Deficits in human trisomy 21 iPSCs and neurons

Isogenic Human Ts21 iPSCs.

Fibroblasts from two individuals diagnosed with DS were reprogrammed to iPSCs. FISH for HSA21 in one fibroblast line showed mosaicism, where ∼90% of cells carried Ts21, whereas ∼10% were euploid (Fig. 1A). Reprogramming of the mosaic fibroblasts by retrovirus (6) resulted in three viable iPSC clones, two clones that carried Ts21 and one euploid (Fig. 1B). Mosaicism in DS individuals is rare, occurring in ∼1–3% of DS cases (7), but it can also emerge in vitro (8), potentially because of nondisjunction events during cell division. Regardless, the generation of euploid isogenic controls is fortuitous, because they become vital controls for a complex multigene disorder such as DS, potentially limiting the need for multiple iPSC lines to control for genetic and epigenetic variation (9⇓–11). SNP analysis of HSA21 in the euploid DS2U iPSC line ruled out uniparental disomy (or isodisomy), which is often associated with trisomy rescue (Fig. 1D). Short tandem repeats at various loci indicated that the Ts21 and euploid lines were isogenic other than the presence of Ts21 (Fig. 1E). To increase statistical power, fibroblasts from a second DS individual were reprogrammed using Sendai virus (12), an RNA virus, which yielded another Ts21 iPSC line (Fig. 1C). All iPSC clones were named according to standardized naming procedures (Fig. 1F) (13) and used in all experiments. Importantly, no chromosomal abnormalities, other than Ts21, were observed in any line throughout the duration of this study, in contrast to the propensity of many iPSC lines to acquire aneuploidy (14, 15). Expression of pluripotent genes and proteins POU class 5 homeobox 1 (Oct4), SRY-box containing gene 2 (Sox2), stage-specific embryonic antigen-4 (SSEA-4), and Tra1-81 and lack of expression of the neuroepithelial marker paired box gene 6 (Pax6) (16) and reprogramming genes verified the pluripotency of the iPSCs (Fig. S1 A–D). In addition, all iPSCs expressed markers of each germ layer after nondirected differentiation (Fig. S1).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Reprogramming of mosaic fibroblasts yields Ts21 iPSCs and an isogenic control. (A) Fibroblast line AG05397 was mosaic for Ts21, with (Upper) ∼89% of cells carrying three copies [runt-related transcription factor 1 (AML1)/Down syndrome critical region (DSCR); orange], whereas (Lower) 10% were disomic for HSA21 (AML1/DSCR; orange). The telomere (TEL) marker (green) was used as a control probe. (B) All iPSC lines had morphological characteristics of pluripotent stem cells, and karyotype analysis showed that DS1 and DS4 iPSCs are trisomic, whereas DS2U is disomic for HSA21 (red circles). (C) 2DS3 iPSCs from a second DS individual carry Ts21. (D) SNP analysis revealed no absence of heterozygosity of HSA21 in the euploid DS2U iPSC line. (E) Short terminal repeat analysis revealed that Ts21 and control lines are isogenic at all loci tested. (F) Table of different iPSC lines used in this study. (G) A heat map shows that genes changed more than fivefold in DS1 and DS4 iPSCs compared with the isogenic DS2U control iPSCs. (H) qPCR verification in all Ts21 iPSC lines of various genes that are changed in microarray results. (I) Ts21 iPSCs did not exhibit increased oxidative stress compared with their respective controls, which were assayed by DHE. (J) The proportion of Ts21 cells that underwent apoptosis was similar to controls, which were assayed by TUNEL⁺ cells. Error bars represent SEM. (Scale bars: 100 µm.)

Theories of the pathophysiology of DS stem from the imbalance of gene expression in critical developmental pathways caused by the presence of an extra HSA21 (17⇓⇓–20). We analyzed global gene expression of DS1 and DS4 Ts21 iPSCs compared with isogenic euploid DS2U to determine if there were gene expression changes that might foreshadow later defects in differentiated tissues. Comparison of gene expression between isogenic cells enabled the identification of changes that were caused by the extra copy of HSA21 and not normal human variation. Both Ts21 iPSC lines (DS1 and DS4) displayed a preferential increase in expression of HSA21 genes compared with genes on other chromosomes (Fig. S2A), consistent with gene expression in Ts21 human ES cells (21). Changes in HSA21 genes seemed evenly distributed across HSA21 (Fig. S2C), and almost 90% were increased (125 of 139) (Dataset S1) with a significantly greater percentage of chromosomal content compared with other chromosomes (Fig. S2A) (n = 3, P < 0.001). Furthermore, mean fold change for altered genes on HSA21 was significantly greater than all other chromosomes and generally reflected the 3:2 ratio of HSA21 genes (Fig. S2B and Datasets S1 and S2). Together, these data suggest that gene expression in this early pluripotent stage is largely based on simple gene dosage.

Functionally, most of the genes with expression that was changed in Ts21 iPSCs were involved with metabolism, but these changes did not cause changes in proliferation of Ts21 iPSCs (Fig. S1G) [Ki67: DS1/4, P = 0.87, n = 4; 2DS3, P = 0.4, n = 3; phospho-histone H3 (PHH3): DS1/4, P = 0.90, n = 4; 2DS3, P = 0.07, n = 3]. Some of the largest expression changes in Ts21 iPSCs were in non-HSA21 genes involved in transcriptional activation and the response to oxidative stress (OS; e.g., catalase, CAT) (Fig. 1 G and H). Similar expression changes were detected in Ts21 neurons (see below). Many of the highly up-regulated transcription factors are members of the zinc finger family that regulate the expression of a variety of downstream genes and have been implicated in numerous developmental disorders affecting different cell lineages (22). Based on the increased expression of OS responsive genes in Ts21 iPSCs, we evaluated the cells for evidence of OS but found no increase in OS, which was assayed by dihydroethidium (DHE) (Fig. 1I) (DS1, P = 0.62; DS4, P = 0.98; 2DS3, P = 0.95; n = 2 each), or cell death in Ts21 iPSCs compared with their control counterparts (Fig. 1J) (DS1, P = 0.79, n = 3; DS4, P = 0.86, n = 2; 2DS3, P = 0.57, n = 2).

Generation of Early-Born Cortical Neurons Is Not Affected by Ts21.

A universal characteristic of DS is the presence of mild to moderate cognitive impairment, potentially because of diminished cortical neuron numbers in DS brains (23⇓⇓–26). However, using our established protocol for cortical neuronal differentiation (27, 28), we found that all iPSC lines robustly differentiated to Pax6⁺ neuroepithelia by day 10 (16) (Fig. 2A), consistent with the best efficiency shown by other iPSC lines (11). Neural progenitors differentiated from Ts21 iPSCs also had increased catalase expression, similar to Ts21 iPSCs, but expressed other HSA21 genes implicated in OS at the expected 1.5-fold (Fig. 2B). PCR analysis and immunostaining revealed no difference in the regional identity of β_III-tubulin⁺ neurons in euploid and Ts21 cultures, where the vast majority expressed dorsal forebrain markers (Fig. 2C). Similarly, quantification of β_III-tubulin⁺ neurons at 6 wk of differentiation revealed no significant differences between the fraction of neurons in euploid and Ts21 cultures (Fig. 2 D and E) (β_III-tubulin: n = 6, P = 0.78; FoxG1: n = 6, P = 0.88; Otx2: n = 6, P = 0.29).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Generation of cortical neurons is not affected by Ts21. (A) FACS analysis reveals no difference in the propensity of Ts21 iPSCs to generate Pax6⁺ neuroepithelia. (B) Ts21 iPSC-derived neural progenitor cells exhibited increases in many HSA21 and oxidative stress genes, which were assayed by qPCR. (C) Transcript expression of dorsal telencephalic markers was evident in neurons differentiated from all lines, but ventral forebrain [NK2 homeobox 1 (Nkx2.1)], hindbrain [gastrulation brain homeobox 2 (Gbx2)], and spinal cord [homeobox B4 (Hoxb4)] markers were not readily observed. (D and E) Immunostaining of cultured cells shows no difference in β_III-tubulin⁺ neurons or forebrain markers [forkhead box G1 (FoxG1) and orthodenticle homeobox 2 (OTX2)] across iPSC lines. (F) Heat map depicts global gene expression changes more than threefold in isogenic Ts21 (DS1 and DS4) vs. control (DS2U) iPSC-derived neuronal cultures. (G) qPCR verification of various genes up-regulated in microarray results in all iPSC-derived neuronal cultures. (H) DS neurons exhibited increased oxidative stress, which was assayed by DHE, and (I) mitochondrial membrane potential. (J) The proportion of Ts21 cells that underwent apoptosis was similar to controls, which were assayed by TUNEL⁺ cells. (Scale bars: 50 µm.) Error bars represent SEM. *P < 0.05. (A, B, D, E, and J) For measures where no significant differences were found between groups, dashed lines indicate the average of the control groups (DS2U and IMR90).

Thus, although studies in both human and mouse have implicated reduced cortical neurogenesis (24, 29⇓⇓⇓–33), our data suggest that early cortical neural progenitors and initial waves of differentiating neurons are unaffected in the presence of Ts21. The reported reductions in cortical neurons in DS brains may affect primarily late-born neurons that are not being evaluated in our study (24, 30, 34, 35).

Gene Expression Changes and Oxidative Stress Vulnerability.

To gain insight into the neuropathophysiology of Ts21 neurons, we analyzed global gene expression of 30-d-old neurons from DS1 and DS4 Ts21 iPSCs compared with those neurons from euploid DS2U iPSCs (Fig. 2F). Ts21 neurons displayed a preferential increase in expression of HSA21 genes, similar to Ts21 iPSCs in number and chromosomal distribution (Fig. S2 D–F). Virtually all of the HSA21 genes with expression that was changed in Ts21 neurons were increased (112 of 113) (Dataset S3), a significantly greater percentage of chromosomal content compared with other chromosomes (Fig. S2D). Furthermore, mean fold change for altered genes on HSA21 was significantly greater than all other chromosomes and generally reflected the 3:2 ratio of HSA21 genes (Fig. S2E and Dataset S4).

A central tenet in DS research is that symptoms are caused by modest increases in expression of trisomic genes and that these expression changes, in turn, cause dysregulation of normal cellular function through alterations in signaling pathways. Our data in both Ts21 iPSCs and neurons largely agree with this principle, whereby gene expression changes were based on gene dosage, but the greatest changes in Ts21 cells were observed for genes on chromosomes other than 21 (Figs. 1G and 2F and Datasets S2 and S4). The largest gene expression changes (more than fivefold) in both Ts21 iPSCs and neurons were of genes associated with transcriptional regulation and OS that we confirmed by quantitative PCR (qPCR) from all Ts21 lines (Figs. 1 G and H and 2 F and G). We did not observe expression of superoxide dismutase 1 (SOD1) and amyloid precursor protein (APP), HSA21 genes known to be involved in OS over the expected amount of 1.5-fold (Figs. 1H and 2G), but these modest increases may be sufficient to drive the response to OS. Ts21 neurons did exhibit increased OS, which was assayed by DHE (Fig. 2H) (DS1/4, P = 0.005; 2DS3, P = 0.03; n = 3 each). Furthermore, the increased expression of v-ets erythroblastosis virus E26 oncogene homolog 2 (ETS2), an HSA21-encoded transcription factor implicated in neuronal death and mitochondrial dysfunction (36), over the expected 1.5-fold in Ts21 neurons (Fig. 2G) prompted us to assess mitochondrial health in these cells. We detected increased mitochondrial membrane potential, a feature reported in human Ts21 neurons (37, 38) (Fig. 2I) (DS1, P = 0.04; 2DS3, P = 0.01; n = 2 each). However, we did not detect a significant increase in cell death in Ts21 neurons (Fig. 2J) (DS1, P = 0.62; 2DS3, P = 0.94; n = 2 each).

DS Neurons Are Deficient in Their Ability to Form Functional Synapses.

Changes in neuronal excitability and synaptic efficacy have been shown to contribute to cognitive impairment in DS mouse models (33, 39, 40). Whole-cell patch clamp recordings were performed on Ts21 iPSC-derived neurons between 5 and 6 wk, a time when human PSC-derived neurons have substantial synaptic activity (41, 42). Results revealed equivalent mean inward sodium and outward potassium currents in response to voltage steps as well as action potential (AP) generation in response to depolarizing current injections (Fig. 3 A and B) (Na⁺ current: DS1: P = 0.33; DS4: P = 0.37; 2DS3: P = 0.29; K⁺ current: DS1: P = 0.18; DS4: P = 0.11; 2DS3: P = 0.97; n = 6 replicates with ∼90 total cells/group) (SI Materials and Methods). Furthermore, no differences were seen in multiple physiological parameters, including cell size (capacitance), resting membrane potential, and AP amplitude in response to current injections (Fig. S3A), suggesting that basic physiological properties are unchanged in Ts21 iPSC-derived neurons at early stages. Interestingly, we found trends in all groups and significant reductions in most groups in the fraction of Ts21 iPSC-derived neurons that displayed spontaneous postsynaptic currents (sPSCs) (Fig. 3D) (DS1: P = 0.04; DS4: P = 0.01; 2DS3: P = 0.06; n = 6 each) as well as the sPSC frequency in Ts21 cells compared with euploid controls (Fig. 3 C–E) (DS1: P = 0.16; DS4: P = 0.04; 2DS3: P = 0.04; n = 6 each). This reduction was mirrored by a decrease in the number of synapsin⁺ punctae on Ts21 neurites (Fig. 3F and SI Materials and Methods) (DS1: P = 0.09; DS4: P = 0.17; 2DS3: P = 0.04; n = 3 replicates with ∼200 total neurites/group).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Forebrain Ts21 neurons display synaptic deficits across transmitter phenotype. (A and B) Representative whole-cell patch clamp traces illustrate that Ts21 (DS) did not show differences in (A) sodium (Na⁺) and potassium (K⁺) currents or (B) the number of APs in response to current injection. (C) Representative voltage clamp (−70 mV) traces show that both control and DS neurons displayed sPSCs. Expanded timescale (traces 3 and 4) illustrates sPSCs with different kinetics in both control and DS neurons. (D) Significantly fewer DS neurons display synaptic activity relative to controls (DS2U: 86 ± 3.9%; IMR90: 81 ± 3.2%). (E) DS neurons also displayed significantly lower frequencies of sPSCs compared with controls (DS2U: 0.48 Hz; IMR90: 0.92 Hz). (F) Representative images of control and DS β_III-tubulin⁺ neurites (red) displaying synapsin⁺ puncta (green; arrows). (Blue) Hoechst. Pooled data revealed fewer synapsin⁺ puncta in DS neurons compared with controls (DS2U: 2.9 ± 0.6/100 µm; IMR90: 4.4 ± 0.6/100 µm). (G) The proportion of excitatory and inhibitory sPSCs was not changed in DS cultures relative to controls [excitatory PSCs (ePSCs): DS2U: 0.54 ± 0.07 Hz; IMR90: 0.55 ± 0.11 Hz; iPSCs: DS2U: 0.46 ± 0.06 Hz; IMR90: 0.45 ± 0.09 Hz]. No differences were observed in (H) the fraction of GABA⁺ neurons compared with controls (DS2U: 47.7 ± 1.9%; IMR90: 46.2 ± 2.2%) or (I) the fraction of VGAT⁺/synapsin⁺ puncta (arrowheads) in DS cultures compared with controls (DS2U: 37.1 ± 6.1%; IMR90: 41.5 ± 4.5%). Error bars represent SEM. *P < 0.05. (Scale bars: F and I, 10 µm; H, 50 µm.)

Studies in DS mouse models have shown impaired synaptic efficacy because of increased inhibition (40, 43), which is potentially caused by overproduction of inhibitory interneurons at the expense of glutamatergic projection neurons (44). However, we found no significant difference in the ratio of excitatory to inhibitory sPSCs (Fig. 3G) (DS1: P = 0.38; DS4: P = 0.83; 2DS3: P = 0.86; n = 6 each), the percentage of neurons that expressed GABA (Fig. 3H) (DS1: P = 0.59; DS4: P = 0.94; 2DS3: P = 0.52; n = 3 each), or the fraction of synapses that stained positive for vesicular GABA transporter (VGAT) (Fig. 3I) (DS1: P = 0.50; DS4: P = 0.96; 2DS3: P = 0.77; n = 3 each). Importantly, the biophysical properties (e.g., amplitude, rise time, and decay constant) of sPSCs did not differ between groups (Fig. S3 C–E), and the proportion of cells that displayed AP firing (Fig. S3A) as well as spontaneous AP frequency (Fig. S3F) remained unchanged between groups. Together, these data suggest that changes in overall excitability of human Ts21 iPSC-derived neurons did not cause reductions in synaptic activity.