Molecular and developmental deficits in Smith-Magenis syndrome human stem cell-derived cortical neural models

Abstract

Smith-Magenis syndrome (SMS) is a genomic disorder caused by the deletion of a chromosomal region at 17p11.2. Individuals with SMS are frequently diagnosed with autism and have profound cortical deficits, including reduced cortex volume, mild ventriculomegaly, and epilepsy. Here, we developed human induced pluripotent stem cell (hiPSC)-derived neuronal models to understand how del(17)p11.2 affects cortical development. Hi-C experiments identified local fusion and global reorganization of topological domains, as well as genome-wide miswiring of chromatin three-dimensional (3D) interactions in SMS hiPSCs and 3D cortical organoids. Single-nucleus RNA sequencing of SMS cortical organoids identified neuropsychiatric disease-enriched transcriptional signatures and dysregulation of genes involved in catabolic and biosynthetic pathways, cell-cycle processes, and neuronal signaling. SMS cortical organoids displayed reduced growth, enlarged ventricles, impaired cell-cycle progression, and accelerated neuronal maturation. Through the use of a complementary hiPSC-derived 2D cortical neuronal model, we report that SMS cortical neurons exhibited accelerated dendritic growth, followed by neuronal hyperexcitability associated with reduced potassium conductance. Our study demonstrates that del(17)p11.2 disrupts multiple steps of human cortical development, from chromatin wiring, transcriptional regulation, cell-cycle progression, and morphological maturation to neurophysiological properties, and hiPSC-derived models recapitulate key neuroanatomical and neurophysiological features of SMS.

Keywords: CNV; Hi-C; RAI1; SMS; Smith-Magenis syndrome; autism; cortical organoids; human stem cells; iPSC; retinoic acid-induced 1.

Graphical abstract

Figure 1

del(17)p11.2 alters global and local chromatin contacts in SMS hiPSCs and hiPSC-derived cortical organoids (A) Experimental overview. Control (Ctrl) and SMS hiPSCs were differentiated into patterned cortical organoids by exposing them to growth factors and allowing them to mature for up to 75 days. (B) Circos plots showing statistically significant (FDR < 0.05) increase (left) and decrease (right) of genome-wide intra- and inter-chromosomal contacts in SMS hiPSCs (SMS 1–4) versus Ctrl hiPSCs (Ctrl 1–4). Each line represents a differential chromosomal contact change. Asterisks indicate chromosome 17. (C) Circos plots exhibiting statistically significant (FDR < 0.05) increase (left) and decrease (right) of genome-wide intra- and inter-chromosomal contacts in SMS cortical organoids (SMS 1–4) versus Ctrl cortical organoids (Ctrl 2–4). Each line represents a differential chromosomal contact change. Asterisks indicate chromosome 17. (D–G) Hi-C heatmaps showing intra- or inter-chromosomal contact frequency in SMS hiPSCs (SMS-2 as an example) versus Ctrl hiPSCs (1–4 combined) (D and E) or SMS-2 cortical organoids versus Ctrl cortical organoids (2–4 combined) (F and G). Each pixel represents one 250-kb region. (D) Centromeres of chromosome 17 are indicated by red triangles; red solid boxes and black lines indicate the boundaries of del(17)p11.2 (the SMS locus). The SMS locus and its surrounding regions (dashed square) are magnified at the bottom, with the red dashed box indicating increased contacts within del(17)p11.2 and the green dashed boxes indicating increased contacts between the SMS locus-flanking regions in SMS hiPSCs. The color scale goes from −2 (blue) to 0 (white) to +2 (red). (E) Heatmap showing that chromosome 1 lacks the changes in intra-chromosomal contacts observed in chromosome 17. (F and G) Similar patterns of chromosomal contact changes were found in SMS cortical organoids. (H and I) Topological domains on chromosome 17 in Ctrl hiPSCs (1–4 combined) and SMS 1–4 hiPSCs (individually shown) (H) and Ctrl cortical organoids (2–4 combined) and SMS 1–4 cortical organoids (individually shown) (I). Black triangles indicate TADs. Note that the newly formed large TADs in most SMS lines (indicated by yellow lines) encompassed del(17)p11.2 boundaries, which are indicated by blue boxes and dashed lines. (J and K) Venn diagrams highlight that differential TADs are associated with a significant subset of DEGs identified in SMS hiPSCs (J) and cortical organoids (K).

Figure 2

snRNA-seq characterization of Ctrl and SMS cortical organoid transcriptomes (A) Classification of cortical organoids against a developing human cortex transcriptome dataset. The snRNA-seq samples were pseudobulked, and the correlation coefficients (Pearson’s r) were color coded. Data obtained from Ctrl 1–3 and SMS 1–3 day-75 organoids (one library for each line, 3–5 cortical organoids per line). (B) Uniform manifold approximation and projection (UMAP) visualization showing different cell clusters, including the cycling radial glia (cycling RG), outer radial glia (oRG), truncating radial glia (tRG), intermediate progenitor cells (IPC), migrating neurons (mi-N), immature neurons (im-N), glutamatergic neurons, and GABAergic neurons. (C) UMAP projection of cell clusters among Ctrl (n = 18,392) and SMS (n = 23,713) lines. Individual samples are shown in Figure S6A. (D) Dotplot of feature expression. TOP2A and MKI67 mark cycling RG; GLI3, VIM, and TNC are expressed in radial glial lineages; HOPX marks outer radial glia (oRG-1 and oRG-2); S100β and NTRK2 are enriched in astroglia; tRG and tRG-like cells express CRYAB; LHX1 and CA8 are expressed in migrating neurons (mi-N); LHX8 and LHX9 are enriched in immature neurons (im-N); EOMES marks intermediate progenitor cells (IPC); DLX5, DLX6, GAD1, and GAD2 are GABAergic markers; GABA-N1 is enriched with a GABAergic projection neuronal marker EBF1; GABA-N2 is enriched with ERBB4, important for GABA release; DLX6-AS1 is expressed in all GABA-N and highest in GABA-N3; SLC17A7 (VGLUT1), TBR1, SATB2, and FEZF2 are expressed in glutamatergic neurons (Glu-N); NFIB marks corticofugal glutamatergic neurons and regulates the radial-glia-to-IPC transition. (E) UMAP plots showing the expression of key genes in different cell clusters (combining both genotypes): RAI1, expressed in all cell clusters; proliferative marker TOP2A; radial glia marker GLI3 and VIM; astroglia marker S100B; tRG marker CRYAB; migrating neuronal marker LHX1; immature neuronal marker LHX8; DLX6-AS1, expressed in GABAergic neurons (GABA-N); and SLC17A7 and FEZF2, expressed in glutamatergic neurons (Glu-N). (F) Bar chart showing the fractions of each cell cluster in Ctrl and SMS cortical organoids. SMS samples have a higher fraction of cyclin-RGs, ^∗FDR < 0.05, calculated with a sum-constrained Beta-binomial model.

Figure 3

Impaired metabolic, cell-cycle, and neuronal signaling pathways in SMS cortical organoids (A) Dotplot showing the number of DEGs (indicated by size) in SMS cortical organoid cell clusters. (B) UMAP plots showing reduced RAI1 expression in SMS cortical organoid cell clusters. (C) Volcano plot showing the downregulation of del(17)p11.2 genes in different cell clusters of SMS cortical organoids. (D) Volcano plot showing the altered expression of non-del(17)p11.2 DEGs in different cell clusters of SMS cortical organoids. DEGs in each cell cluster are shown in Figure S8C. (E) UMAP showing the downregulation of primate-specific cell-adhesion-related genes POTEI and POTEF in multiple cell clusters in SMS cortical organoids. (F) Nested pie charts showing parent and child GO over-represented terms obtained from DEG analysis (p_adj < 0.05) by comparing Ctrl and SMS cortical organoids. (G) Nested pie charts showing upregulated GO terms in SMS tRG and Glu-N1 clusters using gene set enrichment analysis (GSEA). (H) Heatmap showing over-representation of autism spectrum disorder (ASD), bipolar disorder (BD), major depressive disorder (MDD), and schizophrenia (SCZ) related genes in different cell cluster DEGs (FDR < 0.2). The color of the box shows the odds ratio for enrichment. The odds ratios were calculated by Fisher’s exact test.

Figure 4

SMS hiPSC-derived neuronal models showed impaired cell-cycle progression and mild ventriculomegaly (A) Bar graphs of diameter and area of Ctrl and SMS cortical organoids at day 75 of differentiation (divided into three bins by size). SMS organoids in the largest bin were significantly smaller than Ctrl organoids in the same bin. Each dot represents one organoid. Diameter >1,500 μm: U = 2,758, p < 0.0001; area >2 mm²: U = 1,622, p < 0.0001; two-tailed Mann-Whitney test. (B) Bar graphs of diameter and area of Ctrl and SMS cortical organoids at day 25 of differentiation (divided into three bins by size). SMS organoids in the largest bin were significantly smaller than Ctrl organoids in the same bin. Each dot represents one organoid. Diameter >700 μm: U = 7,513, p = 0.0079; area: U = 4,756, p = 0.0314; two-tailed Mann-Whitney test. (C) Representative images showing the overall morphology of Ctrl and SMS organoids at day 25 of differentiation. PAX6 is in green. Orange squares indicate neurogenic zones, which are magnified in (D). Scale bars, 100 μm. (D) Representative images of PAX6⁺ ventricles in Ctrl and SMS organoids at day 25 of differentiation. Ventricles are indicated by white dashed lines; PAX6 is in green. Scale bars, 50 μm. (E) Quantification of PAX6⁺ ventricle area in Ctrl and SMS organoids, including both absolute size (left) and size when normalized to the total organoid surface area (right). Each dot represents averaged data from one organoid, one image per organoid. Mean ventricle size: U = 14, p = 0.0002; ventricle size/total organoid area: U = 5, p < 0.0001; two-tailed Mann-Whitney test. (F) Quantification found that SMS organoids showed an increased number of PAX6⁺ ventricles. Each dot represents averaged data from one organoid, one image per organoid. U = 10.5, p < 0.0001; two-tailed Mann-Whitney test. (G) Quantification found that SMS organoids showed a decreased thickness of PAX6⁺ cells around the ventricles. Each dot represents averaged data from one organoid, one image per organoid. U = 4, p = 0.0019; two-tailed Mann-Whitney test. (H) Representative images of Ctrl (left) and SMS (right) cortical organoids after a 30-min (top) and 24-h (bottom) EdU pulse. EdU-labeled cells are in magenta, and Ki67-labeled cells are in green. White arrowheads indicate Ki67⁺EdU⁺ double-positive cells, and yellow arrowheads indicate Ki67⁻EdU⁺ cells. Scale bars, 20 μm. (I) Top: quantification of the percentages of EdU⁺ cells (U = 4, p = 0.014), Ki67⁺ cells (U = 16, p = 0.5338), EdU labeling index (EdU⁺Ki67⁺ cells/Ki67⁺ cells, U = 6, p = 0.0315), cell-cycle re-entry index (EdU⁺Ki67⁺ cells/EdU⁺ cells, U = 21, p > 0.9999), and cell-cycle exit index (EdU⁺Ki67⁻ cells/EdU⁺ cells, U = 21, p > 0.9999) after a 30-min EdU pulse. Bottom: quantification of the percentages of EdU⁺ cells (U = 27, p = 0.4081), Ki67⁺ cells (U = 23, p = 0.2268), EdU labeling index (EdU⁺Ki67⁺ cells/Ki67⁺ cells, U = 31, p = 0.6520), cell-cycle re-entry index (EdU⁺Ki67⁺ cells/EdU⁺ cells, U = 12, p = 0.019), and cell-cycle exit index (EdU⁺Ki67⁻ cells/EdU⁺ cells, U = 12, p = 0.019) after a 24-h EdU pulse. Each dot represents averaged data from one organoid, 1–3 images per organoid. U and p values by two-tailed Mann-Whitney test. (J) Barplot of cell-cycle profile of SMS and Ctrl NPCs. Three biological replicates per cell line; each dot represents one sample with 4,000–20,000 cells. G₁ phase: t = 43.99, df = 11, p < 0.0001; S phase: t = 43.40, df = 11, p < 0.0001; G₂/M phases: t = 23.41, df = 11, p < 0.0001; one-sample t test. (K) Representative images of Ctrl and SMS NPCs immunostained for Ki67 (cyan). Scale bars, 50 μm. (L) Quantification showing a lower percentage of SMS NPCs expressing Ki67. n = 32 images per genotype with eight images per cell line; each point represents one image colored based on the cell line. U = 23, p < 0.0001, two-tailed Mann-Whitney tests. (M) Expression of γ-H2AX, a DNA-damage marker, in SMS and Ctrl NPCs. Left: representative confocal images of γ-H2AX (cyan) staining in Ctrl and SMS NPCs. Bottom left: outlined nuclei for γ-H2AX quantification. Scale bars, 10 μm. Right: Quantification of γ-H2AX foci per cell in SMS and Ctrl NPCs. n = 400 cells per genotype with 100 cells per cell line. Each dot represents one cell. t = 9.065, df = 798, p < 0.0001; unpaired t test. (N) Expression of p-53BP1 in SMS and Ctrl NPCs. Left: representative images of p-53BP1 (cyan) staining in Ctrl and SMS NPCs. Bottom left: outlined nuclei for p-53BP1 quantification. Scale bars, 10 μm. Right: quantification of p-53BP1 foci per cell in SMS and Ctrl NPCs. n = 400 cells per genotype with 100 cells per cell line. Each dot represents one cell. t = 7.964, df = 798, p < 0.0001; unpaired t test.

Figure 5

SMS hiPSC-derived cortical neurons showed aberrant gene expression and dendritic growth (A) Manhattan plot displaying the genomic landscape of significantly downregulated genes on the chromosome 17 p-arm in SMS cortical neurons (compared to Ctrl neurons) based on −log₁₀ transformed p_adj derived from a one-sided Wald test. The x axis represents the genomic position, and the y axis corresponds to the significance level of differential expression. The blue dots represent individual genes, with their positions on the x axis representing their location on chromosome 17p and their heights on the y axis indicating the significance of their differential expressions. The dark horizontal lines at the bottom indicate chromosomes, with the green segments corresponding to the regions subjected to del(17)p11.2 in SMS cortical neurons. (B) Volcano plot showing the global transcriptomic changes by comparing SMS with Ctrl cortical neurons. Each dot represents a gene. The log₂ fold change of each gene is represented on the x axis, and the −log₁₀ of its p_adj is on the y axis. Upregulated genes in SMS cortical neurons with p_adj less than 0.1 are indicated by red dots. Downregulated genes in SMS cortical neurons with p_adj less than 0.1 are indicated by blue dots. The gray line indicates p_adj = 0.1. (C) GO analysis of SMS cortical neurons. The GO terms for downregulated genes (in blue) and upregulated genes (in red) and the respective −log₁₀(p_adj) are shown. (D) Violin plots of protocadherin gene expression from bulk RNA-seq in SMS and Ctrl cortical neurons. Each dot represents a sample. p_adj calculated by Wald test. (E) Violin plots of expression of ion-exchanger genes from bulk RNA-seq in SMS and Ctrl cortical neurons. Each dot represents a sample. p_adj calculated by Wald test. (F) Soma volume of SMS cortical neurons compared to Ctrl at 4 WPD. Left: representative 4-WPD soma 3D images of Ctrl and SMS cortical neurons transduced with myrGFP lentivirus. Scale bars, 50 μm. Right: barplot of soma volume of Ctrl and SMS neurons at 4 WPD (n = 400 cells per genotype, with 100 cells per cell line; each dot represents one soma and is colored based on cell line). t = 3.412, df = 798, p = 0.0007; unpaired t test. (G) Representative 3D reconstituted neuron images of myrGFP-transduced Ctrl and SMS cortical neurons at 4 WPD. Scale bars, 50 μm. (H) Quantification of total dendrite length in myrGFP⁺ SMS and Ctrl cortical neurons at 4 WPD. n = 40 neurons per genotype with 10 neurons per cell line. Each dot represents one neuron. U = 443, p = 0.0005; two-tailed Mann-Whitney test. (I) Sholl analysis of Ctrl and SMS cortical neurons at 4 WPD. n = 80 neurons per genotype with 20 neurons per cell line. Presented as differences between means (line) ± SEM (shade). Radius 10: t = 3.514, df = 158, p_adj = 0.01247; radius 20: t = 5.203, df = 158, p_adj = 0.000014; radius 30: t = 4.404, df = 158, p_adj = 0.000449; radius 40: t = 3.517, df = 158, p_adj = 0.012467; multiple t tests. (J) Quantification of Sholl critical radius (left) and maximum crossings (right) of Ctrl and SMS cortical neurons at 4 WPD. n = 80 neurons per genotype with 20 neurons per cell line. Each dot represents one neuron. Critical radius: U = 2,743, p = 0.1022; maximum crossing: U = 1,945, p < 0.0001; two-tailed Mann-Whitney test. (K) Quantification of excitatory synapse density (VGLUT1⁺ and PSD95⁺) in Ctrl and SMS cortical neurons at 6 WPD. Left: representative images of Ctrl and SMS neurons stained with MAP2 (gray), VGLUT1 (magenta), and PSD95 (cyan). Scale bars, 10 μm. Right: quantification of the number of puncta (VGLUT1⁺ and PSD95⁺) per μm. Each dot represents one 50-μm segment, n = 40 segments per genotype with 10 segments per cell line. U = 365.5, p < 0.0001; two-tailed Mann-Whitney test.

Figure 6

SMS hiPSC-derived cortical neurons showed altered AP properties and reduced potassium conductance that favored hyperexcitability (A) Representative image of whole-cell patched hiPSC-derived cortical neuron. Scale bar, 20 μm. (B) Percentage of neurons firing spontaneous APs in SMS versus Ctrl neurons (recorded at I_clamp, gap-free mode). Chi-squared = 4.599, df = 1, p = 0.032; chi-squared analysis. Cell numbers for each AP firing category are in parentheses. (C) Evoked APs were elicited from a holding potential of −70 mV followed by 10-pA increment steps to the voltage close to −50 mV. The representative traces of AP number at rheobase, double rheobase (gray), and the voltage close to −50 mV (light gray) are presented. Right: AP frequency of Ctrl and SMS cortical neurons. t = 2.188, df = 66.03, p = 0.0322; Welch’s t test. (D) AP properties analyzed from the first AP evoked by a ramp test: 1-min ramp (nA/ms) followed by a hyperpolarized step (∼−90 mV). From left to right, the representative evoked AP of Ctrl (blue) and SMS (red) cortical neurons. AP threshold: U = 367, p = 0.0063; AP amplitude: U = 361, p = 0.005; AP half-width: U = 392, p = 0.0156; decay slope: U = 360, p = 0.0297; two-tailed Mann-Whitney tests. (E) AP property analyses at a potential of −60 mV. From left to right, the representative AP of Ctrl (blue) and SMS (red) cortical neurons. AP threshold: U = 347.5, p = 0.0126; AP amplitude: U = 227, p < 0.0001; AP half-width: U = 174, p < 0.0001; decay slope: U = 174, p < 0.0001; two-tailed Mann-Whitney tests. (F) Representative traces of whole-cell voltage clamp of Ctrl (left) and SMS (right) cortical neurons. The red circles and lines indicate the fast and slow voltage-gated components, respectively. (G and H) I-V curves show that SMS cortical neurons exhibited decreased voltage-gated potassium fast component (G) and slow component (H) current density. K_fast: genotype = F(1,1151) = 26.25, p < 0.0001; K_slow: genotype = F(1,1278) = 29.55, p < 0.001; two-way ANOVA with Sidak’s multiple comparison tests. (I) I-V curve shows that the current density of voltage-gated sodium components was similar in SMS and Ctrl cortical neurons. Na_v: genotype = F(1,1224) = 2.775, p = 0.096; two-way ANOVA.