Fig. 2: The human-specific astrocyte gene signature is intrinsically programmed.

a Experimental design. Gestational week 18 primary human astrocytes were purified and injected into the brains of neonatal immunodeficient Rag2-knockout mice. After about 8 months, we purified astrocytes from xenografted mouse brains by immunopanning. The astrocytes from both human grafts and mouse hosts were sequenced together and reads were mapped to human and mouse genomes, respectively. GW, gestational week. P, postnatal day. b–d Xenografted human cells in host mouse brains stained with an antibody against human nuclei (green). Scale bar: 100 μm. Dashed lines delineate the corpus callosum. e Xenografted human astrocytes in host mouse brains stained with an anti-GFAP antibody that only reacts with human GFAP but not mouse GFAP. Scale bar: 50 μm. f Species differences in gene expression (shown as percentile ranking in human minus percentile ranking in mouse) in xenografted and acutely purified astrocytes highly correlate. Genes with percentile rankings > 0.33, species differences with FDR < 0.05, and species differences in percentile ranking > 0.4 are shown. r, Pearson’s correlation coefficient. g Non-supervised hierarchical clustering of gene expression in serum-free cultures, acutely purified astrocytes, and transplanted human astrocytes. Genes with significant differences between cultured and acutely purified astrocytes (FDR < 0.05, fold change > 4, average RPKM > 1) are shown. h Non-supervised hierarchical clustering of gene expression of acutely purified astrocytes from patients of different ages and transplanted human astrocytes. Genes with significant differences between age groups (fetal, child, adult; FDR < 0.05, fold change > 2, maximum RPKM > 1) are shown.