Deriving early single-rosette brain organoids from human pluripotent stem cells

Abstract

Brain organoid methods are complicated by multiple rosette structures and morphological variability. We have developed a human brain organoid technique that generates self-organizing, single-rosette cortical organoids (SOSR-COs) with reproducible size and structure at early timepoints. Rather than patterning a 3-dimensional embryoid body, we initiate brain organoid formation from a 2-dimensional monolayer of human pluripotent stem cells patterned with small molecules into neuroepithelium and differentiated to cells of the developing dorsal cerebral cortex. This approach recapitulates the 2D to 3D developmental transition from neural plate to neural tube. Most monolayer fragments form spheres with a single central lumen. Over time, the SOSR-COs develop appropriate progenitor and cortical laminar cell types as shown by immunocytochemistry and single-cell RNA sequencing. At early time points, this method demonstrates robust structural phenotypes after chemical teratogen exposure or when modeling a genetic neurodevelopmental disorder, and should prove useful for studies of human brain development and disease modeling.

Keywords: PCDH19; cortical spheroid; dorsal forebrain; epilepsy; induced pluripotent stem cells; mosaicism; neural tube defects; neurodevelopment; neurulation; valproic acid.

Graphical abstract

Figure 1

Two-dimensional to three-dimensional transition results in SOSR-COs demonstrating characteristics of early cortical development (A) Schematic of SOSR-CO differentiation timeline. The top half describes the media components while the bottom half is the culture format. (B–E) Phase micrographs of SOSR-COs at important stages including neuroepithelial monolayer (B), monolayer cutting (C), early SOSR-CO formation on extracellular matrix (D), and an SOSR-CO in suspension at day 14 (E). (F–L) Confocal micrographs of whole-mount day 6 (F–H), day 8 (I and J), or day 13 (K and L) SOSR-COs immunostained for the designated proteins. (F’ and G’) Orthogonal sections of z series from corresponding confocal micrographs. (G) SOSR-COs labeled by ZO1-EGFP fusion protein and HCS CellMask Deep Red stain to fill all cells. (H) F-actin labeled with Phalloidin-Alexa488. (H and J) Arrowheads indicate mitotic cells at the apical surface labeled either for TPX2 or DNA. (M–P) Confocal micrographs of cryosectioned SOSR-COs immunostained for neural markers on days 20–22. DCX indicates a doublecourtin promoter-driven mCherry. pVIM, phospo-vimentin (S55). Nuclei were stained with bis-benzamide (DNA, blue) for (I)–(P), and days of differentiation are indicated in the upper right corner of each image. All scale bars, 100 μm.

Figure 2

SOSR-COs have neurodevelopmentally consistent cortical lamination patterns (A–C) Three-, 4-, and 6-week SOSR-COs immunostained for the radial glial marker PAX6 (green) and intermediate progenitor marker TBR2 (red). ZO-1 immunolabeling outlines the apical lumen, and bis-benzamide nuclear stain (DNA) is in blue. (D) Four-week SOSR-CO immunostained for the forebrain marker FOXG1 and deep-layer neuronal marker, CTIP2. (E–H) SOSR-COs at several timepoints immunostained for Reelin (green), ZO1 (white in E and F), and HOPX (red in H). Bis-benzamide nuclear stain (DNA) is in blue. (I–K) SOSR-COs immunostained for cortical layer markers CTIP2, SATB2, BRN2, or CUX1. The boxed areas in (G), (J), and (K) are magnified in the insets. (L) Eight-week SOSR-COs begin to express the oRG marker HOPX with radially oriented processes. (M and N) SOSR-COs at 5 months express the astrocyte marker GFAP (M) or S100β (N) on the outer edge and the mature neuronal marker MAP2ab throughout. Distinct phospho-vimentin (S55)-positive, S100β-negative cells are further indication of oRGs in 5-month SOSR-COs (N). Dashed curved lines in (D–F), (I), (K), and (L) mark the apical surface of the lumen, and dotted curved lines present in most panels denote the outer edge of the SOSR-COs. Scale bars, 100 μm.

Figure 3

Single-cell RNA sequencing for four separate 1-month SOSR-COs from line AICS-0023 (A) UMAP plots for 1-month SOSR-COs with color coding of identified clusters. (B) Bar graph showing the percentage of cells found in each cluster of the four 1-month SOSR-COs. The legend at right displays the cell type identity for each cluster. (C) Cartoon of developing telencephalon with the color-coded identity of each cluster labeled. (D–G) UMAP plots for each of the four 1-month SOSR-COs. Red dots are for the specified SOSR-CO while gray dots are from the other three SOSR-COs. (H–E’) Individual UMAP plots of selected genes from each of the four 1-month SOSR-COs. The region each marker identifies is listed above the gene names. Regions that are cluster identities are color-coded to match (A)–(C).

Figure 4

Single-cell RNA sequencing for six separate 3-month SOSR-COs from line AICS-0023 (A) UMAP plots for 3-month SOSR-COs with color coding of identified clusters. (B) Bar graph showing the percentage of cells found in each cluster of the six 3-month SOSR-COs. The legend at right displays the cell type identity for each cluster. (C) Cartoon of developing dorsal telencephalon layers with the color-coded identity labeled for each cluster found in 3-month SOSR-COs. (D–I) UMAP plots for each of the six 3-month SOSR-COs. Red dots are for the specified SOSR-CO while gray dots are from the other five SOSR-COs. (J–G′) Individual UMAP plots of the specified genes for each of the six 3-month SOSR-COs. The region each marker identifies is listed above the gene names. Regions that are cluster identities are color-coded to match (A)–(C).

Figure 5

Single-cell RNA sequencing for three separate 5-month SOSR-COs from line AICS-0023 (A) UMAP plots for 5-month SOSR-COs with color coding of identified clusters. (B) Bar graph showing the percentage of cells found in each cluster of the three 5-month SOSR-COs. The legend at right displays the cell type identity for each cluster. (C) Cartoon of developing dorsal cortex layers with the color-coded identity of each cluster labeled. (D–F) UMAP plots for each of the three 5-month SOSR-COs. Red dots are for the specified SOSR-CO while gray dots are from the other two SOSR-COs. (G–D’) Individual UMAP plots of the specified genes for each of the three 5-month SOSR-COs. The region each marker identifies is listed above the gene name. Regions that are cluster identities are color-coded to match (A)–(C).

Figure 6

Teratogenic compound exposure results in dysmorphic SOSR-COs with enlarged lumens (A) Schematic of cellular and structural changes during neurulation. (B–I) Confocal micrographs of whole-mount day 6 SOSR-COs expressing ZO-1-EGFP fusion protein (green) with either nestin (red) and PAX6 (blue) (B–E) or acetyl-tubulin (red) and SOX2 (blue) (F–I). SOSR-COs were exposed to vehicle (B and F) or inhibitors of the apical constriction pathway, either blebbistatin (C and G, 10 μM), Y-27632 (D and H, 20 μM), or the teratogenic antiseizure medication, valproic acid (E and I, 400 μM). (J) Schematic of important genes in apical constriction and the mode of action for both inhibitors. (K–M) Quantification of normalized lumen area (lumen area/SOSR-CO area, square root transformed). (K) Normalized lumen area was determined for three unrelated control (C) lines with either vehicle or blebbistatin treatment with a single batch for each line. Data are color-coded by line with C1-magenta, C2-green, C3-blue. In (L), data points are individual SOSR-COs from six independent experiments exposed to 0, 200, or 400 μM valproic acid. Each individual experiment also showed a significant increase for both 200 and 400 μM valproic acid. n = 139, 127, and 123, respectively. In (M), data points are individual SOSR-COs from three independent experiments, n = 36 per group, and a valnoctamide (VCD) group was added. Unpaired t test was used for two-group comparisons, and one-way ANOVA test with Tukey post hoc was used for more than two groups. ^∗∗∗∗p < 0.0001, and ns p > 0.05. Error bars are mean ± SD for (K), (L), and (M). Scale bars are 100 μm.

Figure 7

SOSR-COs recapitulate the early cell segregation phenotype found in mouse models of PCE (A–C) Schematics depicting the hypothesized structural outcomes of brain organoids derived from different mixtures (simulating mosaicism in A) of GFP+ and unlabeled isogenic WT and KO cell lines based on previous data from mouse models (Hoshina et al., 2021; Pederick et al., 2018). (D–F) Confocal micrographs of multi-rosette brain organoids generated according to previously published methods (Dang et al., 2021; Qian et al., 2016) for each of the three genotype mixtures: WT^GFP/KO (D), WT^GFP/WT (E), and KO^GFP/KO (F). (G–I) SOSR-COs were generated in parallel with the same three mixtures: WT^GFP/KO (G), WT^GFP/WT (H), and KO^GFP/KO (I). Green arrows and blue arrowheads (in G) indicate segregated stripes of only WT or only KO cells, respectively, in the mosaic SOSR-CO. All organoids were immunostained for GFP (green) and ZO-1 (red), with bis-benzamide nuclear stain (blue). (J) Quantification of the number of GFP stripes in SOSR-COs from imaged slices with each mosaic or non-mosaic mix. n = 24, 28, and 27 SOSR-COs across four independent experiments. (K and L) Six individual SOSR-COs per genotype mixture with WT^GFP/KO mixtures (K) from a single batch showing a clear segregation phenotype, and WT^GFP/WT mixtures (L) showing no segregation phenotype. (M) Quantified coefficient of variation of GFP pixel signal intensity in the center (at one-sixth of total radius) of each SOSR-CO image (as collected in K and L). n = 92 and 122, respectively, across six independent experiments. Each dot is an individual SOSR-CO. Error bars are means ± SD. Statistical analysis performed using Kruskal-Wallis test with Dunn’s multiple comparison (J) or Mann-Whitney (M). ^∗∗∗∗p < 0.0001, ns p > 0.05. All images are from day 20 spin-Ω organoids (D)–(F) or SOSR-COs (G)–(I), (K), (L). Scale bars, 100 μm.