Dysregulation of miRNA expression and excitation in MEF2C autism patient hiPSC-neurons and cerebral organoids

Abstract

MEF2C is a critical transcription factor in neurodevelopment, whose loss-of-function mutation in humans results in MEF2C haploinsufficiency syndrome (MHS), a severe form of autism spectrum disorder (ASD)/intellectual disability (ID). Despite prior animal studies of MEF2C heterozygosity to mimic MHS, MHS-specific mutations have not been investigated previously, particularly in a human context as hiPSCs afford. Here, for the first time, we use patient hiPSC-derived cerebrocortical neurons and cerebral organoids to characterize MHS deficits. Unexpectedly, we found that decreased neurogenesis was accompanied by activation of a micro-(mi)RNA-mediated gliogenesis pathway. We also demonstrate network-level hyperexcitability in MHS neurons, as evidenced by excessive synaptic and extrasynaptic activity contributing to excitatory/inhibitory (E/I) imbalance. Notably, the predominantly extrasynaptic (e)NMDA receptor antagonist, NitroSynapsin, corrects this aberrant electrical activity associated with abnormal phenotypes. During neurodevelopment, MEF2C regulates many ASD-associated gene networks, suggesting that treatment of MHS deficits may possibly help other forms of ASD as well.

Fig. 1. MHS hiPSCs in 2D cultures generate more astrocytes and fewer neurons, but with increased spontaneous activity compared to controls (Ctrls).

A Representative images of GFAP, MAP2, Hoechst (to label cell nuclei), and merged images in 4-week-old hiPSC-derived cultures. Patient-relevant mutation-bearing hiPSCs abbreviated MHS-P1 through P4. The genetic background of MHS-P2 is isogenic to Ctrl1. Scale bar, 100 µm. B Quantification of GFAP intensity for each MHS line normalized to Hoechst and relative to Ctrl1 (at left). Grouped analysis of GFAP expression for Ctrl vs. all MHS patients (at right). C Quantification of MAP2 intensity for each MHS line normalized to Hoechst and relative to Ctrl1 (at left). Grouped analysis of MAP2 expression for Ctrl vs. all MHS patients (at right). D Representative immunoblot of MAP2 expression with GAPDH as loading control. E Quantification of MAP2 protein expression for each MHS line. F Grouped analysis of MAP2 expression showing a decrease in protein levels in MHS lines vs. Ctrl1. G Comparison of MEF2 luciferase reporter gene activity normalized to Renilla at day 7 of differentiation of MHS hiPSCs compared to controls (at left), and at various time points of neuronal differentiation for each MHS line vs. Ctrls (at right). Statistical analysis was performed on area under the curve (AUC). Values are mean ± SEM. Sample size listed above bars (number of fields for immunofluorescence panels in ≥3 independent experiments). Individual datapoints shown on bar graphs wherever possible in this and subsequent figures (for sample sizes <25). *^,#,†p < 0.05, **^,##,††p < 0.01, ***^{, ###, †††}p < 0.001, ****^{,####, ††††}p < 0.0001 by Student’s t-test for pairwise comparisons or ANOVA with Dunnett’s post-hoc test for multiple comparisons to single Ctrl and with Sidak’s test for comparison to multiple Ctrls; comparison to Ctrl1 (*), Ctrl2 (^#), Ctrl3 (^†) (see Methods). Ctrl1 and its isogenic line MHS-P2 designated by (~) in this and subsequent figures.

Fig. 2. ChIP-seq and RNA-seq analyses show MEF2C effects on hiPSC-derived cell types in 2D cultures.

A Top gene ontology (GO) terms for hits found in ChIP-seq analysis of the 198 MEF2C binding targets in hNPCs. B List of miRNAs identified that contain binding sites for MEF2C. C Relative gene expression levels of miRNA identified by ChIP-seq in control hNPCs and hNPCs expressing constitutively active MEF2 containing a VP16 transactivation domain (MEF2CA). D Relative gene expression of miRNA in Ctrl and MHS patient hiPSC-derived cells after 2 weeks in culture. E Relative gene expression of miRNA in MHS patient hiPSC-derived cells exposed to ‘miR-4273 mimic’ or control miR after 2 weeks in culture. F MAP2 neuronal marker expression in MHS hiPSC-derived cells expressing ‘miRNA mimic’ compared to non-target control mimic after 2 weeks in culture. G GFAP astrocytic marker expression in MHS hiPSC-derived cells expressing ‘miRNA mimic’ compared to non-target control mimic after 2 weeks in culture. H Schematic diagram of miRNA effect on neurogenesis and gliogenesis. I Top GO biological process terms based on differentially-expressed genes (DEGs) by RNA-seq after 5 weeks in culture in MHS patient hiPSC-neurons vs. Ctrl showing neuronally-enriched pathways. J Top DEGs from RNA-seq showing higher expression in MHS hiPSC-neurons vs. Ctrl. K Top DEGs from RNA-seq showing lower expression in MHS hiPSC-neurons vs. Ctrl. L Bar Plot of normalized counts for the various genes acquired from RNA seq data on control lines (Ctrl1, 2 and 4) and on patient lines (MHSP1-4). M NRXN3 mRNA expression in each MHS patient vs. Ctrl. N NRXN3 mRNA expression in all MHS patients combined vs. Ctrl. Data are mean + SEM. Sample sizes (n) are listed above bars from at least 3 independent experiments. *p < 0.05, **p < 0.01 by ANOVA with Dunnett’s post-hoc test for multiple comparisons or by two-tailed Student’s t-test for single comparisons.

Fig. 3. MHS hiPSC-derived cerebrocortical neurons show increased excitation and decreased inhibition in 2D cultures.

A Recordings of spontaneous action potential (sAP) at resting membrane potential (RMP). B Quantification of sAP frequency in neurons from each MHS patient (P1-P4) compared to each control (Ctrl1, Ctrl2, Ctrl3) on left; grouped analysis on right. C Representative traces of glutamate- and GABA-evoked currents (each at 100 µM). D, E Quantification of glutamate and GABA current density. F Ratio of glutamate to GABA current densities. G Representative patch-clamp recordings of evoked AMPAR-EPSCs at holding potential (V_h = −70 mV) and NMDAR-EPSCs (V_h = +60 mV). H, I Input-output curves of evoked AMPAR-EPSCs and NMDAR-EPSCs. J, K Quantification of peak current amplitude of evoked AMPAR-EPSCs and NMDAR-EPSCs from each patient. L Quantification of ratio of peak AMPA/NMDA EPSCs from individual neurons for Ctrl and each MHS patient (above); grouped analysis (below). Data are mean ± SEM. Number of neuronal recordings (n) listed above bars from at least 4 experiments in each case. *^,#p < 0.05, **^,##p < 0.01; ***^,###p < 0.001, ****^,####p < 0.0001 by ANOVA for multiple comparisons with Sidak’s post-hoc test in B, C, D or Dunnett’s post-hoc test in H and I; comparison to Ctrl1 (*), Ctrl2 (^#), Ctrl3 (^†) (see Methods).

Fig. 4. MHS hiPSC-derived cerebrocortical neurons exhibit disrupted synaptic transmission in 2D cultures.

A Representative mEPSCs recorded at –70 mV in the presence of 1 µM TTX from Ctrl1 and MHS hiPSC-neurons in culture for 5 weeks. B, C Cumulative probability and quantification of mean mEPSC amplitude and interevent interval (inversely related to frequency). Cumulative probability of MHS mEPSC interevent interval was significantly decreased compared to Ctrl (p < 0.0001 by Kolmogorov–Smirnov test). D Representative mIPSCs recorded at 0 mV. E, F Cumulative probability and quantification of mean mIPSC amplitude and interevent interval. Cumulative probability of MHS mIPSC interevent interval was significantly increased compared to Ctrl (p < 0.0001 by Kolmogorov–Smirnov test). For bar graphs in B, C, E and F, responses of each MHS patient’s neurons vs. each control shown on left, with grouped MHS patient vs. controls shown on right. G Representative images of β3-tubulin, VGLUT1, VGAT, and Hoechst in Ctrl1 and MHS hiPSC-neurons. Scale bar, 100 µm. H Quantification of VGLUT1 in various MHS hiPSC-neurons compared to Ctrl1. I Quantification of VGAT in various MHS patient neurons compared to Ctrl1. J Quantification of the VGLUT1/VGAT ratio in various MHS patients. K Summary of VGLUT1/VGAT in Ctrl1 vs. MHS as a group. Data are mean + SEM. Number of neuronal recordings or imaged fields (n) listed above bars from at least 4 experiments in each case. *^,#,†p < 0.05, **^,##,††p < 0.01, ***^{,###,†††}p < 0.001, ****^{,####,††††}p < 0.0001 by ANOVA with Sidak’s post-hoc test in B, C, E and F for comparison to Ctrl1 (*), Ctrl2 (^#), Ctrl3 (^†); Dunnett’s post-hoc test for multiple comparisons with single Ctrl (see Methods).

Fig. 5. NitroSynapsin normalizes spontaneous calcium transients and neural network activity in MHS hiPSC-derived cerebrocortical neurons in 2D cultures.

A Spontaneous neuronal calcium transients recorded from individual Ctrl1 and MHS hiPSC-neurons loaded with Fluo-4 AM. B Quantification of Ca²⁺ transient frequency for events with rise times <200 ms (individual Ctrl1 and MHS hiPSC-neurons responses in upper panel, and grouped responses in lower panel). C Representative calcium traces showing decrease in spontaneous calcium transient frequency after application of 10 µM NitroSynapsin. D Quantification of calcium transient frequency before and after application of NitroSynapsin. E Quantification of difference in normalized fluorescence (ΔF/F_0Drug–ΔF/F_0Control) as area under the curve (AUC) in response to NitroSynapsin. F Representative heat maps and raster plots of MEA recordings from Ctrl and MHS hiPSC neurons before (w/o) and after treatment with 5 µM NitroSynapsin. Boxes outline examples of network bursts. G–J Quantification of MEA recordings by mean firing rate, electrode burst frequency (representing bursting of individual neurons), network burst frequency (representing bursting of the entire neural network), and synchronous firing. Data are mean ± SEM. Sample size listed above bars represents number of cells (n) analyzed in 5–10 independent experiments. In D, E, G, H, I and J, responses of each MHS patient’s 2D neurons vs. each control shown on left, with grouped MHS patient vs. controls shown on right. *^,#,†p < 0.05, **^,##,††p < 0.01, ^{***, ###, †††}p < 0.001, ****^{,####, ††††}p < 0.0001 by ANOVA by Sidak’s post-hoc test for comparison to Ctrl1 (*) or to Ctrl2 (^#), or within a group (^†) between NitroSynapsin treatment vs. without (w/o) treatment; for panel D, comparison was made by non-parametric Kruskal–Wallis test (see Methods).

Fig. 6. scRNA-seq analysis demonstrates reproducibility of Ctrl and MHS cerebral organoids at 3 months of age.

Uniform Manifold Approximation and Projection (UMAP) analysis of isogenic Ctrl1 and MHS hiPSC-derived cerebral organoids by cell-type (A) and by genotype (B). C Bar charts showing relative cell-type composition of each individual organoid captured from scRNA-seq. D Violin plots showing the distribution of expression of NRXN3 by cell type. Note the increased expression of NRXN3 in MHS GABAergic neurons, in glutamatergic MHS neurons, and in progenitors. E Heatmap of key marker genes to annotated clusters.

Fig. 7. NitroSynapsin abrogates hypersynchronous burst activity in MHS cerebral organoids.

A Representative heat maps and single traces from Ctrl1 and MHS hiPSC-derived cerebral organoids in individual MEA wells at 3–4 months of age. B Representative raster plots of MEA recordings in Ctrl and MHS cerebral organoids. Boxes outline examples of network bursts. C Representative raster plots and heat maps of MEA recordings in Ctrl and MHS cerebral organoids after treatment with NitroSynapsin (NitroSyn). D–I Quantification of MEA mean firing rate, network burst frequency, and synchrony index. Each MHS patient’s cerebral organoids vs. each control shown in D, F and H; or grouped MHS patient cerebral organoids vs. controls shown in E, G and I. Data are mean ± SEM. Sample size is listed above bars from 6 to 7 separate cerebral organoids recorded for each genotype. *^,#,† p < 0.05, **^,##,††p < 0.01, ***^{,###, †††}p < 0.001, ****,^{####,††††}p < 0.0001 by ANOVA with Sidak’s post-hoc test for comparison to Ctrl1 (*) or to Ctrl2 (^#), or within a group (^†) between NitroSynapsin treatment vs. without (w/o) treatment (see Methods).

Fig. 8. Aberrant neurogenesis/gliogenesis and excitation in MEF2C autism patient hiPSC-neurons.

Schematic diagram showing that MHS-hiPSCs generate more astrocytes and fewer cerebrocortical neurons. Compared to control (Ctrl), the MHS neuronal population consists of fewer inhibitory neurons and excitatory neurons, but with more VGLUT1 vesicles, resulting in increased presynaptic glutamate release, increased postsynaptic intracellular Ca²⁺ levels, and hence increased excitability. The novel NMDAR antagonist NitroSynapsin ameliorates this hyperactivity.