Depletion of LMX1A+ progenitors and their derivatives during astrogenic induction in ventral midbrain patterned neural progenitor cultures.

A, Schematic diagram of ventral midbrain neural differentiation and astrogenic induction. B-C, Representative view of immunocytochemistry of ventral midbrain neural progenitor markers and other regional markers in d19 unsorted, sorted BFP+, and sorted BFP- population. Scale bar represents 100 µm. Images shown were cropped to 300 µm×300 µm by randomly selecting the region of interest in the nuclei-only channel (uncropped greyscale images are shown in Figure S1A). D, Quantification of marker expression in unsorted, sorted BFP+, and sorted BFP- population. Error bars represent the standard error of means (SEM) of three independent experiments. E, Flow cytometry quantification of unsorted and BFP+ population during astrogenic induction and progenitor expansion. Each data point represents one biological replicate. The gating strategy used is shown in Figure S1B.

Early astrogenic switch and astrocyte maturation in derivatives of LMX1A+ midbrain progenitors.

A, Representative view of immunocytochemistry of astrogenic marker expression in BFP+ and unsorted progenitors at day 45 and 98. Scale bar represents 100 µm. Images shown were cropped to 462 µm×462 µm by randomly selecting the region of interest in the nuclei-only channel (uncropped greyscale images are shown in Figure S2). B, Representative view of immunocytochemistry of astrocyte marker expression in early and late, BFP+ and BFP- (unsorted) astrocytes. Scale bar represents 100 µm. Images shown were cropped to 300 µm×300 µm by randomly selecting the region of interest in the nuclei-only channel (uncropped greyscale images are shown in Figure S3). C, Quantification of immunocytochemistry of astrogenic marker in BFP+ and BFP- progenitors at day 45 and 98 shown in Panel A. Error bars represent the standard error of means (SEM) of three independent experiments. Two-way ANOVA was performed to compare between lineages (NFIA: p=5.389×10-6, df=1, effect size=3.62; SOX9: p=1.96×10-6, df=1, effect size=4.77) and days of differentiation (NFIA: p=7.82×10-5, df=1, effect size=1.99; SOX9: p=2.62×10-5, df=1, effect size=2.99) D, Quantification of immunocytochemistry of astrocyte marker expression in astrocytes. Error bars represent SEM of three independent experiments. Kruskal-Wallis test results following Bonferroni correction are shown on the top of the figure (AQP4: p.adjust=0.12, df=3, H=8.95; EAAT2: p.adjust=1.00, df=3, H=0.95; GFAP: p.adjust=0.06, df=3, H=10.38; S100B: p.adjust=0.11, df=3, H=9.05). E, Averaged trace of ATP-induced Ca2+ response assayed using FLIPR. Drugs or DMSO were applied at 1 minute of the assay. The line represents the average fluorescence change (ΔF/F0) in at least three independent experiments each with at least three replicate wells. The shaded area represents the SEM across at least three independent experiments. F, Quantitative comparison of the peak amplitude of ATP-induced Ca2+ response among conditions (two-way ANOVA, p<2.2×10-16, df=2, effect size=2.54) and samples (p=2.87×10-14, df=3, effect size=2.17). Error bars represent the SEM across at least independent experiments. G: Quantitative comparison of the rise time of ATP-induced Ca2+ response among conditions (two-way ANOVA, p=2.19×10-13, df=2, effect size =1.958) and samples (p=0.064, df=3, effect size=0.76). Intergroup comparison was performed using Post-hoc Tukey test. Error bars represent the SEM across at least three independent experiments. (****: p<0.0001, ***: p<0.001, **: p<0.01, *: p<0.05, ns: not significant).

Single cell RNA sequencing confirms the authenticity of PSC-derived astrocytes.

A, Uniform manifold approximation and projection plot of unbiased clustering, coloured by clusters. B, Heatmap of the normalised expression of selected markers in different clusters. The assigned identity to each cluster is shown at the top of the plot. C, Sankey plot summarising the result of reference mapping of cells in different clusters to eight published reference human brain scRNAseq datasets. The thickness of the thread is proportional to the number of cells mapped to the same identity in the reference datasets (predicted identity). Detailed results of referencing mapping to each reference datasets are shown in Figure S7A-H and prediction score shown in Figure S7I. (PRE: precursors; N: neurons)

Distinct transcriptome fingerprints of LMX1A+ midbrain floor plate-derived astrocytes

A, Heatmap of the normalised expression of population-specific genes in different populations of astrocytes. B, Violin plots of the normalised expression of selected candidate markers for BFP+, BFP-, and non-patterned (NP) astrocytes. C-D, Representative GO terms significantly enriched in BFP+ (C) and BFP- (D) enriched genes. Semantically similar representative terms were shown with the same colour.

Original images of immunocytochemistry of d19 progenitors and gating strategy of BFP flow cytometry analysis.

A, the original images shown in Figure 1B-C. The gating strategy used for BFP flow cytometry analysis are shown in B-E and described in Methods and Materials. B, scatter plot of SSC-A versus FSC-A. C, scatter plot of FSC-H versus FSC-A. D-E, histogram of BFP fluorescence in the negative control (D) and BFP-expressing (E) samples.

Original images of immunocytochemistry of astrogenic markers shown in Figure 2A.

Original images of immunocytochemistry of astrocyte markers shown in Figure 2B.

Non-patterned astrocytes and fluorescence-activated cell sorting of astrocytes for single cell RNA sequencing.

A, Representative view of immunocytochemistry of astrocyte marker expression in non-patterned astrocytes (scale bar represents 100 µm). B, Quantification of astrocyte marker expression in astrocytes. Error bars represent SEM of three independent experiments. C-E, negative control samples; F-H, one sample of BFP+ astrocytes). C and F shows the dot plot of SSC-A versus FSC-A. D and G shows the dot plot of FSC-H versus FSC-A. E and H shows the dot plot of Alexa Fluor-647-A (labelling CD49f) versus DAPI-A (labelling BFP). P3 was used to isolate CD49f+ population (including both BFP+ and BFP-), while P4 was used to isolate CD49f+/BFP+ population.

Processing of single cell RNA sequencing data.

A-C, Violin plots showing the results of adaptive cell-level filtering based on the number of genes detected per cell (A), percentage of mitochondrial gene counts per cell (B), and total gene count per cell (C). D-E, UMAP plot of all filtered cells before (D) and after integration (E) coloured by chip. F-G, UMAP plot of the subset of sorted BFP+ astrocytes on Chip A and B derived from the same astrocyte differentiation before (F) and after integration (G) coloured by chip. (N: neuron; SBN: sorted BFP-; SBP: sorted BFP+; SNP: sorted non-patterned; UBN: unsorted BFP-; UBP: unsorted BFP+; UNP: unsorted non-patterned; numbers represent independent samples)

Marker expression in scRNAseq.

A-B, UMAP plot of all filtered cells after integration coloured by sample type and sorting status (A) and estimated cell cycle phase (B). C, Raincloud plots of normalised expression of BFP. D, Histogram of the distribution of raw gene count of BFP. E, Violin plots of the normalised expression of endoderm, mesoderm, neuroectoderm, and oligodendrocyte progenitor markers. F, Violin plots of the normalised expression of classic ventral midbrain markers.

Reference mapping to human brain single cell RNA sequencing datasets.

A-H, Sankey plot summarising the result of reference mapping of cells in different clusters to eight published reference human brain scRNAseq datasets. The thickness of the thread is proportional to the number of cells mapped to the same identity in the reference datasets (predicted identity). Cluster IDs in this study are shown on the left. Heatmap in Panel A-E shows the expression of marker genes in different clusters in the re-annotated reference datasets. I, Heatmap of prediction score from Seurat integration.

Antibodies used in this study.

Settings used for processing published datasets.