br Results To investigate the molecular and cellular phenoty
Results To investigate the molecular and cellular phenotypes perturbed in DS, we compared DS and normal ESCs and their neural derivatives. In the past, we have isolated three DS-ESC lines, namely CSES13, CSES20, and CSES21, from PGS-derived embryos with trisomy of chromosome 21 (Figure 1A) (Biancotti et al., 2010). To extend the number of analyzed DS-ESC lines, we have established two additional DS-ESC lines, CSES32 and CSES44, which also carry trisomy 21 (Figure 1A). These DS-ESC lines were characterized in terms of morphology, alkaline phosphatase staining (Figure 1B), and expression of OCT4 (Figure 1C). All five DS-ESC lines showed expression of characteristic markers of pluripotent stem abscisic acid with average expression levels similar to those of wild-type (WT) cells (Figure 1D). Finally, CSES32 and CSES44 DS-ESC lines were differentiated into cells from the three embryonic germ layers upon induction of teratomas in vivo, showing structures of endodermal, mesodermal, and ectodermal tissues (Figure 1E). To better understand the neural phenotype of DS cells compared with normal cells, we differentiated all five DS-ESC lines into NPCs. Gene-expression analysis shows that in DS-ESCs, embryoid bodies (EBs), and NPCs, the relative expression of genes on chromosome 21 is about 1.5-fold higher than that of genes on chromosomes 20 or 22 (Figure 2A). These data suggest that in both undifferentiated and differentiated DS cells, all three copies of chromosome 21 are actively transcribed. This upregulation, however, accounts for only a minority of the differences observed in the global gene-expression profile between normal and DS-ESCs. Notably, the majority of the differentially expressed genes between the two cell types were located on autosomal chromosomes other than chromosome 21. Because DS patients have a striking developmental phenotype related to the CNS, we focused on the neural phenotype of DS-NPCs. To study the neural phenotype, we compared data of expression arrays of NPCs of three different WT cell lines with those of five different DS cell lines. The genes were then sorted according to their expression levels, whereas genes expressed more than 2-fold in DS-NPCs compared with WT-NPCs were considered to be upregulated in DS, while genes expressed less than 0.5-fold in DS-NPCs were considered to be downregulated. Functional annotation analysis of differentially expressed genes between DS- and control NPCs using the Database for Annotation, Visualization and Integrated Discovery (DAVID) (Huang et al., 2009a, 2009b) showed downregulation of genes related to forebrain development and upregulation of genes related to apoptosis (Figure 2B). The forebrain developmental genes downregulated in DS-NPCs include key neuronal genes such as POU3F2 (also known as BRN2) and ASCL1 (Figure 2C). To verify the predicted changes in apoptosis, we performed flow cytometry analysis to quantify the levels of programmed cell death in DS-NPCs derived from the five DS cell lines and compared them with control NPCs derived from three WT cell lines. The results showed an increase in the tendency of DS-NPCs to activate apoptosis when compared with control NPCs, assessed from the populations of both annexin V+/propidium iodide (PI)− and annexin V+/PI+ cells (Figure 2D). We next analyzed whether the differential expression of genes we observed in DS cells results, at least partly, from an extra copy of a transcription factor residing on chromosome 21. For this purpose, we analyzed all upregulated genes with at least 2-fold change of expression in DS-NPCs, the majority of which reside on the autosomes other than chromosome 21 (Figure 3A), using the Promoter Integration in Microarray Analysis (PRIMA) software that searches for binding site enrichments on a given promoter set (Elkon et al., 2003). The analysis found the binding site of the nuclear protein Runt-related transcription factor 1 (RUNX1) to be significantly and highly enriched in the upregulated genes (p < 0.05 with 3.5-fold enrichment) (Figure 3B). RUNX1 is a transcription factor that localizes to the critical region of chromosome 21 (Figure 3B). To better understand the involvement of RUNX1 in the molecular pathology of DS, we used the CRISPR/Cas9 gene-editing system and designed a guide RNA to specifically target all isoforms of RUNX1 in DS-ESCs and create DS-CRISPR-deleted RUNX1 (DSCR) ESCs to observe the maximal effect of RUNX1 dosage differences (Figure 3C). Among the various clones isolated, we identified two clones, DSCR8 and DSCR75, with complete ablation of the RUNX1 protein by western blot analysis (Figure 3D), indicating disruption of all three RUNX1 alleles. Next, we differentiated the two DSCR clones into NPCs and performed global gene-expression analysis by DNA microarrays. We found that 162 genes were downregulated in our DSCR-NPCs compared with their isogenic DS-NPCs. We then analyzed these downregulated genes using the DAVID software and the USCS transcription factor binding site search. The analysis revealed that nearly 70% of the downregulated genes in the DSCR-NPCs (111 genes) were putative targets of RUNX1 with a Benjamini-corrected p value of 0.026. Among the downregulated targets of RUNX1 are several key developmental genes (Figure 4A), with some genes such as IGFBP5, CCL2, LGR5, FBLN5, and TLR4 showing a RUNX1-dosage-dependent expression (Figure 4B). One of these genes, CCL2, showed a much stronger downregulation when analyzed by qRT-PCR (using the primers listed in Table 1) compared with the expression array data, probably due to a less stringent probe set in the expression array. Functional annotation analysis revealed that the downregulated genes in DSCR-NPCs were enriched for neuron/cell migration and regulation of cell growth (Figure 4C). Finally, we analyzed whether the ablation of RUNX1 allowed the correction of the cellular phenotype of apoptosis. Indeed, flow cytometry analysis demonstrated that the DSCR clones had a reduced level of apoptosis when compared with their parental DS lines (Figure 4D).