Heterogeneous subpopulations of GABAAR-responding neurons coexist across neuronal network scales and developmental stages in health and disease

Abstract

Gamma-aminobutyric acid (GABA) is the main inhibitory neurotransmitter in adults. Depolarizing GABA responses have been well characterized at neuronal-population average level during typical neurodevelopment and partially in brain disorders. However, no investigation has specifically assessed whether a mosaicism of cells with either depolarizing or hyperpolarizing/inhibitory GABAergic responses exists in animals in health/disease at diverse developmental stages, including adulthood. Here, we showed that such mosaicism is present in wild-type (WT) and down syndrome (DS) neuronal networks, as assessed at increasing scales of complexity (cultures, brain slices, behaving mice). Nevertheless, WT mice presented a much lower percentage of cells with depolarizing GABA than DS mice. Restoring the mosaicism of hyperpolarizing and depolarizing GABA-responding neurons to WT levels rescued anxiety behavior in DS mice. Moreover, we found heterogeneous GABAergic responses in developed control and trisomic human induced-pluripotent-stem-cells-derived neurons. Thus, a heterogeneous subpopulation of GABA-responding cells exists in physiological/pathological conditions in mouse and human neurons, possibly contributing to disease-associated behaviors.

Keywords: Behavioral neuroscience; Cellular neuroscience; Developmental neuroscience.

Graphical abstract

Figure 1

Mixed subpopulations of neurons with depolarizing or hyperpolarizing GABA signaling are present during in vitro development of neuronal primary cultures (A) Representative calcium traces obtained from live imaging of cultured WT and Ts65Dn hippocampal neurons loaded with the Ca²⁺-sensitive dye Fluo-4 upon bath application of GABA (100 μM) at different time points during in vitro development. ΔF/F0 represents the change in the calcium intensity expressed as variation in fluorescence intensity upon GABA application, relative to the basal fluorescence intensity. (B) Quantification of the mean percentage (±SEM) of neurons showing GABA-induced Ca²⁺ responses (i.e., depolarizing GABA responses) in experiments as in (A). Numbers in parentheses indicate the number of analyzed coverslips for each time point (obtained from 3 to 4 independent neuronal cultures). ∗∗∗p < 0.001; Tukey post hoc test following two-way ANOVA. (C) Representative calcium traces of neurons pretreated with vehicle (0.01% DMSO) or the NKCC1 inhibitor bumetanide (10 μM) upon bath application of GABA at 15 DIV. (D) Quantification of the percentage of neurons showing depolarizing GABA responses in experiments as in (C). In the boxplot, the small square indicates the mean, the central line illustrates the median, the box limits indicate the 25th and 75th percentiles, the whiskers represent the 5th–95th percentiles, and each dot indicates a value obtained from an individual coverslip. The numbers in parentheses indicate the number of analyzed coverslips for each experimental group (obtained from 3 to 5 independent neuronal cultures). ∗p < 0.05, ∗∗∗p < 0.001; Tukey post hoc test following two-way ANOVA. (E) Representative images of WT and Ts65Dn hippocampal neurons during imaging experiments with the chloride-sensitive dye MQAE upon treatment with vehicle (0.01% DMSO) or bumetanide (10 μM) at 15 DIV. The fluorescence intensity of the dye (color-coded at the bottom) is inversely proportional to intracellular chloride concentration ([Cl⁻]_i). Scale bar: 100 μm. (F) Beanplot showing the distribution of MQAE raw fluorescence values for all neurons imaged as in E (WT-vehicle: 840 neurons, Ts-vehicle: 818 neurons, WT-bumetanide: 597 neurons, Ts-bumetanide: 840 neurons, from 3 independent experiments). (G) Quantification of the average [Cl⁻]_i with MQAE in the same dataset described in (F). In the boxplot, the small square indicates the mean, the central line illustrates the median, the box limits indicate the 25th and 75th percentiles, the whiskers represent the 5th to 95th percentiles, and each dot indicates a value obtained from an individual coverslip. The numbers in parentheses indicate the number of analyzed coverslips for each experimental group (obtained from 3 independent neuronal cultures). ∗p < 0.05, ∗∗∗p < 0.001; Tukey post hoc test following two-way ANOVA.

Figure 2

Mixed subpopulations of neurons with hyperpolarizing or depolarizing GABA_AR signaling are present in neuronal cultures grown over MEAs (A) Left: schematic representation of a primary hippocampal neuronal culture grown over a microelectrode array (MEA) for electrophysiological recordings. Center: representative transmitted-light image showing a well of 60 electrodes MEA seeded with hippocampal neurons at 21 DIV. Right: schematic representation of the experimental protocol; neuronal cultures were preincubated with vehicle (0.01% DMSO) or bumetanide (10 μM) for 45 min followed by 30-min recording of spontaneous activity. Neurons were then recorded for an additional 30 min, 10 min after the addition of bicuculline (BIC, 20 μM) or GABA (100 μM). (B, E) Quantification of the mean firing rate (MFR) ratio of the WT and Ts65Dn neuronal cultures upon BIC (B) or GABA (E) bath application. The MFR ratio over the baseline firing (dotted line) was calculated for each electrode and then averaged for each MEA. An MFR ratio higher than 1 (representing the baseline level) indicates an increase in activity upon treatment, whereas a ratio below 1 indicates a decrease in activity. In the boxplot, the small square indicates the mean, the central line illustrates the median, the box limits indicate the 25th and 75th percentiles, the whiskers represent the 5th to 95th percentiles, and each dot represents the MFR ratio for each recorded culture. The numbers in parentheses indicate the number of analyzed MEAs for each experimental group (obtained from 5 independent neuronal cultures). ∗∗p < 0.01; ∗∗∗p < 0.001; Tukey’s post hoc test following two-way ANOVA. (C, F) Scatterplots showing the MFR for each active electrode (plotted as a dot) from all recorded MEAs (in B or E) seeded with WT and Ts65Dn neurons before (x axis) and after (y axis) bath application of BIC (C) or GABA (F). Dark gray dots represent electrodes showing a significant increase in the MFR. Light gray dots represent electrodes showing a significant decrease in the MFR. Black dots represent electrodes showing no significant changes in the MFR. Significant changes in the MFR (numbers with arrows) for each electrode upon BIC or GABA application were evaluated by bootstrap analysis. (D, G) Quantification of the average percentage number (±SEM) of MEA electrodes (in the same experiments in B, C, E, and F), showing significant changes in the MFR by bootstrap analysis after BIC (D) or GABA (G) administration in comparison to their basal conditions in WT (blue) and Ts65Dn (pink) neuronal cultures. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; Tukey’s post hoc test following two-way ANOVA on ranked transformed data.

Figure 3

Mixed subpopulations of neurons with hyperpolarizing or depolarizing GABA signaling are present in acute hippocampal slices from adult WT and Ts65Dn mice (A) Left: schematic representation of an acute brain slice of the hippocampus-entorhinal cortex (EC) region from an adult animal (postnatal day 70 [P70]). The slice is positioned on an MEA for electrophysiological recordings. Center: representative picture of the experimental setup with a 60-electrode MEA and an overlying hippocampus-EC slice. The gray square highlights the analyzed electrodes, which were positioned in the CA1 region. Right: schematic representation of the experimental protocol. Brain slices were first preincubated with vehicle (0.01% DMSO) or bumetanide (10 μM) for 45 min. Then, we recorded 30 min of spontaneous activity in the presence of vehicle or bumetanide followed by 30 min of recording 10 min after the addition of bicuculline (BIC; 20 μM) or GABA (100 μM). (B, E) Representative high-pass filtered electrophysiological traces (>300 Hz) for WT and Ts65Dn slices pretreated with vehicle or bumetanide and recorded before and after bath application of BIC (B) or GABA (E). Scale bars: (B) 80 μV, 2 s; (E) 30 μV, 2 s. (C, F) Quantification of the neuronal mean firing rate (MFR) ratio from WT and Ts65Dn slices upon BIC (C) or GABA (F) bath application. The MFR ratio over the baseline firing (dotted line) was calculated for each electrode and then averaged for each MEA. An MFR ratio higher than 1 (representing the baseline level) indicates an increase in activity upon the treatment, whereas a ratio below 1 indicates a decrease in activity. In the boxplot, the small square indicates the mean, the central line illustrates the median, the box limits indicate the 25th and 75th percentiles, the whiskers represent the 5th to 95th percentiles, and each dot represents the MFR ratio for each recorded slice. The numbers in parentheses indicate the number of analyzed slices for each experimental group. ∗∗p < 0.01; ∗∗∗p < 0.001; Tukey’s post hoc test following two-way ANOVA. (D, G) Quantification of the average percentage number ± SEM of MEA electrodes (in the same experiments in C and F), showing changes of at least 15% in the MFR after BIC (D) or GABA (G) administration in WT (blue) and Ts65Dn (pink) slices. ∗∗p < 0.01; ∗∗∗p < 0.001; Tukey’s post hoc test following two-way ANOVA on ranked transformed data.

Figure 4

Bumetanide treatment differentially affects neuronal activity in vivo in the hippocampus of freely moving adult WT and Ts65Dn mice and partially rescues aberrant GABA_AR signaling in Ts65Dn animals (A) Schematic representation of the experimental setup. Left: two- to three-month-old WT and Ts65Dn mice received a stereotaxic injection in the dorsal hippocampal CA1 region with AAV viruses expressing the Ca²⁺-sensor GCaMP6f under the control of the CamK2a promoter. Ts65Dn and WT mice were then implanted with a microendoscopic probe, and neuronal activity was longitudinally assessed by recording in vivo Ca²⁺ events within the CA1 hippocampal region of freely moving mice with a miniaturized head-mounted microscope. Right: examples of projection maps of recorded neurons in a WT and a Ts65Dn adult animal. (B) Schematic representation of the experimental protocol timeline. Ca²⁺ events were imaged in the same neuron in two consecutive sessions before and after administration of the GABA_AR-positive allosteric modulator diazepam (2 mg/kg), following a subchronic (2 weeks) treatment with either vehicle (2% DMSO in saline) or bumetanide (0.2 mg/kg). Within each imaging session, neuronal activity was recorded in periods of 5 min (“ON”, red) alternated to 5 min of rest (“OFF”, blue) to collect 15 min of neuronal data for each session while avoiding phototoxicity. (C) Quantification of single neuron mean event rates (MERs) of WT and Ts65Dn mice before and after diazepam administration in experiments as in (B). In the boxplot, the small square indicates the mean, the central line illustrates the median, the box limits indicate the 25th and 75th percentiles, and whiskers represent the 5th to 95th percentiles. Each dot indicates the binned value of the MER (bin size = 0.005 events/sec) obtained from individual neurons (from 5 WT to 8 Ts65Dn adult mice). ∗p < 0.05; ∗∗∗p < 0.001; Tukey’s post hoc test following a linear mixed effects model (with two factors). (D) Scatterplots showing the MER for each active (MER>0.01 events/s in vehicle) neuron (plotted as a dot) in (C) before (x axis) and after (y axis) administration of diazepam. Dark gray dots represent neurons showing a significant increase in the MER. Light gray dots represent neurons showing a significant decrease in the MER. Black dots represent neurons showing no significant changes in the MER. Significant changes in the MER (number with arrows) for each neuron upon diazepam application were evaluated by bootstrap analysis. (E) Quantification of the percentage of neurons (in the same experiments in C), showing significant changes in the MER by bootstrap analysis after diazepam administration in comparison to their basal conditions in WT (blue) and Ts65Dn (pink) mice. ∗p < 0.05, ∗∗∗p < 0.001; chi-squared test with Sidak adjustment for multiple comparisons.

Figure 5

Mixed subpopulations of neurons with hyperpolarizing or depolarizing GABA_AR signaling respond differently to bumetanide administration in vivo (A) Schematic representation of the experimental protocol timeline (top) together with representation of the analysis approach (bottom). (B) Schematic representation of the possible distribution of neuronal subpopulations based on MER variation in response to bumetanide or diazepam treatment (Ratio 1 vs. Ratio 2, see main text). Cells showing MER ratio changes below 10% were excluded from the analysis (shadowed gray area). The numbers in the black dots indicate the four quadrants representing the possible different responses (increasing or decreasing arrows) of neuronal activity upon diazepam or bumetanide treatment over that elicited after vehicle treatment. (C) Scatterplot showing the neuronal population distribution based on MER variation in response to bumetanide or diazepam treatment for Ts65Dn neurons in the same experiments described in Figure 4. Pearson correlation showed a significant positive correlation between the responses to bumetanide or diazepam treatment in neurons from Ts65Dn mice. (D) Quantification of the percentage of neurons for each quadrant on the same dataset described in (C). ∗∗p < 0.01; chi-squared test with Sidak adjustment for multiple comparisons. (E) Schematic representation of the support vector machine (SVM) classifier used for discrimination between the two classes of variation (i.e., decrease [<1] or increase [>1] of activity after bumetanide plus diazepam when compared with vehicle plus diazepam). (F) Left: percentage of Ts65Dn neurons used as a training set or a test set for the SVM classification. Right: the upper pie chart reports the percentage of neurons that fell into the two categories (>1 and <1) for the training set. The lower pie chart reports the percentage of neurons that fell in the two categories (>1 and <1) after running the SVM classification on the test set. (G) Left: confusion matrix showing the performance accuracy (correct green and incorrect white) of the SVM classifier in Ts65Dn neurons based on the two subpopulations of diazepam-responding neurons (<1 decrease upon diazepam plus bumetanide or >1 increase upon bumetanide plus diazepam when compared with diazepam plus vehicle). Right: percentage of neurons in the test set correctly (green boxes) or incorrectly (white boxes) classified into the two categories (<1 or >1).

Figure 6

Bumetanide treatment rescues aberrant behavioral responses to the benzodiazepine diazepam in adult Ts65Dn mice (A) Schematic representation of the protocol for the diverse experimental groups. Adult WT and Ts65Dn mice were treated with bumetanide (0.2 mg/kg i.p.) or the corresponding vehicle (2% DMSO in saline) for two weeks and then assessed in two different anxiety tests. On the day of testing, mice were pretreated with bumetanide or vehicle and then tested 30 min after diazepam (2 mg/kg i.p.) or saline administration. (B) Quantification of time spent in the light zone of the dark-light test for WT and Ts65Dn mice. (C) Quantification of the total number of transitions between the two zones of the dark-light test for WT and Ts65Dn mice. (D) Quantification of the time spent in the center of the arena during the open-field test for WT and Ts65Dn mice. (E) Quantification of the total distance traveled in the open field test for WT and Ts65Dn mice. In all boxplots, the small square indicates the mean, the central line illustrates the median, the box limits indicate the 25th and 75th percentiles, whiskers represent the 5th to 95th percentiles, and each dot indicates a value obtained from individual animals. ∗∗∗p < 0.001; Tukey’s post hoc test following two-way ANOVA or two-way ANOVA on ranked transformed data.

Figure 7

Mixed subpopulations of neurons with depolarizing or hyperpolarizing GABA signaling are present in human isogenic control and trisomic iPSC-derived neurons obtained from a person with DS (A) We analyzed trisomic neurons and corresponding isogenic control (euploid) neurons derived from iPSCs obtained from an individual with a mosaic form of DS. Representative calcium traces of iPSC-derived control and trisomic (DS) neurons pretreated with vehicle (0.01% DMSO) or the NKCC1 inhibitor bumetanide (10 μM) upon bath application of GABA (100 μM) at 60 days following plating for final differentiation. ΔF/F0 represents the change in the calcium intensity expressed as variation in fluorescence intensity upon GABA application, relative to the basal fluorescence intensity. (B) Quantification of the percentage of neurons showing depolarizing GABA responses in experiments as in (A). In the boxplot, the small square indicates the mean, the central line illustrates the median, the box limits indicate the 25th and 75th percentiles, the whiskers represent the 5th to 95th percentiles, and each dot indicates a value obtained from an individual coverslip. The numbers in parentheses indicate the number of analyzed coverslips for each experimental group (obtained from 5 independent neuronal differentiation experiments). ∗p < 0.05, ∗∗∗p < 0.001; Tukey post hoc test following two-way ANOVA. (C) Representative immunoblots for NKCC1 and KCC2 protein extracts from control and trisomic (DS) neurons. Actin was used as an internal standard. Full blots are shown in Figure S16A. (D) Quantification of average NKCC1 protein levels (±SEM; expressed as the percentage of control neurons, dotted line) in same experiments as in (C). Actin was used as an internal standard. Dots indicate values of individual coverslips (obtained from 2 neuronal independent differentiation experiments). ∗∗p < 0.01; Student’s t test. (E) Quantification of average KCC2 protein levels (±SEM) in the same experiments described in (D). Dots indicate values of individual coverslips (obtained from 2 neuronal independent differentiation experiments).