Human Astrocyte Maturation Captured in 3D Cerebral Cortical Spheroids Derived from Pluripotent Stem Cells

Abstract

There is significant need to develop physiologically relevant models for investigating human astrocytes in health and disease. Here, we present an approach for generating astrocyte lineage cells in a three-dimensional (3D) cytoarchitecture using human cerebral cortical spheroids (hCSs) derived from pluripotent stem cells. We acutely purified astrocyte-lineage cells from hCSs at varying stages up to 20 months in vitro using immunopanning and cell sorting and performed high-depth bulk and single-cell RNA sequencing to directly compare them to purified primary human brain cells. We found that hCS-derived glia closely resemble primary human fetal astrocytes and that, over time in vitro, they transition from a predominantly fetal to an increasingly mature astrocyte state. Transcriptional changes in astrocytes are accompanied by alterations in phagocytic capacity and effects on neuronal calcium signaling. These findings suggest that hCS-derived astrocytes closely resemble primary human astrocytes and can be used for studying development and modeling disease.

Keywords: RNA-seq; astrocyte; cerebral cortex; hCS; human; iPSC; maturation; organoids; spheroids; transcriptome.

FIGURE 1. Purification of Astrocytes from hCS

(A) Schematic for generating hCS from iPSCs. Individual colonies are enzymatically dissociated and suspended in low-attachment plates to form neural spheroids. (B) GFAP immunostaining of astrocytes in a 10 μm hCS cryosection at 363 days in culture. Scale bar = 50 μm. (C) GFAP staining of an astrocyte isolated from a hCS at 295 day and cultured for 3 days in monolayer. Scale bar = 30 μm. (D) hCS can be immunopanned after single cell dissociation to isolate neurons with an anti–Thy1 antibody and astrocytes with an anti–HepaCAM antibody. Representative images are shown for cultured samples of (E) unpurified cells, (F) Thy-1 panned neurons, and (G) HepaCAM panned astrocytes. Cells are immunostained with an anti–TUJ1 antibody (red) for neurons and anti–GFAP antibody (cyan) for astrocytes. Scale bar = 150 μm. (H) RNA-seq expression data showing enrichment of neuronal and astrocyte-specific genes in bulk Thy1– and HepaCAM– immunopanned samples. (Left) Variability in immunopanned samples from a single iPSC line across multiple differentiations (HepaCAM: 3–15 hCS per time-point from one iPSC line in 11 differentiation experiments; Thy1: 3–15 hCS per time-point from one iPSC line from 4 differentiations experiments). (Right) Variability in immunopanned samples across multiple iPSC lines (HepaCAM: 3–15 hCS per time-point from 3 iPSC lines in 1–11 differentiations per line; Thy1: 3–15 hCS per time-point from 2 iPSC lines in 4 differentiations per line). (I) PCA using the top 2 principal components and showing bulk RNA-seq of primary human fetal and adult CNS cell type samples along with hCS-derived neurons and astrocytes. The top 5000 over-dispersed genes were used for analysis. hCS-derived cells are labeled by in vitro differentiation stage (d, day); 3–15 hCS were collected from 2 iPSC lines across 18 differentiation experiments.

FIGURE 2. Maturation of hCS-derived Astrocyte Lineage Cells

(A) Schematic predicting the time-course of astrogenesis and maturation in hCS. (B) (Top) Phase images of dissociated astrocytes on the immunopanning plate at various ages (scale bar = 12 μm), and (Bottom) after immunostaining with an anti–GFAP antibody following 7 days in monolayer culture (scale bar = 50 μm). (C) GFAP immunostain of a cryosection from a hCS at day 350 demonstrates branched morphology of astrocytes within the 3D cytoarchitecture. Scale bar = 10 μm. (D) Quantification of the number of primary branches in GFAP⁺ cells following 7 days in monolayer culture. One-way ANOVA, F(5,143) = 15.68, P < 0.0001; n = 15–38 cells/time point). (E) Heatmap indicating expression of the top 100 fetal and top 100 mature astrocyte-specific genes in primary fetal and astrocyte samples. (F) Heatmap for the expression of these same 200 genes in hCS-derived astrocytes over in vitro culture from day 96 to day 495. The heatmaps are normalized across each gene (row). Cells at the extreme early and late timepoints express the highest levels of fetal and mature astrocyte genes, respectively, whereas the transitional timepoints (day 150–300) express more intermediate levels of these genes. Data derived from one iPSC line in 11 differentiation experiments (3–15 hCS per time-point). (G) Spearman correlations between hCS-derived astrocytes of varying in vitro stages and primary fetal (magenta) or mature (green) astrocytes. Values represent Spearman rank correlations on the 200 genes between panels (F) and (G). (H) FPKM values of selected mature (top row, green) and fetal (bottom row, purple) astrocyte genes in hCS-astrocytes purified at different in vitro stages. Boxed graphs highlight the decline in expression of proliferative markers as hCS-derived astrocytes mature. (I) Representative images from the EdU proliferation assay. Astrocytes were purified from hCS and grown in monolayer culture with 10 μM EdU for 48 hours and then fixed at 7 days. Scale bar = 100 μm. Data derived from two iPSC lines in 7 differentiation experiments (3–15 hCS per time-point). (J) Quantification of proliferation in culture. Percentages represent the number of EdU positive cells per total number of DAPI⁺ nuclei. One-way ANOVA, F(5,11) = 49.48, P < 0.0001; n = 3 wells/timepoint).

FIGURE 3. Single Cell RNA-seq of iPSC-derived hCS

(A) hCS were dissociated at different in vitro stages and single cells were isolated by FACS into 96-well plates (n= 710 cells). Cells were either randomly selected without any immunolabeling or gated by the presence of FITC-conjugated HepaCAM antibodies. Single cells were derived from two iPSC lines from 5 differentiation experiments (3–7 hCS per time-point). (B) Glial population (cyan) indicated by SOX9 gene expression in the t-SNE space of all hCS cells. Dark circles indicate high expression, grey circles indicate low expression. (C) Neuronal population (red) indicated by STMN2 gene expression in the t-SNE space of all hCS cells. (D) Unsupervised hierarchical clustering (top 1000 over-dispersed genes) showing separation of glial and neuronal populations. (E) Violin plots demonstrating expression patterns of generic cell type-specific markers. (F) Unsupervised hierarchical clustering of the glial population in (B, D) (top 1000 over-dispersed genes) revealing three Clusters (1–3). Glial population (from B, D) are colored by Cluster identity as shown in (E). (G) Distribution of cells from each in vitro differentiation stage across Clusters 1–3. (H) Expression of enriched genes within Clusters 1–3. (I) The proportion of the top 100 genes in Clusters 1–3 that were enriched in adult astrocytes (orange, > 4–fold expression increase in primary mature versus primary fetal astrocytes), fetal astrocytes (cyan, > 4–fold expression increase in primary fetal versus mature primary astrocytes), or indeterminate (grey).

FIGURE 4. Functional Changes during hCS-Astrocyte Maturation in vitro

(A) Synaptosomes were harvested from the mouse brain and labeled with pHrodoRed, a pH sensitive indicator that fluoresces only at acidic pH (< 6). (B) A representative astrocyte from a hCS (day 150) cultured in monolayer with pHrodoRed-labeled mouse synaptosomes for 2 hours. Phagocytosed synaptosomes fluoresce red and can be seen in multiple locations within the astrocyte. Scale bar = 20 μm. (C) Quantification of phagocytosis over 16 hours. Synaptosomes were added at t = 1 hour, and images were taken every 15 minutes. Data from three batches of hCS (3–15 hCS per batch) derived from one iPSC line in one differentiation. (D) (Left) Array tomography from an intact hCS that was previously labeled with the hGFAP::eGFP reporter. (Right) Synaptogram showing the colocalization of pre- and post-synaptic puncta within an astrocyte process. (E) Time course of hCS-astrocyte purification for functional studies. (F) Representative images of phagocytosis in hCS-astrocytes (day 100 to day 590) after 16 hours. Scale bar = 100 μm. Cells derived from two iPSC lines in 5–10 differentiation experiments (3–15 hCS per time-point). (G) Quantification of pHrodoRed phagocytosis. One-way ANOVA, F(10, 273)=)= 2.83, P = 0.002; n = 9 fields/condition. (H) Astrocyte lineage cells were purified from hCS and grown on inserts. RGCs were simultaneously purified and cultured either beneath an insert with hCS-glia or without any cells on the insert above. (I) Quantification of co-localized synaptic puncta per cell. Co-localization defined by overlap between the Homer and Bassoon signal as determined by in-house MatLab image processing software. Two-tailed t-test, P = 0.006. Cells derived from one iPSC line in 6 differentiation experiments (3–15 hCS per time-point). (J) Synaptic immunostainings for the pre– (Bassoon) and post-synaptic (Homer) markers in RGCs after 14 days in culture without (H) or with (I) inserts containing hCS-derived astrocytes. Scale bar = 12 μm. Insets, scale bar = 2 μm (K) Neurons purified from day 83-old hCS were co-cultured with hCS-derived astrocyte lineage cells at three in vitro differentiation stages, and calcium signaling was assessed using the calcium dye Fura-2. (L) Depolarization-induced calcium responses in Syn1⁺ neurons co-cultured with astrocyte-lineage cells isolated from hCS at various in vitro stages (M) Peak calcium amplitude in Syn1⁺ cells; one-way ANOVA; F(2, 126)= 15.69, P< 0.0001; posthoc Bonferroni for day 379–437 versus the other two time points, ***P< 0.001, ****P< 0.0001.