool of Imaris. A mask of VGAT+ puncta was created in either the CB1 or PV channel, to mark all double positive synapses. A second automatic detection of VGAT+ and CB1+ or PV+ puncta was again performed with the spot analysis tool. Density of VGAT+ and double positive VGAT+ and CB1+ or PV+ puncta was calculated by dividing the total number of puncta by the volume of the whole z-stack.

Data analysis

The analysis of patch clamp data was performed with the software Stimfit (https://neurodroid. Github.io/stimfit) (Guzman et al., 2014) and scripts written in Python. On mean traces, the amplitude of the two responses to stimulation was identified by measuring the peak over 5 sample points and subtracting the baseline formed by the mean over 1,280 sample points or 64 ms, respectively.

Input resistance and Rs were determined directly after break-in and through-out the experiment, respectively, by applying a − 5 mV step pulse. The baseline for those measurements was calculated as the mean over 1,800 sample points or 90 ms before the start of the −5 mV voltage step. To evaluate Rin, the mean was calculated over a defined 50 ms time window, starting 100 ms after the onset of the voltage pulse. The difference between the calculated mean value and the described baseline was computed and used to calculate Rin. To evaluate the Rs, the onset peak amplitude was determined and subtracted from the baseline. Another evaluated cellular property was the decay time constant τ(break-in). It was measured for the onset peak amplitude, in response to a − 5 mV voltage step directly after break-in. The time constant τ(break-in) was estimated by the weighted average of time constants from a biexponential fit starting at 95% of the amplitude and ending approximately 20 ms after the onset of the voltage step. The estimate of the cell capacitance (Cp) was calculated with τ(break-in) and Rs directly after break-in.

We analyzed the risetime and decay time constant τ for IPSCs in response to a double pulse stimulation. The risetime was calculated from 20 to 80% of the peak amplitude for the first IPSC. A biexponential fit as for τ(break-in) was used to determine the time constant τ of the second IPSC starting at 95% of the amplitude and ending 250 ms after the stimulation. τ represents the weighted average of the fit.

Inclusion criteria for electrophysiological recordings

Neurogenesis in the DG continues well into adulthood, but declines with age (Ben Abdallah et al., 2010). New-born and immature GCs show striking differences of cellular and synaptic electrophysiological properties including the cellular input resistance compared to mature GCs (see, e.g., Heigele et al., 2016). We used the following strategy to obtain a homogenous sample of GCs to enable robust statistical evaluation of genotype-specific effects on evoked GABAergic inputs: 1. GCs were recorded from the outer GCL adjacent to the ML where primarily mature GCs reside. 2. Based on recordings of outer GCs obtained from 8-week-old mice kept in standard housing, we used the 95-percentile of the cumulative contribution of Rin of 550 MΩ as the cut-off threshold for the inclusion of GCs in our analysis (Supplementary Figure S1). Rin measurements were neither affected by the intracellular solution nor by the genotype, allowing us a pooling of Rin from all experiments (Supplementary Figure S1). In this GC population with Rin ≤ 550 MΩ, GCs can be considered to be relatively mature. Table 1 shows the cell parameters Rin, τ(break-in) and Cp for all cells recorded in 6- and 8-week-old mice. There were no significant differences between Eu and Ts65Dn mice at a comparable level of cellular maturity (p ≥ 0.11, unpaired t-tests two-tailed).

EuTs65DnRin ≤ 550 MΩ (n = 102)Rin > 550 MΩ (n = 26)Rin ≤ 550 MΩ (n = 89)Rin > 550 MΩ (n = 23)Rin [MΩ]341.5 ± 0.01818.6 ± 51.3322.3 ± 12.7874.9 ± 73.1τ(break-in) [ms]1.51 ± 0.031.51 ± 0.061.50 ± 0.041.60 ± 0.09Cp [pF]119.7 ± 2.55119.2 ± 3.84126.3 ± 3.14122.2 ± 5.71

Cellular properties of GCs recorded from Eu and Ts65Dn mice in the DG.

Values are shown as mean ± SEM. Including experiments done in all conditions and both ages of mice recorded from in the whole study.

Experimental design and statistical analysis

Statistical analysis was performed in Prism 9 (GraphPad). Unpaired two-tailed t-tests were used to detect significant effects between two groups. We used a mixed-effect analysis to test for differences in the input–output relation with repeated measures and two-way ANOVA for non-repeated measures (e.g., comparison between the two groups of ages) or for comparison between stimulations of different interneurons. As there were no genotype-specific differences in synaptic properties, data was pooled for comparison between targeted interneurons. For data pooled across genotypes, one-way ANOVA was used. Multiple comparisons were performed using a Šídák-Test, which minimizes the occurrence of type one errors by a weighting of the p-value. The significance level was set to p = 0.05 for all tests. All data is shown as mean ± SEM. The number (n) of observations reflects for electrophysiological experiments the number of cells recorded from and for immunohistochemical experiments the number of animals. For electrophysiological recordings, blinding of the experimenter to the genotype was not possible, because Ts65Dn mice have phenotypic characteristics, for example a smoother scalp, smaller brain and body sizes. Although a formal randomization of experiments was not performed, Eu control mice were littermates of Ts65Dn mice and were always used in an interleaved fashion. For the 3D-reconstruction, the analysis images were randomized and the experimenter was fully blinded.

ResultsIncreased proximal dendritic inhibition evoked by iML stimulation in Ts65Dn mice

Previous studies demonstrated enhanced GABAergic synaptic transmission in the hippocampus of Ts65Dn mice (Kleschevnikov et al., 2012; Schulz et al., 2019). Kleschevnikov and colleagues showed increased inhibitory synaptic signaling in the DG of Ts65Dn mice when they stimulated electrically in the molecular layer (ML) while recording from GCs. The identity of the interneurons mediating the increased inhibition in Ts65Dn mice has not been determined. In order to stimulate specifically local inhibitory inputs onto the proximal dendrites of GCs, we used electrical stimulation at low intensity in the inner ML (iML) in two cohorts of mice of different ages (Figure 1; Supplementary Figure S2). These experiments as well as subsequent experiments employing optogenetic stimulation were performed under the assumption that there were no major changes in axonal excitability of targeted interneurons in the DG of Ts65Dn mice.

Electrical stimulation in the iML evokes IPSCs with an increased amplitude in Ts65Dn mice. (A) Schematic illustration of the stimulation in the iML and recording site in the GCL (modified from Lodge et al., 2020). (B) Representative mean IPSCs recorded in GCs of Eu and Ts65Dn 6-week-old mice. The holding potential was −80 mV. Recordings were performed in the presence of 3 mM Kyn. (C) Group data for the input–output relation. IPSC amplitude is plotted over stimulating current strength. Mixed-effect analysis (repeated measure) indicated a significant effect of the genotype (p = 0.0176, F(1, 17) = 6.917, nEu = 10, nTs65Dn = 9). (D) Combined data of IPSCs from 6- and 8-week-old mice evoked with a stimulation intensity of 30 μA. Significant effect of the genotype reveals increased IPSC amplitude in Ts65Dn mice (p = 0.002, F(1, 36) = 11.46, two-way ANOVA, nEu = 19, nTs65Dn = 21). (E) Representative traces for IPSCs evoked by injecting 20 μA. Wash-in of 10 μM gabazine for Eu and Ts65Dn mice completely blocked the response. (F) The IPSC amplitude is plotted over time. Time of the wash-in of gabazine is indicated. After approximately 2 min the response was suppressed. (G) Evaluation of the reduction of IPSC amplitudes in presence of gabazine for Eu (n = 4) and Ts65Dn mice (n = 4). A two-way ANOVA demonstrated a significant effect between control and presence of gabazine (p < 0.0001, F(1, 12) = 751.30). The response was significantly reduced by up to 95% in the presence of gabazine (Eu = 95.2 ± 2.4%; Ts65Dn = 90.5 ± 6.3%; p < 0.0001, Sidak’s multiple-comparisons test). Ctrl, control; Eu, euploid; GC, granule cell; GCL, granule cell layer; H, hilus; iML, inner molecular layer; IPSC, inhibitory postsynaptic current; Kyn, kynurenic acid; mML, medial molecular layer; oML, outer molecular layer.

Inhibitory postsynaptic currents were evoked at increasing current intensities ranging from 10 μA to 30 μA. Amplitudes of evoked IPSCs increased with higher stimulation intensities in Eu and Ts65Dn mice (p < 0.0001, F(2, 34) = 57.01, mixed-effect analysis, nEu = 10, nTs65Dn = 9) (Figure 1C). There was a significant effect of the genotype indicating larger IPSCs in Ts65Dn mice (p = 0.0176, F(1, 17) = 6.917, mixed-effect analysis). To test for differences in fully mature mice, experiments were repeated in a cohort of 8-week-old mice. The results were qualitatively very similar (Supplementary Figure S2), with a strong trend toward an effect of the genotype by stimulation intensity interaction (p = 0.08, F(1, 18) = 3.39, mixed-effect analysis; Supplementary Figure S2). Combining the IPSCs evoked at 30 μA from both datasets supported a significant effect of the genotype (p = 0.002, F(1, 36) = 11.46, two-way ANOVA, nEu = 19, nTs65Dn = 21; Figure 1D). Experiments were performed in the presence of Kyn to block glutamatergic neurotransmission. Wash-in of the GABAA receptor blocker gabazine after the experiment caused a full suppression of the electrically evoked IPSCs, confirming that the recorded currents were GABAergic (Figures 1E–G). Together, these results support the hypothesis of an increased inhibition in the DG of Ts65Dn mice and point to proximal dendritic inhibitory inputs as an important source.

Age-dependent reduction of optogenetically activated SOM and PV interneuron inhibition in the dentate gyrus of Ts65Dn mice

A previous study in the PFC indicated that SOM interneuron-mediated inhibition to more distal compartments could also cause enhanced GABAergic transmission in Ts65Dn mice (de Zorilla San Martin et al., 2020). To directly test whether SOM interneuron-mediated dendritic inhibition is altered in the DG of Ts65Dn mice, we optogenetically stimulated these interneurons expressing ChR2. The field of illumination was directed to the outer ML for dendritic inhibition to selectively stimulate synaptic inputs of SOM interneurons innervating the distal dendrites of GCs (Figures 2A–D). Dendritic IPSCs of GCs were measured in the presence of Kyn.

Optogenetically evoked SOM inputs onto the distal dendrites of GCs are not enhanced in Ts65Dn mice. (A) Experimental design. A patch-clamp recording was obtained from a GC. Optogenetic stimulation in the oML was used to activate synaptic inputs from SOM interneurons indicated by a blue schematic with its soma in the hilus and an axon targeting the distal dendrites of GCs in the oML. GCs were recorded at a holding potential of −80 mV in presence of 3 mM Kyn. (B) Biocytin-filled GC. Neuronal processes of SOM interneurons are shown by the YFP signal in green. (C) Putative SOM interneuron synapses enclose the pictured distal dendrite of the GC in the oML. (D) The negligible YFP signal in the GCL indicates that the soma of GCs is not the main synaptic target of SOM interneurons. (E) Mean traces of GC recordings after optical activation of SOM interneuron synapses in Eu and Ts65Dn 6-week-old animals. (F) Averaged data representing the input–output relation. The amplitude of IPSCs increases with increasing laser intensity. Mixed-effect analysis (repeated measure) revealed an effect of the laser intensity and genotype interaction (p = 0.0012, F(3, 48) = 6.175, two-way ANOVA, nEu = 8, nTs65Dn = 10). (G) Data combined from recordings in two cohorts of 6- and 8-week-old mice in response to the highest laser stimulation intensity of 6.43 mW. SOM inputs are similar in both genotypes (two-way ANOVA, p = 0.2627, F(1, 37) = 1.293, nEu-6w = 8, nEu-8w = 11, nTs65Dn-6w = 10, nTs65Dn-6w = 12). Eu, euploid; GC, granule cell; GCL, granule cell layer; H, hilus; iML, inner molecular layer; IPSC, inhibitory postsynaptic current; Kyn, kynurenic acid; mML, medial molecular layer; oML, outer molecular layer; SML, stratum lacunosum moleculare; SOM, somatostatin.

Figures 2E,F represent the input–output relation of dendritic IPSCs from SOM interneurons in response to increasing laser intensity in 6-week-old mice. There was a significant main effect of the laser light intensity on the amplitude (p < 0.0001, F(3, 63) = 72.30, mixed-effect analysis, nEu = 11, nTs65Dn = 12). In addition, there was significant light intensity by genotype interaction (p = 0.0012, F(3, 48) = 6.175, mixed-effect analysis, nEu = 8, nTs65Dn = 10; Figure 2F). However, IPSC amplitudes recorded in 8-week-old were not different between Eu and Ts65Dn mice (laser intensity genotype interaction, p = 0.6, F(3, 63) = 0.61, mixed-effect analysis, nEu = 11, nTs65Dn = 12; Figure 2G; Supplementary Figure S3). Thus, while the IPSC amplitude from SOM inputs were reduced in 6-week-old animals, this phenomenon was no longer present in 8-week-old mice. This result may be explained by a delayed maturation of SOM inputs to distal dendrites of GCs in Ts65Dn mice. However, these results do not support the hypothesis that SOM interneuron inputs to the distal dendritic tree of GCs contribute to increased GABAergic transmission in the DG of Ts65Dn mice.

The perisomatic compartment receives prominent inhibition that is primarily mediated by PV interneurons (Kraushaar and Jonas, 2000; Hu et al., 2014). PV interneuron-mediated synaptic inputs onto GCs have not been directly investigated in the DG of Ts65Dn mice previously. In mice expressing ChR2 in PV interneurons, we targeted the field of illumination to the GCL with the recorded GC in the center (Figures 3A–D). The addition of 100 μM PTX fully blocked the evoked IPSCs confirming that they were GABAergic (Figures 3H–J).

Reduced input of PV interneurons onto GCs in the DG of Ts65Dn mice. (A) Experimental design. Optogenetic stimulation in the GCL was used to activate synaptic inputs from PV interneurons indicated by a red schematic with both its soma and axon in the GCL. PV interneurons were optically stimulated in the GCL and IPSCs were recorded from GCs at a holding potential of −80 mV. (B) A biocytin-filled GC is pictured in red and PV interneurons which express YFP, are shown in green. The green YFP signal mainly labels the dendrites of PV interneurons in the ML and PV axons and synapses in the GCL. (C) The distal dendrite of the GC in the oML is shown in detail. Fibers of PV interneurons do not enclose the GC dendrite. (D) In the GCL, the YFP signal labels the dense axon network of PV interneurons including putative synapses that are formed onto the soma of the GC. (E) Representative IPSCs evoked by optical stimulation of PV interneurons and recorded from GCs in 8-week-old Eu and Ts65Dn mice. (F) Input–output relationship is shown by plotting IPSC amplitude over the strength of stimulation. Mixed-effect analysis (repeated measure) demonstrated a pronounced effect of the genotype × laser intensity interaction (p = 0.0097, F(3, 51) = 4.221, nEu = 8, nTs65Dn = 11). (G) IPSCs measured in 6- and 8-week-old mice in response to the same laser intensity. Two-way ANOVA indicated a significant effect of the genotype (p = 0.0340, F(1, 48) = 4.763, nEu-6w = 11, nEu-8w = 17, nTs65Dn-6w = 13, nTs65Dn-8w = 11). (H) Representative IPSCs before and after application of 100 μM PTX, blocking GABAARs. IPSCs were evoked at an intensity of 0.35 mW. (I) PTX reduces the IPSC amplitude over time in Eu mice. After approximately 5 min the response was completely suppressed. (J) Presence of PTX reduced the response up to 94% (Ctrl = 100%; +PTX = 93.8 ± 2.1%, n = 6). Ctrl, control; DG, dentate gyrus; Eu, euploid; GC, granule cell; GCL, granule cell layer; H, hilus; iML, inner molecular layer; IPSC, inhibitory postsynaptic current; mML, medial molecular layer; oML, outer molecular layer, PTX, picrotoxin; PV, parvalbumin; SLM, stratum lacunosum moleculare.

Inputs were evoked with increasing stimulation intensity in slices from 8-week-old Eu and Ts65Dn littermates (Figures 3E,F). There was a significant effect of the laser light intensity on IPSC amplitude (p < 0.0001, F(3, 51) = 33.50, mixed-effect analysis, nEu = 8, nTs65Dn = 11), which, however, appeared rather weak in Ts65Dn mice. Interestingly, IPSCs evoked by an optogenetic stimulation of PV interneurons were reduced in Ts65Dn mice (Figure 3F). Thus, there was a significant interaction between laser intensity and genotype (p = 0.0097, F(3, 51) = 4.22, mixed-effect analysis, nEu = 8, nTs65Dn = 11). Similar results were obtained in a second independent set of data recorded in 6-week-old animals (Figure 3G; Supplementary Figure S4). Combining both datasets confirmed a significant effect of the genotype (p = 0.03, F(1, 48) = 4.7, two-way ANOVA, nEu-6w = 11, nEu-8w = 17, nTs65Dn-6w = 13, nTs65Dn-6w = 11). Hence, these results indicate that PV interneurons do not contribute to increased GABAergic transmission in the DG of Ts65Dn mice and that PV interneuron-mediated perisomatic inhibition may be reduced instead.

Properties of proximal dendritic GABAergic inputs are clearly distinct from SOM and PV interneuron-mediated inhibition

The results so far suggest that electrically evoked proximal dendritic inputs are enhanced in Ts65Dn mice, while optogenetically evoked inputs from SOM and PV interneurons were either unchanged or reduced. To investigate whether electrically evoked proximal dendritic inputs were indeed distinct from inputs from SOM and PV interneurons, we investigated the kinetics of the inputs and their short-term plasticity. These properties are characteristic for specific sets of synapses and depend on the presynaptic neuron of a synapse, postsynaptic receptor composition and location within the dendritic tree (Savanthrapadian et al., 2014; Schulz et al., 2018).

We used two-way ANOVAs to assess genotype-specific differences in synaptic properties. For none of the properties, we observed significant genotype specific effects (Figure 4; Supplementary Table S1). Therefore, we pooled data across genotypes for further comparisons of synaptic properties between different targeted presynaptic interneurons. First, we evaluated the rise time of the optically and electrically evoked IPSCs from 20 to 80% of the peak amplitude to baseline (Figures 4A–D). IPSCs elicited by the three different targeted stimulations showed significantly different rise times (p < 0.0001, F(2, 57) = 2.64, one-way ANOVA) (Figure 4D). Electrical stimulation in the iML evoked IPSCs with significantly slower rise time (Eu = 4.92 ± 0.17 ms, n = 8; Ts65Dn = 4.50 ± 0.41 ms, n = 11) than by PV interneurons (Eu = 1.88 ± 0.17 ms, n = 8; Ts65Dn = 2.21 ± 0.34 ms, n = 10; p < 0.0001, Sidak’s multiple-comparisons test). They also tended to be slower than IPSCs mediated by SOM interneurons (Eu = 3.95 ± 0.23 ms, n = 11; Ts65Dn = 4.04 ± 0.07 ms, n = 12; p = 0.03, Sidak’s multiple-comparisons test).

Pronounced difference in the kinetics and short-term plasticity of GABAergic inputs recruited by optogenetic and electrical stimulation. (A) Representative IPSCs in response to a double-pulse stimulation evoked by electrical stimulation in the iML, optogenetic stimulation of SOM interneurons in the oML (B) and PV interneurons in the GCL (C). (D) Group means of the rise time of the first IPSCs for the two genotypes and the three stimulations. Two-way ANOVA reveals a significant main effect of the targeted interneuron population (p < 0.0001, F(2, 54) = 48.12, iML: nEu = 8, nTs65Dn = 11, SOM: nEu = 11, nTs65Dn