A human isogenic iPSC-derived cell line panel identifies major regulators of aberrant astrocyte proliferation in Down syndrome

Abstract

Astrocytes exert adverse effects on the brains of individuals with Down syndrome (DS). Although a neurogenic-to-gliogenic shift in the fate-specification step has been reported, the mechanisms and key regulators underlying the accelerated proliferation of astrocyte precursor cells (APCs) in DS remain elusive. Here, we established a human isogenic cell line panel based on DS-specific induced pluripotent stem cells, the XIST-mediated transcriptional silencing system in trisomic chromosome 21, and genome/chromosome-editing technologies to eliminate phenotypic fluctuations caused by genetic variation. The transcriptional responses of genes observed upon XIST induction and/or downregulation are not uniform, and only a small subset of genes show a characteristic expression pattern, which is consistent with the proliferative phenotypes of DS APCs. Comparative analysis and experimental verification using gene modification reveal dose-dependent proliferation-promoting activity of DYRK1A and PIGP on DS APCs. Our collection of human isogenic cell lines provides a comprehensive set of cellular models for further DS investigations.

Fig. 1. Generation of an isogenic iPSC panel for disease modelling of DS.

Schematic overview depicting the strategy used to generate isogenic iPSC lines by combining XIST-induced chromosome silencing and genome/chromosome-editing technologies. a Cord blood mononuclear cells were derived from a male patient with DS. Then, a DS-specific iPSC line (Tri21) was directly established using Sendai virus. A corrected-disomy cell line (cDi21) was generated by eliminating a single copy of chromosome 21 from a Tri21 iPSC line. b A 4-Mb region corresponding to a ‘Down syndrome critical region’ was selectively deleted only from the paternal chromosome 21 in Tri21 iPSCs to generate Partial-Tri21 iPSC. NPCs differentiated from Partial-Tri21 iPSCs were transfected with piggyBac (PB) transposon vectors containing human PIGP and/or DSCR3 cDNA under the regulation of the Tet-inducible system. NPCs were differentiated into APCs for further analysis. c Dox-inducible human XIST cDNA was inserted into one copy of chromosome 21 in Tri21 iPSCs (XIST-Tri21 iPSC) for chromosome silencing. rtTA was additionally introduced into NPCs differentiated from XIST-Tri21 iPSCs, using a PB vector. Transcriptional and phenotypic analysis was conducted in Dox-untreated (D−) and Dox-treated (D+) cells, and in NPCs or APCs at 3 weeks after Dox removal (D^remov). The location of the XIST transgene is indicated by the red rectangle. The black and striped chromosomes indicate a silenced and a reverted chromosome 21, respectively. d The DYRK1A gene was targeted in one or two chromosomes in the XIST-Tri21 iPSC line to generate DY^+/+/m- and DY^+/m/m-XIST-Tri21 iPSCs, respectively. Three types of genome-edited cell lines were generated in each single- or double-targeted clone, based on the parental origin of the three chromosomes. Black stars indicate the DYRK1A target sites. Two maternal copies and one paternal copy of chromosome 21 are indicated as M1/M2 and P, respectively. e DYRK1A-targeted iPSC lines were differentiated to APCs and subjected to phenotypic analysis. Light yellow panels and light green panels indicate iPSC lines and NPC/APC lines, respectively.

Fig. 2. XIST-mediated chromosome silencing affects the overproliferative phenotype of DS APCs.

a Schematic depiction of the differentiation protocol for NPCs and APCs. EBs embryoid bodies. b Immunocytochemistry of XIST-Tri21 APCs using GFAP and S100β (astrocyte markers), CD44 and vimentin (APC markers), and SOX1 (NPC marker). Nuclei were stained with Hoechst 33342. Scale bars: 100 μm. c Relative expression levels of XIST RNA in XIST-Tri21 APCs. Expression was normalised to that of D− APCs (n = 3 experiments per cell line). d Immunocytochemistry of XIST-Tri21 NPCs using an H3K27me3-specific antibody. Nuclei were stained with Hoechst 33342. Scale bars: 50 μm. e Percentage of H3K27me3-positive cells in XIST-Tri21 APCs (n = 4 experiments per cell line). f Relative expression levels of genes on chromosome 21 in XIST-Tri21 APCs. Expression levels were normalised to those of cDi21 cells (n = 5 experiments per cell line). g EdU assay in XIST-Tri21 APCs. EdU-positive cells were stained with an Alexa Fluor 488-conjugated antibody (light green). Nuclei were stained with Hoechst 33342. Scale bars: 100 μm. h Percentage of EdU-positive APCs (n = 3 independent differentiation experiments per cell line, performed in triplicate). i Relative numbers of APCs 1 day after seeding. Cell numbers were normalised to those of cDi21 lines (n = 3 independent differentiation experiments per cell line, performed in triplicate). Data were obtained from three lines (c, e, f, h, i). Error bars represent the SEM. Data were analysed by Student’s t-test (e, f), Welch’s two-sample t-test (c), or one-way ANOVA with Bonferroni’s correction (h, i). NS not significant (P > 0.05). EdU 5-ethynyl-2′-deoxyuridine.

Fig. 3. Forced XIST expression causes whole-chromosome inactivation, which is overall maintained after Dox removal.

a Violin plots showing relative log-transformed expression ratios of genes on each chromosome in XIST-Tri21 APC lines. Gene-expression levels (determined by RNA-seq) were normalised to those of cDi21 lines (n = 3 per cell line). The upper orange dashed lines indicate a ratio of 1.5, whereas the lower black dashed lines indicate a ratio of 1.0. Transcription from chromosome 21 was effectively suppressed in D+ cells, and inactivation was retained overall in D^remov cells. A small hump was observed in chromosome 21 of a D^remov cell line (black arrow). The plots show mean expression levels, with error bars indicating the SD. The data shown were analysed by the Kruskal–Wallis test with Bonferroni’s correction. ***P = 4.2 × 10⁻⁵, ****P = 5.1 × 10⁻⁹; NS not significant (P > 0.05). b Heatmap with a hierarchical-clustering diagram depicting the transcriptional dynamics of chromosome 21. We analysed 178 genes with positive read counts on chromosome 21. Expression levels were normalised to those of cDi21 cell lines (Supplementary Data 1 and 2). Genes are ordered according to chromosomal position. c Overview of the RNA-seq and ChIP-seq data for chromosome 21. The distribution of 178 genes with positive read counts (red plots), 56 genes with an D−/cDi21-expression ratio ≥ 1.5 (light green), 50 genes in which Dox treatment decreased expression (blue), 21 genes in which Dox treatment significantly decreased expression (Welch’s two-sample t-test, P < 0.05, yellow), and 18 genes in which Dox removal reverted the expression levels (purple). (Red bars) The distributions of regions with significantly higher accumulation of H3K27me3 are shown. False-discovery rates (FDRs) were calculated using the Benjamini–Hochberg method. P and FDR < 0.05 were considered to reflect significant differences. d Integrated analysis of the ChIP-seq and RNA-seq data. H3K27me3 accumulation within each gene body and 2 kb upstream of the transcription start site were analysed. The density of all paired points is shown with a blue contour line. Genes with a significant difference in H3K27me3 accumulation between two lines are indicated (red points). The green dashed ellipse emphasises the distribution of inactivated genes with increased H3K27me3 accumulation after Dox treatment. The orange dashed line indicates a two-thirds decrease in gene expression.

Fig. 4. Identification of candidate marker genes for the aberrant proliferation of DS APCs using PCA and Partial-Tri21 cell line.

a Two-dimensional PCA of the expression profiles of 178 genes with positive read counts on chromosome 21 (n = 3 per cell line). Note that the values of the D^remov cell lines in component 1 were relatively close to those of the D− cell lines. b Scheme for identifying genes responsible for APC overproliferation using RNA-seq data and PCA. c Schematic illustration of chromosome 21 in the Partial-Tri21 cell line. The 4-Mb region between RUNX1 and ETS2 was deleted from a single copy of chromosome 21. d The percentage of EdU-positive APCs (n = 3 independent differentiation experiments per cell line, each performed in triplicate). e Relative numbers of APCs 1 day after seeding. The cell numbers were normalised to that of the cDi21 cell lines (n = 3 independent differentiation experiments per cell line, each performed in triplicate). f Relative expression levels of candidate genes involved in the overproliferation of DS APCs (n = 3 experiments per cell line), normalised to that of cDi21 cell lines. g Map showing the distribution of H3K27me3 modifications in DYRK1A in XIST-Tri21 APCs. Integrative Genomics Viewer screenshot (http://software.broadinstitute.org/software/igv/) of H3K27me3 ChIP-seq track peaks on DYRK1A (hg19) in D−, D+, and D^remov lines. The Y-axis shows the number of fragments per base pair per million reads. (Red bars) The distributions of regions with significantly higher H3K27me3 accumulation in the indicated cell lines. Data were obtained from three lines (d–g). The error bars represent the SEM. The data shown were analysed using Student’s t-test (d–f). The FDR was calculated using the Benjamini–Hochberg method (g). NS not significant (P > 0.05).

Fig. 5. Targeted deletion of DYRK1A promotes APC proliferation in a gene dosage-dependent manner.

a Schematic overview of the strategy used to target DYRK1A. b Comparative STR analysis of the DYRK1A locus was performed between a patient with DS and his parents’ genomes. c STR analysis of DYRK1A-targeted iPSC lines to identify targeted alleles (F2/R3 primers) and non-targeted alleles (F2/R2 primers). d Relative expression levels of DYRK1A in DY^+/+/m- and DY^+/m/m-XIST-Tri21 APCs. The expression levels were normalised to that of cDi21 cell lines (n = 3 experiments per cell line). e, f Percentage of EdU-positive cells and the relative numbers of targeted APCs. Cell numbers were normalised to those of the cDi21 lines (n = 3 experiments for the cDi21 cell lines and the other cell lines, performed in triplicate). g Schematic illustration depicting the expected changes resulting from knocking out different DYRK1A alleles and chromosome silencing. Red rectangle, XIST transgene; black stars, DYRK1A target sites; black chromosomes, inactivated chromosome 21. The expected copy numbers of active DYRK1A genes in each APC line are shown. h, i Percentage of EdU-positive, DYRK1A-targeted cell lines, with or without Dox treatment. The percentage of EdU-positive cells was normalised to that of each Dox-untreated cell line (n = 3 experiments per cell line). j, k Relative cell numbers of the DYRK1A-targeted lines, with or without Dox treatment. The cell numbers were normalised to that of each Dox-untreated line (n = 3 experiments per cell line). l–o Percentage of EdU-positive cells and the relative numbers of Tri21 APCs (l, m) or Tri21 NPCs (n, o) after a 2-day treatment with FINDY (n = 3 experiments per cell line). The cell numbers were normalised to those of Tri21 APCs or NPCs without FINDY treatment (n = 3 experiments per cell line). p Immunoblot analysis of p-STAT3 (Ser 727) and STAT3 in DYRK1A-targeted XIST-Tri21 APCs. Expression levels were normalised to that of the XIST-Tri21 line (n = 3–6 experiments per cell line). Data were obtained from three lines (p). The error bars represent the SEM. The data shown were analysed by Student’s t-test to compare two independent groups or one-way ANOVA with Bonferroni’s correction (d, e, f). NS not significant (P > 0.05).

Fig. 6. PIGP promotes DS APC proliferation in an expression level-dependent manner.

a Schematic depicting the transfection of a Dox-inducible PIGP or DSCR3 transgene using a PB transposon vector and a hyperactive PB transposase into Partial-Tri21 NPCs. Note that chromosome 21 was trisomic except for the 4-Mb segment, which corresponds to the Down syndrome critical region. Transduced NPCs were differentiated to APCs. b Partial-Tri21 APCs stably expressing the PIGP-overexpression vector. PIGP overexpression was induced by a 6-week treatment with Dox. Expression levels were normalised to that of untreated APCs (n = 3 experiments per condition). c, d Percentage of EdU-positive cells (c) and relative cell numbers 1 day after seeding (d) in PIGP-overexpressing Partial-Tri21 APCs (n = 3 experiments per condition). Cell numbers were normalised to that of untreated APCs. e Relative PIGP expression in siRNA-treated Tri21 APCs. Expression levels were normalised to those in siCtrl-treated Tri21 APCs (n = 3 experiments per condition). f, g Percentage of EdU-positive cells (f) and relative cell numbers 1 day after seeding (g) in siRNA-treated Tri21 APCs (n = 3 experiments per condition). The percentage of EdU-positive cells and cell numbers were normalised to those of the untreated Tri21 line. siPIGP, PIGP siRNA. The error bars represent the SEM. The data shown were analysed by Student’s t-test (c–g) or Welch’s two-sample t-test (b).