Aberrant neurodevelopment in human iPS cell-derived models of Alexander disease

Abstract

Alexander disease (AxD) is a rare and severe neurodegenerative disorder caused by mutations in glial fibrillary acidic protein (GFAP). While the exact disease mechanism remains unknown, previous studies suggest that mutant GFAP influences many cellular processes, including cytoskeleton stability, mechanosensing, metabolism, and proteasome function. While most studies have primarily focused on GFAP-expressing astrocytes, GFAP is also expressed by radial glia and neural progenitor cells, prompting questions about the impact of GFAP mutations on central nervous system (CNS) development. In this study, we observed impaired differentiation of astrocytes and neurons in co-cultures of astrocytes and neurons, as well as in neural organoids, both generated from AxD patient-derived induced pluripotent stem (iPS) cells with a GFAP^R239C mutation. Leveraging single-cell RNA sequencing (scRNA-seq), we identified distinct cell populations and transcriptomic differences between the mutant GFAP cultures and a corrected isogenic control. These findings were supported by results obtained with immunocytochemistry and proteomics. In co-cultures, the GFAP^R239C mutation resulted in an increased abundance of immature cells, while in unguided neural organoids and cortical organoids, we observed altered lineage commitment and reduced abundance of astrocytes. Gene expression analysis revealed increased stress susceptibility, cytoskeletal abnormalities, and altered extracellular matrix and cell-cell communication patterns in the AxD cultures, which also exhibited higher cell death after stress. Overall, our results point to altered cell differentiation in AxD patient-derived iPS-cell models, opening new avenues for AxD research.

Keywords: Alexander disease; GFAP; iPS cells; neural organoids.

FIGURE 1

Overview of the scRNA‐seq dataset derived from astrocyte‐neuron co‐cultures. (a) Experimental design explaining co‐culture compositions and conditions entering the scRNA‐seq analysis. (b) UMAP plot showing three main cellular populations and their clusters across all six samples. AxD cluster: n = 2558, iAs: n = 6855, iNs: n = 4604 (c) Dotplot with selected marker genes of the cluster of the AxD cluster, iAs, and iNs. Wilcoxon test with Bonferroni correction was used to determine marker genes. (d, e) Proportional barplots showing higher abundance of the population of the AxD cluster and less mature cells in astroAxD/neuroC and astroAxD/neuroAxD samples. Oxygen–glucose deprivation (OGD) challenge did not significantly reflect in population proportions. (f) Violin plot showing that GFAP was expressed predominantly in the ASTRO 2 cluster. (g) Overrepresentation analysis results showing top five upregulated and downregulated GO terms distinguishing the AxD cluster from iAs and iNs (p _adj < 0.1, false discovery rate (FDR) was used to correct for multiple comparisons). astroC/neuroC, corrected co‐cultures; astroAxD/neuroC, co‐cultures with AxD astrocytes and corrected neurons; astroAxD/neuroAxD, co‐culture with AxD astrocytes and neurons; AxD, Alexander disease; CTRL, control without OGD challenge; GO, gene ontology; iAs, induced astrocytes; iNs, induced neurons; OGD, oxygen–glucose deprivation.

FIGURE 2

Immunocytochemistry identified less mature cells in astroAxD/neuroC co‐cultures. (a) Co‐cultures of iAs and iNs were labeled with antibodies against MAP2 (red), vimentin (green), and nuclei were visualized with DAPI (blue). Arrows point to undifferentiated cells; scale bar, 50 μm. (b) Number of undifferentiated cells, MAP2⁺, and vimentin⁺ cells. n = 5, 4, and 5 independent sets of astroC/neuroC and astroAxD/neuroC cultures. Wilcoxon test (undifferentiated cells) and t‐test (MAP2, vimentin) were used for statistical comparison. (c) Representative image of GFAP⁺ astrocytes (GFAP, green; DAPI, blue); scale bar, 20 μm. (d) Number of GFAP⁺ astrocytes, n = 3 independent sets of astroC/neuroC and astroAxD/neuroC cultures. T‐test was used for statistical comparison. (e) Circularity and perimeter of GFAP⁺ signal for individual astrocytes (astroC/neuroC: n = 55, astroAxD/neuroC: n = 55). Wilcoxon test was used for statistical comparison. ns: not significant, **p ≤ 0.01; ****p ≤ 0.0001. In b, d, and e Shapiro–Wilk test was used for normality assessment. astroC/neuroC, corrected co‐cultures; astroAxD/neuroC, co‐cultures with AxD astrocytes and corrected neurons; iAs, induced astrocytes; iNs, induced neurons.

FIGURE 3

Results of differential expression analysis of iAs and iNs and cell–cell interaction analysis. (a) Volcano plot showing DEGs between astroC/neuroC and astroAxD/neuroC astrocytes and neurons (|log₂FC| > 0.65 and p _adj < 0.05, t‐test was used to calculate DEGs, Bonferroni correction was used to correct for multiple comparisons). (b) Heatmap shows expression of DEGs shared between comparisons of astroC/neuroC with astroAxD/neuroC and with astroAxD/neuroAxD. (c) Differential number of interactions (top) and differential interaction strength (bottom) between astroC/neuroC and astroAxD/neuroC defined by CellChat analysis (red—increase in astroAxD/neuroC, blue—decrease in astroAxD/neuroC). (d) Information flow chart from the CellChat analysis of cell–cell interactions presents the pathways identified in both conditions. Significantly dysregulated pathways (paired Wilcoxon test, p‐value < 0.05) are highlighted in colors (red—increase in astroAxD/neuroC, blue—decrease in astroAxD/neuroC). (e) Heatmap divided to outgoing (ligands) and incoming (receptors) signaling patterns shows selected neurodevelopmental and astrogenesis‐related pathways that were changed in astroAxD/neuroC compared to astroC/neuroC co‐cultures. astroC/neuroC, corrected co‐cultures; astroAxD/neuroC, co‐cultures with AxD astrocytes and corrected neurons; astroAxD/neuroAxD, co‐culture with AxD astrocytes and neurons; AxD, Alexander disease; DEGs, differentially expressed genes; iAs, induced astrocytes; iNs, induced neurons.

FIGURE 4

The OGD challenge enhanced the effect of the GFAP mutation in the AxD cluster. (a) A scheme of the timeline of the OGD challenge. (b) LDH assay measuring cell death showed increased LDH release in OGD astroAxD/neuroC co‐culture compared to OGD astroC/neuroC and astroAxD/neuroC without stress, n = 6 sets of cultures of CTRL and OGD, astroC/neuroC and astroAxD/neuroC. T‐test was used for statistical comparison, Shapiro–Wilk test was used for normality assessment, *p < 0.05. (c) Heatmap summarizing numbers of DEGs showing that the AxD cluster was the most affected by the OGD challenge (|log₂FC| > 0.65 and p _adj < 0.05; t‐test with Bonferroni correction). (d, e) Information flow chart showing pathways affected by the OGD challenge in astroC/neuroC (d) and astroAxD/neuroC (e) co‐cultures. Significant change is marked by colors: red—increased in OGD, blue—decreased in OGD; paired Wilcoxon test, p‐value < 0.05. Arrows highlight pathways included in panels f–h. (f) Heatmap of selected pathways that are enriched in astroAxD/neuroC co‐cultures after OGD challenge. (g) Ligand‐receptor pairs of three pathways uniquely enriched in astroC/neuroC co‐cultures after OGD. Clusters of senders and receivers are distinguished by colors. (h) Detailed heatmap of PDGF signaling pathway shows roles of individual clusters in this signaling based on statistical and network analysis by CellChat. astroC/neuroC, corrected co‐cultures; astroAxD/neuroC, co‐cultures with AxD astrocytes and corrected neurons; AxD, Alexander disease, CTRL, control co‐culture without stress; DEGs, differentially expressed genes; iAs, induced astrocytes, iNs, induced neurons; LDH, lactate dehydrogenase; L‐R pairs, ligand‐receptor pairs; OGD, oxygen–glucose deprivation.

FIGURE 5

GFAP mutant unguided neural organoids partially diverged their differentiation into other than neuroectodermal lineage and did not develop astrocyte‐like cells. (a) UMAP plot showing various cell types that were identified in 165 days old unguided organoids, split to conditions. Legend includes a barplot depicting abundance (absolute cell counts) of clusters in each condition (top bar = CTRL, bottom bar = AxD). (b) Selected markers highlighting clusters of neuroectodermal lineage, as well as off‐target populations that were overrepresented in AxD organoids. (c) DEA was performed on clusters of radial glia (in a highlighted with a magenta rectangle) and neuronal clusters (in a highlighted with a green rectangle) comparing CTRL and AxD. Volcano plots show DEGs (|log₂FC| > 0.65 and p _adj < 0.05; t‐test with Bonferroni correction). GO overrepresentation analysis was performed on the DEGs, with FDR used to correct for multiple comparisons and p _adj < 0.1 used as significance threshold for the results. (d) Immunofluorescent microscopy images showing a Hoechst (nuclei), GFAP, and S100B staining of Day 165 CTRL and AxD unguided organoids. Scale bar, 50 μm. AxD, Alexander disease; CTRL, control; DEA, differential expression analysis; DEGs, differentially expressed genes; FDR, false discovery rate; GO, gene ontology; log₂FC, log₂ fold change; p _adj, adjusted p‐value; pre‐OPCs, pre‐oligodendrocyte progenitor cells.

FIGURE 6

GFAP mutant cortical organoids showed delayed development and were enriched for mesoderm‐derived cell populations. (a) UMAP plot showing various cell types that were identified in 165 days old cortical organoids, split to conditions. Legend includes barplot depicting abundance (absolute cell counts) of clusters in each condition (top bar = CTRL, bottom bar = AxD). (b) Selected markers highlighting clusters of neuroectodermal lineage, as well as off‐target populations that were overrepresented in AxD organoids. (c) Expression plot of telencephalic marker FOXG1 showing only limited expression in neuronal clusters in the AxD organoids. (d) DEA was performed on the cluster of oRG/astrocytes (in a highlighted with magenta rectangle) comparing CTRL and AxD. Volcano plot shows DEGs (|log₂FC| > 0.65 and p _adj < 0.05; t‐test with Bonferroni correction). GO overrepresentation analysis was performed on DEGs with FDR used to correct for multiple comparisons and p _adj < 0.1 used as significance threshold for the results. Top five upregulated and downregulated terms are shown in the dotplot. (e) DEGs (p _adj < 0.05) and DEPs (FDR < 0.05) were plotted against each other, with fitted line, Pearson's correlation coefficient (r) and statistical significance. Myosin genes are highlighted in red, selected neuronal genes are highlighted in blue, and other genes of interest are highlighted in green. (f) Normalized intensities of selected proteins were compared in CTRL and AxD organoids using t‐test; ns: not significant, *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001. (g) Immunofluorescent microscopy images showing SOX9, DCX, FOXG1, and TNNT2 for Day 165 CTRL and AxD cortical organoids. Nuclei are stained by Hoechst; scale bar, 100 μm. (h) Quantification of immunofluorescent signal of SOX9 and FOXG1 proteins in ratio to nuclei staining with Hoechst. Data points represent three images taken from three individual organoids. DCX integrated density was measured from four organoids per genotype and presented as integrated density per μm². No difference in the Hoechst signal was observed. Shapiro–Wilk test was used to determine normal distribution of the data Wilcoxon test (SOX9, FOXG1) and t‐test (DCX) were used to compare CTRL and AxD samples. AxD, Alexander disease; CTRL, control; DEA, differential expression analysis; DEGs, differentially expressed genes; DEPs, differentially expressed proteins; FDR, false discovery rate; GO, gene ontology; log₂FC, log₂ fold change; oRG, outer radial glia; p _adj, adjusted p‐value; pre‐OPCs, pre‐oligodendrocyte progenitor cells.