SEARCH

SEARCH BY CITATION

Keywords:

  • Induced pluripotent stem cells;
  • Disease modeling;
  • Down syndrome;
  • Trisomy 21;
  • Neural development;
  • Gene dosage

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Down syndrome (DS) is the most frequent cause of human congenital mental retardation. Cognitive deficits in DS result from perturbations of normal cellular processes both during development and in adult tissues, but the mechanisms underlying DS etiology remain poorly understood. To assess the ability of induced pluripotent stem cells (iPSCs) to model DS phenotypes, as a prototypical complex human disease, we generated bona fide DS and wild-type (WT) nonviral iPSCs by episomal reprogramming. DS iPSCs selectively overexpressed chromosome 21 genes, consistent with gene dosage, which was associated with deregulation of thousands of genes throughout the genome. DS and WT iPSCs were neurally converted at >95% efficiency and had remarkably similar lineage potency, differentiation kinetics, proliferation, and axon extension at early time points. However, at later time points DS cultures showed a twofold bias toward glial lineages. Moreover, DS neural cultures were up to two times more sensitive to oxidative stress-induced apoptosis, and this could be prevented by the antioxidant N-acetylcysteine. Our results reveal a striking complexity in the genetic alterations caused by trisomy 21 that are likely to underlie DS developmental phenotypes, and indicate a central role for defective early glial development in establishing developmental defects in DS brains. Furthermore, oxidative stress sensitivity is likely to contribute to the accelerated neurodegeneration seen in DS, and we provide proof of concept for screening corrective therapeutics using DS iPSCs and their derivatives. Nonviral DS iPSCs can therefore model features of complex human disease in vitro and provide a renewable and ethically unencumbered discovery platform. STEM CELLS2013;31:467–478


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Down syndrome (DS) is the most frequent cause of human congenital mental retardation and affects approximately 1 in 700 live births [1]. All individuals with DS exhibit mild to severe abnormalities in cognitive and language development, learning and memory impairments, and behavioral alterations [2, 3]. Although it has been over 100 years since the first description of DS by Langdon Down, and 50 years since Jérôme Lejeune found that an extra chromosome 21 was present in individuals with DS [4], the developmental aberrations and genetic mechanisms underlying the large range of often debilitating conditions experienced by individuals with DS remain poorly understood [5].

This is particularly true in the nervous system where human tissues are largely unavailable for the systematic study of development. Studies to date using post mortem or aborted fetal DS brains and fetal brain derived neurospheres indicate DS is associated with neurite and synapse formation defects [6–8], over-representation of glial lineages [9–11], and a reduction in total brain volume [12], in part due to reduced neurogenesis [13, 14] and increased propensity for neuronal apoptosis [11, 15–17]. However, DS fetal brains are generally not available at all developmental time points (particularly early and late gestation), which prevents pinpointing of the developmental origin of brain phenotypes in DS. Moreover, despite particular chr21 regions that are sufficient for DS brain phenotypes being identified by human segmental trisomies [18, 19], lack of a renewable source of developing tissues makes functional interrogation of these candidate genes in DS difficult or impossible. Model systems enabling studies linking causative genetic factors to precisely defined developmental and functional defects will be essential for the design of rationally targeted corrective strategies for DS.

Mouse models are currently the primary tools used to study DS etiology [20, 21]. However, brain development differs dramatically between mice and humans, presenting obstacles for using mouse models to study neurological features of human diseases such as DS [22]. Moreover, even the most recently developed DS mouse models, which are either trisomic for the complete human somatic autosome (HSA21) (Tc1) [23] or contain single chromosome duplications for all three mouse chromosomal regions orthologous to HSA21 [24], have a fundamentally nonequivalent genotype to human DS. It is widely acknowledged that, for Tc1 mice, it is likely that human genes will behave differently in a mouse compared to human cellular context [21]; for triple single chomosome duplication mice, the orthologous chromosomal regions still have many differences to HSA21 [25], a problem which is further compounded by complex genetic interactions involving trisomic genes [5]. Overall, these issues limit the utility of mouse models in dissecting the developmental and genetic mechanisms underlying DS pathogenesis, in particular within the nervous system.

Disease-specific induced pluripotent stem cells (iPSCs) have recently emerged as a new technology allowing human disease to be modeled in vitro, providing a powerful new system for research into pathogenic disease mechanisms, and the development of corrective therapeutics [26]. These models have proven to be particularly useful in investigating single gene disorders, however, it is largely yet to be seen how useful iPSCs will be in modeling the phenotypes of complex human diseases, such as DS [27]. We report the generation of multiple nonviral DS and wild-type (WT) iPSC lines by episomal reprogramming, which, to our knowledge, are the first reported nonviral complex human disease iPSC lines. We further use nonviral DS iPSCs to gain insight into the genetic and neural developmental features of DS etiology and provide proof of concept for using DS iPSCs to test corrective therapeutic strategies for DS in vitro. Our results indicate that disease-specific iPSCs are indeed able to model phenotypes of complex human diseases in vitro. DS iPSCs thus offer a renewable and ethically unencumbered source of early DS tissues, allowing interrogation of previously inaccessible windows of DS development.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Cell Culture

All work was carried out with the informed consent of patients under the approval of the Human Research Ethics Committee (HREC: 2008001651). Human embryonic stem cells (ESCs) and iPSCs were maintained on irradiated mouse embryonic fibroblasts (MEFs) in Dulbecco's modified Eagle's medium (DMEM)/F12 culture medium supplemented with 20% knockout serum replacement (KOSR), 0.1 mM nonessential amino acids, 1 mM L-glutamine, 0.1 mM β-mercaptoethanol, and 100 ng/ml human basic fibroblast growth factor (bFGF) (all from Invitrogen, Carlsbad, CA, http://www.invitrogen.com). Feeder-free culture on Matrigel (BD Biosciences, San Diego, CA, http://www.bdbiosciences.com) with conditioned medium was carried out in the presence of 100 ng/ml human bFGF, as described in [28]. Human male CRL2429 and female CRL1502 fibroblasts and male DS CCL-54 fibroblasts (ATCC) were cultured in DMEM supplemented with 10% heat-inactivated fetal bovine serum, 0.1 mM nonessential amino acids, and 2.0 mM Glutamax (all from Invitrogen).

Episomal Reprogramming of Human Fibroblasts

The oriP/EBNA1-based pCEP4 episomal vectors pEP4EO2SCK2MEN2L (7.3 μg) (Addgene plasmid 20924) and pEP4EO2SET2K (3.2 μg) (Addgene plasmid 20927) outlined in condition 4 of experiment 4 of Supporting Information Table 2 by Yu et al. [29] were transfected into 106 human fibroblasts using nucleofection (NHDF-VPD-1001 with U020 program, Amaxa). After 2 days of culture in fibroblast medium, 50,000 fibroblasts were seeded onto MEF feeder plates and adapted to KOSR medium containing 100 ng/ml b-FGF over a course of 3 days. The cultures were next supplemented with 20 mg/ml pifithrin-α and cultured for 3–4 weeks (first week with unconditioned KOSR + 100 ng/ml b-FGF and week 2–4 with MEF conditioned KOSR + 100 ng/ml b-FGF [Invitrogen]) under low oxygen (2%) conditions with daily medium changes. Subsequently, 20 colonies of each line were manually picked and cultured on fresh MEF feeder plates for 4 weeks before bulk expansion by Collagenase IV (Gibco, Grand Island, NY, http://www.invitrogen.com) passaging (as described [28]). Ultimately, two iPSC clones from each fibroblast donor that conformed to the criteria of iPSCs were established (Supporting Information Table S1).

Neural Differentiation of hiPSCs

hiPSC cultures were grown for approximately 5 days after mechanical passage in conditioned medium (as above) and changed directly into KOSR supplemented with 10 μM SB431542 (Sigma, Sydney, Australia, http://www.sigmaaldrich.com) and 5 μM dorsomorphin (Stemgent, Cambridge, MA, http://www.stemgent.com) for the first 6 and 12 days of differentiation, respectively, with media changes every 2 days (essentially as described in [30, 31]) to initiate neural conversion. KOSR was gradually substituted with N2B27 medium (neurobasal medium supplemented with Glutamax, N2 and B27 supplements [all from Gibco]): 25%, 50%, 75%, and 100% N2B27 in KOSR on days 4, 6, 8, and 10, respectively. Neurospheres were formed on day 6 of differentiation by 10 minutes incubation in 1 mg/ml Collagenase IV (Gibco) at 37°C and dislodging of large pieces of colonies by use of a cell scraper and P1000 pipette. Neuralized colony fragments were seeded into Ultra-low Cluster plates (Costar, Acton, MA, http://www.corning.com/lifesciences) where they aggregated into tight spheres. Neurospheres were subsequently expanded in Ultra-low Cluster plates, or were seeded onto Matrigel (BD). N2B27 media was changed every 3–4 days. Adherent cultures were passaged weekly by cell dissociation buffer (Sigma) at a 1:2–1:3 ratio, eventually leading to the complete dissociation of neurosphere aggregates.

Immunocytochemistry

Manually cut iPSC colony pieces grown on Matrigel-coated chamber slides (Thermo Fisher, Scientific, Waltham, MA, http://www.thermofisher.com) in conditioned medium for 3 days were fixed and permeabilized in 70% (v/v) ethanol (Sigma) for 20 minutes at −20°C. Plated neurospheres, or dissociated neural cultures, were grown on Matrigel-coated tissue culture plates for 5–7 days after seeding or passaging, respectively, and fixed in 4% paraformaldehyde (Sigma) in Dulbecco's phosphate-buffered saline (DPBS) (Gibco) for 20 minutes at RT, and permeabilized in 0.05% Triton X-100 (Thermo Fisher Scientific) in DPBS for 15 minutes at RT. All immunos were blocked for 30–60 minutes with 10% goat serum (Invitrogen) and 0.01% Triton X-100 in DPBS. Primary antibodies OCT4 (1:50, Millipore, Billerica, MA, http://www.millipore.com), SOX2 (1:50, Millipore), NANOG (1:50, Millipore), SSEA-4 (1:50, Millipore), TRA-1-81 (1:200, Millipore), TRA-1-60 (1:200, Millipore), TUBB3 (1:400, Millipore), MAP2 (1:400, Millipore), PAX6 (1:500, DSHB, Iowa City, IA, http://www.uiowa.edu/∼dshbwww), GFAP (1:1,000, DAKO, Glostrup, Denmark, http://www.dako.com) were applied for 3–4 hours at RT or overnight at 4°C. Species and isotype matched Alexa-Fluor conjugated secondary antibodies (1:1,000, Invitrogen) were applied for 1–2 hours at RT. Cells were washed in DPBS and stained with Hoechst or 4′,6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com). Fresh DPBS was replaced and cells were imaged immediately using an Olympus IX51 fluorescent microscope (Olympus, Tokyo, Japan, http://www.olympus-global.com) equipped with MicroPublisher 3.3 RTV CCD camera (QImaging, Surrey, BC, Canada, http://www.qimaging.com) using Q-Capture Pro v6.0 software.

Image Analysis

Neurosphere diameter, cell counts, and image scale bars were produced using standard features of ImageJ. Average neurite extension was calculated by measuring the length of every neurite (from cell body to the growth cone) in each image using a multimeasure ImageJ plug-in. See Supporting Information Text 1 for a full description of experimental design for each assay.

Flow Cytometry Analysis, Karyotyping, and DNA Fingerprinting

Flow cytometry was performed as previously described [32], using OCT4 (Millipore, 1:100), NANOG (Millipore, 1:100), SSEA-4 (Millipore, 1:50), and TRA-1-60 (Millipore, 1:250) primary antibodies. Species and isotype matched Alexa-Fluor conjugated secondary antibodies (1:2,000, Invitrogen) were used. Stained cells were analyzed by flow cytometry using a BD CSampler Accuri C6 fluorescent activated cell sorting (FACS) analyzer. Standard G-banding karyotype analysis was carried out by Sullivan Nicolaides Pathology (Taringa, Queensland, Australia, http://www.snp.com.au). To confirm the origin of iPS clones, short tandem repeat DNA fingerprinting analysis was carried out by DNA solutions (Wantirna, Victoria, Australia, http://www.dnasolutions.com.au).

RNA Isolation and cDNA Synthesis

Total RNA was extracted using the Qiagen RNeasy Mini RNA extraction kit (Qiagen, Hilden, Germany, http://www1.qiagen.com) according to the manufacturer's protocols. For microarray experiments, cDNA was synthesized using the TotalPrep RNA Amplification Kit (Illumina, San Diego, CA, http://www.illumina.com), and for neural conversion experiments, cDNA was synthesized using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, http://www.bio-rad.com), both according to manufacturer's protocol.

DNA Isolation

Total high molecular weight DNA was extracted by a traditional phenol/chloroform method. Briefly, cells were lysed with Proteinase K digestion buffer (100 mM Tris-HCl, pH 8.0, 5 mM EDTA, 0.5% SDS) at 55°C overnight. DNA was then precipitated in ethanol, and the precipitate was spooled, dried, and resuspended in 10 mM Tris-HCl, pH 8.0.

Reverse Transcriptase Polymerase Chain Reaction and Quantitative PCR

Reverse transcriptase polymerase chain reaction (RT-PCR) was performed with Taq DNA polymerase (Invitrogen). Two microliters out of 100 μL whole cell lysate produced by heating approximately 5 × 105 cells to 95°C for 10 minutes in lysis buffer (100 mM KCL, 50 mM Tris-HCL, pH 9.2, 0.2% v/v Triton X-100) was used as template. 30 cycles were performed before half of the reaction mixture was separated on a 1% Tris-acetate-EDTA (TAE) agarose gel. Quantitative PCR (qPCR) reactions used Ssofast Evagreen (Bio-Rad) with cDNA template (from ∼ 5 ng RNA equivalent/reaction) according to manufacturer's instructions using a C1000 Thermal Cycler (Bio-Rad) and analyzed as described in [33]. All primer sequences are listed in Supporting Information Table S2.

Southern Hybridization

Two probes specific to both episomal vectors, and one probe specific to each individual episomal vector created using the Rediprime II random prime labeling kit (GE Healthcare, Pittsburgh, PA, http://www.gehealthcare.com) were hybridized overnight in Ambion ULTRAHyb buffer (Life Technologies, Carlsbad CA, http://www.lifetech.com) to gel separated digested genomic DNA blotted onto Amersham Hybond-XL positively charged nylon membranes (GE Healthcare). Membranes were exposed to Fujifilm Super RX medical x-ray film at −80°C (Fujifilm, Tokyo, Japan, http://www.fujifilm.com). For more information including details about probe design and digestion strategies used, see Supporting Information Materials and Methods.

Bisulfite Sequencing of CpG Islands in OCT4 and NANOG Promoters

Genomic DNA (0.5–1 μg) was bisulfite converted using MethylEasy DNA Bisulfite Modification Kit (Human Genetic Signatures, Randwick, NSW, Australia, http://www.geneticsignatures.com) as per manufacturer instructions. OCT4 and NANOG promoter CpG insland sequences were amplified using platinum Taq DNA Polymerase (Invitrogen). For the NANOG promoter, a nested PCR amplification strategy was used. PCR conditions were those used in [29]. The resultant PCR products were cloned into a pGEM-Teasy vector (Promega, Madison, WI, http://www.promega.com) and sequenced. All primer sequences are listed in Supporting Information Table S2.

Microarray Analysis

cDNA was hybridized to Illumina HT12 v4 BeadChip microarrays. For each probe, raw signal intensity was calculated using GenomeStudio software (Illumina). Normalization and statistical analysis of microarray data were performed in R using the Bioconductor packages Lumi and Limma. Statistically significant differential gene expression was calculated using moderated B statistic methods or using linear models to fit the data and empirical Bayes B statistic methods (as described in [34–36]). GeneSpring GX-7.3 (Agilent Technologies, Palo Alto, CA, http://www.agilent.com) was used for data visualization. Hierarchical cluster analyses were carried out with 1 - Pearson correlation coefficient as the distance measurement. The maximum distance between cluster members was used as the basis to merge lower-level clusters (complete linkage) into higher-level clusters. Microarray data are available for download from the Gene Expression Omnibus repository under the accession numbers GSE42956, at http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE42956.

Heat Map Generation

The log intensities of the 500 most differentially expressed genes between pluripotent cells and fibroblasts were standardized, so that their expression values across all samples have mean 0 and standard deviation 1. The standardized values were reordered and displayed in a heat map, with the spectrum ranging from green (low level) to red (high level).

Chromosomal Distribution of Transcript Enrichment

Probe level lists of differential expression for each comparison were annotated with chromosomal positions using Galaxy (build: $Rev 5054:1c184746177e$) to match probe IDs from differential expression data with those from the Illumina annotation file (HumanHT-12_V4_0_R1_15002873_B.txt). 100% of probes were annotated successfully. Genes represented on the microarray by more than one probe were collapsed into lists of unique enriched genes. Chromosomal gene enrichment was calculated as the number of unique enriched genes mapping to a chromosome normalized by the number of unique genes represented on the microarray for that chromosome.

Teratoma Formation

iPSCs expanded on MEFs were injected into hind limb muscles of 6-week-old severe combined immunodeficient (SCID) mice (approximately 5 × 106 cells per mouse). After 8–10 weeks, teratomas were dissected and fixed in 4% paraformaldehyde. Samples were embedded in paraffin, processed with hematoxylin and eosin staining, and examined for the presence of representatives of the three germlayers by an independent pathologist (Stemcore, Australian Stem Cell Centre, St Lucia, Queensland, Australia). All mouse procedures were conducted under local ethical guidelines and after gaining permission from the local animal ethics committee (AIBN/084/09/ASCC/LEJEUNE; The University of Queensland, QLD, Australia).

Quantification of Reactive Oxygen Species Production

Reactive oxygen species (ROS) production was quantified in live day 28 neural cultures, grown for 24 hours prior in N2B27 with standard B27 supplement substituted for custom B27 without antioxidants (Gibco), using Image-iT LIVE Green Reactive Oxygen Species Detection Kit (Molecular Probes, Eugene, OR, http://probes.invitrogen.com). Staining was carried out as per manufacturer's instructions, and stained cells were dissociated from the plate using TrypLE (Gibco) and analyzed by flow cytometry using a BD CSampler Accuri C6 FACS analyzer.

H2O2 Sensitivity Assays and Terminal Deoxynucleotidyl Transferase dUTP Nick End Labeling Staining

Dissociated day 28 neural cultures were grown to approximately 60% confluence in 24 well plates, and H2O2 was titrated across individual wells in all cells lines, diluted in 10% N2B27 medium in Neurobasal medium (Gibco) but with B27 replaced with custom B27 supplement minus antioxidants (Gibco). For rescue, 100 μM N-acetylcysteine (NAC) (Sigma) was added along with 100 μM H2O2. Cells were incubated for 24 hours in a 37°C cell culture incubator before terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) labeling with APO-BrdU TUNEL Assay Kit (Invitrogen), as per manufacturer's instructions but adapted to stain adherent rather than suspended cells.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Generation and Characterization of Nonviral, Integration-Free DS and WT iPSCs

DS and WT fibroblast cell lines were reprogrammed by transfection with the pCEP4 episomal vectors pEP4EO2SCK2MEN2L and pEP4EO2SET2K (from [29]) using AMAXA nucleofection and, therefore, did not rely on viral particles. Two pluripotent iPSC lines were generated from each of two WT and one DS donors as determined by immunocytochemistry detection of pluripotency markers OCT4, NANOG, SOX2, SSEA-4, TRA-1-81, and TRA-1-60 (Fig. 1A; Supporting Information Fig. S1A), and formation of teratomas composed of derivatives of all three germ layers (Fig. 1B; Supporting Information Fig. S1B). Standard karyotype analysis confirmed normal or trisomy 21 karyotypes in WT and DS iPSCs, respectively (Fig. 1E; Supporting Information Fig. S1D), and DNA fingerprint analysis confirmed that iPSC clones were indeed derived from their donor cell lines. In total four WT (denoted WT1-4-iPS) and two DS lines (denoted DS1-2-iPS) were generated (Supporting Information Table S1), and of these, two WT (WT1- and WT2-iPS) and two DS lines (DS1- and DS2-iPS) were further characterized and used for neural conversion experiments.

thumbnail image

Figure 1. Characterization and transcriptome profiling of bona fide virus- and integration-free DS and WT iPSCs. (A): Immunocytochemical detection of pluripotency markers OCT4, SOX2, NANOG, SSEA-4, Tra-1-81, and Tra-1-60 in DS and WT iPSCs. (B): Histology images of derivatives of the three germ layers found in teratomas formed after injection of DS and WT iPSCs into severe combined immunodeficient (SCID) mice demonstrating trilineage competence of nonviral DS and WT iPSCs. (C): Absence of free or integrated episomal vectors from DS and WT iPSCs; no product after 35 cycles of polymerase chain reaction using whole cell lysate as template compared to GAPDH and episomally transfected fibroblast positive controls. (D): Bisulfite sequencing showing OCT4 and NANOG promoter CpG island demethylation in H9 hESCs, DS iPSCs, and WT iPSCs compared to donor fibroblast lines. (E): Trisomy 21 karyotype in DS iPSCs. (F): Hierarchical clustering of 10,837 genes in Mel 1 and H9 hESCs, DS and WT iPSCs, and donor fibroblast lines using PCC as a measure of distance, with heatmap representing the expression levels of the 500 most differentially expressed genes across all cell lines. DS and WT iPSCs show the greatest differential expression relative to fibroblasts, and cluster closely with hESCs as a function of genotype. (G): Genome-wide transcriptome deregulation in DS iPSCs and DS fibroblasts compared to WT, illustrated by the distribution of enriched transcripts belonging to each chromosome. A clear bias of enriched transcripts mapping to chr21 is evident. Microarray data are available for download from the Gene Expression Omnibus (GEO) repository under the accession numbers GSE42956, at http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE42956. Scale bars = 100 μm. See also Supporting Information Figures S1–S5 for further characterization of virus- and integration-free DS and WT iPSCs including additional clones, Supporting Information Figure S6 for overlap of chr21 gene overexpression in DS iPSCs and previous array analyses in DS, Supporting Information Table S1 for a summary of generated DS and WT iPSC lines, and Data File S1 for a list of differentially expressed genes in DS and WT iPSCs. Abbreviations: DS, Down syndrome; hESC, human embryonic stem cells; iPSC, induced pluripotent stem cells; PCC, Pearson's correlation coefficient; WT, wild type.

Download figure to PowerPoint

In addition to these core iPSC lines, another iPSC line was generated from DS fibroblasts that was surprisingly found to possess a normal euploid genotype (denoted euDS-iPS). Since many DS individuals are mosaic for trisomy 21, this line seemed to be the result of reprogramming a euploid cell from a mosaic DS fibroblast population. DNA profiling demonstrated that this line was indeed generated from the same donor fibroblasts as the other two DS iPSC lines, and the robust pluripotent phenotype of this line was confirmed by the same assays used for the core iPSC lines (Supporting Information Fig. S2). Importantly, due to its euploid but otherwise isogenic genotype, euDS-iPSCs served as a perfect control for analyses comparing DS and WT iPSC lines.

FACS analysis demonstrated that greater than 90% of DS and WT iPSCs express OCT4, NANOG, SSEA-4, and TRA-1-60 when grown in maintenance conditions (Supporting Information Fig. S1F), and embryoid body differentiation of iPSCs into cells representing all three germ layers (Supporting Information Fig. S1E) further confirmed the robust pluripotent phenotype of our lines. The absence of episomal vector sequence from our iPSCs (including both free and integrated episomes) was confirmed by PCR analysis of whole cell lysates using primers specific to each episomal plasmid (Fig. 1C) and Southern blot analysis of genomic DNA with multiple probes complementary to several sites on each episomal vector (Supporting Information Fig. S3). Bisulfite sequencing of CpG islands within OCT4 and NANOG promoters in our iPSC lines and their fibroblast donor lines confirmed that reprogramming was associated with pluripotency promoter demethylation to levels resembling H9 hESCs (Fig. 1D; Supporting Information Fig. S1C). Collectively, these data confirmed that we had generated multiple bona fide virus and integration-free DS and WT iPSC lines, and an isogenic control euploid DS iPSC line (Supporting Information Table S1).

Chr21 Gene Dosage Effects and Global Transcriptome Deregulation of DS iPSCs

We next aimed to test whether episomal reprogramming resulted in a typical pluripotent transcriptome in DS and WT iPSCs. To do this, we profiled gene expression in each of four WT and two DS iPSC lines, their parent fibroblasts, and two hESC lines using Illumina HT12 v4 Beadchip microarrays. Differential gene expression analysis revealed that hESCs, DS-, and WT- iPSCs displayed the greatest differential expression when compared to their parent fibroblast cell lines, and, notably, the top 50 pluripotency-associated genes were differentially expressed between iPSCs and fibroblasts (Supporting Information Data File S1), consistent with robust transformation of episomally reprogrammed fibroblasts into a pluripotent phenotype. Hierarchical clustering further revealed that both DS and WT iPSCs cluster closely with Mel1 and H9 hESC lines, and all DS and WT iPSC lines robustly passed pluritest (Supporting Information Fig. S4), indicating that episomally reprogrammed DS and WT iPSCs have indeed been reprogrammed effectively and express a comparable pluripotent transcriptome to ESCs (Fig. 1F).

Since it is a prerequisite for DS iPSCs to be useful as a model system that they retain a trisomy 21-associated transcriptional output following reprogramming, our next question was whether DS and WT iPSCs were distinguishable by their respective transcriptomes. Indeed, DS and WT iPSCs clustered distinctly, suggesting that transcriptome differences between our iPSC lines were a function of genotype (Fig. 1F). Moreover, euDS-iPSCs clustered between WT-iPSCs, and well away from the otherwise isogenic trisomic DS-iPSCs, providing strong proof that transcriptome differences between DS and WT iPSCs were indeed the result of trisomy of chromosome 21, and not due to the influence of polymorphisms, genetic background, age or sex differences between lines (Supporting Information Fig. S5). We then further sought to investigate the nature of the transcriptional deregulation caused by trisomy 21 in DS iPSCs and fibroblasts. Differential expression analysis surprisingly revealed that a total of 2,413 genes were significantly differentially expressed by more than 1.5-fold (linear) between DS and WT iPSCs, indicating that trisomy 21 is associated with significant global transcriptional deregulation in DS compared to WT iPSCs (Supporting Information Data File S1). Notably, these genes were commonly overexpressed (32.8% of probes) or underexpressed (67.2% of probes). Overall, this analysis revealed a surprising complexity in the transcriptional deregulation caused by trisomy 21 in naïve DS cells, which is likely to underlie DS developmental phenotypes.

We next focused on comparing chromosome 21 gene expression between DS- and WT- iPSCs. Only 63 out of 431 chr21 genes represented by the array were significantly overexpressed in DS iPSCs, and 7 were significantly underexpressed, indicating that many chr21 genes in fact appear to escape dosage effects (Supporting Information Data File S1). This may be attributable to the complex transcriptional interactions that occur between these genes, the many other deregulated genes, and their associated protein products. Our data showed some overlap with previously reported array analyses of DS tissues, with between 13% and 64% of chr21 overexpressed genes identified by our analysis also appearing in the other tissue-specific datasets, and our data containing between 31% and 46% of the chr21 overexpressed genes identified in other individual datasets (Supporting Information Fig. S6). Differences between gene lists in these analyses likely result from tissue-specific overexpression profiles, as no studies to date have examined pluripotent DS cells. Importantly, we found a fourfold higher proportion of overexpressed genes that map to chr21, after normalizing for the gene content represented by the array for each chromosome, than to any other chromosome in both DS iPSCs and fibroblasts, compared to WT (Fig. 1G). This finding indicated that although a fraction of chr21 genes escape dosage effects, chr21 genes are nonetheless selectively overexpressed in DS iPSCs and fibroblasts compared to genes from other chromosomes. Collectively, these data demonstrate that nonviral DS and WT iPSCs were effectively reprogrammed to an ESC like pluripotent state, that DS iPSCs were distinguishable from WT iPSCs by their transcriptome, and that DS iPSCs selectively overexpress a subset of chr21 genes within a globally deregulated transcriptome.

Similar Potency and Timing of Neural Conversion of DS and WT iPSCs

To model DS neural developmental phenotypes, we directed differentiation of DS and WT iPSCs into neurons using a modified dual SMAD inhibition protocol [30]. We replaced Noggin with the small molecule dorsomorphin, which has been shown to improve robustness against cell line to cell line variability in neural conversion potency [31]. We further modified the protocol by forming neurospheres from PAX6 enriched day 6 cultures (Supporting Information Fig. S7A), which we found reduced the impact of cell density on neural conversion efficiency, and allowed for the efficient expansion of neural cells.

DS and WT neurospheres plated on Matrigel on day 12 produced abundant neurons at the sphere periphery by day 17 (Fig. 2A), and when dissociated over two passages to total day 28 both DS and WT neurospheres were found to contain comparable proportions of PAX6 positive neural progenitors (Fig. 2B) and TUBB3 positive immature neurons (p > .05, Student's t test) (Fig. 2C), overall indicating a potency of approximately 95% neural conversion, and demonstrating similar amounts of neurogenesis in early DS and WT neural cultures. These cultures also expressed similar levels of OTX1 and OTX2 compared to HOXB4, indicating a similar anterior regional identity in DS- and WT-iPSC derived neural cultures (Supporting Information Fig. S7B).

thumbnail image

Figure 2. Similar potency and timing of neural conversion in DS and WT induced pluripotent stem cells. (A): Immunocytochemical detection of neuronal markers TUBB3 and MAP2 in day 17 cultures, 5 days after neurospheres were plated on Matrigel. (B): Immunocytochemical detection of neural progenitor marker PAX6 in single cell dissociated day 28 cultures. (C): Quantification of PAX6 and TUBB3 positive cell numbers in day 28 single cell dissociated cultures as detected by immunocytochemistry shows no significant difference between DS and WT. (D): Quantitative polymerase chain reaction (qPCR) profiling of core neural gene expression over the first 25 days of neural conversion in DS and WT cultures shows no significant differences. (E): qPCR showing consistent overexpression of chr21 genes over the first 25 days of neural conversion in DS compared to WT cultures. Scale bars = 100 μm. All graphs present the pooled average value for each data point from multiple replicate experiments, and error bars represent SEM. # indicates lack of statistical significance (p > .05) by Student's t test in (C) or two-way ANOVA with Bonferroni post-test in (D). See also Supporting Information Figure S7 for additional data comparing DS and WT neural conversion. Abbreviations: DS, Down syndrome; DCX, doublecortin; WT, wild type.

Download figure to PowerPoint

To further test for potentially more subtle differences in neural conversion kinetics in DS and WT iPSCs, we collected total RNA across six time points spanning the first 25 days of directed neural differentiation and profiled the expression of core neurogenic genes. We found robust induction of genes involved in neural lineage specification (PAX6, SOX1), neural stem cell maintenance (SOX9, SOX2), and neuronal lineage entry and neurogenesis (MASH1, DCX) in DS and WT cultures, with no significant differences at any time points (p > .05, two-way ANOVA with Bonferroni post-test) (Fig. 2D). These data were consistent with immunocytochemical quantification of PAX6 and TUBB3 positive cells and thus further support the previously observed similar potency of DS and WT iPSC neural conversion. Moreover, these data indicate a highly similar kinetic profile of early neural specification, neural stem cell maintenance, and neurogenesis in DS and WT neural cultures. It has previously been suggested that overexpression of genes such as APP, DYRK1A, DSCR1, and SOD1, belonging to the so-called “Down syndrome critical region,” is sufficient to disturb normal developmental processes in DS. However, our analyses found consistent 1.5–3-fold overexpression of these genes in DS (Fig. 2E), despite the overall highly similar neural differentiation in DS and WT cultures, indicating that overexpression of these genes is not in fact sufficient to disturb early neural lineage differentiation in DS.

Similar Proliferation and Neurite Extension of DS and WT Neural Cultures

We next sought to test whether neural stem cell proliferation defects reported in fetal DS brains were seen in DS iPSC derived neural cultures at early developmental time points. To do this, we grew individual DS and WT neurospheres in separate wells of 96-well cell culture plates in maintenance medium, and calculated neurosphere volume as a measure of cell number every 2 days for up to 10 weeks of culture. Over this period, both DS and WT neurospheres stably expanded, and we detected no significant difference in proliferation between DS and WT neurospheres (Fig. 3A; Supporting Information Fig. S8A) (p > .05, two-way ANOVA with Bonferroni post-test, n = 40 per line). Combined with the similarity in neurogenesis, PAX6 protein and neural stem cell maintenance gene expression in early DS neural cultures, these results indicate that proliferation defects in DS might arise later during development and are not present at early stages. It has been reported that DS neurons display defective neurite extension, however, we also detected no difference in the average neurite extension of DS and WT neurospheres seeded onto Matrigel after 24 hours at four different time points spanning 10 weeks in culture (p > .05, Student's t test) (Fig. 3B, for individual time points see Supporting Information Fig. S8B), indicating that these defects are also not detectable at early stages of development. Collectively, these data (including the previous results section) indicate that the early stages of DS neuronal differentiation in vitro are surprisingly normal, not showing many of the phenotypes that have been observed in DS brains and fetal brain derived neurospheres, suggesting that these phenotypes may arise later during development.

thumbnail image

Figure 3. DS neurospheres show no proliferation or neurite extension defect at early time points. (A): Representative day 60 neurospheres that were used to calculate sphere volume as a measure of cell number, and the resulting cumulative cell number doublings over 60 days in culture. DS and WT neurospheres both stably expanded, and DS neurospheres showed no proliferation defect at early time points. (B): Representative images of neurite extensions produced 24 hours after plating neurospheres on Matrigel, and pooled data showing no difference in the average neurite extension in DS and WT neural cultures. Scale bars = 100 μm. Error bars represent SEM. # indicates lack of statistical significance (p > .05) by Student's t test in (B) or two-way ANOVA with Bonferroni post-test in (A). See also Supporting Information Figure S8 for average neurosphere doubling time, and neurite extension data for individual time points. Abbreviations: DS, Down syndrome; WT, wild type.

Download figure to PowerPoint

DS Neural Cultures Show a Developmental Bias Toward Glial Lineages

Given the similarity of early neural differentiation in DS iPSCs, we next asked whether glial lineage development was altered during in vitro differentiation of DS neural cultures, reflecting differences in glial cell representation seen in fetal and adult DS brains and biases toward glial lineages reported in studies using DS fetal brain derived neurospheres. To do this, we seeded DS- and WT-iPSC derived neurospheres onto Matrigel and stained for neuronal (TUBB3, MAP2) and glial (GFAP) lineage markers after 1 week (Fig. 4A; Supporting Information Fig. S8E). Neuronal cells were detected at all time points in DS and WT cultures, and glial cells were detected only after 60 days of culture in both DS and WT cultures (Fig. 4B), indicating that gliogenesis occurs with normal timing in DS. Notably, however, DS neurospheres produced an approximately twofold higher proportion of glia than WT cultures at both day 60 (p < .05 by two-way ANOVA with Bonferroni post-test) and day 80 (p < .0001 by two-way ANOVA with Bonferroni post-test) (Fig. 3B). This overproduction of glia appeared to occur at the expense of neurons, since a smaller total number of neurons, and a higher total number of glia were detected in DS cultures at these time points (Supporting Information Fig. S8C). This is suggestive of a glial-lineage bias in DS, where differentiating cells are prone toward glial, as opposed to neuronal cell fates. Alternatively this result could reflect increased proliferation of glia or selective degeneration of neurons in DS. However, we found no evidence of elevated proliferation in DS neurospheres up to day 60 that might reflect overproduction of glia, and we found no evidence of selective degeneration of DS neurons or progenitors under standard culture conditions (see next section).

thumbnail image

Figure 4. DS neurospheres undergo a normally timed but exaggerated gliogenic switch. (A): Immunocytochemical detection of neuronal (TUBB3) and glial (GFAP) markers in D60 neurospheres, 1 week after plating on Matrigel, showing a significant bias toward glial lineages in DS neurospheres. (B): Quantification of the proportion of glia produced by DS and WT neurospheres at D40, D60, and D80. Glia appear at the same time in DS and WT cultures, with increasingly more glia at D60 and D80, respectively. Notably DS neurospheres produce a twofold to threefold higher proportion of glia than WT, suggestive of a developmental bias toward glial lineages in DS. (C): Quantitative polymerase chain reaction quantification of GFAP mRNA expression in neurospheres correlates with immunocytochemistry data, showing GFAP mRNA is detected in both DS and WT cultures around D50 and is more dramatically induced in DS. Scale bars = 100 μm. Error bars represent SEM. # indicates lack of statistical significance (p > .05); *, p < .05; ****, p < .0001 by two-way ANOVA with Bonferroni post-test. See also Supporting Information Figure S8 for total counts of neuronal and glial cells in DS and WT cultures, showing DS neurospheres produced a higher total number of glia, and lower total number of neurons than WT neurospheres. Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; DS, Down syndrome; WT, wild type.

Download figure to PowerPoint

qPCR analysis of RNA from neurospheres collected over a time course of 60 days in culture confirmed the normal kinetics of gliogenesis in DS, demonstrating that GFAP mRNA was upregulated similarly in DS and WT neurospheres around D50 (Fig. 5C), consistent with our immunocytochemical detection of GFAP positive cells at day 60, but not day 40. Moreover, this analysis revealed a consistent overexpression of GFAP mRNA in DS during gliogenesis (Fig. 5C; Supporting Information Fig. S8G), with a 2.41-fold overexpression of GFAP at day 60, supporting our detection of a twofold glial bias in DS by immunocytochemistry. We also detected 1.4–3-fold overexpression of other glial development associated genes including S100B, GLAST, and NFIA from the beginning of gliogenesis in DS cultures (Supporting Information Fig. S8D, S8F). Importantly, since RNA was sampled from entire pooled neurospheres, this indicates that the glial bias detected by immunocytochemistry reflects the overall composition of DS neural cultures. Therefore, we conclude that DS neural development is associated with a significant developmental bias toward glial lineages, during normally timed gliogenesis, and this occurs at the expense of neurons.

thumbnail image

Figure 5. DS neural cultures are sensitive to oxidative stress-induced apoptosis and this can be pharmacologically prevented. (A): Representative immunocytochemistry images with TUNEL labeled (green) apoptotic cells and PI stained (red) live cells in DS and WT day 28 neural cultures, following exposure to the indicated concentration of H2O2 and NAC (100 μM). (B): Quantification of TUNEL positive cells revealed that DS and WT neural cultures underwent apoptosis in a dose-dependent fashion in response to H2O2 insult. Notably, DS neural cultures are significantly more sensitive to 100 μM H2O2 than WT, and this can be prevented by the addition of 100 μM NAC. (C): Pooled data from FACS-based quantification of ROS production in DS and WT day 28 neural cultures, showing no significant difference between DS and WT basal ROS levels, suggesting that oxidative stress sensitivity in DS arises from a ROS independent mechanism. Scale bars = 100 μm. Error bars represent SEM. # indicates lack of statistical significance (p > .05) by two-way ANOVA with Bonferroni post-test in (B), or Student's t test in (C). ****, p < .0001 by two-way ANOVA with Bonferroni post-test in (B). Abbreviations: DS, Down syndrome; NAC, N-acetylcysteine; ROS, reactive oxygen species; TUNEL, terminal deoxynucleotidyl transferase dUTP nick end labeling; WT, wild type.

Download figure to PowerPoint

DS Neural Cultures Are Sensitive to Oxidative Stress-Induced Apoptosis and This Can Be Pharmacologically Prevented

Increased propensity for neuronal apoptosis and redox defects have been reported in DS, and so we set out to test whether DS iPSC derived neurons representing early developmental stages also displayed these deficiencies. We exposed day 28 dissociated DS and WT neural cultures (which are of equivalent cellular composition [Fig. 2]) to a titration of H2O2 for 24 hours, and identified apoptotic cells by fluorescent TUNEL labeling (Fig. 5A). Quantification revealed similar levels of apoptosis in DS and WT cultures under normal culture conditions, and that both DS and WT neural cultures underwent apoptosis in a dose-dependent fashion in response to H2O2 insult (Fig. 5B). Notably, DS cells were twice as sensitive to 100 μM H2O2, with 83.1% TUNEL positive cells in DS, compared to 44.4% TUNEL positive cells in WT neural cultures (p < .0001 by two-way ANOVA with Bonferroni post-test). It has previously been reported that DS neurons produce higher amounts of ROS, and so we asked whether DS sensitivity to H2O2 was potentially related to differences in basal ROS production. We quantified ROS production in DS and WT neural cultures by FACS analysis after staining with a ROS sensitive dye. However, we found no evidence of significantly elevated ROS production in DS neural cultures compared to WT (p > .05 by Student's t test) (Fig. 5C). Collectively, these data indicated that DS neural cells are up to twice as sensitive to H2O2 insult compared to WT neural cells, and that this sensitivity arises from a ROS production independent mechanism.

We next tested whether sensitivity to H2O2 insult could be prevented in DS cultures, and found that apoptosis in the presence of 100 μM H2O2 could be restored to near baseline levels by addition of the antioxidant NAC (100 μM) (p < .0001 by two-way ANOVA with Bonferroni post-test) (Fig. 5B). In the presence of NAC no significant difference was seen between the TUNEL positive fraction of cells in DS (25.9%) and WT (18.8%) neural cultures, despite 100 μM H2O2 insult (p > .05 by two-way ANOVA with Bonferroni post-test). Importantly, these data provide proof of concept for testing corrective strategies aiming to offset functional defects in DS using DS iPSCs, with significant implications for future works aiming to use disease-specific iPSCs for high throughput screening of drugs to correct complex disease phenotypes in vitro.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Disease-specific iPSCs have proven to be powerful model systems of single gene human disorders, however, their usefulness in modeling complex disease phenotypes, such as DS, is less well-established [27]. To address this report the generation of bona fide virus-free DS iPSCs by episomal reprogramming and demonstrate their usefulness in modeling early DS genetic and neural developmental phenotypes. While directed neural differentiation somewhat unexpectedly revealed that there are no major differences between DS and WT iPSC early in vitro neuronal development, we identified a twofold developmental bias toward glial lineages originating with the onset of gliogenesis in DS neural cultures. DS iPSC derived neural cultures were also up to twice as sensitive to H2O2 insult, and this was pharmacologically preventable by the addition of NAC. We additionally provide evidence that DS iPSCs selectively overexpress a subset of chr21 genes within a globally deregulated transcriptome.

This latter result is consistent with previous array-based transcriptome profiling experiments in human DS [37–41], and DS mouse model [42–44] tissues, which have generally found evidence of 1.5-fold overexpression of trisomic genes reflecting gene dosage, that some trisomic genes appear to be compensated or underexpressed [41, 44], and that trisomy may additionally be associated with misexpression of significant numbers of other genes throughout the genome [38, 39, 42, 43]. Since DS iPSCs represent a very early, inner-cell-mass-equivalent stage of development, our results imply that DS developmental phenotypes may result from not only elevated dosage of a subset of chr21 genes, but also altered expression of thousands of non-chr21 genes, adding significant additional complexity to our understanding of the potential molecular genetic etiology of DS phenotypes. It also follows from these results that deregulated non-chr21 genes may in some cases be appropriate therapeutic targets in DS, and that non-chr21 genetic background is likely to contribute to the variable manifestation of DS phenotypes. It remains to be determined what mechanisms control which chr21 genes are overexpressed, compensated, or underexpressed in DS, and furthermore whether non-chr21 gene deregulation is stochastic, or the result of chr21 dosage related mechanisms influencing specific non-chr21 genes, such as trans-regulatory-effects of chr21 gene products. Due to the large array of genetic tools available for the manipulation of human pluripotent cells, DS iPSCs provide a powerful model system with which to dissect such molecular mechanisms of trisomy in DS. DS iPSCs are also likely to prove useful in dissecting the potential contribution of candidate genes identified by human segmental trisomies to specific DS phenotypes.

We found that DS iPSC derived neural progenitors did not display evidence of the proliferation defects that have previously been reported in DS [6, 13, 14]. However, our results are consistent with studies on fetal DS brain derived neurospheres, which have found that these neurospheres only show proliferation defects after extended culture (>10 weeks) [45], and seem to similarly indicate that proliferation defects may only arise later during DS neural development. We additionally found that early DS iPSC derived neural cultures did not display evidence of reduced neurogenesis associated with DS brain development [13, 14]. In fact, DS iPSCs had very similar kinetics of early in vitro neural development to WT, indicating that these phenotypes too likely only arise later during DS development. Despite this lack of early neuronal phenotype in DS neural cultures, we found consistent overexpression of Down syndrome critical region genes, indicating that overexpression of these genes is not in fact intrinsically sufficient to disturb early neural development. However, it remains possible that proliferation and neurogenesis defects were not seen by our analysis due to limitations of DS iPSCs as a model system. For example, it is possible that the regional identity of our neurospheres does not correspond to a hypocellular brain region in DS, or that these phenotypes arise from more complex 3D structural defects not captured by our assays.

Overrepresentation of glia has previously been reported in DS brains [9, 10, 14, 15], Ts65dn mice brains [14], and DS fetal brain derived neurospheres [11]. However, whether this phenotype arises from a developmental mechanism, an inflammatory mechanism triggering glial overproduction, or simply reflects reduced neurogenesis or neuronal cell death, has remained unclear. Neural development naturally proceeds with an initial exclusively neurogenic phase, which is followed by a transition into a more gliogenic period, often referred to as a “gliogenic switch” [46]. We found that during in vitro neural differentiation, DS iPSCs displayed a twofold bias toward glial over neuronal lineages during gliogenesis. Combined with our observation that there was no defect in proliferation or basal neuronal apoptosis levels during early in vitro differentiation of DS iPSCs, this suggests that DS cultures underwent a normally timed but exaggerated gliogenic switch, and therefore that over-representation of glia in DS brains originates with an early developmental mechanism. It is noteworthy that exaggerated gliogenesis likely causes additional defects in DS brain development. As such, glial bias could be related to neurogenesis and proliferation defects seen in DS brains, possibly by reducing neuronal precursor specification or overall cell-cycling speeds, respectively, which would explain the absence of these defects at early time points by our analysis. The contribution of glial bias to neuronal phenotypes could be tested by assays comparing pregliogenesis and postgliogenesis neuronal phenotypes in DS iPSC derived neural cultures. The mechanism responsible for glial bias in DS remains to be determined but could be related to perturbations of Notch signaling by chr21 genes such as BACE2, a β-secretase gene involved in Notch processing, or overexpression of chr21 genes such as APP and S100B, both of which have been shown to promote gliogenesis.

DS is invariably associated with early onset Alzheimer's disease (AD) [47], and it was recently reported that DS iPSC derived neurons showed hallmarks of AD pathology within months of culture, including elevated amyloid beta (Aβ) 40/42 production, amyloid plaque formation, and hyperphosphorylation, mislocalization and elevated secretion of tau [48]. We found that DS iPSC derived neurons recapitulated the sensitivity to oxidative stress-induced apoptosis seen in DS fetal brain derived neurons [16], and that this seemed to result from a ROS production independent mechanism. Notably, this is likely to result in vulnerability to oxidative stress arising from progressive amyloid and tau pathologies, and normal aging, and could contribute to the accelerated neurodegeneration responsible for the invariable early onset AD pathology seen in DS. We were also able to offset oxidative stress sensitivity by the addition of NAC, suggesting that antioxidants may offer some benefits to DS patients. Previous studies have suggested that predisposition to apoptosis in DS neurons may be related to overexpression of the chr21 gene ETS2 activating p53-mediated mitochondrial apoptosis [49, 50], however, the mechanism responsible for oxidative stress sensitivity in DS iPSC derived neurons remains to be determined. Overall, these results offer promise that DS iPSCs will prove useful in future investigations into the mechanisms underlying progressive and accelerated AD pathologies in DS, and testing corrective therapeutic strategies combating neurodegeneration in DS [26].

Collectively, our results indicate that iPSCs can indeed model phenotypes of complex human disease in vitro and demonstrate that nonviral DS iPSCs are a powerful discovery platform allowing interrogation of phenotypes affecting previously inaccessible early stages of DS development. Ease of genetic manipulations in pluripotent cells will facilitate future studies interrogating the relationship between chromosome 21 candidate genes and DS phenotypes. Moreover, future large-scale efforts using multiple DS iPSC lines with documented clinical histories, and potentially partial trisomy patients, will allow interrogation of complex disease phenomena associated with DS, such as diverse clinical manifestations and variable phenotype penetrance. Finally, by providing a renewable and ethically unencumbered source of disease relevant human cells, DS iPSCs will also prove valuable in the development and testing of corrective therapeutics for DS.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

We conclude that integration-free iPSCs can be used effectively to model complex human diseases such as DS, and to identify phenotypes affecting early stages of DS neural development that are difficult to access with existing DS models. Our data show that DS iPSCs, pluripotent cells equivalent to the inner cell mass of the blastocyst, express a globally deregulated transcriptome that is likely to arise from numerous intra- and inter-chromosomal interactions far more complex than simple gene-dosage effects. Neural differentiation of these DS iPSCs also revealed increased oxidative stress induced neural cell death, and a marked developmental bias towards glial lineages as early brain developmental phenotypes of DS potentially amenable to therapeutic intervention.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

We acknowledge the Stemcore facility at AIBN for cell culture support and Katia Nones (QLD Centre for Medical Genomics) and Drew Titmarsh (Stem Cell Engineering Group, AIBN) for expert technical assistance. Othmar Korn (Wells laboratory AIBN) is greatly acknowledged for software design and bioinformatic support. J.A.B. and E.J.W. designed the experiments; J. Sun generated and characterized the iPSC lines with assistance from S.Y.S, S.P.N, L.P.K, C.A.M, D.A.O, and N.Y.T.; T.L.C. prepared microarray samples; and J. Shepherd ran and analyzed the microarrays; J.A.B performed neural conversion and phenotype modeling experiments, with assistance in analysis from S.P.N.; J.A.B. prepared the final figures and supplementary data; and J.A.B. and E.J.W. wrote the final manuscript. This work was supported by funding from the Queensland Government and Lejeune Foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
sc-12-0689_sm_SupplData_FileS1.xls15392KData File S1. Differentially expressed genes in DS iPSCs compared to WT iPSCs.
sc-12-0689_sm_SupplFigure1.pdf2615KFig. S1. Additional characterization of DS and WT virus and integration free iPSC clones. (A) Immunocytochemistry detection of pluripotency markers OCT4, SOX2, NANOG, SSEA-4, TRA-1-81 and TRA-1-60 in DS and WT iPSCs. (B) Histology images of derivatives of the three germ layers found in teratomas formed after injection of DS and WT iPSCs into immunocompromised SCID mice. (C) Bisulfite sequencing showing OCT4 and NANOG promoter CpG island demethylation in H9 hESCs, DS iPSCs and WT iPSCs compared to parent fibroblast lines. (D) WT and trisomy 21 karyotpe in WT and DS iPSCs respectively. (E) Immunocytochemistry detection of germ layer specific markers FOXA2 (endoderm), MAP2 (ectoderm) and SMA (mesoderm) in dissociated iPSC derived embryoid bodies plated on Matrigel following differentiation for two weeks in KOSR without bFGF, further demonstrating tri-lineage competence of DS and WT iPSCs. (F) FACS analysis of pluripotency markers OCT4, NANOG, SSEA-4 and Tra-1-60 expression across populations of DS and WT iPSCs, showing robust pluripotency marker expression profile. Scale bars: 100μm in (A) or 50μm in (E). Please note that data presented in the main text figure represent WT2- and DS1-iPSCs. Data presented in (A), (B), (C) and (D) of this figure represent iPSC lines not included in the main text figure. Data in (E) and (F) of this figure constitute further characterization of the core DS and WT iPSCs that were used subsequently for neural phenotype modeling experiments.
sc-12-0689_sm_SupplFigure2.pdf3039KFig. S2. Characterization of euDS-iPSCs. (A) Immunocytochemical detection of pluripotency markers Tra-1-60, Tra-1-81, OCT4, SOX2 and NANOG in euploid DS iPSCs. (B) Absence of free or integrated episomal vectors from euploid DS iPSCs; no product after 35 cycles of PCR using whole cell lysate as template compared to GAPDH and episomally transfected fibroblasts positive controls. (C) G-banding karyotype analysis of euploid DS iPSC showing a normal karyotype and the absence of trisomy 21. (D) Histology images of derivatives of the three germ layers found in teratomas formed after injection of euploid DS and WT iPSCs into immuno-compromised SCID mice demonstrating trilineage competence. (E) Short tandem repeat DNA fingerprinting analysis of CCL54 DS fibroblasts, DS1-iPSC and euploid DS iPSC demonstrates the identical genetic background of the cell lines as well as the absence of trisomy 21 in the euploid DS iPSC (pentaD is on HSA21).
sc-12-0689_sm_SupplFigure3.pdf629KFig. S3. Southern validation of integration free episomal reprogramming of DS and WT iPSCs. (A-B) Southern probe detection of undigested episomal plasmids E1 (pEP4EO2SCK2MEN2L, Addgene plasmid 20924) and E2 (pEP4EO2SET2K, Addgene plasmid 20927), and transgene fragments (indicated ‘Tg’) from virally generated iPSC positive control (viP) but not DS and WT episomally reprogrammed iPSCs. Additional bands result from probe binding of related genomic sequence that is not transgenic (present in negative control H9 and Hes4 hESC (H9 and He4) and DS and WT donor fibroblast (DF and WF) lines). Note that fragments denoted Tg* appear to result from low copy partial integrated transgenic sequences in viPs as they deviate from expected transgene sizes. (C) Southern probe detection of digested episomal plasmid, but not DNA from DS and WT iPSCs. Note EBNA1 is not present in the viral sequence present used to reprogram viP and therefore viP is not detected by this probe. (D) Southern probe detection of SspI digested episomal plasmid, and transgene (indicated ‘Tg’) from virally generated iPSC positive control (viP) but not DS and WT episomally reprogrammed iPSCs. Note the episomal vector is digested with a different enzyme to genomic DNA in this blot. ∼40pg of episomal DNA was loaded in each blot. See also Supplementary Materials and Methods for expected Tg fragment sizes for each digestion.
sc-12-0689_sm_SupplFigure4.pdf834KFig. S4. Pluritest results for DS and WT iPSCs (A) Pluripotency Score: A score that is based on all samples (pluripotent cells, somatic cells and tissues) in the stem cell model matrix (http://www.pluritest.org). Samples with positive values are more similar to the pluripotent samples in the model matrix than to all other classes of samples in the matrix. The area between the red lines indicates the range that contains approximately 95 percent of the pluripotent samples tested. (B) Novelty Score: A score that is based on wellcharacterized pluripotent samples in the stem cell model matrix. Samples are color-coded green (pluripotent), orange, red (not-pluripotent) based on the probabilities given from the logistic regression model. (C) Combination of the Pluripotency Score on the y-axis with the Novelty Score on the x-axis. The red and blue background indicate the empirical distribution of the pluripotent (red) and non-pluripotent samples (blue) in the Scripps test data set.
sc-12-0689_sm_SupplFigure5.tif2272KFig. S5. Hierarchical clustering analysis including euDS-iPSC microarray data Hierarchical clustering of 31927 probes with sd/mean > 0.1 in Mel 1 and H9 hESCs, DS and WT iPSCs, euploid DS iPSC and donor fibroblast lines using Pearson's Correlation Coefficient (PCC) as a measure of distance. Euploid DS iPSC cluster with WT iPSC and hESC lines, and far from their otherwise isogenic but trisomic DS iPSC counterparts, demonstrating that transcriptome differences in DS iPSCs are the direct consequence of trisomy 21.
sc-12-0689_sm_SupplFigure6.pdf741KFig. S6. Overlap of chromosome 21 overexpressed genes in DS iPSCs with previous DS array analyses (A) Figure shows a list of the 69 overexpressed genes identified in DS iPSCs, and the number of genes also found to be overexpressed in other array data sets (no genes / total genes reported to be over expressed in that study). Coloured boxes indicate which genes were found to be commonly overexpressed in each given data set. Between 13% and 64% of chr21 overexpressed genes identified by our analysis also appeared in the other tissue specific data sets, and our data contained between 31% and 46% of the chr21 overexpressed genes identified in other individual data sets. Reference studies: chorion villi [1], heart [2], fibroblasts [3], and brain [4].
sc-12-0689_sm_SupplFigure7.tif1410KFig. S7. Supporting information for DS and WT neural conversion. (A) Schematic summarizing neural conversion protocol followed in all experiments. Note that for later time points in neurite extension assays, and glial emergence quantification assays, neurospheres were expanded and then seeded at later time points. (B) Relative expression of anterior neural markers OTX1 and OTX2 compared to caudal neural marker HOXB4 at day 17 as determined by qPCR, indicating a similar anterior phenotype in DS and WT neural cultures.
sc-12-0689_sm_SupplFigure8.pdf1390KFig. S8. Additional data comparing DS and WT neurosphere growth, neurite extension and glial differentiation (A) Comparison of average neurosphere cell number doubling time in DS and WT neurospheres, calculated from data spanning 10 weeks in culture. (B) Average neurite extension showing no difference between DS and WT at four time points spanning ten weeks in culture. (C) Counts of the total numbers of neurons (TUBB3 or MAP2 positive) and glia (GFAP positive) found on average per 5 randomly sampled images. It can be seen that DS neurospheres produced a lower total number of neurons, and a higher total number of glia at both day 60 and day 80. (D) Overexpression of glial development associated genes S100B, GLAST and NFIA in day 42 DS neural cultures initiating gliogenesis. (E) Immunocytochemistry detection of MAP2 and GFAP at D80, illustrating overrepresentation of glia in DS compared to WT cultures. (F) Consistent overexpression of glial marker GLAST in DS cultures, starting at day 24 and continuing throughout gliogenesis. (G) Alternate presentation of GFAP mRNA expression, showing the average folds expression in DS compared to WT, revealing that DS neurospheres consistently overexpressed GFAP during gliogenesis (Fig. 4C). Scale bars: 100μm. Error bars represent S.E.M. # denotes lack of significance (P>0.05) as determined by Student's t-test in (A), or Twoway ANOVA with a Bonferroni post-test in (B). References: 1. Altug-Teber O, Bonin M, Walter M, et al. Specific transcriptional changes in human fetuses with autosomal trisomies. Cytogenet Genome Res 2007;119:171-184. 2. Conti A, Fabbrini F, D'Agostino P, et al. Altered expression of mitochondrial and extracellular matrix genes in the heart of human fetuses with chromosome 21 trisomy. BMC Genomics 2007;8:268. 3. Prandini P, Deutsch S, Lyle R, et al. Natural gene-expression variation in Down syndrome modulates the outcome of gene-dosage imbalance. Am J Hum Genet 2007;81:252-263. 4. Mao R, Zielke CL, Zielke HR, et al. Global up-regulation of chromosome 21 gene expression in the developing Down syndrome brain. Genomics 2003;81:457-467.
sc-12-0689_sm_SupplMethods.pdf69KSupplementary Data
sc-12-0689_sm_SupplTable1.pdf80KTable S1. Summary of generated DS and WT iPSC lines.
sc-12-0689_sm_SupplTable2.pdf12KTable S2. List of primer and probe sequences.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.