Restoring thalamocortical circuit dysfunction by correcting HCN channelopathy in Shank3 mutant mice

Abstract

Thalamocortical (TC) circuits are essential for sensory information processing. Clinical and preclinical studies of autism spectrum disorders (ASDs) have highlighted abnormal thalamic development and TC circuit dysfunction. However, mechanistic understanding of how TC dysfunction contributes to behavioral abnormalities in ASDs is limited. Here, our study on a Shank3 mouse model of ASD reveals TC neuron hyperexcitability with excessive burst firing and a temporal mismatch relationship with slow cortical rhythms during sleep. These TC electrophysiological alterations and the consequent sensory hypersensitivity and sleep fragmentation in Shank3 mutant mice are causally linked to HCN2 channelopathy. Restoring HCN2 function early in postnatal development via a viral approach or lamotrigine (LTG) ameliorates sensory and sleep problems. A retrospective case series also supports beneficial effects of LTG treatment on sensory behavior in ASD patients. Our study identifies a clinically relevant circuit mechanism and proposes a targeted molecular intervention for ASD-related behavioral impairments.

Keywords: ASDs; HCN2; Shank3; autism spectrum disorders; lamotrigine; neurodevelopmental disorders; sleep fragmentation; tactile hypersensitivity; thalamocortical circuits.

Graphical abstract

Figure 1

Shank3 deletion leads to neuronal hyperactivity and a mismatched TC locking relationship (A) Schematic of head-fixed recordings in mice with multielectrode drive targeting the VPM to record spontaneous neuronal activity and tactile-evoked responses. (B) Representative recording traces of VPM neurons in Shank3 WT and KO. Blue shades mark the identified bursts. (C) Cumulative probability plot of VPM spontaneous firing rates in Shank3 WT and KO (left) and average firing rates of VPM neurons in Shank3 WT and KO (right). (D) Cumulative probability plot of VPM spontaneous burst rates in Shank3 WT and KO (left) and average burst rates of VPM neurons in Shank3 WT and KO (right). (E) Example peri-stimulus time histograms (top) and rasters (bottom) showing tactile-evoked (20 ms, green bar) responses in Shank3 WT and KO. (F) Quantification of the area under curves (AUC) representing elevated neuronal responses of VPM neurons with 0.3-mm whisker deflection in Shank3 KO mice. (G) Probability of the burst occurrence in a trial, showing an increased proportion of trials featuring bursts in Shank3 KO mice. (H) Quantification of burst ratio, displaying increased burst numbers in all trials featuring bursts in Shank3 KO mice. (C–H) n = 147 neurons from 5 mice for WT; n = 123 neurons from 5 mice for KO. (I) Schematic of VPM recordings coupling with EEG and EMG recordings with multielectrode drive in vivo (top) and Example traces of EEG and EMG during NREM natural sleep (bottom). (J) Cumulative probability plot of VPM firing rates during NREM sleep in Shank3 WT and KO (left) and average firing rates of VPM neurons in Shank3 WT and KO (right). n = 147 neurons from 5 mice for WT; n = 123 neurons from 5 mice for KO. (K) Cumulative probability plot of VPM burst rates during NREM sleep in Shank3 WT and KO mice (left) and average burst rates of VPM neurons in Shank3 WT and KO (right). (L) Cumulative probability plot of NREM sleep bout length in Shank3 WT and KO (left) and NREM sleep bout length of individual animals of Shank3 WT and KO (right). n = 13 mice for WT; n = 12 mice for KO. (M) Schematic of multielectrode drive targeting the VPM and somatosensory cortex (S1). (N) Example traces of S1-LFP raw data (top), delta waves (1–4 Hz, center), and VPM spikes (bottom) during typical NREM sleep, showing the temporal alignment of VPM spikes to the preferred phase of the delta activity in WT (left) and KO mice (right). Blue and red dots mark the identified bursts. (O) Spike distribution histogram for an example WT VPM neuron (left) and quantification of spike numbers with temporal alignment to different phases of LFP-delta of the S1 (right). (P) Spike distribution histogram for an example KO VPM neuron (left) and quantification of spike numbers with temporal alignment to different phases of LFP-delta of the S1 (right). (Q) Quantification of locked phase, showing a mismatched relationship between VPM spikes and S1-Delta LFP in Shank3 KO mice. (R) Quantification of locking strength in Shank3 WT and KO mice. For statistical comparisons, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Two-tailed unpaired t test (C, D, F–H, J, K, Q, and R) and Kolmogorov-Smirnov test (C, D, J, and I). Detailed statistical information can be found in Data S1. Data are presented as median ± 95% CI.

Figure 2

Enhanced dendritic integration and altered intrinsic properties of VPM neurons contribute to neuronal hyperactivity in Shank3 KO mice (A) Schematic of virus and Retrobead injections. (B) Example traces of EPSPs with 40-Hz photoactivation. (C) Quantification of EPSP summation ratio, showing that VPM cells show enhanced integration capability to cortical excitatory synaptic inputs. n = 11 neurons from 4 mice for WT; n = 15 neurons from 6 mice for KO. (D) Schematic of virus and Retrobead injections. (E) Example traces of EPSPs with 40-Hz photoactivation. (F) Quantification of EPSP summation ratio, showing that VPM cells show intact integration capability to excitatory synaptic inputs from the brain stem. n = 7 neurons from 4 mice for WT; n = 9 neurons from 4 mice for KO. (G) Summary of resting membrane potential (RMP) of VPM cells, showing that Shank3 KO cells are more hyperpolarized compared with WT cells. n = 25 neurons from 7 mice for WT; n = 20 neurons from 5 mice for KO. (H) Summary of membrane resistance (Rm) of VPM cells displaying a higher Rm of Shank3 KO cells compared with WT cells. n = 25 neurons from 7 mice for WT; n = 20 neurons from 5 mice for KO. (I) Representative traces of depolarization-induced bursts of VPM cells of Shank3 WT (left) and KO (right) mice. (J) Quantification analysis showing a decreased VPM burst threshold in Shank3 KO cells. (K) Quantification analysis denoting an unaltered burst latency in Shank3 KO cells. (L) Representative traces of rebound bursts of VPM cells of Shank3 WT (left) and KO (right). (M) Quantification analysis showing a reduced VPM rebound burst threshold in Shank3 KO cells. (N) Quantification analysis denoting an increased rebound burst latency in Shank3 KO cells. (J, K, M, and N) n = 21 neurons from 7 mice for WT; n = 20 neurons from 5 mice for KO. For statistical comparisons, ∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Friedman’s M test (C), repeated-measurement ANOVA (F), two-tailed unpaired t test (G, J, and M); and Mann-Whitney U test (H, K, and N). Detailed statistical information can be found in Data S1. Data are presented as mean ± SEM.

Figure 3

Shank3 deletion impairs the function and expression of HCN2 channels in the VPM (A) Representative traces of Ih currents in Shank3 WT (left) and KO (right) cells. (B) Summary of Ih current density with different voltage steps, showing decreased HCN mediated current density in Shank3 KO cells. (C) Quantification plot representing reduced maximal Ih current density in Shank3 KO cells. (B and C) n = 9 neurons from 3 mice for WT; n = 9 neurons from 3 mice for KO. (D) Representative images of immunoblotting for HCN1, HCN2, and HCN4 of the VPM total protein in Shank3 WT and KO. (E) Quantification plot showing that the expression level of HCN2 is decreased in Shank3 KO mice. (F) Representative images of immunoblotting for HCN2 of the VPM synaptoneurosomal protein in Shank3 WT and KO. (G) Quantification plot showing that the expression level of HCN2 is reduced in Shank3 KO mice. (E and G) n = 3 mice for WT; n = 3 mice for KO. (H) Representative immunogold micrographs showing the subcellular location of HCN2 (red arrows point to HCN2 in the membrane) in somatic (left) and dendritic regions (right) in Shank3 WT (top) and KO (bottom). Scale bar: 0.5 μm. (I) Density of HCN2-positive particles in somata (left) and dendrites (right). (J) Linear density of HCN2-positive particles in the neuronal membranes in somata (left) and dendrites (right). (I and J) n = 20 somata and 99 dendrites from 3 mice for WT; n = 16 somata and 95 dendrites from 3 mice for KO. For statistical comparisons, ∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Friedman’s M test (B) and two-tailed unpaired t test (C, E, G, I, and J). Detailed statistical information can be found in Data S1. Data are presented as mean ± SEM (B, C, E, and G) and median ± 95% CI (I and J).

Figure 4

HCN2 overexpression in the VPM corrects electrophysiological alterations in Shank3 KO mice (A) Schematic of virus injection. (B) Representative traces of Ih currents. (C) Summary of Ih current density with different voltage steps, showing that the HCN2 overexpression (KO+HCN2) group displays increased Ih current density compared with the scramble (KO+Scr) group. (D) Quantification plot representing increased maximal Ih current density in the KO+HCN2 overexpression (OE) group. (C and D) n = 5 neurons from 3 mice for WT+Scr; n = 9 neurons from 3 mice for KO+HCN2; n = 11 from 3 mice for KO+Scr. (E) Representative images of immunoblotting, showing the HCN2 expression level. (F) Quantification of protein level, showing the effects of HCN2 OE in Shank3 KO mice. n = 3 mice for WT+Scr; n = 3 mice for KO+HCN2; n = 3 mice for KO+Scr. (G) Immunofluorescence images showing the HCN2 expression level in mCherry-positive (yellow arrows) and -negative cells (white arrows). Scale bar: 20 μm. (H) Quantification of fluorescence intensity in mCherry-positive and -negative cells. n = 127 mCherry-positive neurons and n = 136 mCherry-negative neurons from 3 mice. (I–M) Quantification of RMP (I), Rm (J), burst threshold (K), rebound burst threshold (L), and rebound burst latency (M), showing that HCN2 OE could rescue the intrinsic properties of VPM cells in Shank3 KO mice. n = 16 neurons from 3 mice for WT+Scr; n = 12 neurons from 3 mice for KO+HCN2; n = 19 from 3 mice for KO+Scr. (N) Schematic of virus injection and patch-clamp recordings. (O) Representative images showing EGFP and mCherry expression in the VPM. Scale bar: 10 μm. (P) Example traces of EPSPs with 40-Hz photoactivation. (Q) Quantification of the EPSP summation ratio, showing that VPM cells show reduced integration capability to cortical excitatory synaptic inputs with HCN2 OE in Shank3 KO mice. n = 7 neurons from 3 mice for WT+Scr; n = 6 neurons from 3 mice for KO+HCN2; n = 10 from 3 mice for KO+Scr. For statistical comparisons, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Friedman’s M test (C), one-way ANOVA (D, F, and I–M), two-tailed unpaired t test (H), and repeated-measurement ANOVA (Q). Detailed statistical information can be found in Data S1. Data are presented as mean ± SEM.

Figure 5

Acute administration of LTG rescues sensory hypersensitivity without normalization of sleep fragmentation and TC phase locking (A) Representative traces showing the effect of LTG on Ih currents. (B) LTG could increase the maximal Ih current density in Shank3 KO VPM cells. (C–E) LTG rescues passive membrane properties in Shank3 KO VPM cells. (B–E) n = 13 neurons from 7 mice. (F) Schematic and workflow of in vivo recording design with LTG acute injection. (G–I) Effects of LTG on spontaneous firing rate and burst rate in awake Shank3 KO mice. n = 142 neurons from 5 mice for Veh; n = 110 neurons from 5 mice for LTG. (J) Fraction plot showing the LTG effects on whisker-induced neuronal responses (left). Quantification of AUC denotes that LTG could decrease the hyper-reactivity of VPM cells in Shank3 KO mice. (K) Fraction plot showing the LTG effects on whisker-induced burst response (left). Quantification of AUC denotes that LTG could decrease the burst ratio of VPM cells in Shank3 KO mice. (L–N) Effects of LTG on spontaneous firing rate and burst rate in NREM sleep in Shank3 KO mice. n = 142 neurons from 5 mice for Veh; n = 110 neurons from 5 mice for LTG. (O) Quantification of locked phase, representing no significant difference between LTG and vehicle groups. (P) Quantification of locking strength, denoting that LTG could increase the locking strength in Shank3 KO mice. (Q) Schematic of behavioral test designs. (R) Quantification of the discrimination score for the textured zone, showing decreased exploration preference in the textured zone in Shank3 KO mice after LTG acute administration. (S) Quantification of response scores following whisker stimulation in Shank3 KO mice after LTG acute administration. (T) Average response scores assigned to WT and Shank3 KO mice during trials. (R–T) n = 9 mice for all groups. (U) Summary of NREM sleep bout length, showing no difference following LTG acute injection in Shank3 KO mice. n = 8 mice for all groups. For statistical comparisons, ∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Wilcoxon signed-rank test (B), two-tailed paired t test (C–E), two-tailed unpaired t test (H–K and M–P), Kolmogorov-Smirnov test (J and K), Kruskal-Wallis H test (R and S), Friedman’s M test (T), and one-way ANOVA (U). Detailed statistical information can be found in Data S1. Data are presented as mean ± SEM (B–E and R–U) and median ± 95% CI (H–K and M–P).

Figure 6

Early developmental treatment of LTG exerts long-lasting rescue effects in Shank3 KO mice (A) Schematic of experimental procedures. (B) The Ih current density, showing the effects of LTG treatment during P7–P11 in Shank3 WT and KO mice. n = 11 neurons from 3 mice for WT+Veh; n = 8 neurons from 3 mice for KO+Veh; n = 16 neurons from 3 mice for WT+LTG; n = 14 neurons from 3 mice for KO+LTG. (C) Quantification of maximal Ih current density, showing that LTG treatment could boost Ih current in Shank3 KO mice. (D and E) Effects of a 5-day regimen of LTG on spontaneous firing rate and burst rate in awake Shank3 KO mice. n = 213 neurons from 5 mice for WT+Veh; n = 222 neurons from 5 mice for KO+Veh; n = 235 neurons from 5 mice for KO+LTG. (F) Fraction plot showing the effects of a 5-day regimen of LTG treatment on whisker-induced neuronal responses (left). Quantification of AUC denotes that LTG could decrease the hyper-reactivity of VPM cells in Shank3 KO mice. (G) Fraction plot showing the effects of a 5-day regimen of LTG on whisker-induced burst response (left). Quantification of AUC denotes that LTG could decrease the burst ratio of VPM cells in Shank3 KO mice. (H and I) Effects of a 5-day regimen of LTG on spontaneous firing rate and burst rate in NREM sleep in Shank3 KO mice. n = 321 neurons from 5 mice for WT+Veh; n = 328 neurons from 5 mice for KO+Veh; n = 327 neurons from 5 mice for KO+LTG. (J and K) Spike distribution histograms (J) and quantification of the locked phase, representing that LTG corrects the locked phase in Shank3 KO mice (K). (L) Quantification of locking strength, denoting that LTG could increase the locking strength in Shank3 KO mice. (M) Quantification of the discrimination score for the textured zone, showing decreased exploration preference in the textured zone in Shank3 KO mice receiving a 5-day regimen of LTG treatment. (N) Quantification of response scores following whisker stimulation in Shank3 KO mice receiving a 5-day regimen of LTG treatment. (O) Average response scores during trials. n = 8 mice for WT+Veh; n = 9 mice for KO+Veh; n = 8 mice for KO+LTG. (P) Summary of NREM sleep bout length, showing that a 5-day regimen of LTG treatment rescues sleep fragmentation in Shank3 KO mice. n = 9 mice for WT+Veh; n = 11 mice for KO+Veh; n = 11 mice for KO+LTG. For statistical comparisons, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Friedman’s M test (B and O), one-way ANOVA (C–J and L–N), Kruskal-Wallis H test (P), and Jonckheere-Terpstra test (F, G, and K). Detailed statistical information can be found in Data S1. Data are presented as mean ± SEM (B, C, and M–P) and median ± 95% CI (D–L).

Figure 7

LTG treatment shows beneficial effects on hypersensitivity of ASDs patients (A) Summary of patient characteristics. (B) Forest plot showing the WMD value of the difference between the treatment and no-treatment group on ABC sensory score. (C) Forest plot showing the WMD value of the difference between the treatment and no-treatment group on CARS (item 9) score. Mann-Whitney U test. Data are presented as mean ± SD (A).