Persistent input- and cell-type-specific synaptic alterations in the somatosensory thalamus of Dravet syndrome mice

Abstract

Dravet syndrome is an epileptic encephalopathy most often caused by loss-of-function mutations in the SCN1A gene, leading to haploinsufficiency of the voltage-gated sodium channel Na_V1.1. Seizures begin during infancy and wane throughout childhood, but behavioral symptoms, including intellectual disability, motor impairments, and autistic features, remain through adulthood. Seizures primarily stem from inhibitory neuron hypo-excitability in the cortex, hippocampus, and thalamus, but circuit abnormalities underlying persistent behavioral symptoms are poorly understood. Prior work showed synapse dysfunction in thalamocortical neurons in four-week-old DS mice, but elucidating the timing and progression of these alterations is necessary to understand the disease stages that synapse dysfunction may contribute to circuit and behavioral phenotypes. We investigated synapse function in the ventral posterolateral (VPL) and ventral posteromedial (VPM) thalamus prior to seizure onset (P13-P17), after the period of highest seizure burden (P28-P32), and in adulthood (P58-P63). VPL and VPM synaptic activity showed excitatory input to the VPL was reduced after seizure onset and this reduction persisted through adulthood, while VPM excitatory input was unaffected. We further showed a selective reduction in the function and number of excitatory sensory synapses in the VPL, with no change to cortical synapses. VPL and VPM neurons both showed inhibitory synapse dysfunction at four weeks, which persisted in adult DS mice only in VPL neurons. These results revealed persistent input- and cell-type-specific alterations to thalamic synapses that develop after seizure onset and are maintained into adulthood, suggesting synaptic deficits could contribute to ongoing somatosensory thalamocortical circuit dysfunction and behavioral deficits in DS.

NEW & NOTEWORTHY

This study reveals that synaptic dysfunction in the somatosensory thalamus of DS mice emerges after seizure onset and persists into adulthood. Using electrophysiology and high-resolution imaging, we demonstrate a selective, persistent reduction of sensory input to VPL neurons, while cortical input remains intact. Inhibitory input to VPL thalamus is similarly impaired after seizure onset and persists in adulthood. These input- and cell-type-specific deficits may contribute to the enduring sensory and sleep impairments characteristic of DS.

Graphical Abstract

INTRODUCTION

Dravet syndrome (DS) is a severe form of myoclonic epilepsy that is most often caused by mutations in the SCN1A gene, which encodes the voltage-gated sodium channel Na_V1.1{Citation}(Claes et al., 2001; Escayg et al., 2000). DS is characterized by frequent seizures and a high incidence of sudden unexplained death in epilepsy (SUDEP) (1–3). Na_V1.1 facilitates action potential generation and propagation via expression at the axon initial segment and nodes of Ranvier (4, 5). The Scn1a gene is highly expressed in GABAergic neurons in the cortex, hippocampus, and reticular thalamus (nRT) all of which display significant reduction in excitability in DS mouse models (6–9). The resulting disinhibition leads to circuit hyperexcitability beginning in the third postnatal week contributing to spontaneous convulsive seizures, heat-induced seizures, and spike-and-wave discharges (6, 9, 10). This phenotype recapitulates the human disease involving myoclonic jerks, tonic-clonic seizures, febrile seizures, and absence seizures with a median onset at six months of age (2, 7, 11, 12).

Seizures lessen throughout adolescence into adulthood in patients with DS, but debilitating behavioral afflictions continue including motor impairments, intellectual disability, sleep impairment, and autistic behaviors (Lagae et al., 2018). While much has been learned about aberrant spike generation underlying seizures in DS, far less is known about the enduring mechanisms contributing to behavioral impairments throughout the lifespan. Given that the highest seizure burden occurs during postnatal development, it is conceivable that activity-dependent synapse formation is disrupted in DS, which could result in miswiring of brain circuits and perpetual brain dysfunction. Notably, cortical parvalbumin interneuron spike generation normalizes by adulthood in DS mouse models, but parvalbumin GABAergic synaptic transmission remains disrupted (13, 14). Impaired action potential propagation in axons contributes to this sustained disruption in synapse function, but other studies have suggested that altered synaptic transmission or development may contribute to disrupted inhibition in DS (15, 16). Recent work in a human induced pluripotent cell line from a patient with DS demonstrated that network dysfunction involves reduced excitatory synapse strength, and computational modeling further supports the likelihood of homeostatic downscaling of excitatory synapses in DS models (17). On the other hand, enhanced excitatory input to dentate gyrus granule cells contributes to persistent excessive excitation in hippocampal circuits (18). While several reports indicate that altered synaptic transmission may contribute to circuit dysfunction in DS, the direction, magnitude, and duration of these changes as well as the specific synaptic connections affected remain largely unresolved.

In the thalamus, Na_V1.1 is expressed in glutamatergic thalamocortical (TC) neurons in the ventral posterolateral (VPL) and ventral posteromedial (VPM) thalamus as well as the GABAergic nRT beginning in the second postnatal week and continuing through adulthood (15, 19). In addition to the profound alterations in inhibitory nRT neuron excitability (5, 15, 20), TC neurons exhibit cell-type- and synapse-specific alterations that may further disrupt thalamocortical network function in DS (15). VPL and VPM neurons receive glutamatergic input from corticothalamic (CT) neurons, glutamatergic sensory input from the spinal cord and brainstem, respectively, and GABAergic input from the nRT (Figure 1A). In a DS mouse model, VPL neurons showed reduced excitability as well as reduced glutamatergic sensory input, while VPM neurons exhibited increased excitability and no apparent changes in synaptic input (15). Notably, CT input to the nRT was reduced, while CT input to VPL and VPM was unchanged. This cell-type- and synapse-specific pathophysiology within the somatosensory CT circuitry indicates that hyperexcitable circuits do not undergo a global homeostatic scaling of synaptic transmission in DS. However, the underlying mechanisms and timeline of cell-type- and input-specific dysfunction remain unknown. Elucidating this timeline is important because somatosensory thalamocortical network dysfunction persists into adulthood in a DS mouse model (20). Moreover, patients and mouse models display enduring sensorimotor and sleep deficits that could be underpinned by this thalamocortical circuitry (21–27).

Figure 1. Circuit diagram and experimental framework for VPL and VPM recordings.

A. Circuit diagram shows excitatory and inhibitory synaptic inputs to VPL and VPM neurons (ST: spinothalamic tract; TT: trigeminothalamic tract).

B. Images of VPL and VPM recordings in acute thalamic slices with a representative location of recording electrode in VPL.

C. Representative locations for VPL and VPM neuron recordings determined from images of recording electrode placement (as depicted in panel B) or from posthoc labeling of biocytin-filled neurons.

D. Morphological reconstructions for streptavidin labeling of representative biocytin-filled VPL and VPM neurons.

E. A survival timeline for WT and DS mice illustrates the timing of three age groups in this study: 2W, 4W, and 8W relative to the period of highest seizure burden (P17-P24).

To better understand how synaptic transmission may be disrupted within the somatosensory thalamus, we investigated if alterations to synapse function or number occur before seizure onset, potentially contributing to circuit dysfunction underlying seizures, or after the period of highest seizure burden. We further examined if thalamic synapse dysfunction was maintained into adulthood. In this study, we utilized electrophysiology and high-resolution imaging to selectively quantify glutamatergic CT and sensory inputs as well as GABAergic inputs to VPL and VPM neurons. This work uncovered that input- and cell-type-specific synapse dysfunction in VPL neurons developed after seizure onset and some, but not all, of the altered synaptic connections showed enduring changes in adult DS mice.

MATERIALS AND METHODS

Mouse model

Mouse studies were performed according to protocols approved by the Institutional Animal Care and Use Committee at Virginia Polytechnic Institute and State University and in accordance with the National Institutes of Health guidelines. Scn1a^tm1Kea mice (037107-JAX), which are on a 129S6/SvEvTac background, were purchased from the Mutant Mouse Resource and Research Center (28). The Scn1a^tm1Kea colony was maintained by crossing heterozygous 129S6/SvEvTac-Scn1a^tm1Kea/WT mice with wild type 129S6/SvEvTac mice. Experimental mice were F1 hybrids generated by crossing heterozygous 129S6/SvEvTac-Scn1a^tm1Kea/WT mice with C57BL/6J mice (000664, JAX). Genotyping was performed by Transnetyx using real-time PCR. Mice were housed in a 12 hour light/dark cycle with ad libitum access to food and water. Scn1a^WT/WT and Scn1a^WT/- (hereafter referred to as WT and DS) littermates of both sexes were used at the age of P13-P17 (25 mice total), P28-P32 (24 mice total), and P58-P62 (21 mice total). The number of mice of each sex are reported in the figure legends for P28-P32 and P58-P62 age groups; sex was undetermined for P13-P17 mice.

Electrophysiology

For recording synaptic activity, mice were euthanized by isoflurane overdose and then perfused with ice-cold sucrose-based artificial cerebrospinal fluid (aCSF) containing (in mM) 230 sucrose, 24 NaHCO3, 10 glucose, 3 KCl, 10 MgSO4, 1.25 NaH2PO4, and 0.5 CaCl2 saturated with 95% O2 / 5% CO2(pH 7.4, 320 mOsm). The brain was removed and sliced (300 µm) with the vibratome (Leica VT1200S). Slices were cut horizontally in the oxygenated cold sucrose aCSF. Slices were incubated in the NaCl-based aCSF containing (in mM) 130 NaCl, 24 NaHCO3, 10 glucose, 3 KCl, 4 MgSO4, 1.25 NaH2PO4, and 0.5 CaCl2 saturated with 95% O2 / 5% CO2 (pH 7.4, 305 mOsm) at room temperature. One recording was performed per slice and no more than two cells per mouse were included in a single dataset. Recordings were made using a Multiclamp 700B amplifier (Molecular Devices), sampled at 20 kHz (Digidata 1550B, Molecular Devices), and low pass filtered at 10 kHz using Axon pClamp 11 software (Molecular Devices). The extracellular recording solution contained (in mM) 130 NaCl, 24 NaHCO3, 10 glucose, 3 KCl, 1 MgSO4, 1.25 NaH2PO4, and 2 CaCl2 saturated with 95% O2 / 5% CO2 (pH 7.4, 305 mOsm) and was maintained at 32 ◦C for all recordings.

Whole-cell voltage-clamp recordings were made using borosilicate glass recording electrodes (4–5 MΩ) filled with (in mM) 120 CsMeSO3, 15 CsCl, 8 NaCl, 10 tetraethylammonium chloride, 10 HEPES, 1 EGTA, 3 Mg-ATP, 0.3 Na-GTP, 1.5 MgCl2, 1 QX 314, and 0.1% biocytin (pH 7.3, 295 mOsm). After a 10-minute equilibration period, sEPSCs and sIPSCs were recorded in three two-minute epochs at a holding potential of −70 mV and 0 mV, respectively. Recordings of mEPSCs and mIPSCs were performed similarly with 1 μM TTX added to the extracellular recording solution. For NASPM recordings, VPL neurons were held at –80 mV, and mEPSCs were continuously recorded for 6 min with 1 μM TTX in the recording solution (baseline) followed by 10 min while 1 μM TTX and 100 μM NASPM were applied to the slice. These mEPSC recordings were performed at –80 mV to increase driving force and ensure limited NMDAR activation as at high concentration NASPM can block NMDAR channels (29, 30). Holding commands were adjusted for a 10 mV liquid junction potential during the recordings. Series resistance and cell capacitance were monitored throughout the experiment but were not compensated, and cells were excluded if either parameter changed > 20% within the first two recording epochs. Cell location was confirmed by posthoc biocytin labeling and/or acquiring an image of the recording electrode immediately after the recording and/or posthoc labeling of biocytin-filled cells (Figure 1B-D).

Electrophysiology data analysis

For sE(I)PSCs and mE(I)PSCs analysis, recordings of 6 minutes were analyzed by MiniAnalysis software (Synaptosoft) to determine the frequency and amplitude of the events. Four minutes of recording data with a stable E(I)PSC frequency were used for each cell. Only four minutes were used because some cells were not stable in either the first (12/440 cells) or last (23/440 cells) recording epoch. Data were digitally filtered at 1 kHz. Automated detection identified events with amplitude ≥7 pA (~5 × RMS noise), and then events ≥5 pA were manually detected, and automated detection accuracy was assessed. The amplitude and frequency of the recordings averaged in 30-second bins in each recording. The decay time for all mEPSC events were measured by fitting 30 – 70% decay in MiniAnalysis and plotted as a frequency distribution for each cell independently. All distributions were multimodal (p < 0.05). Each distribution was interpolated using the Akima spline method (GraphPad Prism 9), and the first derivative of the spline was plotted to determine the x-intercept between the two modes, which is the decay time corresponding to the critical value used to separate the bimodal distribution two groups of mEPSC events. The events were separated into Type 1 (shorter, putative sensory events) and Type 2 (longer, putative cortical events) mEPSCs.

Immunohistochemistry and imaging

For counting the number of glutamatergic synapses, mice were euthanized by isoflurane overdose, transcardially perfused with 4% paraformaldehyde (PFA, pH 7.4), and post-fixed by 4% PFA for 24 hours. Brains were then transferred into 30% sucrose in 1× phosphate-buffered saline (PBS, pH 7.4) for 48 h and frozen in OCT. Brains were cryosectioned (20 μm) and slide-mounted, and immunostained as previously described (31). Briefly, brain slices were treated with 0.8% sodium borohydride in 1× Tris-buffered saline (TBS, pH 7.5) at room temperature, and then 0.01 M sodium citrate buffer at 100°C, pH 6.0 for 10 min. Permeabilized mounted brains with 0.5% Triton X-100 in 1×TBS, and blocked with FAB fragments and 10% normal donkey serum (NDS). Sections were incubated with antibodies against the following proteins overnight at 4°C: VGLUT1 (1:200, Synaptic Systems 135–511, RRID AB_887879), VGLUT2 (1:400, Synaptic Systems 135–402, AB_2187539), Homer (1:200; Synaptic Systems 160–003, RRID AB_887730; 1:200, Synaptic Systems 160–025, RRID AB_2744655), and Bassoon (1:800, Abcam ab82956, AB_1860018) by using antibodies. The following secondary antibodies were applied for 1 hr at RT: Alexa 488-conjugated anti-guinea pig (AB_2534117), Alexa 555-conjugated anti-mouse (AB_2535769), and Alexa-647-conjugated anti-rabbit (AB_2535813) to detect homer, bassoon, and VGLUT2, respectively; Alexa 488-conjugated anti-rabbit (AB_2762833), Alexa 555-conjugated anti-mouse IgG2a (AB_2535776), and Alexa-647-conjugated anti-mouse IgG1 (AB_2535809) to detect homer, bassoon, and VGLUT1, respectively.

Images were acquired using a Nikon CSU-W1 super resolution by optical reassignment (SoRa) spinning disk confocal microscope with a 10x objective for thalamus overview images and a 60x objective with 4x SoRa disk (27 nm pixel size) for near super-resolution synaptic images. Fluorescence signals were excited using 488, 561, and 640 nm lasers and detected using GFP, RFP, and Cy5 filters and a Hamamatsu back-thinned Orca Fusion CMOS camera. Z-stacks were acquired with Nikon Elements software with a 0.15 μm interval. Three 60x images from different fields of view (FOV) within the VPL and VPM thalamus were taken for each mouse. Images of WT and DS tissue were taken in each imaging session using the same acquisition parameters. Images were processed with Nikon Elements Denoising software and Blind Deconvolution algorithm. All further image processing and analysis was performed using Fiji on a single z-plane from each image series.

Image analysis

The SynapseJ plugin was used to detect glutamatergic synapses defined as overlapping homer and bassoon puncta as well as VGLUT1- and VGLUT2-positive bassoon puncta to label CT and sensory inputs, respectively (31, 32). The minimum and maximum threshold intensities to detect each synaptic marker as separate particles were set empirically for each imaging session. The minimum and maximum puncta size parameters, set at 0.02 and 5 for both bassoon and homer, were kept constant across all images. VGLUT-positive synapses were defined as bassoon puncta overlapping VGLUT1 or VGLUT2 signal with at least one pixel at the maximum intensity threshold. VGLUT2 is highly expressed in sensory axon terminals, but also visible in VPL and VPM neurons, so to be sure we specifically detected bassoon puncta overlapping sensory terminals we performed a median blur on VGLUT2, which ensured thresholding eliminated weak VGLUT2 signal in VPL and VPM neurons and maintained strong VGLUT2 signal in axon terminals. The data from three FOV were averaged for each mouse. Experimenters were blind to genotype during imaging and analysis.

Statistical analysis

A priori power analyses were performed in GPower 3.1 to estimate required sample sizes given appropriate statistical tests with α = 0.05, power (1 – β) = 0.8, and a moderate effect size or effect size based on pilot data. GraphPad Prism 9.4.1 was used to perform all statistical analysis. Normality was tested by the D’Agostino-Pearson test and equal variances was tested by the Spearman’s tests with the results reported in Supplemental Table S1. Normal datasets with equal variances were analyzed by two-way ANOVA with Sidak’s corrections for multiple comparisons; these data were reported in the text and plotted as the group mean with standard deviation (SD). All frequency data for synaptic currents had log-normal distributions and most exhibited unequal variances; therefore, these data were log-transformed for statistical analysis. If any frequency data point was ≤ 1.0 Hz, then a constant value = 1.0 was added to each data point so that log transformed values were greater than zero. All transformed datasets passed normality and equal variances tests and were analyzed by two-way ANOVA with pairwise Sidak’s tests to correct for multiple comparisons. For transformed datasets, the geometric mean of the raw (untransformed) data was reported in Table 1 and Table 2 with the 95% confidence interval [95 CI], while both the raw data with geometric mean [95 CI] and log-transformed data with mean [95 CI] were plotted in the figures. The mean log-transformed data [95 CI] group summary data are reported in Supplemental Table S2. ANOVAs and pairwise comparisons were only performed on the transformed data and are reported with the plots of transformed data. The ANOVA test statistics, p values, and degrees of freedom were reported in Supplemental Tables S3 and S4, and pairwise Sidak’s test p values were reported in the figures and/or figure legends. Cumulative distributions were generated by automatic binning for each cell and then averaged to obtain a mean cumulative distribution with SD for each group, and WT and DS group data were compared by Kolmogorov tests.

Table 1.

sEPSC and mEPSC frequency and amplitude data.

	Frequency						Amplitude
	Raw data: Geometric mean [95 CI]						Raw data: Mean (SD)
	WT			DS			WT			DS
	GM	UL	LL	GM	UL	LL	M	SD	n	M	SD	n
VPL sEPSC
2W	3.05	4.05	2.30	3.22	4.99	2.07	16.89	2.60	10	16.20	0.91	6
4W	3.59	4.85	2.66	2.11	2.87	1.55	17.70	4.30	11	15.03	3.49	9
8W	3.82	5.35	2.73	2.50	3.87	1.62	15.91	2.62	11	17.35	2.97	8
VPM sEPSC
2W	1.55	1.93	1.24	1.57	2.21	1.11	14.21	1.45	9	14.03	2.32	8
4W	2.30	3.40	1.56	2.23	2.87	1.73	17.35	4.15	9	16.03	2.86	11
8W	2.13	2.97	1.52	2.28	3.59	1.45	16.72	2.66	7	17.30	2.55	7
VPL mEPSC
2W	3.34	4.84	2.30	3.27	4.26	2.51	12.14	1.80	9	13.12	1.87	6
4W	3.52	4.42	2.80	1.84	2.30	1.46	17.70	1.55	8	14.28	1.93	9
8W	3.23	4.91	2.13	2.00	2.55	1.57	17.12	2.83	8	18.58	2.86	10
VPM mEPSC
2W	1.60	2.23	1.15	1.90	2.80	1.29	12.50	2.25	11	14.07	1.80	8
4W	1.83	2.65	1.27	1.82	2.43	1.36	13.93	2.60	7	13.10	1.98	8
8W	2.12	3.07	1.47	2.43	3.53	1.67	13.90	2.87	7	12.88	2.00	9
VPL Type 1 mEPSC
2W	1.16	1.59	0.84	1.02	1.28	0.82	12.36	1.33	6	11.71	1.37	9
4W	1.94	2.57	1.47	0.76	0.93	0.62	16.07	2.53	8	13.45	1.52	8
8W	1.61	2.87	0.91	1.06	1.44	0.78	14.77	2.13	7	16.40	2.13	8
VPL Type 2 mEPSC
2W	1.81	2.69	1.21	1.80	2.44	1.33	14.15	2.47	6	12.85	1.56	9
4W	1.68	2.19	1.29	1.19	1.66	0.85	17.11	2.40	8	16.10	2.12	7
8W	1.52	2.45	0.95	1.41	1.72	1.16	16.89	2.90	9	18.70	2.49	8
VPM Gp 1 mEPSC
2W	0.63	0.91	0.44	0.69	1.11	0.42	12.20	2.34	11	12.68	1.65	7
4W	0.54	0.73	0.40	0.72	0.98	0.53	13.20	1.15	7	14.93	2.54	8
8W	0.72	1.05	0.49	0.70	0.97	0.50	14.77	2.85	7	14.56	1.71	7
VPM Gp 2 mEPSC
2W	0.43	0.78	0.24	0.67	1.38	0.32	13.87	2.26	11	15.40	1.78	8
4W	1.10	1.43	0.85	1.15	1.69	0.78	16.81	2.87	5	17.64	2.69	8
8W	0.70	1.52	0.32	1.12	1.73	0.73	19.24	3.98	6	17.41	1.18	7

Table 2.

Number of total, VGLUT1-positive, and VGLUT2-positive glutamatergic synapses.

	Mean (SD)
	WT			DS
	M	SD	n	M	SD	n
VPL glutamatergic synapse #
2W	943	189	8	908	133	8
4W	1706	216	8	1498	277	8
8W	1637	275	8	1568	280	8
VPM glutamatergic synapse #
2W	995	172	8	923	154	8
4W	1535	199	8	1417	210	8
8W	1396	258	8	1267	201	8
VPL VGLUT1 synapse #
2W	907	217	7	1042	197	7
4W	1772	311	8	1631	312	8
8W	1637	170	8	1508	281	8
VPM VGLUT1 synapse #
2W	785	176	7	1069	145	7
4W	1346	286	7	1327	283	8
8W	1335	187	8	1479	119	8
VPL VGLUT2 synapse #
2W	148	30	7	135	30	7
4W	187	34	8	123	25	8
8W	191	46	8	117	35	8
VPM VGLUT2 synapse #
2W	97	30	7	113	27	7
4W	108	16	8	101	17	8
8W	107	29	7	112	27	8

RESULTS

DS mice have reduced glutamatergic input to VPL neurons after seizure onset and in adulthood

To investigate synapse dysfunction across DS disease stages in VPL and VPM neurons, we recorded spontaneous synaptic transmission in acute brain slices from mice aged approximately 2 weeks (2W, P13-P17, before highest seizure burden), 4 weeks (4W, P28-P32, after seizure onset), and 8 weeks (8W, P58-P62, adult; Figure 1). We first evaluated glutamatergic synapse function by measuring the amplitude and frequency of sEPSCs (Figure 2; Table 1; Supplemental Table S1-S3). The mean frequency of VPL sEPSCs in DS mice was similar to WT in 2W and 8W mice , but was significantly reduced compared to WT in 4W mice (Figure 2A-C). Whereas, mean sEPSC amplitude was similar between WT and DS mice for VPL neurons at all ages (Figure 2D). In VPM neurons, mean sEPSC frequency and amplitude were similar between DS and WT mice in all age groups ages (Figure 2E-H). Comparisons between cumulative distributions for sEPSC frequency and amplitude yielded similar results (Supplemental Figure S1). Comparisons between cumulative distributions for sEPSC frequency and amplitude yielded similar results (Supplemental Figure S1). These data suggest that reduced spontaneous glutamatergic input to VPL neurons in DS mice appears after seizure onset and normalizes by adulthood, while spontaneous glutamatergic input to VPM neurons is unaffected in DS mice.

Figure 2. VPL neuron sEPSC frequency is reduced in DS mice after seizure onset.

A. Representative voltage-clamp recordings from VPL neurons show sEPSCs in WT and DS mice aged 2W (WT: n = 9 cells, 7 mice; DS: n = 6 cells, 5 mice), 4W (WT: n = 10 cells, 6 mice, 3 female (F)/3 male (M); DS: n = 9 cells, 9 mice, 3F/6M), and 8W (WT: n = 11 cells, 9 mice, 4F/5M; DS: n = 8 cells, 6 mice, 1F/5M).

B,C. The mean sEPSC frequency data were plotted as E) raw values with the geometric mean ± 95CI, which had a non-normal distribution, and F) log-transformed values with group mean ± 95CI, which were used for statistical analyses.

D. Mean sEPSC amplitudes were plotted for each mouse with the group mean ± SD.

E. Representative VPM recordings show sEPSCs in WT and DS mice aged 2W (WT: n = 9 cells, 8 mice; DS: n = 8 cells, 8 mice), 4W (WT: n = 9 cells, 7 mice, 2F/5M; DS: n = 11 cells, 9 mice, 3F/6M), 8W (WT: n = 7 cells, 7 mice, 5F/2M; DS: n = 7 cells, 6 mice, 3F/3M).

F-H. The mean sEPSC frequency and amplitude data were plotted as described for VPL. The data were analyzed by two-way ANOVA with pairwise Sidak’s comparisons (p-values shown in plots) between WT and DS at each age.

The observed differences in sEPSC frequency could have been due to varying activity levels in presynaptic axons, probability of release, and/or synapse number. To further understand how synaptic transmission was affected, we measured mEPSC frequency and amplitude in the absence of action potential driven synaptic release (Figure 3; Table 1; Supplemental Table S1-S3). In VPL neurons, we found no significant differences in mean mEPSC frequency in 2W mice. Whereas, the mean VPL mEPSC frequency was reduced in DS mice at 4W and 8W compared to WT mice (Figure 3A-C). The mean VPL mEPSC amplitude was reduced in DS mice relative to WT at 4W, but not 2W or 8W mice (Figure 3D). Consistent with no change in VPM sEPSCs, we also found no change in mEPSC frequency or amplitude in VPM neurons within any age group (Figure 3E-H). Comparisons between cumulative distributions for mEPSC frequency and amplitude yielded similar results (Supplemental Figure S2). Altogether, these data suggest that specifically VPL neurons, but not VPM neurons, have a reduced number of glutamatergic inputs or reduced probability of glutamate release, which appears after seizure onset in DS mice and is maintained into adulthood. In addition, the strength of glutamatergic input to VPL neurons is reduced immediately after the period of highest seizure burden, but normalizes by adulthood.

Figure 3. DS mice show reduced mEPSC frequency in VPL neurons after seizure onset.

A. Representative voltage-clamp recordings from VPL neurons show mEPSCs in WT and DS mice aged 2W (WT: n = 9 cells, 6 mice; DS: n = 10 cells, 7 mice), 4W (WT: n = 8 cells, 7 mice, 3F/4M; DS: n = 7 cells, 6 mice, 1F/5M), 8W (WT: n = 8 cells, 7 mice, 3F/4M; DS: n = 10 cells, 6 mice, 1F/5M).

B,C. The mean mEPSC frequency was plotted as B) raw data (geometric mean ± 95CI) and C) log-transformed values (mean ± 95CI) as the data had a non-normal distribution.

D. Mean mEPSC amplitude was plotted for each mouse with the group mean ± SD.

E. Representative VPM recordings show mEPSCs in WT and DS mice aged 2W (WT: n = 11 cells, 9 mice; DS: n = 7 cells, 7 mice), 4W (WT: n = 8 cells, 8 mice, 3F/5M; DS: n = 8 cells, 7 mice, 2F/5M), 8W (WT: n = 6 cells, 6 mice, 3F/3M; DS: n = 7 cells, 5 mice, 2F/3M).

F-H. The mean mEPSC frequency and amplitude data were plotted as described for VPL. VPL and VPM data were analyzed by two-way ANOVA with pairwise Sidak’s comparisons (p-values shown on plots) between WT and DS at each age.

Sensory synapse number is selectively reduced in VPL neurons in DS mice

VPL and VPM neurons receive descending glutamatergic input from CT neurons and ascending glutamatergic input carrying sensory information from the spinal cord and brainstem, respectively. To determine whether one or both of these inputs were affected in DS mice, we quantified the number of CT and sensory synapses in the VPL and VPM utilizing immunohistochemistry and high-resolution imaging. Bassoon immunostaining was used to identify presynaptic release sites, and CT inputs were labeled by immunostaining for VGLUT1 (Figure 4A-C; Supplemental Figures S3 and S4), while sensory terminals were labeled with VGLUT2 immunostaining (Figure 5A-C; Supplemental Figures S5 and S6). We compared the number of VGLUT1- and VGLUT2-positive bassoon puncta to approximate CT and sensory synapse density in WT and DS mice. The number of VGLUT1-positive bassoon puncta in the VPL was not significantly different between WT and DS mice at 2W, 4W, or 8W (Figure 4D,E; Table 2; Supplemental Table S4). We also detected similar numbers of VGLUT1-positive bassoon in VPM neurons across WT and DS mice at 2W, 4W, or 8W (Figure 4D,F). The numbers of VGLUT2-positive bassoon puncta in the VPL neurons were similar between WT and DS mice aged 2W, but were reduced in 4W and 8W DS mice compared to WT (Figure 5D,E; Table 2; Supplemental Table S4). No significant differences in the number of VGLUT2-positive bassoon puncta were detected in the VPM between WT and DS mice across all three age groups (Figure 5D,F).

Figure 4. VGLUT1-positive CT synapse numbers are unaffected in VPL and VPM thalamus of DS mice.

A. 10x image of the thalamus including the VPL and VPM immunostained for VGLUT1.

B. 60x images maximum intensity projection of VGLUT1 and bassoon staining in the VPM thalamus.

C. Boxed region from panel B shows individual bassoon puncta (arrowheads) overlapping VGLUT1 signal in a single z-plane. Scale: A) 200 µm, B) 5 µm, and C) 2 µm.

D. Representative ROIs from 60x VPL and VPM images show VGLUT1-positive bassoon puncta in WT and DS mice aged 2W, 4W, and 8W.

E,F. The number of VGLUT1-positive bassoon puncta were averaged across three images from each mouse in the E) VPL [2W (WT: n = 7 mice; DS: n = 7 mice), 4W (WT: n = 8 mice, 3F/5M; DS: n = 8 mice, 5F/3M), and 8W (WT: n = 8 mice, 3F/5M; DS: n = 8 mice, 4F/4M)] and F) VPM [2W (WT: n = 7; DS: n = 7), 4W (WT: n = 7 mice, 2F/5M; DS: n = 8 mice, 5F/3M), and 8W (WT: n = 8 mice, 3F/5M; DS: n = 8 mice, 4F/4M)]. The group mean and SD were plotted with data points from all mice. Data were analyzed by one-way ANOVA with posthoc Sidak’s tests between genotypes at each age (p-values shown in plots).

Figure 5. VGLUT2-positive synapse number in the VPL thalamus is reduced in DS mice.

A. 10x image of the thalamus including the VPL and VPM immunostained for VGLUT2.

B. 60x images maximum intensity projection of VGLUT2 and bassoon staining in the VPM thalamus. Arrows highlight a single sensory axon with a large VGLUT2-positive terminal. Arrowheads highlight VGLUT2-postive presynaptic terminals on the soma and proximal dendrites of a VPM thalamus neuron.

C. Boxed region from panel B shows individual bassoon puncta (arrowheads) overlapping VGLUT2 signal in a sensory axon terminal in a single z-plane. Scale: A) 200 µm, B) 10 µm, and C) 1 µm.

D. Representative regions of interest (ROIs) from 60x VPL and VPM images show VGLUT2-positive bassoon puncta in WT and DS mice aged 2W, 4W, and 8W.

E,F. The number of VGLUT2-positive bassoon puncta were averaged across three images from each mouse in the E) VPL [2W (WT: n = 7 mice; DS: n = 7 mice), 4W (WT: n = 8 mice, 4F/4M; DS: n = 8 mice, 6F/2M), and 8W (WT: n = 8 mice, 3F/5M; DS: n = 8 mice, 3F/5M)] and F) VPM [2W (WT: n = 7 mice; DS: n = 7 mice), 4W (WT: n = 8 mice, 4F/4M; DS: n = 8 mice, 6F/2M), and 8W (WT: n = 7 mice, 3F/4M; DS: n = 8, 3F/5M)]. The group mean and SD were plotted with data points from all mice. Data were analyzed by one-way ANOVA with posthoc Sidak’s tests between genotypes at each age (p-values shown in plots).

We also quantified total glutamatergic synapses in VPL and VPM, with single synapses identified as overlapping bassoon and homer (postsynaptic) puncta (Supplemental Figures S7 and S8). The total number of synapses was not significantly different across WT and DS mice in any age group as measured by quantifying all overlapping pairs of homer and bassoon in VPL neurons or VPM neurons (Table 2; Supplemental Table S4). These data are consistent with our CT and sensory synapse quantification, because sensory synapses make up a small portion (<10%) of total glutamatergic synapses in somatosensory thalamus; thus, any change in this population would not necessarily be detectable when assessing the whole population of glutamatergic synapses. Although few in number, sensory inputs are large, proximal synapses with multiple release sites with high release probability that contribute a substantial portion of the total excitatory input to VPL and VPM neurons. Altogether, these data suggest that show a cell-type- and input-specific reduction of sensory synapses onto VPL neurons.

DS mice have reduced frequency of putative sensory mEPSCs

As a second means of assessing whether one or both inputs were affected in DS mice, we took advantage of the well-established differences in decay time between sensory and CT EPSCs. Several studies have shown that evoked sensory EPSCs have significantly faster decay kinetics than evoked cortical EPSCs (33–36). Multiple factors contribute to these differences including input-specific glutamate receptor expression (31, 37, 38) and somatodendritic distribution of the inputs (39–41). For the data evaluated here, namely mEPSCs recorded at –70 mV or –80 mV and thus primarily mediated by AMPARs, the difference in kinetics is primarily driven by input location. Sensory synapses are located exclusively on the soma and proximal dendrites and thus maintain fast AMPAR-mediated EPSCs; while CT EPSCs are slowed by dendritic filtering as CT inputs are distributed across distal dendrites. We and others have taken advantage of input-specific EPSC properties to separate mEPSC or sEPSC events into two populations that can then be analyzed individually to determine how the populations differ (42–48). We showed previously (Studtmann et al., 2022) and herein (Supplemental Figure S9) that plotting the frequency histogram of decay times for all mEPSC events recorded from each VPL or VPM neuron yields a bimodal distribution that can be separated at a mathematically determined critical value. This allowed us to then analyze the amplitude and frequency of Type 1 mEPSCs, the population with faster decay kinetics (putative sensory), and Type 2 mEPSCs, the population with slower decay kinetics (putative CT), to determine if differences observed in the total mEPSC population resulted from one or both populations.

To test whether these two populations indeed represent putative sensory and CT mEPSCs, we used the knowledge that calcium-permeable AMPARs are only expressed at CT synapses onto VPL and VPM neurons, not sensory synapses (34). We reasoned that if our bimodal decay time distribution is indeed made up of sensory and CT events, then NASPM, an antagonist selective for calcium-permeable AMPARs, should decrease the amplitude of only Type 2 events (putative CT) and have no effect on Type 1 events (putative sensory). Calcium-permeable receptors contribute about half of the peak evoked AMPAR EPSC at CT inputs; therefore, we further reasoned that NASPM would reduce the amplitude of many mEPSCs below detection level and thus also lower Type 2 mEPSC frequency. We analyzed mEPSCs at baseline and in the presence of 100 µM NASPM, grouped the events into Type 1 and Type 2, and then analyzed the effect of NASPM on mEPSC frequency and amplitude (Supplemental Figure S10A-C). NASPM significantly decreased the frequency and amplitude of Type 2 events, with no significant effect on Type 1 mEPSC frequency or amplitude (Supplemental Figure S10D-G), further supporting the idea that Type 1 mEPSCs are putative sensory events and Type 2 mEPSCs are putative CT events.

To understand whether one or both of these populations were affected in DS mice, we compared the amplitude and frequency of putative sensory (Type 1) and CT (Type 2) mEPSCs between WT and DS mice within each age group. In VPL neurons, the frequency of Type 1 mEPSCs at 2W was not different between WT and DS mice (Figure 6A,B; Table 1; Supplemental Table S1-S3). Notably, the Type 1 mEPSC frequency was significantly reduced in DS mice at 4W compared to WT, and this reduction was maintained in adult mice aged 8W (Figure 6C,D). Interestingly, Type 2 mEPSC frequency was not statistically different between WT and DS mice at 2W, 4W, or 8W. The amplitude of Type 1 mEPSCs in VPL neurons were significantly reduced in 4W DS mice relative to WT, but not at 2W or 8W (Figure 6E). There were no statistically significant differences in Type 2 mEPSC amplitude in VPL neurons in any age group. Altogether, these electrophysiology and synapse imaging data suggest that VPL neurons display reduced glutamatergic transmission in a synapse-specific manner due to a reduction in ascending sensory input, while cortical input remains unchanged. There were no changes in the amplitude or frequency of Type 1 and Type 2 mEPSCs in VPM neurons further suggesting that this reduced glutamatergic input to somatosensory thalamus is cell-type-specific (Table 1; Supplemental Figure S11; Supplemental Table S1-S3).

Figure 6. DS mice show a selective reduction in Type 1 mEPSC frequency in VPL neurons after seizure onset.

A. Representative voltage-clamp recordings from VPL neurons show mEPSCs in WT and DS mice aged 2W (WT: n = 6 cells, 4 mice; DS: n = 9 cells, 6 mice), 4W (WT: n = 7 cells, 6 mice, 3F/3M; DS: n = 7 cells, 6 mice, 1F/5M), and 8W (WT: n = 7 cells, 6 mice, 3F/3M; DS: n = 8 cells, 5 mice, 1F/4M) with each mEPSC denoted as Type 1 (fast, putative sensory) or Type 2 (slow, putative CT) based on their decay kinetics. B. Representative ensemble average traces show Type 1 and Type 2 mEPSCs from WT and DS mice. Histograms show representative bimodal frequency distributions of mEPSC decay kinetics for all mEPSCs from representative WT and DS recordings.

C,D. The mean mEPSC frequency was plotted as D) raw data (geometric mean ± 95CI) and E) transformed log10 values (mean ± 95CI) as the data had a non-normal distribution.

E. Mean mEPSC amplitude were plotted for each mouse with the group mean ± SD. The data were analyzed by two-way ANOVA with pairwise Sidak’s comparisons (p-values shown on plots) between WT and DS at each age.

DS mice show persistent inhibitory synaptic deficits in VPL neurons, but not VPM neurons

VPL and VPM neurons receive dense GABAergic inhibitory input from nRT neurons, which show reduced tonic firing and excess burst firing in DS mice (5, 15, 20). To determine how GABAergic inhibitory synaptic transmission was affected across development in DS, we measured spontaneous inhibitory postsynaptic currents (sIPSCs) across the three age groups (Figure 7; Table 3; Supplemental Tables S1-S3). Although VPL and VPM neurons also receive inhibitory glycinergic input (49, 50), all IPSCs we recorded from VPL and VPM neurons were inhibited by gabazine and were therefore mediated by GABA_A receptors (Supplementary Figure S12), which is consistent with other reports (51–54). At 2W, we found no significant changes in mean sIPSC frequency in VPL; however, DS mice showed reduced sIPSC frequency compared to WT mice at 4W and 8W (Figure 7A-C). Mean sIPSC amplitude was similar for VPL neurons at 2W, 4W, and 8W (Figure 7D). In VPM neurons, sIPSC frequency in 4W DS mice was significantly reduced compared to WT, while sIPSC frequency was similar at 2W and 8W (Figure 7E-G). Mean sIPSC amplitude in VPM neurons did not differ between WT and DS mice at 2W, 4W, or 8W (Figure 7H). Comparisons between cumulative distributions for sIPSC frequency and amplitude yielded similar results (Supplemental Figure S13). These data suggest that DS mice have reduced spontaneous inhibitory synaptic input that arises after seizure onset in both VPL and VPM neurons, but is only maintained into adulthood in VPL neurons.

Figure 7. Reduced sIPSC frequency persists in adult DS mice in VPL neurons, but normalizes in VPM neurons.

A. Representative voltage-clamp recordings from VPL neurons show sIPSCs in WT and DS mice aged 2W (WT: n = 11, 8 mice; DS: n = 9 cell, 8 mice), 4W (WT: n = 9 cells, 7 mice, 3F/4M; DS: n = 8 cells, 7 mice, 3F/4M), and 8W (WT: n = 8 cells, 7 mice, 2F/5M; DS: n = 11 cells, 8 mice, 3F/5M).

B,C. The mean sIPSC frequency data were plotted as B) raw values with the geometric mean ± 95CI, which had a non-normal distribution, and C) transformed log₁₀ values with group mean ± 95CI, which were used for statistical analyses.

D. Mean sIPSC amplitudes were plotted for each mouse with the group mean ± SD.

E. Representative VPM recordings show sIPSCs in WT and DS mice aged 2W (WT: n = 9 cells, 7 mice; DS: n = 8 cells, 8 mice), 4W (WT: n = 10 cells, 7 mice, 1F/5M; DS: n = 8 cells, 8 mice, 4F/4M), and 8W (WT: n = 7 cells, 5 mice, 3F/2M; DS: n = 6 cells, 4 mice, 2F/2M).

F-H. The mean sIPSC frequency and amplitude data were plotted as described for VPL. The data were analyzed by two-way ANOVA with pairwise Sidak’s comparisons (p-values shown in plots) between WT and DS at each age.

Table 3.

sIPSC and mIPSC frequency and amplitude data.

	Frequency						Amplitude
	Raw data: Geometric mean [95 CI]						Raw data: Mean (SD)
	WT			DS			WT			DS
	GM	UL	LL	GM	UL	LL	M	SD	n	M	SD	n
VPL sIPSC
2W	6.14	8.81	4.28	7.32	9.96	5.38	25.31	4.33	11	25.23	5.74	9
4W	7.29	10.13	5.24	3.66	5.18	2.58	25.84	5.42	9	26.83	6.25	8
8W	8.22	11.37	5.95	4.81	6.39	3.63	29.90	4.29	8	27.34	2.98	11
VPM sIPSC
2W	8.69	10.98	6.88	6.55	10.28	4.18	27.11	6.58	10	22.13	2.57	9
4W	9.19	13.31	6.35	4.14	5.99	2.86	28.47	6.24	11	26.07	4.56	9
8W	7.45	11.35	4.88	6.57	9.24	4.67	28.78	1.58	7	28.68	7.74	6
VPL mIPSC
2W	6.07	8.95	4.11	5.53	7.68	3.98	23.20	3.61	12	21.97	2.13	10
4W	7.79	10.19	5.96	4.03	5.32	3.06	26.90	3.92	12	25.28	5.42	13
8W	5.36	7.36	3.91	3.37	4.82	2.36	25.63	3.92	10	25.32	3.20	9
VPM mIPSC
2W	4.72	6.54	3.40	6.25	9.33	4.18	21.29	3.18	11	21.46	2.83	10
4W	4.85	6.53	3.60	6.24	10.34	3.76	22.77	3.71	11	25.14	4.81	10
8W	3.71	4.40	3.13	4.84	6.89	3.40	27.33	4.06	8	28.25	5.51	10

We further recorded mIPSCs to investigate whether WT and DS mice exhibit any differences in synapse number or release probability that could have been obscured by differences in presynaptic neuron firing in sIPSC recordings (Figure 8A,E; Table 2; Supplemental Tables S1-S3). In VPL neurons, the mean mIPSC frequency was decreased in 4W DS mice relative to WT, but did not differ between 2W and 8W mice (Figure 8B,C). The mean mIPSC amplitude was similar for VPL neurons at 2W, 4W, and 8W (Figure 8D). There were no significant changes in mean mIPSC frequency for VPM neurons at 2W, 4W, or 8W (Figure 8F,G). Likewise, mIPSC amplitude was similar between WT and DS mice in VPM neurons at 2W, 4W, and 8W (Figure 8H). Comparisons between cumulative distributions for mIPSC frequency and amplitude yielded similar results (Supplemental Figure S14). Together, these data show that both cell types exhibit reduced frequency of spontaneous inhibitory input, but this deficit is only maintained through adulthood in the VPL thalamus. Whereas, the strength of inhibitory synapses does not differ between WT and DS mice in either cell type across the age groups.

Figure 8. DS mice show reduced mIPSC frequency in VPL neurons after seizure onset.

A. Representative voltage-clamp recordings from VPL neurons show mIPSCs in WT and DS mice aged 2W (WT: n = 12 cells, 7 mice; DS: n = 10 cells, 8 mice), 4W (WT: n = 11 cells, 7 mice, 3F/4M; DS: n = 12 cells, 8 mice, 2F/6M), and 8W (WT: n = 10 cells, 8 mice, 3F/5M; DS: n = 9 cells, 6 mice, 2F/4M).

B,C. The mean mIPSC frequency was plotted as B) raw data (geometric mean ± 95CI) and C) transformed log10 values (mean ± 95CI) as the data had a non-normal distribution.

D. Mean mIPSC amplitude were plotted for each mouse with the group mean ± SD.

E. Representative VPM recordings show mIPSCs in WT and DS mice aged 2W (WT: n = 11 cells, 8 mice; DS: n = 9 cells, 8 mice), 4W (WT: n = 9 cells, 6 mice, 2F/4M; DS: n = 7 cells, 6 mice, 3F/3M), and 8W (WT: n = 8 cells, 6 mice, 3F/3M; DS: n = 10 cells, 7 mie, 3F/4M).

F-H. The mean mIPSC frequency and amplitude data were plotted as described for VPL. VPL and VPM data were analyzed by two-way ANOVA with pairwise Sidak’s comparisons (p-values shown on plots) between WT and DS at each age.

DISCUSSION

This study demonstrates that excitatory and inhibitory synaptic dysfunction in the VPL thalamus of DS mice emerge after seizure onset and persist into adulthood, whereas VPM synaptic input remains largely intact. These findings underscore a region-specific vulnerability within the somatosensory thalamus that may contribute to persistent sensory and sleep deficits in DS patients. This is particularly relevant given that some Na_V1.1-dependent changes in neuronal excitability normalize in adult DS mice, indicating that synaptic deficits may play an increasingly important role with disease progression.

We uncovered reduced spontaneous and miniature EPSC frequency in VPL neurons in DS mice at 4 and 8 weeks of age, with amplitude reduced only at 4 weeks. Inhibitory input to VPL neurons was also impaired and followed the same developmental trajectory. Na_V1.1 expression increases rapidly during the second postnatal week (55) and seizures begin in the third postnatal week (9, 28). Therefore, it is conceivable that Na_V1.1 haploinsufficiency could impact synapse formation before seizure onset. However, synapse number and function were normal in 2-week-old DS mice, indicating that synaptic dysfunction arises after the period of highest seizure burden (P17 – P24). These results extend prior work showing deficits to excitatory and inhibitory inputs to VPL neurons in 4-week-old DS mice (15) and clarify that sensory input impairments develop after seizure onset and persist into adulthood.

Establishing this timeline provides mechanistic insight into how circuit dysfunction progresses and when synaptic abnormalities may contribute to behavioral phenotypes or be therapeutically targeted. This is particularly important for understanding the behavior-specific emergence and progression of sensorimotor and cognitive deficits in DS. Several early-appearing phenotypes arise in the second postnatal week in DS mice and can thus be uncoupled from seizures, including proprioception and grasping deficits (56) as well as impaired olfactory discrimination and cognitive deficits associated with altered social information processing (57). Notably, some early deficits, such as impaired proprioception, grasping, and hyperactivity, largely normalize by adulthood. In contrast, other behaviors, including blunted acoustic startle and visuospatial navigation deficits, emerge only after seizure onset and persist into adulthood (25, 56, 58, 59). Sleep impairments such as sleep fragmentation and increased sleep latency, follow a similar trajectory, worsening after seizure onset and into adulthood in both DS mouse models (7, 22, 26) and patients (21, 23, 24, 27, 60, 61). Together, these findings indicate that behavioral phenotypes are not static across disease stages, implying ongoing molecular or structural changes within affected circuits that may exacerbate deficits or compensate to restore function.

Immunohistochemistry revealed a selective reduction in VGLUT2-positive sensory synapses in VPL neurons at 4 and 8 weeks, while VGLUT1-positive CT synapses were preserved. Reduced mEPSC frequency at putative sensory inputs (Type 1) further supports this input specificity. Sensory synapse number remained reduced in adult DS mice, whereas synapse strength normalized, suggesting altered synapse maintenance rather than persistent functional weakening of synapses. However, we cannot rule out changes in release probability at these synapses. The mechanisms underlying selective vulnerability of sensory input to VPL neurons, but not cortical inputs that are more involved in seizure propagation, remain unresolved. High Scn1a expression in ascending sensory neurons, but not CT neurons, could selectively impair sensory signals to VPL neurons in DS (62, 63), disrupting activity-dependent maintenance mechanisms at sensory synapses. Alternatively, combined effects of circuit hyperexcitability, altered presynaptic neuron function, and postsynaptic properties could underlie input-specific vulnerabilities. Indeed, prior work shows impaired corticoreticular synapses in this Na_V1.1-haploinsufficient DS model (15) and similar reductions in nRT sEPSCs in another DS mouse model (20). Additional seizure mouse models also show that nRT dysfunction can disrupt corticoreticular input but not CT input to VPL and VPM thalamus (64–68). The same cortical neurons send axon collaterals to nRT and VPL/VPM neurons, suggesting that specific postsynaptic molecular signatures such as distinct glutamate receptor expression may confer input-specific vulnerability. Future work using input- and cell-type-specific manipulations will be required to ascribe these vulnerabilities to specific circuit components and mechanisms.

The combined loss of excitatory sensory input and inhibitory control in VPL neurons could disrupt thalamocortical processing in DS. VGLUT2-positive sensory synapses are large, proximal, and high-release-probability, so their loss likely reduces excitatory drive, while reduced nRT-mediated inhibition may impair temporal precision and gain control of thalamic output (20, 69). These persistent deficits may contribute to the sensory abnormalities in DS patients, including altered pain perception, sleep, and sensorimotor integration (70–73). These findings complement evidence that the somatosensory thalamocortical circuit continues to generate hypersynchronous oscillatory activity in adult DS mice (20), suggesting that both oscillatory abnormalities and disrupted sensory input may contribute to adult-stage dysfunction. Altered intrinsic excitability in nRT, VPL, and VPM neurons has been reported in juvenile DS mice (5, 15, 20) and determining whether these changes persist into adulthood will be essential for distinguishing synaptic from cellular contributions to circuit pathology. Another limitation of this study is that we did not assess how these synaptic changes alter sensory responses or CT modulation including feed-forward inhibition via the nRT. In addition, VPL and VPM neurons receive inhibitory glycinergic input arising from the brainstem, whose status in DS remains unknown. Future work addressing these gaps will be necessary to determine how input- and cell-type-specific synaptic deficits influence circuit-level functions and sensory information processing.

VPM synaptic input was largely unaffected at all ages, suggesting cell-type-specific vulnerability. VPL and VPM differ in excitatory drive, firing properties (36), inhibitory connectivity (74), and ascending sensory innervation. Both VPL- and VPM-projecting ascending sensory pathway express Na_V1.1 (75, 76), and Na_V1.1 expression in dorsal root ganglion neurons is important for mechanosensation and proprioception (77–79); however, sensory neuron functional susceptibility to Na_V1.1 haploinsufficiency has not been compared. Moreover, VPL neurons show reduced excitability in DS mice, whereas VPM neurons show modestly increased excitability (15), which could contribute to divergent synaptic phenotypes. Because altered sensory processing and sleep deficits persist throughout the lifespan of DS patients (Anwar et al., 2019; Brunklaus et al., 2012; Lagae et al., 2018), determining which circuit components are directly impacted by Na_V1.1 haploinsufficiency versus secondary to circuit hyperexcitability will be critical for understanding of how ascending and central sensory pathways are affected across disease stages.

Overall, these results emphasize the need to determine how impaired excitatory and inhibitory synaptic integrity in the VPL thalamus contributes to thalamocortical dysfunction and whether restoring these inputs can improve DS-associated sensory symptoms. Identifying the molecular basis of sensory synapse vulnerability, and the apparent resilience of CT input and VPM neurons, may reveal new therapeutic targets and clarify whether these underlying vulnerabilities exist only during development or persist through adulthood. Future work should also examine whether similar synapse-specific susceptibility occurs in other epilepsy or neurodevelopmental disorders.

Supplementary Material

Supplemental Figs. S1-S14: https://doi.org/10.7294/32326587.

Supplemental Tables S1-S4: https://doi.org/10.7294/32326587.

DATA AVAILABILITY

Data underlying this manuscript are made available through the Virginia Tech Data Repository at https://doi.org/10.7294/32326587.