Author contributions: J.T.D.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; J.Y.J.: conception and design, collection and/or assembly of data; D.W.H.: collection and/or assembly of data, data analysis and interpretation; M.J.A.-B.: data analysis and interpretation; M.J.K.: collection and/or assembly of data; B.G.: data analysis and interpretation; H.Z. and I.S.-K.: provision of study material or patients; H.M.C.: provision of study material or patients, financial support; H.R.S.: conception and design, financial support, manuscript writing, final approval of manuscript.
First published online in STEM CELLS EXPRESS October 8, 2009.
Disclosure of potential conflicts of interest is found at the end of this article.
Somatic cells can achieve a pluripotent cell state in a process called pluripotential reprogramming. Multipotent stem cells can differentiate into cells of only one lineage, but pluripotent stem cells can give rise to cells of all three germ layers of an organism. In this study, we generated induced pluripotent stem (iPS) cells from bimaternal (uniparental) parthenogenetic neural stem cells (pNSCs) by transduction with either four (4F: Oct4, Klf4, Sox2, and c-Myc) or two (2F: Oct4 and Klf4) transcription factors. The resultant maternal iPS cells, which were reprogrammed directly from pNSCs, were capable of generating germ line-competent chimeras. Interestingly, analysis of global gene expression and imprinting status revealed that parthenogenetic iPS cells clustered closer to parthenogenetic ESCs than to female ESCs, with patterns that were clearly distinct from those of pNSCs. STEM CELLS 2009;27:2962–2968
Cells inherit a stable genetic program partly through various epigenetic marks, such as DNA methylation and histone modifications. This cellular memory, however, gets erased as cells undergo “genetic reprogramming”, and the cellular program reverts to that of an earlier developmental stage. Somatic cells can therefore re-establish a pluripotent cell state in a process called pluripotential reprogramming. Induced pluripotent stem (iPS) cells have recently been generated by reprogramming somatic cells by transduction with a transcription-factor cocktail . Such iPS cells have been derived from somatic cells of all three germ layers: mesoderm (fibroblasts, B lymphocytes, and stomach cells) [2, 3], endoderm (liver and pancreatic beta cells) [2, 4], and ectoderm (neural cells) [5–7]. These iPS cells all originate from biparental tissue. In this study, we have generated iPS cells from bimaternal (uniparental) parthenogenetic neural stem cells (pNSCs) by transduction with either four (4F) or two (2F) transcription factors. To date, parthenogenetic embryos and ES cells (pESCs) have been used to investigate the biological significance of genomic imprinting during embryonic development. Partial loss of imprinting has been observed in parthenogenetic blastocysts and their derivative pESCs but not in control blastocysts and their derivative ESCs . However, it is not clear why the imprinting status of parthenogenetic blastocysts and pESCs differs, as the constituent cells originate from totipotent fertilized embryos. Here we first generated parthenogenetic somatic cell lines and then used them to directly derive parthenogenetic maternal pluripotent stem (miPS) cells. We then compared these miPS cells with ESCs and biparental iPS cells and determined the similarity of their parthenogenetic imprinting pattern to that of pESCs or control ESCs. With this unique system it is now possible to study the molecular events that alter the genomic imprint during reprogramming of parthenogenetic somatic cells into their pluripotent derivatives.
MATERIALS AND METHODS
Generation of Parthenogenetic Embryos
OG2+/− mice were induced to superovulate by serial injections of 5 IU of pregnant mare serum gonadotropin (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) and 5 IU of human chorionic gonadotropin (hCG) (Sigma-Aldrich), administered 48 hours apart. Fourteen hours after the last hCG injection, cumulus-oocyte complexes were recovered from the mouse oviducts, and cumulus cells were removed by treatment with 0.1% hyaluonidase (Sigma-Aldrich) in M2 medium. For parthenogenetic activation, denuded oocytes were cultured in M16 medium supplemented with 10 mmol/l SrCl2 and 5 μg/ml cytochalasin B. Activated oocytes were cultured first in G1.3 medium (Vitrolife, Göteborg, Sweden, http://www.vitrolife.com) for 2 days and later in G2.3 medium (Vitrolife) under mineral oil at 37°C, 5% CO2. Blastocysts were transferred into one uterine horn of 2.5-dpc (days postcoitum) pseudopregnant mice; 10.5-dpc parthenogenetic embryos were recovered from the pregnant mice. OG2+/− mice were also used for derivation of parthenogenetic ESCs and female ESCs. All mouse strains were bred and housed at the mouse facility of the Max Planck Institute for Molecular Biomedicine or were bought from Harlan Winkelmann (Harlan GmbH, Borchen, Germany, http://www.harlan.com).
Generation of Parthenogenetic NSCs
Brain tissue was collected from 10.5-dpc parthenogenetic embryos (OG2+/−). Neurospheres were cultured as thoroughly described in our previous article . Primary neurospheres were re-plated onto gelatin-coated dishes in NSC expansion medium. Outgrowing cells were trypsinized, re-plated, and cultured in NSC expansion medium. NSCs were established by dissociation and were re-plated onto gelatin-coated dishes in NSC expansion medium, comprising NS-A media (Euroclone, Siziano, Italy, http://www.euroclonegroup.it) supplemented with N2 supplement, 10 ng/ml of each epidermal growth factor and basic fibroblast growth factor (Invitrogen, Carlsbad, CA, http://www.invitrogen.com), 50 μg/ml of bovine serum albumin (Fraction V; Gibco-BRL, Gaithersburg, MD, http://www.gibcobrl.com), 1× penicillin/streptomycin/glutamine, and 1× nonessential amino acids (Gibco BRL). Passage 10-12 pNSCs were used as donor cells for reprogramming.
Generation and Culture of Induced Pluripotent Stem Cells
pMX-based retroviral vectors encoding the mouse cDNAs of Oct4, Sox2, KLf4, and c-Myc  were separately co-transfected by packaging defective helper plasmids into 293T cells using Fugene 6 transfection reagent (Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com). Forty-eight hours after infection, virus supernatants were collected, filtered, and concentrated as previously described . Female NSCs (fNSCs) or pNSCs (OG2+/−) were seeded at a density of 1 × 105 cells per 6-well plate and incubated with virus-containing supernatants of the four factors (4F; 1:1:1:1) supplemented with 6 μg/ml of protamine sulfate (Sigma-Aldrich) for 24 hours. Cells were re-plated onto mitomycin C (MMC)-treated mouse embryonic fibroblast (MEF) feeders in ESC medium. Oct4-GFP+ iPS cells were sorted by fluorescence-activated cell sorting and subcultured onto MMC-treated MEF feeders. Passage 12-15 miPS cells were used for further analysis.
Aggregation of iPS Cells with Zona-Free Embryos
iPS cells were aggregated and cultured with denuded postcompacted 8-cell-stage mouse embryos. Morula embryos were flushed from 2.5-dpc B6C3F1 female mice. Clumps of iPS cells (10–20 cells per clump) were selected after a brief period of trypsinization and transferred into microdrops of KSOM medium (10% fetal calf serum) under mineral oil containing zona-free morula embryos. Aggregates were cultured overnight at 37°C, 5% CO2. After 24 hours of culture, the aggregated blastocysts were transferred into one uterine horn of a 2.5-dpc pseudopregnant recipient.
Whole Genome Expression Analysis
RNA samples for microarrays were prepared using Qiagen RNeasy columns with on-column DNA digestion. Three hundred nanograms of total RNA per sample was used as the starting material for the linear amplification protocol (Ambion, Austin, TX, http://www.ambion.com), which involved synthesis of T7-linked double-stranded cDNA and 12 hours of in vitro transcription incorporating biotin-labeled nucleotides. Purified and labeled cRNA was then hybridized for 18 hours onto MouseRef-8 v2 gene expression BeadChips (Illumina, Inc., San Diego, http://www.illumina.com) following the manufacturer's instructions. After being washed, the chips were stained with streptavidin-Cy3 (GE Healthcare, Chalfont St. Giles, UK, http://www.gehealthcare.com) and scanned using the iScan reader and accompanying software. Samples were exclusively hybridized as biological replicates.
The intensities for each bead were mapped onto gene information using BeadStudio 3.2 (Illumina). Background correction was performed using the Affymetrix Robust Multi-array Analysis background correction model, variance stabilization was performed using the log2 scaling, and gene expression normalization was calculated with the quantile method implemented in the lumi package of R-Bioconductor. Data postprocessing and graphics were performed with in-house developed functions in Matlab. Microarray data can be assessed at Gene Expression Omnibus (GEO) under the following link: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi.
Generation of Parthenogenetic MEFs and Neural Stem Cells
To derive parthenogenetic mouse embryonic fibroblasts (pMEFs) and pNSCs, parthenogenetic blastocysts (Fig. 1A) were obtained from parthenogenetically activated OG2+/− heterozygous embryos and transferred into pseudopregnant mice. The pregnant mice were then sacrificed at 10.5 dpc, as mouse parthenogenetic embryos die by day 10 of gestation [11, 12]. One fetus, which exhibited growth retardation, was recovered from a foster mother (Fig. 1A). The fetal brain was dissected and cultured for neurosphere culture; the rest of the fetus was cultured for pMEFs. pMEFs exhibited slow proliferation and appeared to undergo senescence after passage 2, indicating a pathenogenetic defect in proliferation (data not shown). However, parthenogenetic neurospheres were observed to form normally until day 6 of culture (Fig. 1B), when pNSCs were derived under adherent culture conditions. The pNSCs were morphologically similar to control fNSCs and stained positive for the two major NSC markers Sox2 and Nestin (Fig. 1B). Moreover, pNSCs and fNSCs expressed similar levels of Sox2, Nestin, and Olig2 (Fig. 1C). Interestingly, however, pNSCs exhibited a much slower proliferation rate than fNSCs (Fig. 1D). Since maternally imprinted genes, such as Impact, Ndn, and Snrpn, are associated with the development of the mouse brain [13–15], it is possible that the slower proliferation rate of pNSCs may be due to the lower expression of maternally imprinted genes in pNSCs compared with those in control fNSCs (supplemental online Fig. 1). But pNSCs derived from parthenogenetic embryos can still self-renew for more than 30 passages (3 months).
Generation of miPS Cells
We first attempted to derive miPS cells from passage 2 pMEFs, which had already undergone senescence. However, we failed to generate Oct4-GFP+ or ESC-like colonies from pMEFs by transduction with four transcription factors (Oct4, Sox2, c-Myc, and Klf4) in two independent experiments.
Next, we used pNSCs to generate miPS cells. NSCs have been successfully reprogrammed by transduction with fewer than four transcription factors [6, 7, 16], and thus appear to be a somatic cell type ideal for reprogramming. We attempted to establish miPS cells from pNSCs using retroviral vectors transducing either four or two transcription factors and, in parallel, transduced normal fNSCs as a control. During the reprogramming of control fNSCs, ESC-like colonies were first observed on day 4 postinfection (supplemental online Fig. S2) and Oct4-GFP+ colonies on day 9 postinfection (Fig. 2A). However, during the reprogramming of pNSCs, ESC-like colonies were first observed on day 6 postinfection, 2 days later than in control fNSCs (supplemental online Fig. S2), and Oct4-GFP+ colonies were observed on day 17 postinfection (Fig. 2A), 8 days later than in fNSCs. The number of Oct4-GFP+ ESC-like colonies on day 17 was calculated to be 95 ± 21.2 (mean ± SD) and 3 ± 1.0 for a starting cell population of 1 × 105 fNSCs and pNSCs, respectively (Fig. 2B). The reprogramming efficiency was estimated by flow cytometric analysis for Oct4-GFP+ cell count on day 21 postinfection. The percentage of Oct4-GFP+ cells obtained after the reprogramming of pNSCs was approximately 20-fold lower than that obtained after the reprogramming of fNSCs (0.1% vs. 2.1%) (Fig. 2C). There is no obvious explanation for the lower reprogramming efficiency of pNSCs compared with that of fNSCs, but the slow proliferation rate of pNSCs may partially account for this, as viral infection-induced reprogramming may depend on the active proliferation of somatic cells . Next, we generated miPS cells from pNSCs by transduction with two transcription factors: Oct4 and Klf4. Oct4-GFP+ colonies were first observed on day 30 postinfection in both pNSCs and control fNSCs (Fig. 3A). The efficiency of reprogramming of pNSCs and fNSCs using these two factors was similarly reduced, indicating that the use of fewer than four transcription factors reduces the reprogramming rate of both cell types in a similar fashion. These data suggest that miPS cells can be generated from pNSCs by transduction with at least two transcription factors.
Developmental Potential of miPS-2F Cells In Vitro and In Vivo
To determine whether miPS-2F cells have a differentiation potential comparable to pluripotent ESCs, we induced the differentiation of miPS-2F cells in vitro using embryoid body (EB) formation. After the cultured EBs (miPS-2F cell aggregates) were plated onto gelatin-coated dishes, miPS-2F cells were found to differentiate into endodermal (Sox17+), mesodermal (Brachyury+), and ectodermal (Tuj1+) tissues—that is, derivatives of all three germ layers (Fig. 3B). In addition, we assessed the in vivo developmental potential of the miPS-2F cells through aggregation with wild-type morula embryos. miPS-2F cells were found to efficiently incorporate into the inner cell mass of the blastocysts and contribute to germ line, which were observed in 13.5-dpc embryos (Fig. 3C). After genotyping the organs of chimeric embryos for the Oct4-GFP cassette (miPS-2F cell origin), we demonstrated that the miPS-2F cells contributed to the formation of ectodermal (skin and brain), endodermal (liver and lung), and mesodermal (kidney) tissues in vivo (Fig. 3D). We next assessed whether the miPS-2F cells were capable of contributing to the formation of adult chimeras. We found that three of the five pups born to two surrogate mothers were indeed chimeras (Fig. 3E). Bisulfite DNA sequencing analysis showed that the Nanog promoter region, which was hypermethylated in pNSCs, was completely demethylated in miPS-2F cells and also in pESCs (supplemental online Fig. S3). These data confirm that the miPS-2F cells, which were reprogrammed from pNSCs, possess phenotypical characteristics and functional properties characteristic of pluripotent cells. Interestingly, miPS-2F cells exhibited a global gene profile clustering closer to pESCs clustering than to female ESCs (fESCs) clustering and that was clearly distinct from that of pNSCs (Fig. 4A–4C and supplemental online Fig. S4).
Genomic Imprinting Pattern in miPS Cells
To determine whether the parthenogenetic imprinting pattern of miPS cells is similar to that of pESCs or control ESCs, we compared the genomic imprinting status of pNSCs with that of miPS-2F cells. The imprinting status of the maternally imprinted genes Peg1 (also known as Mest) and Peg3 and of the paternally imprinted gene H19 was determined by combined bisulfite restriction analysis . As expected, pNSCs exhibited a parthenogenetic imprinting pattern. However, loss of methylation in the Peg1 and Peg3 regions and gain of methylation in the H19 region were observed in the miPS-2F cells, indicating that pNSCs lose the parthenogenetic imprinting patterns during pluripotential reprogramming.
Two clonal lines of the four factor-induced miPS cells and pESCs (passage 7) also exhibited the same imprinting patterns as miPS-2F cells (Fig. 4D), which is consistent with observations in pESCs . Our results were confirmed by DNA bisulfite sequencing analysis. pNSCs exhibited hypomethylation of the paternally imprinted genes H19 and Igf2 and hypermethylation of the maternally imprinted genes Peg1 and Peg3 (Fig. 4E). Interestingly, however, de novo DNA methylation was observed in the H19 and Igf2 regions of pNSC-2F cells (Fig. 4E), confirming that loss of parthenogenetic imprinting occurs during pluripotential reprogramming. Changes in the parthenogenetic imprinting patterns observed during the reprogramming process are reminiscent of those observed during the establishment of induced pluripotent stem cells from germ cells . Therefore, loss of parthenogenetic imprinting appears to be a phenomenon specific to parthenogenetic pluripotent stem cells, such as pESCs and miPS cells. Collectively, miPS cells and pESCs display similar imprinting patterns and global gene expression profiles.
Here, we first generated pNSCs and then used them to directly derive parthenogenetic miPS cells by transduction with reprogramming factors. The direct reprogramming of parthenogenetic somatic cells renders possible a direct comparison between the original parthenogenetic somatic cells and their pluripotent derivative miPS cells, which would not be possible using pESCs.
Interestingly, pNSCs exhibited a slower proliferation rate than control biparental fNSCs, whereas miPS cells and pESCs showed comparable proliferation rates. This may be due to the re-establishment of maternally imprinted genes, which are associated with the development of the mouse neural system, during the pluripotential reprogramming of miPS cells. Also conceivable is the possibility that maternally imprinted genes are not necessary for the self-renewal of pluripotent cells or that maternally imprinted genes are reprogrammed during the induction of pluripotency. pMEF cells had a very severe proliferation defect, however, exhibited by slow proliferation and eventual senescence after passage 2, which may have ultimately resulted in their failure to generate iPS cells by transduction with four transcription factors. Hanna et al. reported that terminally differentiated B cells could not be induced to become iPS cells unless the B-cell-specific gene Pax5 had been inactivated by overexpressing CCAAT/enhancer-binding-protein-α (C/EBPα) . Therefore, successful reprogramming of pMEFs by a four transcription factor-transduction process may actually require the presence of yet additional factors. This senescence barrier also might be overcome by hypoxia condition  or manipulation of the p53 pathway [21–25] since these systems remarkably increase the reprogramming efficiency. Park et al. recently reported the derivation of disease-specific iPS cells from individual patients with different genetic diseases . These immortalized cells directly induced from patients' somatic cells represent a potential alternative to the sparse cell lines used for research into the etiology and treatment of various diseases. Consistent with this approach, we provide two alternative resources for investigations into the parthenogenetic and genomic imprinting status of cells: parthenogenetic somatic cells and pluripotent stem cells.
In this study, we found that miPS cells are capable of generating germ line chimeras and exhibit a global gene expression profile clustering closer to that of pESCs than to that of control fESCs. Interestingly, miPS cells, which were directly reprogrammed from pNSCs, exhibited changes in genomic imprinting patterns characteristic of the donor cells; pNSCs containing only maternal alleles regain the biparental imprinting patterns after reprogramming. To date, ESCs derived from parthenogenetic embryos have been widely used to study parthenogenetic development. pNSCs, which are also capable of self-renewal, maintain their imprinting pattern; yet once the cells are reprogrammed to pluripotent miPS cells, their parthenogenetic imprinting pattern becomes similar to that of pESCs. Thus, changes in the genomic imprinting of parthenogenetic cells occur concomitantly with the induction of pluripotency in multipotent stem cells. Here, we suggest that pluripotential reprogramming induced the loss of parthenogenetic imprinting and postulate that changes in genomic imprinting are a marker for pluripotency but not for multipotency. This phenomenon may be explored in an effort to discover treatments for a number of diseases, such as Prader-Willi and Angelman syndrome, which are caused by an abnormality in imprinted genes . It could be envisioned that cells from a patient afflicted with an imprinting-related disease that exhibit abnormal imprinting patterns could be subjected to in vitro reprogramming and differentiation and then transplanted back to the patient to replace cells of the impaired tissue. Therefore, pNSCs and miPS cells represent suitable ex vivo model systems for exploring how genomic imprinting is regulated in both somatic and pluripotent cell types.
We are grateful to Claudia Ortmeier for assisting with the real-time RT-PCR experiments. We thank Dr. Toshio Kitamura for providing the MX retroviral vector.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.