The Senescence-Related Mitochondrial/Oxidative Stress Pathway is Repressed in Human Induced Pluripotent Stem Cells

Abstract

The ability of stem cells to propagate indefinitely is believed to occur via the fine modulation of pathways commonly involved in cellular senescence, including the telomerase, the p53, and the mitochondrial/oxidative stress pathways. Induced pluripotent stem cells (iPSCs) are a novel stem cell population obtained from somatic cells through forced expression of a set of genes normally expressed in embryonic stem cells (ESCs). These reprogrammed cells acquire self-renewal properties and appear almost undistinguishable from ESCs in terms of morphology, gene expression, and differentiation potential. Accordingly, iPSCs exhibit alterations of the senescence-related telomerase and p53 signaling pathways. However, although treatments with antioxidants have been recently shown to enhance cellular reprogramming, detailed information regarding the state of the mitochondrial/oxidative stress pathway in iPSCs is still lacking. Mitochondria undergo specific changes during organismal development and aging. Thus, addressing whether somatic mitochondria within iPSCs acquire ESC-like features or retain the phenotype of the parental cell is an unanswered but relevant question. Herein, we demonstrate that somatic mitochondria within human iPSCs revert to an immature ESC-like state with respect to organelle morphology and distribution, expression of nuclear factors involved in mitochondrial biogenesis, content of mitochondrial DNA, intracellular ATP level, oxidative damage, and lactate generation. Upon differentiation, mitochondria within iPSCs and ESCs exhibited analogous maturation and anaerobic-to-aerobic metabolic modifications. Overall, the data highlight that human iPSCs and ESCs, although not identical, share similar mitochondrial properties and suggest that cellular reprogramming can modulate the mitochondrial/oxidative stress pathway, thus inducing a rejuvenated state capable of escaping cellular senescence.

iPS, Reprogramming, Aging, Gene expression, Differentiation, Embryonic stem cells, Induced pluripotency

Introduction

Human embryonic stem cells (hESCs) are derived from the inner cell mass of blastocysts and retain the ability to propagate indefinitely and differentiate into all the derivates of the three primary germ layers [1]. hESCs are characterized by specific self-renewal pathways, such as TGF-beta and AKT signaling [2, 3], which appear highly overlapping with cellular senescence mechanisms. These mechanisms, which play a relevant role in the context of aging and age-associated disorders [4], include the telomerase pathway, the p53 pathway, and the mitochondrial/oxidative stress pathway [5]. The capability of hESCs to escape senescence may indeed occur through the selective modulation of these specific pathways. In fact, hESCs exhibit high telomere length and telomerase activity [1], reduced p53 activity [6], and low levels of reactive oxygen species (ROS) and mitochondrial activity [7–8].

Recently, human somatic cells have been successfully reprogrammed to a hESC-like state by ectopically forcing the expression of a combination of genes commonly present in hESCs. Induced pluripotent stem cells (iPSCs) appear almost identical to hESCs and share the same cardinal features of self-renewal and pluripotency [9–12]. Genome-wide analysis revealed specific gene expression differences between iPSCs and hESCs [13], suggesting that the degree of molecular similarities between the two cell types still needs to be fully clarified. Nonetheless, their isogenic nature and their derivation methods, which do not require the use of pre-implantation embryos, put forward iPSCs as promising candidates for patient-specific cell therapy and in vitro modeling of complex human disorders, including aging and neurodegeneration [14]. When compared to their parental differentiated cells, iPSCs displayed telomere elongation and telomerase activity induction [9, 15]. In addition, the efficiency of iPSC generation has been found significantly increased by inhibition of p53 signaling [16–17]. However, although the mitochondrial structure within murine iPSCs appeared similar to murine ESCs [18] and antioxidant treatment has been recently associated with enhanced iPSC generation [19], detailed information on the impact of reprogramming on the mitochondrial/oxidative stress pathway still needs to be provided.

Structural analyses of mitochondria within hESCs have demonstrated an immature network characterized by few organelles with poorly developed cristae with peri-nuclear polarization [20–22]. Studies of organelle number and functionality confirmed a low content of mitochondrial DNA (mtDNA) and low levels of intracellular ATP and ROS [7, 8, 23]. Hence, hESCs in their self-renewal state appear to rely more on anaerobic rather than aerobic mitochondrial respiration, in agreement with the hypoxic environment to which early-stage mammalian embryos are exposed [24] and the growth improvement displayed by hESCs under low oxygen tension [25–27]. Accordingly, inhibition of mitochondrial respiratory chain has been recently found associated with enhancement of hESC pluripotency [26].

During cellular differentiation and organismal development, mitochondria become mature and functionally active, as the energy metabolism switch from anaerobic glycolysis to oxidative phosphorylation [8, 28]. Several mitochondrial modifications occur within differentiated somatic cells during aging [29]. Indeed, transcriptome analysis of human skin aging has identified specific alterations in pathways related to mitochondria and oxidative stress [30]. mtDNA undergoes a high rate of mutation, due to elevated ROS levels and the lack of histones and efficient repairing mechanisms. mtDNA mutations can thus accumulate over time [31, 32], causing multiple cellular dysfunction, including defective protein degradation or cellular secretion [33–34]. To overcome the detrimental effects of toxic free radicals, cells protect themselves using antioxidants, such as glutathione peroxidase [35–36]. Perturbation of this fine equilibrium is implicated in numerous diseases and life-span regulation, as also suggested by the metabolic stability of regulatory networks [37].

Herein, we report the first extensive characterization of mitochondrial properties within human iPSCs and hESCs. Unlike other reprogramming methods, such as cell fusion and somatic cell nuclear transfer, mitochondria within iPSCs are exclusively of somatic origin [38]. Thus, addressing iPSC mitochondrial modifications, such as potential acquisition of hESCs features or persistence of somatic cell-like characteristics, is an essential and unresolved question. Our data show that somatic mitochondria within human skin fibroblasts reacquire immature-like features typical of hESCs upon reprogramming to pluripotency. Following spontaneous differentiation into fibroblast-like cells, iPSCs and hESCs show a similar pattern of anaerobic-to-aerobic metabolic transition. Specific similarities and differences are reported and the findings are discussed with respect to their relevance for cellular senescence and in vitro modeling of human aging and associated diseases such as neurodegeneration.

Materials and Methods

Cell Culture Conditions

Human ES and iPS cells were cultured in hESCs media containing knockout Dulbecco's modified Eagle's medium (DMEM) supplemented with 20% knockout serum replacement, nonessential amino acids, L-glutamine, penicillin/streptomycin, sodium pyruvate, 0.1 mM β-mercaptoethanol (Invitrogen, Carlsbad, CA, http://www.invitrogen.com), and 8 ng/ml bFGF (Peprotech, Rocky Hill, NJ, http://www.peprotech.com). Cultures were maintained on mitomycin C-inactivated mouse embryonic fibroblast (MEFs) and passaged using dispase treatment followed by cut-and-paste technique [39]. Feeder-free conditions were used for most of the experiments, and iPSCs and hESCs were grown on dishes coated with Matrigel (BD Biosciences, San Diego, http://www.bdbiosciences.com) in MEF-conditioned media (MEF-CM) [2].

Retroviral Production and iPSC Generation

OCT4, KLF4, SOX2, and c-MYC retroviruses were generated using pMX vectors (kindly donated by V. Broccoli) as described before [9]. Briefly, 293T cells were plated at 8 × 10⁶ cells per 150 mm dish, incubated overnight and transfected with 32 μg of vectors, according to conventional CaCl₂ transfection protocol. After 30 hours, medium was collected, centrifuged at 20,000 rpm for 2 hours, filtered, and supplemented with 4 μg/ml of Polybrene (Sigma, St. Louis, http://www.sigmaaldrich.com). For generation of human iPSCs, 200.000 neonatal foreskin HFF1 fibroblasts (ATCC, #ATCC-SCRC-1041, http://www.atcc.org) were seeded and transduced with the retroviral cocktail on days 1 and 2. The media was changed on day 3 to DMEM supplemented with 10% fetal bovine serum (FBS), nonessential amino acids, L-glutamine, penicillin/streptomycin, and sodium pyruvate. On day 5, cells were split and seeded on MEFs on matrigel-coated dishes and cultured in ES medium for about four weeks. ES-like colonies were then manually picked and expanded for characterization. Experiments were performed with iPS cells between passage 8 (p8) and passage 20 (p20). We currently have in culture iPS2 and iPS4 lines passaged more than 25 times (p25).

In Vitro and In Vivo Differentiation

For in vitro differentiation, embryoid bodies (EBs) were generated from iPSCs by harvesting the cells and seeding them onto low-attachment dishes in differentiating medium without bFGF supplementation. One week later, EBs were plated onto gelatin-coated tissue culture dishes and grown for an additional 2-3 weeks. In vivo differentiation experiments were performed by EPO-Berlin Gmbh. iPSCs were collected by trypsinization, washed, and injected subcutaneously into NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ mice, commonly known as NOD scid gamma. Teratomas were collected 50-70 days after injection and processed according to standard procedures for paraffin embedding and hematoxylin and eosin staining. Histological analysis was performed by an expert pathologist.

Fibroblast-like differentiated cells were derived from hESCs (hESC-DF) and from iPSCs (iPSC-DF), as previously described [2, 12]. Briefly, undifferentiated cells were cultured in differentiating conditions (ES media without bFGF supplementation) for two weeks. Undifferentiated colonies were then manually removed, and differentiated fibroblast-like cells were grown in DMEM medium with 10% FBS and passaged with trypsin for at least three passages.

DNA Fingerprinting and Karyotyping

Somatic origin of iPSCs was confirmed by fingerprinting analysis, as previously described [12]. Briefly, genomic DNA was isolated with FlexiGene DNA kit (Qiagen, Hilden, Germany, http://www1.qiagen.com) and PCR was amplified to detect genomic intervals containing variable numbers of tandem repeats. A total of 50 ng of DNA was amplified at 94°C for 1 minute, 55°C for 1 minute, and 72°C for 1 minute for a total of 40 cycles using Dyad thermal cycler (Bio-Rad, Hercules, CA, http://www.bio-rad.com) and then run on 2.5% agarose gel. Primer sets used were D7S796, D10S1214, D21S2055; sequences are provided in supporting information Table S6. For detection of possible karyotype abnormalities in iPSCs, chromosomal analysis after GTG-banding was performed at Human Genetic Center of Berlin. For each line, 20 metaphases were counted and 12 karyograms were analyzed.

Illumina Bead Chip Hybridization

Total RNA was quality checked by Nanodrop analysis (Nanodrop Technologies, Wilmington, DE, USA, http://www.nanodrop.com), and a quantity of 400 ng was used as input. Biotin-labeled cRNA was produced using a linear amplification kit (Ambion, Austin, TX, http://www.ambion.com). Hybridizations, washing, Cy3-streptavidin staining, and scanning were performed on the Illumina BeadStation 500 platform (Illumina, San Diego, CA, http://www.illumina. com), according to manufacturer's instruction. cRNA samples were hybridized onto Illumina human-8 BeadChips version 3. All basic expression data analysis was carried out using the BeadStudio software 3.0. Raw data were background-subtracted and normalized using the “rank invariant” algorithm and then filtered for significant expression on the basis of negative control beads. Pathway analysis was determined according to Gene Ontology terms or mapped to KEGG pathways with DAVID 2006 (http://david.abcc.ncifcrf.gov), by using GenBank accession numbers represented by the corresponding chip oligonucleotides as input.

Quantitative Real-Time Polymerase Chain Reaction

Quantitative Real-Time polymerase chain reaction (QPCR) was performed in 384-well optical reaction plates (Applied BioSystems, Foster City, CA, http://www.appliedbiosystems.com) using SYBR Green PCR Master Mix (Applied Biosystems). Reactions were carried out on the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems) as previously described [37]. Triplicate amplifications were carried out for each target gene with three wells serving as negative controls. Quantification was performed using the comparative Ct method (ABI instruction manual) normalized with the housekeeping genes GAPDH or ACTB and presented as a percentage of biological controls. Human Embryonic Stem Cells StellARray qPCR Array (Lonza, Basel, Switzerland, http://www.lonza.com) was used to confirm the expression of key ESC-related genes in our ES and iPS cell lines (supporting information Figure S2). Quantification of exogenous transgenes was performed as previously described [9]. As positive controls, we transduced HFF1 cells with single factors (respectively, HFF1-O, HFF1-K, HFF1-S, and HFF1-M) and isolated RNA after 72 hours. All primer sequences are provided in supporting information Table S6.

Quantification of mtDNA Copy Number

mtDNA copy number was quantified as previously described [40]. Briefly, for the mitochondrial genome, a mitochondrially encoded subunit of complex I, ND5, was amplified, whereas, for the nuclear genome, cystic fibrosis (CF) gene was used. A standard curve was made for both the mitochondrial and nuclear genes, in order to calculate the number of copies of two genes in a given amount of DNA. ND5 gene was amplified from 0 to 1 ng of K562 DNA (Promega, Madison, WI, http://www.promega. com), and CF was amplified from 0 to 100 ng of K562 DNA. Each sample was run in triplicate, and quantitative PCR (QPCR) analysis was performed as described above. Primer sequences are reported in supporting information Table S6.

Transmission Electron Microscopy

Differentiated and undifferentiated cells were grown on matrigel-coated Thermanox plastic coverslips (Nunc, Rochester, NY, http://www.nuncbrand.com) until 70% confluent. Cells were then fixed on coverslips with 2.5% glutaraldehyde in 50 mM sodium cacodylate buffer (pH 7.4) supplemented with 50 mM sodium chloride for at least 30 minutes at room temperature (RT). Specimens were washed in the same buffer and postfixed for 1.5 hours in 0.5% osmium tetroxide at RT, followed by 0.1% tannic acid for 30 minutes and 2% uranyl acetate for 1.5 hours. Samples were dehydrated in a graded series of ethanol, embedded in Spurr's resin (Low Viscosity Spurr Kit, Ted Pella, Redding, CA, http://www.tedpella.com) and polymerized at 60°C. Ultrathin sections (70 nm) were prepared with an ultramicrotome (Reichert Ultracut E, Leica, http://www.leica-microsystem.com) and mounted on electron microscopy copper grids, 300 mesh. Sections were counterstained with uranyl acetate and lead citrate for 20 seconds. Micrographs were made with a Philips CM100 using a 1K charge-coupled device camera (Tietz Video and Image Processing Systems, Gauting, Germany http://www.tvips.com). Measurement of mitochondrial diameters was performed using the EMMENU4 software (Fastscan, TVIPS).

Immunofluorescence, Alkaline Phosphatase, and Mitochondrial Staining

For immunocytochemistry, cells were fixed with 4% paraformaldehyde (Science Services) for 20 minutes at RT, washed two times with phosphate-buffered saline (PBS), and blocked with 10% chicken serum (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com) and 0.1% Triton X-100 (Sigma). Nuclei were counterstained with DAPI (200 ng/ml, Invitrogen, Carlsbad, CA, http://www.invitrogen.com# H3570). Primary antibodies included SSEA1, SSEA4, TRA-1-60, and TRA-1-81 from the ES cell characterization tool (all 1:100, Millipore, Billerica, MA, http://www.millipore.com, #SCR004), NANOG (1:100, Abcam, Cambridge, U.K., http://www.abcam.com, #ab62734), OCT4 (1:100 Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com, #sc-5279), SOX2 (1:100, Santa Cruz, #sc-17320), Smooth Muscle Actin (1:100, DakoCytomation, Glostrup, Denmark, http://www.dakocytomation.com, #M0851), Alpha Feto Protein (1:100, Sigma, #WH0000174M1), SOX17 (1:50, R&D, #AF1924), PAX6 (1:300, Covance, #PRB-278P), Nestin (1:200, Chemicon, Temecula, CA, http://www.chemicon.com, #MAB5326), and TUJ-1 (1:1000, Sigma, #T8660), 8-OHdG (1:100, Millipore, #AB5830). Secondary antibodies used were conjugated with either Alexa 488 or Alexa 594 (Invitrogen, #A11001, A11055, A21201, A21468, A11005, A21442). Alkaline phosphatase (AP) staining was performed following the manufacturer's instructions (Millipore, #SCR004). For mitochondria labeling, undifferentiated and differentiated cells were washed twice with PBS and incubated at 37°C for 1 hour with 50 nM mitochondrion-selective dye MitoTracker Green (MTG) (Molecular Probes Inc., Eugene, OR, http://probes.invitrogen.com, #M7514). The solution was then replaced with fresh prewarmed media, and cells were then fixed and permeabilized. Coverslips were mounted using Dako fluorescent mounting medium (Dako, #S3023) and visualized using a confocal microscope LSM 510 (Zeiss).

Western Blotting and OxyBlot Analysis

Total cell protein extracts were obtained using a modified RIPA buffer (50 mM Tris pH 7.4, 100 mM NaCl, 10 mM EDTA, 1 mM PMSF, 1% IGEPAL) supplemented with complete protease inhibitor cocktail (Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com) added just before use. Protein concentration was determined according to the Bradford method. Equal quantities of proteins were separated by electrophoresis in a 10% sodium dodecyl sulfate–polyacrylamide gel (Bio-Rad, Hercules, CA, http://www.bio-rad.com). Primary antibodies anti-GPX1 (1:500, Abcam #ab22640) and GAPDH (1:5000, Ambion #4300) were used with the suitable HRP-conjugated secondary antibodies. Quantification of total oxidized proteins was performed using the OxyBlot Protein Oxidation Detection Kit (Millipore), as previously described [41]. Bound antibodies on nitrocellulose membrane (GE Healthcare, Piscataway, NJ, http://www.gelifesciences.com) were detected using the chemiluminescent substrate ECL (GE Healthcare). Membranes were stripped using Restore Western Blot (Thermo Fisher Scientific, Fremont, CA, http://www.thermofisher.com). Densitometry analysis was performed using ImageJ software (NIH, Washington, DC, http://www.nih.gov, ImageJ); the results were normalized by GAPDH expression and presented as area under the curve (AUC) values.

ATP Measurement

Cellular ATP content was determined using the ATPLite bioluminescence luciferase-based assay (Perkin Elmer), as previously described [34]. Briefly, 100.000 cells were resuspended in 100 μl of PBS, added to 50 μl of cell lysis per well of a 96-well microplate, and shaken for 5 minutes in an orbital shaker. Fifty μl of substrate solution was then added, and the plates were shaken for 5 minutes and incubated for 10 minutes in the dark. Luminescence was measured with a luminometer (Berthold Technologies, Bad Wildbad, Germany, http://www.berthold.com). A standard curve was prepared for each measurement using the provided ATP standard solution by making a dilution series in water from a concentration of 1 × 10⁻⁵ M to blank. Every sample was measured in triplicate, and the experiment was repeated three times. The results are presented as nM of ATP per cell.

Lipid Hydroperoxidase Assay

The measurement of lipid hydroperoxides (LPO) was used to determine ROS-related lipid modifications and was performed using the LPO assay kit (Calbiochem, San Diego, http://www.emdbiosciences.com) as previously described [42]. Prior to the experiment, chloroform and methanol were deoxygenated by bubbling nitrogen through the solvent for 45 minutes. Briefly, 400.000 cells were homogenized by sonication and lipid hydroperoxides were extracted with chloroform. Five hundred μl of the chloroform extracts were transferred with 450 μl of a chloroform–methanol mixture in glass cuvettes. Fifty μl of freshly-prepared chromogen was added, and every sample was incubated for 5 minutes at RT. The assay was completed by reading the absorbance at 500 nm using the Ultrospec 3,100 (GE Healthcare). Results were calculated using a standard curve and presented as pmol of LPO per cell.

Measurement of Lactic Acid Production

Lactate production rate was measured using a colorimetric-based Lactate assay kit (Bio Vision, Mountain View, Californiahttp://www.biovision.com), as previously described [43]. Briefly, cells in six-well plates were replenished with fresh medium and incubated overnight. The supernatants were collected and concentrated with Amicon 10kd molecular weight filters (Millipore) to prevent lactate degradation by lactate dehydrogenase. Protein amount was quantified according to the Bradford method. Several dilutions were utilized to ensure the readings were within the standard curve range. Samples were diluted with the Lactate reagent and incubated for 30 minutes at RT. Measurement of outer diameter 570 nm was performed in a 96-well plate using the DTX 880 Multimode detector fluorimeter (Beckman Coulter, Fullerton, CA, http://www.beckmancoulter.com) and corrected by subtracting the background readings, as indicated by the manufacturer. Results were presented as nm of lactate per μg of proteins.

Statistical Analysis

Data are expressed as mean and standard deviation. Comparisons between two groups were performed by two-tailed unpaired student's t-test and p values of ≤.05 were considered statistically significant. Data were analyzed using GraphPad-Prism software (GraphicPad, La Jolla, CA, http://www.graphpad.com) and Windows XP Excel (Microsoft, Redmond, WA, http://www.microsoft. com).

Results

Derivation of Human iPSCs

Human iPSCs were derived from neonatal foreskin fibroblasts (HFF1) transduced with the Yamanaka retroviral cocktail [9] (Fig. 1A). Two iPSC lines (iPS2 and iPS4) were fully characterized and exhibited hESC-like properties including morphology and growth rate, alkaline phosphatase (AP) activity, and expression of undifferentiated markers NANOG, OCT4, SOX2, SSEA4, TRA-1-60, and TRA-1-81 (Fig. 1B). The pluripotency abilities of iPS2 and iPS4 were assayed by in vitro and in vivo differentiation into the three germ layers (supporting information Fig. S1). QPCR analysis demonstrated an almost complete silencing of the exogenous transgenic factors (Fig. 1C) and confirmed the expression of pluripotency-associated genes including NANOG, GDF3, and CRIPTO (Fig. 1D and supporting information Fig. S2). Both iPSC lines displayed a normal karyotype, and their somatic relatedness to HFF1 cells was confirmed by DNA fingerprinting analysis (Fig. 1D).

1

Generation of human induced pluripotent stem cells (iPSCs) and characterization of iPS2 and iPS4 lines. iPS cells were derived by retroviral transduction of the Yamanaka cocktail (OCT4, KLF4, SOX2, and c-MYC) in neonatal foreskin fibroblasts HFF1. Two iPSC lines (iPS2 and iPS4) were fully characterized. (A): Pictures showing the essential steps of iPSC reprogramming. Upper lane, from left to right: HFF1 fibroblasts at passage 4 used for the reprogramming experiments, cellular clone 28 days after transduction, AP staining 4 weeks after transduction. Lower lane, from left to right: iPS4 line at passage 10, iPS2 line at passage 8, iPS4 line at passage 12 stained with AP. (B): Immunofluorescence imaging of embryonic stem (ES) cell markers in iPS4 cells. (C): Retroviral silencing in iPS2 and iPS4 determined by quantitative polymerase chain reaction (QPCR). As positive control, HFF1 cells were transduced with a single factor (respectively, HFF1-O, HFF1-K, HFF1-S, and HFF1-M) and harvested 72 hours later for RNA isolation. Immunofluorescence pictures confirming the expression of the transduced factors are reported on the right side. (D): Upper left panel, DNA fingerprinting analysis confirmed the somatic origin of iPS2 and iPS4. Upper right panel, karyotype analysis of the iPS4 line. Lower panel, QPCR analysis of pluripotency-associated genes in H1, H9, iPS2, and iPS4. Results were normalized with GAPDH and expressed in comparison to H1.

Similar Mitochondrial-Related Transcriptional Signature in iPSCs and hESCs

Mitochondria possess their own DNA, which encodes for 13 peptides of the electron transport chain (ETC). However, the majority of mitochondrial proteins are encoded by nuclear DNA [44]. Thus, in order to examine the mitochondrial-related nuclear gene expression signature in human iPSCs, we first compared the transcriptional profiles of iPSC lines iPS2 and iPS4 and hESC lines H1 and H9 to somatic fibroblasts HFF1.

As previously shown [13], the iPSC transcriptomes are distinct from somatic fibroblasts (r² = 0.6782), although not completely identical to hESCs (r² = 0.9045) (Fig. 2A and supporting information Fig. S2). The cellular pathways down- or upregulated in iPSCs compared to fibroblasts were analogous to those down- or upregulated in hESCs compared to fibroblasts (supporting information Tables S1–S4). We then evaluated the distinct and overlapping transcriptional signatures of the three cell types, based on expression p values ≤ .01. Different signatures were identified, including a self-renewal signature, which comprises genes (OCT4, NANOG, LIN28, SOX2) shared between ES and iPS cells, a donor cell signature, containing genes (HOXD11, ECM2, COL8A1) shared between iPSCs and their original cell source, and a housekeeping signature, with genes (GAPDH, ACTB, POLG) expressed in all three cell types (Fig. 2B and supporting information Table S5). Pathway analysis revealed, as part of the common self-renewal signature, pathways known to be associated with pluripotency, including WNT and MAPK signaling. Interestingly, functions related to mitochondria were mainly represented within the housekeeping signatures, including oxidative phosphorylation, TCA cycle, pyruvate, and fatty acid metabolism (Fig. 2C). Indeed, within the lists of pathways referred to the down- or upregulated transcripts in iPSCs or hESCs in comparison to HFF1, no mitochondrial-related pathway could be identified. This suggests that nuclear transcription of mitochondrial-related genes is persistent both in somatic and pluripotent cells.

2

Mitochondrial-related nuclear transcriptional profiling in hESCs and iPSCs. Transcriptome analysis of fibroblasts (HFF1), hESC (H1 and H9), and iPSCs (iPS2 and iPS4) was performed using illumina bead chips. (A): Upper panel, hierarchical clustering. Lower panel, table showing the correlation coefficient (r²) between the samples. Low r² values (in gray) were detected between fibroblasts and ESCs and between fibroblasts and iPSCs. ES and iPS cells exhibited higher r² values (in light blue). (B): Venn diagram showing the distinct and overlapping transcriptional signatures between the three cell types. Identified signatures are reported. (C): Pathway analysis of the 7,302 genes comprising the housekeeping transcriptional signature revealed several mitochondrial-related pathways. (D): Heatmap figures depicting transcripts related to the oxidative stress response and to mitochondrial biogenesis. Values represent the ratio of the array average signal of the given gene divided by the average signal of HFF1 line (fold change 1.5, p value ≤ .01). Downregulated genes are shown in green, and upregulated genes are shown in red. Abbreviations: hESCs, human embryonic stem cells; iPSCs, induced pluripotent stem cells.

Finally, investigation of specific mitochondrial-related transcripts revealed that, although nuclear encoded ETC complexes did not appear to be consistently up- or downregulated in undifferentiated iPSCs compared to somatic cells (Figure S3), genes involved in the response to oxidative stress were mainly downregulated in both ES and iPS cells in comparison to fibroblasts (Fig. 2D). The expression of main antioxidant enzymes was reduced in iPS and ES cells compared to HFF1, including catalase, GPX1, SOD1-2-3, and ATM, which is commonly associated with the DNA damage response [45]. On the other hand, mitochondrial biogenesis nuclear factors appeared upregulated in iPSCs as well as in hESCs compared to somatic fibroblasts (Fig. 2D).

Low Mitochondrial Content and mtDNA Copy Number in iPSCs and hESCs

In order to confirm the microarray results and to better compare the mitochondrial properties in undifferentiated and differentiated cells, we used spontaneous differentiation of iPSCs and ESCs to obtain fibroblast-like cells, as previously described [2, 12]. hESC-derived fibroblasts (H1-DF, H9-DF) and iPSC-derived fibroblasts (iPS2-DF and iPS4-DF) exhibited similar growth culture and morphology as somatic fibroblasts and were negative for OCT4 expression (Fig. 3A).

3

Mitochondrial biogenesis and mitochondrial DNA (mtDNA) copy number in undifferentiated and differentiated human embryonic stem cells (hESCs) and induced pluripotent stem cells (iPSCs). To monitor mitochondrial biogenesis in the self-renewal and differentiated states, fibroblast-like cells were generated from human ESCs and iPSCs. H1- and H9-derived fibroblasts (respectively, H1-DF and H9-DF) and iPS2 and iPS4-derived fibroblasts (respectively, iPS2-DF and iPS4-DF) were obtained as previously described [2, 12]. (A): Quantitative polymerase chain reaction analysis of nuclear factors involved in mitochondrial biogenesis. Expression values were normalized over the expression of actin beta (ACTB) and presented as relative changes compared to HFF1. (B): Quantification of mtDNA copy number within HFF1 and undifferentiated and spontaneously differentiated ES and iPS cells. ∗∗∗, p ≤ .01, two-tailed unpaired student's t-test, H1 versus HFF1, H9 versus HFF1, iPS2 versus HFF1, and iPS4 versus HFF1. #, p ≤ .05, two-tailed unpaired student's t-test, H1-DF versus H1. ##, p ≤ .05, two-tailed unpaired student's t-test, H9-DF versus H9. ###, p ≤ .05, two-tailed unpaired student's t-test, iPS2-DF versus iPS2. ####, p ≤ .05, two-tailed unpaired student's t-test, iPS4-DF versus iPS4. (C): Mitochondria content was assayed using the mitochondrion-selective dye MitoTracker Green (MTG). Nuclei were counterstained with DAPI and OCT4 immunofluorescence was used to monitor the relationship between mitochondrial mass and self-renewal capacity.

Active replication of mtDNA is controlled by nuclear-encoded factors, such as POLG and TFAM, that are subsequently translocated to the mitochondria [44]. The expression of these factors, including POLG, TFAM, and PGC1, was induced in undifferentiated iPSCs compared to somatic fibroblasts, in agreement with the array data, and mostly decreased upon differentiation into fibroblast-like cells (Fig. 3A). A similar pattern was detected in spontaneously differentiated hESCs, in agreement with previous results [8].

Induction of nuclear factors involved in the biogenesis of mitochondria has been described in response to mtDNA depletion [46]. Accordingly, mtDNA content in undifferentiated iPSCs appeared significantly decreased in comparison to somatic cells (Fig. 3B). The reduction of mtDNA copy number in iPSCs was comparable to that in hESCs (Fig. 3B). This finding recapitulates the low amount of mtDNAs described in human and murine undifferentiated ESCs [8, 28]. As further confirmation, both iPSC-DFs and hESC-DFs displayed a significant induction of mtDNA copy number compared to pluripotent cells, although mtDNA levels did not appear as high as in HFF1 somatic fibroblasts (Fig. 3B).

Modifications of the mitochondria content during reprogramming and differentiation were confirmed using MitoTracker Green (MTG), a marker of mitochondrial mass regardless of membrane potential [47]. MTG labeling in somatic fibroblasts showed mitochondria as scattered dots throughout the cytoplasm, whereas ES and iPS cell colonies were poorly stained, except for the periphery of the colonies (Fig. 3C). A similar MTG pattern has been described in undifferentiated hESCs and explained by the fact that cells at the colony periphery might show specific epitheliod appearance and may be on the verge of differentiation [8, 20]. Upon differentiation into DF cells, MTG labeling revealed mitochondria exhibiting a dense tubular network rather than a scattered-dot pattern as in the parental cells (Fig. 3C). This may be due to the fact that fibroblast-like cells could be slightly different compared to somatic fibroblasts. Accordingly, transcriptional profiling confirmed that hESC-DFs and iPSC-DFs clustered close to HFF1 but were still not entirely overlapping with the somatic cells (data not shown).

Overall, the content of mitochondria and mtDNA show a similar profile in undifferentiated hESCs and iPSCs, suggesting that reprogramming might decrease the amount of mitochondria within somatic cells to a level comparable to ES cells. On the other hand, differentiation of both pluripotent cells induced a similar increase in the overall number of the organelles.

Somatic Mitochondria Acquire hESC-like Ultrastructural Morphology and Distribution upon Cellular Reprogramming

Ultrastructural morphological features of the mitochondrial network within undifferentiated and differentiated cells were investigated by employing transmission electron microscopy (TEM). Human iPSCs exhibited few mitochondria with underdeveloped cristae, all features of an immature-like morphology characteristic of hESCs (Fig. 4A, arrowheads). The maximal (D1) and minimal (D2) axis of mitochondria were measured (Fig. 4B). When compared to HFF1, iPSCs showed a significantly decreased D1 length associated with an increased D2 length—in a similar fashion as hESCs—suggesting a likely change from an elongated tubular morphology to a round-shaped one (Fig. 4B). Round-shaped mitochondria have been recently detected within undifferentiated murine iPSCs as well, making this a potentially non-species-specific feature [18]. Finally, we observed that mitochondria within iPS and ES cells were assuming a bipolar clustering at the two sides of the nuclei (Fig. 4C, arrows), a typical intracellular distribution previously described in mouse and human ESCs [20–22].

4

Mitochondrial morphology and distribution within hESCs and iPSCs. Transmission electron microscopy (TEM) was used to investigate the morphological features of the mitochondrial network within differentiated and undifferentiated cells. (A): TEM pictures at low and high magnification of HFF1 somatic fibroblasts, undifferentiated H1 and iPS4 lines, and their derived fibroblast-like cells, H1-DF and iPS4-DF, respectively. Asterisks show mitochondria with tubular elongated morphology; arrowheads indicate round-shaped mitochondria with poorly developed cristae. This latter feature was observed as the most common one in undifferentiated ES and iPS cells. (B): Graphs reporting the measurement of mitochondrial diameters. D1 and D2 represent, respectively, the major and minor axes, as shown in the scheme on the left. D1 diameter appeared significantly lower in undifferentiated ES and iPS cells compared to HFF1, whereas it was increased in derived-fibroblasts in comparison to their respective undifferentiated cells. An inverse pattern could be observed for the D2 diameter. ∗∗∗, p ≤ .001, two-tailed unpaired student's t-test, hESCs versus HFF1 and iPSCs versus HFF1. #, p ≤ .01, two-tailed unpaired student's t-test, hESC-DF versus hESCs. ##, p ≤ .05, two-tailed unpaired student's t-test, iPSC-DF versus iPSCs. Fifty mitochondria were counted for each cell type. Bars represent standard deviation. (C): TEM overview images showing mitochondria localization within undifferentiated H1 and iPS4 cell lines. Arrows indicate round-shaped mitochondria clustering at the two sides of the nuclei. Abbreviations: DF, fibroblast-like differentiated cells; ESCs, embryonic stem cells; iPSCs, induced pluripotent stem cells.

Upon differentiation into fibroblast-like cells, the mitochondrial network in iPSCs showed parallel modifications to ESC-differentiated cells and reacquired characteristics comparable to somatic cells, displaying significantly elongated mitochondria with developed cristae (Fig. 4A, asterisks, and Fig. 4B). Some round-shaped organelles with underdeveloped cristae persisted in both iPS4-DF and H1-DF (Fig. 4A, arrowheads). These findings confirmed an incomplete overlapping between somatic fibroblasts and DF cells. Taken as a whole, mitochondrial network within human iPSCs appeared highly similar to hESCs both in the undifferentiated and differentiated state.

iPSCs and hESCs Share a Common Dependence on Anaerobic Glycolysis

Because the structural study suggests that the mitochondrial network within iPSCs might be underdeveloped, we aimed to determine organelle functionality. Undifferentiated iPSCs exhibited a significantly lower content of cellular ATP than somatic fibroblasts, a feature identical to hESCs (Fig. 5A). Upon differentiation, ATP levels significantly increased in both iPSC- and ESC-derived cells (Fig. 5A). Conversely, undifferentiated iPSCs, as well as hESCs, displayed a higher production of lactate in comparison to somatic fibroblasts or differentiated cells (Fig. 5B). Taken together, the results suggest that undifferentiated human iPS and ES cells might rely less on mitochondrial respiration and predominantly produce energy by anaerobic glycolysis followed by lactate fermentation.

5

Aerobic and anaerobic energy generation in human embryonic stem cells (hESCs) and iPSCs. Cellular energy can be obtained either by aerobic respiration, which produces high amount of ATP and requires oxygen and active mitochondria, or by anaerobic respiration followed by lactic acid formation. To test energy generation in undifferentiated and differentiated iPS cells, ATP content and lactate generation were measured. (A): Cellular ATP level in HFF1 somatic fibroblasts, undifferentiated hESC H1 and H9 lines, undifferentiated iPSC iPS2 and iPS4 lines, and derived fibroblasts H1-DF, H9-DF, iPS2-DF, and iPS4-DF cells. ∗∗∗, p ≤ .01, H1 versus HFF1, H9 versus HFF1, iPS2 versus HFF1, and iPS4 versus HFF1. #, p ≤ .05, H1-DF versus H1. ##, p ≤ .05, H9-DF versus H9. ###, p ≤ .05, iPS2-DF versus iPS2. ####, p ≤ .05, iPS4-DF versus iPS4. (B): Levels of extracellular lactate in undifferentiated and differentiated cells in comparison to HFF1. ∗∗∗, p ≤ .01, H1 versus HFF1, and H9 versus HFF1. ∗∗, p ≤ .05, iPS2 versus HFF1 and iPS4 versus HFF1. #, p ≤ .01, H1-DF versus H1. ##, p ≤ .01, H9-DF versus H9. ###, p ≤ .05, iPS2-DF versus iPS2. ####, p ≤ .05, iPS4-DF versus iPS4. Abbreviations: DF, fibroblast-like differentiated cells; iPS, induced pluripotent stem.

Reprogramming Restores a Cellular State Characterized by Low Oxidative Stress Levels

Cells relying less on mitochondrial function may exhibit a lower presence of free radical damage [48]. Accordingly, OxyBlot analysis, which detects the presence of oxidatively modified proteins [41], revealed a significantly decreased content of protein damage in undifferentiated iPSCs and hESCs compared to HFF1 (Fig. 6A). Upon differentiation, the amount of oxidatively modified proteins was found to be significantly increased in comparison to undifferentiated cells, in a similar fashion for both hESC-DFs and iPSC-DFs (Fig. 6A).

6

Oxidative-mediated damage in human embryonic stem cells (hESCs) and iPSCs. Highly active mitochondria may generate reactive oxygen species (ROS) as byproducts. ROS-related damage to proteins, DNA, and lipids was evaluated in undifferentiated and differentiated cells. (A): Left panel, representative image for OxyBlot assay, which detects the level of oxidatively modified proteins. Membranes were stripped and reprobed with GAPDH as a loading control. Right panel, average densitometry analysis normalized over GAPDH expression and presented as area under the curve (AUC) values. ∗∗, p ≤ .05, H1 versus HFF1, H9 versus HFF1, iPS2 versus HFF1, and iPS4 versus HFF1. #, p ≤ .05, H1-DF versus H1. ##, p ≤ .05, H9-DF versus H9. ###, p ≤ .05, iPS2-DF versus iPS2. ####, p ≤ .05, iPS4-DF versus iPS4. (B): Immunofluorescence staining with 8-OHdG of untreated cells and cells exposed to 500 uM of H₂0₂ for 2 hours. Nuclei were counterstained with DAPI. (C): Levels of oxidatively modified lipids were monitored by measuring lipid hydroperoxidases (LPO), as previously shown [37]. ∗∗∗, p ≤ .01, H1 versus HFF1, H9 versus HFF1, iPS2 versus HFF1, and iPS4 versus HFF1. #, p ≤ .05, H1-DF versus H1. ##, p ≤ .05, H9-DF versus H9. ###, p ≤ .05, iPS2-DF versus iPS2. ####, p ≤ .05, iPS4-DF versus iPS4. (D): Upper panel, representative immunoblot image of glutathione peroxidase 1 (GPX1), the most abundant glutathione peroxidase isozyme in mammalian cells. Membranes were stripped and reprobed with GAPDH as a loading control. Lower panel, average densitometry analysis normalized over GAPDH expression and presented as AUC values. ∗∗∗, p ≤ .01, H1 versus HFF1, H9 versus HFF1, iPS2 versus HFF1, and iPS4 versus HFF1. ###, p ≤ .01, iPS2-DF versus iPS2. ####, p ≤ .01, iPS4-DF versus iPS4. Abbreviations: DF, fibroblast-like differentiated cells; iPS, induced pluripotent stem.

Oxidative damage to DNA was measured by staining cells against 8-hydroxy-2′-deoxyguanosine (8-OHdG), an oxidation derivative of deoxyguanosine, both at the basal level and after two hours of treatment with 500uM H₂0₂, a known pro-oxidant agent [49]. In untreated somatic HFF1 fibroblasts, few cells were positively stained with 8-OHdG and the number increased after H₂0₂ treatment (Fig. 6B). A similar staining pattern could be observed in both hESC-DFs and iPSC-DFs (Fig. 6B). However, 8-OHdG staining was almost undetectable in undifferentiated ES and iPS cells even after exposure to H₂0₂, suggesting a low presence of DNA damage and a higher resistance to oxidative stress (Fig. 6B).

The levels of lipid hydroperoxides (LPO), a biomarker of ROS-related lipid modification [42], were also significantly reduced in undifferentiated iPSCs and hESC in comparison to somatic fibroblasts and differentiated cells (Fig. 6C), implying a similar low amount of oxidative damage to lipids in both ES and iPS cells.

Finally, the levels of glutathione peroxidase one (GPX1) were detected by immunoblotting (Fig. 6D). GPX1 is the most abundant glutathione peroxidase isozyme in mammalian cells, and its expression has been shown to augment in response to increased ROS levels, as revealed by the gene regulatory network proposed to be required to maintain metabolic stability in aged tissues in mice [37]. GPX1 expression was significantly decreased in undifferentiated ESCs and iPSCs compared to somatic HFF1 fibroblasts (Fig. 6D). The data are in agreement with the microarray results, which showed a trend of downregulation for several genes related to the antioxidant response in both ES and iPS cells related to HFF1 (Fig. 2D). Moreover, GPX1 expression was induced in iPSC-DFs in comparison to undifferentiated iPSCs (Fig. 6D). A similar induction upon differentiation of hESCs, although previously demonstrated by others [8], could not be observed, possibly due to a different time point used, and might suggest a not completely identical redox behavior of iPS and ES cells upon differentiation into fibroblast-like cells.

Overall, iPSCs in the self-renewal state exhibited a decreased level of free radical damage to proteins, DNA, and lipids, coupled with a decreased expression of the antioxidant GPX1, all features of reduced mitochondria functionality that could also be observed in undifferentiated hESCs (Fig. 7).

7

General scheme of mitochondrial modifications upon reprogramming and differentiation of human iPS cells. The experimental findings herein presented suggest that several mitochondrial changes occur upon reprogramming of human neonatal somatic fibroblasts to pluripotency. Mitochondria become fewer, less developed, and less active. Lower levels of ATP generation are associated with increased lactate production, implying a likely dependence on anaerobic mechanism for energy supply to overcome the decreased mitochondria functionality. Upon differentiation, iPSCs exhibit mitochondria maturation and an anaerobic-to-aerobic metabolic transition. The mitochondria increase in number and become more developed. Mitochondrial activity also appears to be induced, as suggested by higher ATP generation and higher oxidative-mediated damage in differentiated cells compared to cells in their self-renewing undifferentiated state. Abbreviations: hESCs, human embryonic stem cells; iPSCs, induced pluripotent stem cells; mtDNA, mitochondrial DNA.

Discussion

Repression of the Mitochondrial/Oxidative Stress Pathway in Undifferentiated Human iPS Cells

Prolonged self-renewal may take place in undifferentiated pluripotent cells via specific modulation of pathways normally implicated in cellular senescence [5]. Accordingly, analysis of the telomerase and the p53 signaling pathways in hESCs and in somatic cell-derived iPSCs showed similar modification [1, 6, 9, 15–17]. Thus, cellular reprogramming has the ability to counteract the mechanisms of cellular aging and bring the cells to a self-renewing rejuvenated state. In this report, we aimed to investigate the mitochondrial/oxidative stress pathway, which is commonly associated with cellular and organismal aging, in human iPS and ES cells. We demonstrated that mitochondria derived from somatic cells acquire immature hESC-like features upon reprogramming into iPSCs and are then able to regain the characteristics of differentiated cells, in a similar fashion as mitochondria within hESCs (Fig. 7).

Mitochondrial architecture within iPSCs appeared characterized by underdeveloped cristae, which may imply the inhibition of mitochondrial energy production and is commonly linked with hypoxic conditions [50, 51]. Consequently, undifferentiated iPSCs exhibited low ATP/cell content and increased lactate production rates compared to somatic fibroblasts. Similar results were observed for undifferentiated hESCs, suggesting that both pluripotent cells rely on anaerobic rather than aerobic glycolysis for energy supply. Hypoxic culture conditions have been indeed associated with improvement of hESC cultivation and more recently of iPSCs generation [25, 27].

The mitochondria amount and mtDNA content in iPSCs were significantly decreased compared to somatic fibroblasts and similar to that of hESCs, suggesting a comparable level of mitochondrial replication. Accordingly, nuclear factors involved in mitochondrial biogenesis were similarly expressed in undifferentiated hESCs and iPSCs. Interestingly, steady-state expression of POLG has been found to be essential for the maintenance of pluripotency in mouse ESCs [28]. Since the whole process of active replication of mtDNA is regulated by nuclear encoded factors [44, 52], it is tempting to speculate that genetic reprogramming, which resets the nuclei to an embryonic-like epigenetic state, might also relay information to the mitochondria, reverting them to an analogous immature state. Indeed, inhibition of p53 signaling not only increases the efficiency of iPSC generation [16, 17] but also regulates mitochondrial biogenesis and causes mtDNA depletion [53–54].

Finally, the amount of oxidatively modified proteins, lipids, and DNA in iPSCs appeared similar to that in hESCs and significantly reduced compared to that in somatic fibroblasts. Thus, cellular reprogramming impacts the mitochondria-related oxidative stress pathway and is capable of restoring a cellular state characterized by low oxidative stress level.

Similar But Not Identical Activation of the Mitochondrial/Oxidative Stress Pathway upon Spontaneous Differentiation in Human iPS and ES Cells

After spontaneous differentiation into fibroblast-like cells, the mitochondrial network within human iPSCs and ESCs showed a parallel evolution leading to the acquisition of mature somatic fibroblast features, including tubular shape and fully developed cristae. Functional mitochondrial activation is necessary for efficient differentiation of mouse and human ESCs [8, 43, 55–57]. Indeed, upon spontaneous differentiation, a similar pattern of anaerobic-to-aerobic metabolic transition could be observed in both iPSCs and hESCs, as ATP levels and lactate generation were increased and reduced, respectively, together with an augmented content of mtDNA copies.

Some minor differences could be identified between iPSC- and hESC-derived fibroblasts. The expression of the mitochondrial biogenesis factor PGC-1alpha in iPSC-DFs was not highly inhibited as in the case of hESC-DFs. In addition, the expression of the antioxidant GPX1 was found to be significantly increased in iPSC-DFs but not in hESC-DFs compared to their undifferentiated cells. These findings may suggest that the hESCs and the iPSC-derived fibroblasts may be not identical in their ability to counteract the detrimental consequences of free radical generation. Additional studies with a higher number of iPSC lines are needed in order to provide a more definite answer.

Implications for the Use of Human iPS Cells for In Vitro Modeling of Aging and Neurodegeneration

Human iPSCs have enormous potential for regenerative medicine and specifically for the generation of in vitro models of complex human disorders. Several patient-specific iPSC lines have already been generated, including genetic and neurodegenerative diseases such as Parkinson's disease (PD) [58–60]. However, the distinction of reprogramming-induced phenomena from the actual disease phenotype is critical [14]. To this end, the study of mitochondrial biology and function in human iPSCs in comparison to ESCs becomes highly relevant. Indeed, altered mitochondrial activity and oxidative-related damage play significant roles in the pathogenesis and therapy of several human diseases, including sporadic and familial PD [61–63]. Hence, in iPSC-based in vitro PD models, the occurrence of disease-related mitochondrial modifications will have to be evaluated to ensure that these modifications are not a consequence of the reprogramming process.

Moreover, mitochondrial dysfunctions in aging and neurodegenerative disorders such as PD can be detected also in peripheral somatic cells [64–66]. Whether the pre-existing perturbation of mitochondrial functionality can be corrected upon iPSC generation is, thus, an essential question that needs to be addressed.

Conclusion

Overall, our findings demonstrate the repression of the somatic mitochondrial/oxidative stress pathway upon cellular reprogramming and highlight significant similarities between the mitochondrial properties of human ES and iPS cells, both in the undifferentiated and differentiated state. However, further studies are needed to better define the extent of these similarities. In fact, mtDNA mutations might persist upon reprogramming and then negatively affect the differentiation into specific cell types [67]. Finally, as early progenitor cells have been found to be more easily induced to pluripotency [68–70], age-related mitochondrial dysfunctions might influence the process of reprogramming, and it will thus be essential to study the same mitochondrial parameters in iPSCs derived from aged and diseased individuals.

Acknowledgements

The authors would like to thank Dr. M.F. Beal for critical reading, all colleagues in the Adjaye lab for support and fruitful discussion, A. Wulf-Goldenberg for the teratoma assay, and A. Sabha at the microarray facility. We are grateful to B. Di Stefano and Dr. V. Broccoli at the San Raffaele Institute of Milan for kindly providing the pMX viral vectors.

Disclosure of Potential Conflicts of Interest

The authors declare no competing financial or commercial interests and acknowledge support from the BMBF (01GN0807) and the Max Planck Society.

Author notes

Author contributions: A.P.: conception and design, data collection and interpretation, manuscript writing; B.F. data collection; R.L. data interpretation; H.L infrastructure support; J.A. conception, data interpretation, manuscript writing.

First published online in STEM CELLS EXPRESS March 3, 2010.

Disclosure of potential conflicts of interest is found at the end of this article.

Telephone: [phone]; Fax: [phone]