Reprogramming of Neural Progenitor Cells into Induced Pluripotent Stem Cells in the Absence of Exogenous Sox2 Expression


  • Sarah Eminli,

    1. Massachusetts General Hospital Cancer Center and Center for Regenerative Medicine, Boston, Massachusetts, USA
    2. Harvard Stem Cell Institute, Free University of Berlin, Berlin, Germany
    3. Department of Biology, Chemistry, and Pharmacy, Free University of Berlin, Berlin, Germany
    Search for more papers by this author
  • Jochen Utikal,

    1. Massachusetts General Hospital Cancer Center and Center for Regenerative Medicine, Boston, Massachusetts, USA
    2. Harvard Stem Cell Institute, Free University of Berlin, Berlin, Germany
    Search for more papers by this author
  • Katrin Arnold,

    1. Massachusetts General Hospital Cancer Center and Center for Regenerative Medicine, Boston, Massachusetts, USA
    2. Harvard Stem Cell Institute, Free University of Berlin, Berlin, Germany
    Search for more papers by this author
  • Rudolf Jaenisch,

    1. Whitehead Institute and Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA
    Search for more papers by this author
  • Konrad Hochedlinger

    Corresponding author
    1. Massachusetts General Hospital Cancer Center and Center for Regenerative Medicine, Boston, Massachusetts, USA
    2. Harvard Stem Cell Institute, Free University of Berlin, Berlin, Germany
    • Konrad Hochedlinger, Ph.D., Massachusetts General Hospital, 185 Cambridge Street, CPZN 4242, Boston, Massachusetts 02114, USA. Telephone: 617-643-2075; Fax: 617-643-3170===

    Search for more papers by this author


Expression of the transcription factors Oct4, Sox2, Klf4, and c-Myc in mesodermal and endodermal derivatives, including fibroblasts, lymphocytes, liver, stomach, and β cells, generates induced pluripotent stem (iPS) cells. It remains unknown, however, whether cell types of the ectodermal lineage are equally amenable to reprogramming into iPS cells by the same combination of factors. To test this, we have isolated genetically marked neural progenitor cells (NPCs) from neonatal mouse brains and infected them with viral vectors expressing Oct4, Sox2, Klf4, and c-Myc. Infected NPCs gave rise to iPS cells that expressed markers of embryonic stem cells, showed demethylation of pluripotency genes, formed teratomas, and contributed to viable chimeras. In contrast to other somatic cell types, NPCs expressed high levels of endogenous Sox2 and thus did not require viral Sox2 expression for reprogramming into iPS cells. Our data show that in addition to mesoderm- and endoderm-derived cell types, neural progenitor cells of the ectodermal lineage can be reprogrammed into iPS cells, suggesting that in vitro reprogramming is a universal process. These results also imply that the combination of factors necessary for reprogramming is dependent on cellular context.

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


Author contributions: S.E. and J.U.: conception and design, provision of study material, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; S.E. and J.U. contributed equally to this work. K.A.: provision of study material, collection and/or assembly of data, data analysis and interpretation, final approval of manuscript; R.J.: provision of study material, final approval of manuscript; K.H.: conception and design, financial support, administrative support, data analysis and interpretation, manuscript writing, final approval of manuscript.

Induced pluripotent stem (iPS) cells have been generated from fetal and adult fibroblast cultures in mouse and human following ectopic expression of the transcription factors Oct4 and Sox2, combined with either Klf4 and c-Myc or Lin28 and Nanog [1, [2], [3], [4], [5], [6]–7]. Recently, murine liver and stomach cells [8], β cells [9], and lymphocytes [10] have been reprogrammed into iPS cells. Interestingly, the reprogramming of mature B lymphocytes but not that of progenitor B cells required the introduction of a fifth transcription factor in addition to Oct4, Sox2, Klf4, and c-Myc, suggesting that certain less differentiated cells are more easily reprogrammed than fully differentiated cells.

iPS cells are molecularly and functionally indistinguishable from embryonic stem (ES) cells. For example, they show reactivation of pluripotency genes, telomerase activity, and the silent X chromosome in female cells, as well as genome-wide transcriptional and epigenetic patterns that are characteristic of ES cells. Moreover, iPS cell chimeras give rise to offspring, indicating their ability to contribute to the germline. Many chimeras do, however, succumb to tumors, likely because of the reactivation of retroviral transgenes or insertional mutagenesis [5].

The contribution of the individual factors to the reprogramming process remains largely unknown. The observation that Klf4 and c-Myc can be replaced by Nanog and Lin28 during the generation of human iPS cells [7] suggests that Oct4 and Sox2 may be essential for reprogramming, whereas c-Myc and Klf4 may enhance the process. Indeed, recent evidence indicated that omission of c-Myc from the reprogramming cocktail gave rise to iPS cells from fibroblasts, albeit at extremely low frequencies and with delayed kinetics [11, 12]. Omission of Sox2, on the other hand, gave rise to ES cell-like lines, which were unable to differentiate and resembled nullipotent embryonic carcinoma cells [2].

Fibroblast cultures are heterogeneous and contain both mesenchymal and nonmesenchymal cell types, including keratinocytes, muscle cells, endothelial cells, and hematopoietic cells. Because of this, the exact cell type that gave rise to iPS cells in previous experiments remains unknown. Likewise, explanted liver and stomach tissue may contain nonepithelial cell types, such as mesenchymal and hematopoietic cells, which may serve as selective donors for iPS cells. To assess the identity of the cells of origin of iPS cells, either prospective identification with reporters or surface antigens or retrospective identification by genetic lineage tracing is required. For example, albumin-producing liver cells, which were genetically marked by Cre-lox technology, have been shown to give rise to iPS cells, although their developmental potential has not been evaluated [8]. Recently, genetically marked β cells [9] and lymphocytes carrying differentiation-associated DNA rearrangements [10] have been reprogrammed to pluripotency, demonstrating that defined mature cell types remain competent to give rise to iPS cells.

Given that mesodermal (fibroblasts, blood) and endodermal (liver, stomach, pancreas) cells had been reprogrammed into iPS cells in previous experiments, we set out to test the questions of (a) whether cells of the third germ layer (ectoderm) are equally amenable to reprogramming into iPS cells and (b) whether ectodermal cells require the same combination of factors. Addressing these questions is important for drawing conclusions about the generality of factor-induced reprogramming across cell types from all three germ layers and for identifying cell types that may require fewer factors.

Here we attempted the reprogramming of neural progenitor cells (NPCs), as they are a well-defined ectodermal cell type that can be marked genetically using Cre-lox technology. In addition, their serum-free growth conditions [13] select against non-neural cell types, such as fibroblasts, and they already express one of the key reprogramming factors, Sox2 [14, 15], and thus may need less genetic manipulation to become iPS cells.

Materials and Methods

Viral Production

Viral infections were performed with retroviral vectors or with replication-defective doxycycline-inducible lentiviral vectors and a lentiviral vector constitutively expressing the reverse tetracycline-controlled transactivator (rtTA). For retroviral production, cDNAs for Oct4, Sox2, c-Myc (T58A mutant), and Klf4 were cloned into the retroviral pMX vector and transfected into 293T cells together with packaging plasmids PCG-ori-gagpol and PCG-ori-VSV-G using Fugene transfection reagent (Roche Diagnostics, Basel, Switzerland, Viral supernatants were harvested on three consecutive days starting 24 hours after transfection, concentrated by centrifugation at 20,000 rpm for 1.5 hours at 4°C, resuspended in phosphate-buffered saline (PBS), and stored at −80°C. Infections were carried out in 1 ml of NPC medium using 10 μl of each viral concentrate per 35-mm plate. The generation and structure of tet-inducible lentiviral vectors (LV-tetO) expressing c-Myc, Klf4, Sox2, and Oct4 have been described in detail elsewhere [14]. Briefly, a tet-inducible lentiviral vector termed LV-tetO was generated by replacing the ubiquitin promoter elements from the FUΔGW vector [16] with tetO sequences. cDNAs for c-Myc (T58A mutant), Klf4, and Oct4 were isolated from pMIG or pMX vectors and cloned into LV-tetO. The lentiviral vector containing rtTA is under control of an ubiquitin promoter [14]. To produce infectious viral particles, 293T cells cultured on 10-cm dishes were transfected with the LV-tetO vectors (11 μg) together with the packaging plasmids VSV-G (5.5 μg) and Δ8.9 (8.25 μg) using Fugene. Viral supernatants were harvested on three consecutive days starting 24 hours after transfection, yielding a total of ∼30 ml of supernatant per viral vector. Viral supernatant was concentrated approximately 100-fold by ultracentrifugation at 20,000 rpm for 1.5 hours at 4°C and resuspension in 300 μl of PBS. Infections were carried out in 1 ml of NPC medium using 5 μl of each viral concentrate per 35-mm plate.

Cell Culture

Neural progenitor cells were isolated from newborn brains of wild-type mice, Sox2-green fluorescent protein (GFP) and Oct4-GFP indicator mice (all mixed background of C57Bl/6 and 129SvJae), and Nestin-Cre × Rosa26R-enhanced yellow fluorescence protein (EYFP) (mixed background of C57Bl/6 and 129SvJae) lineage-tracing mice and expanded in DMEM/F12 medium containing 20 μg/ml insulin, 100 μg/ml transferrin, 20 nM progesterone, 60 μM putrescine, 30 nM sodium selenite (all from Sigma-Aldrich, St. Louis,; 0.154% d-glucose; and 1% l-glutamine and penicillin-streptomycin (Gibco, Grand Island, NY, Epidermal growth factor (EGF; Sigma-Aldrich) and basic fibroblast growth factor (bFGF; Sigma-Aldrich) were added every 2 days to the culture medium at 20 ng/ml. NPCs were grown adhesion-free for the first six passages to eliminate non-sphere-forming cells. Viral vector infections were usually carried out after passage 6 on 10-cm plates at a density of 200,000 cells per plate in supplemented F12 medium containing 5 μg/ml polybrene. Medium changes were performed after 12–24 hours with supplemented F12 medium containing 1 μg/ml doxycycline. Fresh medium with doxycycline was added every other day until iPS colonies developed, followed by the addition of 1,000 U/ml leukemia inhibitory factor to supplemented F12 medium. Three days later, cell culture conditions were switched to ES cell medium without doxycycline. iPS colonies were picked into 96-well cells containing PBS without magnesium and calcium using a 10-μl pipette. Trypsin was added to each well and incubated for 5 minutes, and single-cell suspension was transferred into 24-well dishes containing mouse embryonic fibroblasts (MEFs). Picked iPS cells were grown on MEFs under ES cell conditions. For blastocyst injections, iPS cells were marked with an FUΔGW lentiviral vector constitutively expressing GFP.

Determination of iPS Cell Formation Efficiency with Three and Four Factors

The efficiency of iPS cell formation is based on the number of Oct4-GFP-positive iPS colonies, the initial cell number, and the infection efficiency of cells. Cells (1 × 106) were seeded on polyornithine-coated plates and infected with either three or four doxycycline-inducible lentiviral vectors in addition to a lentiviral vector expressing rtTA. We have calculated the infection efficiency of NPCs and tail fibroblasts with tetO-GFP lentiviral vector in the presence of the rtTA-expressing lentiviral vector (80.6% and 90.7%, respectively). The fraction of cells receiving all viral vectors in the various combinations was determined by multiplying the infection rate of GFP-positive cells as has been done previously [2]. The final percentage of efficiency was determined by dividing the number of GFP-positive colonies by the fraction of cells calculated to express either three or four viral transgenes.


To test the differentiation potential of NPCs, cells were cultured in serum-containing medium on pretreated coverslips. Cells were fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton X-100. The cells were then stained with primary antibodies against S100β (S2532; Sigma-Aldrich) and βIII-tubulin (ab18207; Abcam, Cambridge, MA,, followed by staining with the respective secondary antibodies conjugated to Alexa Fluor 546 (Invitrogen, Carlsbad, CA, Nuclei were counterstained with 4,6-diamidino-2-phenylindole (Invitrogen). Cells were imaged using a Leica DMI4000B inverted fluorescence microscope (Leica, Heerbrugg, Switzerland, equipped with a Leica DFC350FX camera.

Alkaline Phosphatase Staining

Alkaline phosphatase staining was performed with the Vector Red substrate kit (Vector Laboratories, Burlingame, CA, according to the manufacturer's instructions.

Viral Vector Integration Analysis

Retroviral vector integrations were determined by Southern blot analysis. The genomic DNA was digested with BamHI (Oct4, Klf4), HindIII (Sox2), and BglII (c-Myc) and hybridized with the respective cDNA probes as described previously [4]. To confirm the results of the Southern blot for viral integration sites, polymerase chain reaction (PCR) analysis was performed. DNA of iPS cells was isolated using standard procedures. Primer sequences have been described elsewhere [4].

Real-Time PCR Analysis

To test expression of pluripotency genes from the endogenous locus, total RNA was treated with the DNA-free Kit (Ambion, Austin, TX, and reverse-transcribed with SuperScript First-Strand Synthesis System (Invitrogen) using oligo(dT) primers according to the manufacturer's instructions. Primer sequences are listed in supplemental online Table 1.

Quantitative PCR

Real-time quantitative PCRs were set up in triplicate with the Brilliant II SYBR Green QPCR Master Mix (Stratagene, La Jolla, CA, and run on an Mx3000P QPCR System (Stratagene). Primer sequences are listed in supplemental online Table 2.

Bisulfite Sequencing

Bisulfite treatment of DNA was performed with the EpiTect Bisulfite Kit (Qiagen, Hilden, Germany, according to the manufacturer's instructions. Primer sequences were as previously described for Oct4 and Nanog [2]. Amplified products were purified by using gel filtration columns, cloned into the pCR4-TOPO vector (Invitrogen), and sequenced with M13 forward and reverse primers.

Generation of Teratomas and Chimeras

For teratoma induction, 2 × 106 cells of each iPS cell line were injected subcutaneously into the dorsal flank of isoflurane-anesthetized SCID mice. Teratomas were recovered 3–4 weeks postinjection, fixed overnight in 10% formalin, paraffin-embedded, and processed with hematoxylin and eosin. For chimera production, female BDF1 mice were superovulated with pregnant mare serum and human chorionic gonadotropin (hCG) and mated to BDF1 stud males. Zygotes were isolated from plugged females 24 hours after hCG injection. After 3 days of in vitro culture in KSOM media, blastocysts were injected with iPS cells and transferred into day 2.5 pseudopregnant recipient females. Caesarean sections were performed 17 days later, and pups were fostered with lactating Swiss females.


We isolated NPCs from neonatal mice and initially expanded them as spheres to eliminate adherent, non-sphere-forming cells in serum-free medium containing EGF and bFGF as described previously [13]. After six passages, NPCs were switched to adherent cultures on polyornithine-coated plates. To characterize NPCs, we induced differentiation into glia and neuronal cells by adding serum to the cultures. Indeed, immunofluorescence for the neuronal marker βIII-tubulin and the glial marker S100β showed positive cells, indicating that our NPC cultures can give rise to these two main neural lineages (supplemental online Fig. 1).

Our initial attempts to produce iPS cells from NPCs following single rounds of infection with pMX retroviral vectors expressing Oct4, Sox2, c-Myc, and Klf4 were unsuccessful, even though control infections of fibroblasts consistently yielded iPS cells (data not shown), suggesting that NPCs are more difficult to infect or reprogram than fibroblasts. We therefore used a modified protocol, which is depicted in Figure 1A. Specifically, NPCs were infected with either concentrated retroviral or replication-incompetent doxycycline-inducible lentiviral vectors, which have been described previously (additional information is given in Materials and Methods), on days 1 and 3 after plating. At day 7, ES-like colonies first appeared (Fig. 1B) and cultures were switched to ES cell medium containing serum, which induced instant differentiation of NPCs and favored growth of ES-like cells (Fig. 1B, day 9). By day 11, ES-like colonies were picked and expanded into stable iPS lines (Fig. 1B, day 13). Lentivirally produced iPS lines continued to grow in the absence of doxycycline, indicating successful reactivation of the endogenous pluripotency program and independence of transgene expression. In agreement, staining of established iPS lines for the ES cell marker alkaline phosphatase yielded positive colonies, and real-time (RT) PCR for pluripotency markers showed expression of endogenous Oct4, Sox2, Eras, Cripto, and Nanog, confirming that the virally infected NPCs had reacquired an embryonic phenotype (Fig. 1B, 1C). Moreover, the Oct4 and Nanog promoter regions, which were heavily methylated in NPCs, were demethylated in iPS cells, demonstrating faithful epigenetic reprogramming (Fig. 1D).

Figure Figure 1..

Generation of iPS cells from NPCs. (A): Experimental outline scheme for generating iPS cells from NPCs. (B): Untreated NPCs and virally vector-infected NPCs at days 7, 9, and 13. Shown are NPCs on polyornithine-coated plates (upper left panel), developing iPS colonies at day 7 after infection (upper center panel), developing iPS colony after addition of mouse embryonic fibroblasts at day 9 (lower right panel), and picked NPC-derived iPS cell colonies on feeder cells at day 13 (lower center panel). iPS cell colonies derived from NPCs showed positive AP staining. (C): Reverse transcriptase polymerase chain reaction analysis of marker gene expression in V6.5 ES cells, NPCs, and two iPS cell lines derived by 3F and 4F from NPCs. Nat1 was used as a loading ctrl. (D): Bisulfite sequencing of the Oct4 and Nanog promoters in NPCs, iPS cells produced with 3F and 4F, and ES cells. Open circles represent unmethylated CpGs; filled circles denote methylated CpGs. Abbreviations: 3F, three factors; 4F, four factors; AP, alkaline phosphatase; ctrl., control; dox, doxycycline; ES, embryonic stem; iPS, induced pluripotent stem; LIF, leukemia inhibitory factor; NPC, neural progenitor cell; RT, real-time.

NPCs have previously been shown to express high levels of Sox2 [15], which maintains neural progenitor identity by inhibiting differentiation [17]. We verified this by RT-PCR (Fig. 1C) and by fluorescence-activated cell sorting (FACS) analysis of NPCs that we established from Sox2-GFP reporter mice [14, 15] (Fig. 2). As shown in Figure 2A, NPCs grown as adherent cultures expressed Sox2-GFP homogeneously under serum-free culture conditions (F12 media, supplemented with EGF and bFGF) but instantly lost GFP signal upon induction of differentiation by the addition of serum. Likewise, Sox2-GFP ES cells showed strong expression of GFP in undifferentiated colonies but lower expression in colonies with differentiated morphology (Fig. 2A, dashed circle). Quantification of GFP fluorescence by FACS analysis confirmed this observation and indicated that ES cells and iPS cells derived from NPCs expressed comparable levels of Sox2-GFP (Fig. 2B). Interestingly, tail-tip fibroblasts isolated from newborn Sox2-GFP pups also expressed Sox2-GFP, which was, however, an order of magnitude lower than that in ES cells and NPCs (Fig. 2B).

Figure Figure 2..

Analysis of NPCs and ES cells from Sox2-GFP mice. (A): Sox2-GFP NPCs grown in serum-free supplemented Ham's F-12 medium showed Sox2-GFP expression, whereas NPCs differentiated in ES cell medium lost Sox2-GFP expression. Sox2-GFP ES cells showed strong expression of Sox2-GFP in undifferentiated colonies and lower expression in colonies with differentiated morphology (circled colony). (B): Sox2-GFP expression in NPCs, iPS cells, TT fibroblasts, and ES cells. (C): iPS cells derived by three factors from Sox2-GFP NPCs showed Sox2 expression. Shown are a bright-field image (left) and a GFP image (right). Abbreviations: ES, embryonic stem; GFP, green fluorescent protein; iPS, induced pluripotent stem; Max, maximum; NPC, neural progenitor cell; TT, tail-tip; WT, wild-type.

On the basis of the observation that abundant Sox2 expression was detectable in more than 97% of NPCs, we reasoned that Oct4, Klf4 and c-Myc alone might be sufficient to induce pluripotency. We therefore infected NPCs with retroviral vectors expressing Oct4, Klf4, and c-Myc and, indeed, were able to recover iPS colonies that grew into stable cell lines (data not shown). In addition, NPCs carrying the Sox2-GFP reporter were infected with doxycycline-inducible lentiviral vectors [14] expressing Oct4, Klf4, and c-Myc, as well as with a fourth lentiviral vector constitutively expressing rtTA in the presence of doxycycline. Similar to the retrovirally derived iPS cells, lentivirally transduced NPCs gave rise to ES-like colonies, which grew into stable iPS lines upon discontinuation of doxycycline. Together, these observations demonstrate that exogenous Sox2 expression is not required for the reprogramming of NPCs into pluripotent cells (Fig. 2C). Similarly to four-factor iPS cells, three-factor iPS cells expressed endogenous pluripotency genes and showed demethylation of the Nanog and Oct4 promoter regions (Fig. 1C, 1D). Moreover, RT-PCR analysis confirmed that the retroviral and lentiviral transgenes became silenced, whereas the corresponding endogenous Oct4, Klf4, and c-Myc loci became reactivated in iPS cells (supplemental online Fig. 2). Attempts to produce stable iPS cell lines from tail-tip fibroblasts in the absence of exogenous Sox2 expression using the doxycycline-inducible system were unsuccessful in our hands (data not shown), which is in contrast to a previous report suggesting that nullipotent iPS cells can be generated from fibroblasts following retroviral overexpression of Oct4, Klf4, and c-Myc alone [2].

To assess the efficiency of reprogramming NPCs to pluripotency, we generated iPS cells from NPCs and tail-tip fibroblasts carrying an Oct4-GFP reporter [9, 18] and counted the number of GFP-positive colonies. A representative Oct4-GFP-positive iPS colony is shown in supplemental online Figure 3. iPS cells derived from tail fibroblasts were obtained at an efficiency of approximately 0.04%, which is consistent with previous observations (supplemental online Fig. 4) [2, [3]–4]. In contrast, NPCs infected with four factors gave rise to iPS cells at efficiencies of 0.0002% and 0.0008%, respectively, which is two orders of magnitude lower than that of fibroblasts and may explain our initial failure at generating NPC-derived iPS cells. Interestingly, iPS cells produced with only the three factors Oct4, Klf4, and c-Myc gave rise to iPS cells at frequencies of 0.002% and 0.001%, which is roughly four times higher than the efficiency of obtaining iPS cells in the presence of exogenous Sox2 expression. We did not observe a difference in the appearance of ES-like colonies, however, between virally infected fibroblasts or NPCs. Together, these data suggest that the efficiency of obtaining iPS cells from NPCs is significantly lower than that from fibroblasts and that overexpression of Sox2 has adverse effects on NPC reprogramming.

To test the differentiation potential of NPC-derived iPS cells, we first aggregated them into embryoid bodies (EBs) and then explanted them in culture. Eighteen of 18 tested iPS cell lines derived with three or four factors, irrespective of the viral system, gave rise to contracting muscle fibers (“beating hearts”) in culture after 7 days, which is indicative of differentiation into the cardiomyocyte lineage (data not shown). Furthermore, eight of nine three-factor iPS clones and six of six four-factor iPS clones produced with either lentiviral vectors or retroviral vectors gave rise to differentiated teratomas upon injection into the flanks of SCID mice, thus confirming their pluripotency (Fig. 3A). To assess the developmental potential of NPC iPS cells, a three-factor iPS line produced from Sox2-GFP NPCs using doxycycline-inducible lentiviral vectors was labeled with another lentiviral construct constitutively expressing GFP, FACS-sorted for the GFPhigh-expressing population, and injected into blastocysts to produce chimeric animals. Three viable newborn pups were obtained that showed widespread green fluorescence (Fig. 3B). All three mice developed into adulthood and showed obvious coat color chimerism (Fig. 3C), indicating that NPC-derived iPS cells, like iPS cells from mesodermal and endodermal cell types, contributed to postnatal development.

Figure Figure 3..

Differentiation potential of three-factor neural progenitor cell (NPC)-derived induced pluripotent stem (iPS) cells. (A): Hematoxylin and eosin staining of a teratoma derived from an NPC-iPS line shows differentiation into cell types from all three germ layers: epithelial structures (left), cartilage, bone (center), and glandular structures (right). (B): Viable newborn chimeras from iPS cells derived from Sox2-GFP NPCs. Chimeric pup (right) and nonchimeric littermate (left) are shown under regular light (left) and UV light (center and right). (C): Adult chimeras derived from three-factor Sox2-GFP NPC-iPS cells. Abbreviation: GFP, green fluorescent protein.

Although it is highly unlikely, there is the remote chance that iPS cells may have originated from a non-neural cell type contaminating the NPC culture. To unequivocally demonstrate that iPS cells derived from NPC cultures are of neuroectodermal origin, we produced NPCs from neonates in which neural progenitors had been permanently marked by genetic lineage tracing. Specifically, Nestin-Cre mice, in which Cre recombinase is driven by the neural progenitor-marker Nestin [19], were crossed with ROSA26-EYFP reporter mice, and NPCs were isolated from newborn offspring as described before. FACS analysis of NPC cultures showed that roughly 98% of the cells were EYFP-positive (Fig. 4A), which is similar to the fraction of Sox2-GFP-positive cells (97%; Fig. 2B). Twenty-four of 24 iPS clones produced with retroviral vectors expressing Oct4, Klf4, and c-Myc were EYFP-positive, demonstrating their origin from neural cells (Fig. 4A, 4B). In a different set of experiments, tail fibroblasts from Nestin-Cre/ROSA26-EYFP mice were reprogrammed into iPS cells, yielding rare EYFP-positive colonies (2 of 185), which corresponded to the fraction of EYFP-positive cells found in the tail-tip starting population (0.5%; M. Stadtfeld and K. Hochedlinger, unpublished results). Thus, our observations that (a) 97% of NPCs express the NPC marker Sox2, (b) 98% of NPCs from Nestin-Cre/ROSA26-EYFP mice were EYFP-positive, and (c) 100% of iPS colonies derived from these NPCs, in contrast to fibroblasts from the same mouse model, were EYFP-positive, demonstrate that the Nestin-Cre transgene specifically labels NPCs and does not become spuriously activated during the reprogramming process.

Figure Figure 4..

Generation of three-factor iPS cells from genetically marked NPCs. (A): Nestin-EYFP expression in Nestin-Cre/ROSA26R-EYFP NPCs, Nestin-Cre/ROSA26R-EYFP TT fibroblasts, and WT NPCs. (B): Derivation of iPS cells from NPCs of the lineage-tracing mouse model Nestin-Cre/ROSA26R-EYFP; shown are adherent culture growth of NPCs on polyornithine-coated plates, development of iPS cells 5 d after infection (arrows), and picked iPS colonies on feeder cells. (C): Viral vector integration analysis of genomic DNA from NPC iPS via Southern blot analysis shows integration sites for pMX-c-Myc, -Oct4, and -Klf4 (red frame). (D): Lineage-marked iPS cells showed positive AP staining. Differentiated iPS cells did not show AP staining but remained EYFP-positive (marked by arrow). (E): EB formation of iPS cells derived from the NPCs. (F): Teratoma formation after injection of iPS cells in SCID mice. HE staining shows contribution to all three germ layers (cartilage (mesoderm), keratinized epithelium (ectoderm), and glandular structures (endoderm). Teratomas were also positive for anti-EYFP staining. Abbreviations: AP, alkaline phosphatase; d, days; EYFP, enhanced yellow fluorescence protein; HE, hematoxylin and eosin; iPS, induced pluripotent stem; Max, maximum; NPC, neural progenitor cell; TT, tail-tip; WT, wild-type.

Analysis of genomic DNA from iPS cells produced with three factors via Southern blot analysis showed integration sites for pMX viral vectors containing c-Myc, Oct4, and Klf4 and confirmed the absence of the pMX-Sox2 viral vector (Fig. 4C). On average, 11 retroviral copies were present per iPS cell line, which is consistent with our previous findings in fibroblast-derived iPS cells [4]. iPS cells derived from these genetically marked NPCs expressed alkaline phosphatase (Fig. 4D) and differentiated in vitro into embryoid bodies (Fig. 4E) and into mesodermal, ectodermal, and endodermal derivatives in the context of teratomas (Fig. 4F), thus corroborating the notion that neural progenitor cells are amenable to reprogramming into pluripotent cells.


Our data allow two major conclusions. First, we provide genetic evidence that a defined ectodermal cell type can be reprogrammed into iPS cells by the same combination of transcription factors as previously used for reprogramming of mesodermal and endodermal derivatives. NPC-derived iPS cells expressed markers similar to those expressed by ES cells and gave rise to teratomas and chimeras. These observations suggest that reprogramming is not restricted to cells from a particular germ layer and that most, if not all, diploid cells are amenable to reprogramming by the same factors. This notion does not exclude the possibility, however, that certain cell types or differentiation stages may be refractory to reprogramming for technical or biological reasons. Thus, although our data show that an ectodermal progenitor cell type can be reprogrammed into iPS cells, it remains to be tested whether terminally differentiated ectodermal cells can give rise to iPS cells. Consistent with this, a recent report indicated that mature B lymphocytes, but not B cell progenitors, are refractory to reprogramming into iPS cells by ectopic expression of Oct4, Sox2, Klf4, and c-Myc alone, suggesting that the differentiation state of a cell can affect its potential to be reprogrammed [10]. Terminally differentiated cells can be reprogrammed into iPS cells, however, by these four factors when using β cells as donors [9] or lymphocytes lacking the B cell-specific transcription factor Pax5 [10].

The other main conclusion from our work is that cell types that already express one of the four reprogramming factors at appropriate levels do not require its ectopic expression. Specifically, we found that NPCs and ES cells express comparable levels of Sox2, whereas tail fibroblasts exhibit low-level expression of Sox2. Consistent with this, we found that NPCs can be reprogrammed into pluripotent cells in the absence of exogenous Sox2 expression, whereas we were unable to generate iPS cells from tail-tip fibroblasts without ectopic Sox2 expression. iPS cells produced with three factors were molecularly and functionally indistinguishable from iPS cells produced with four factors, including their ability to generate high-degree chimeric mice. Interestingly, the overexpression of Sox2 during NPC reprogramming further reduced the overall low reprogramming efficiency, suggesting that elevated levels of Sox2 are toxic or induce the differentiation of iPS cells, as seen previously [20]. This observation reinforces the notion that the expression levels and the stoichiometry of the four transcription factors are critical during reprogramming, which is in agreement with the finding that fibroblasts that have pre-existing expression of Klf4 and c-Myc still require viral overexpression of these genes to efficiently generate iPS cells.

It has been shown previously that viral expression of Oct4, Klf4 and c-Myc alone is sufficient to generate iPS cells from embryonic fibroblasts, which, in contrast to four-factor iPS cells, are unable to differentiate [2]. Our inability to produce stable iPS cells from tail-tip fibroblasts with Oct4, Klf4 and c-Myc expression alone may be due to differences in the iPS selection strategy (drug selection vs. morphological selection used here) or to differences in cell types (embryonic fibroblasts vs. neonatal fibroblasts used here). Another possible explanation for our failure to produce three-factor iPS cells may be our use of the doxycycline-inducible system, which eliminates cells that have not reactivated the endogenous pluripotency program [14], whereas iPS cells produced with retroviral vectors can contain both fully reprogrammed and partially reprogrammed cells. Indeed, the three-factor iPS cells reported by Takahashi and Yamanaka showed continuous viral transgene expression of some of the factors and incomplete reactivation of the corresponding endogenous loci [2].

Although c-Myc is highly expressed in NPCs, we consistently failed to reprogram NPCs in the absence of both Sox2 and c-Myc (data not shown). It may be that different cell types require different levels of c-Myc for efficient reprogramming and the levels expressed in neonatal NPCs are not sufficient to generate iPS cells. Alternatively, the efficiency of obtaining iPS cells in the absence of both Sox2 and c-Myc may be feasible but too low to be detected with our current culture and infection protocols.

It seems puzzling that the overall efficiency of reprogramming NPCs by transcription factors is roughly 100-fold lower compared with that of fibroblasts, since NPCs have previously been shown to be more easily reprogrammed than differentiated cell types by nuclear transfer [21]. NPCs, in contrast to fibroblasts, are more difficult to grow and manipulate, and we have noticed increased apoptosis following the three rounds of viral infections when producing iPS cells. This increased sensitivity and cell death may eliminate a large proportion of reprogrammed cells and could result in the low overall efficiency. Thus, one has to be careful when interpreting differences in efficiencies between fibroblasts and NPCs or other published cell types without considering these cell type-specific differences.


Our findings imply that cell types that express one or more of the reprogramming factors at levels similar to ES cells should be more easily reprogrammed than cells with low or no expression of the factors and thus may require less genetic manipulation. Such cell types, including NPCs, may be an appropriate source of cells for attempts to replace viral vector gene delivery systems with transient delivery systems or small compounds to produce iPS cells without genetic manipulation.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


We thank Jennifer Shay and members of the Hochedlinger laboratory for critical review of the manuscript. We also thank Laura Prickett, Kathryn E. Folz-Donahue, Adlen Foudi, and Matthias Stadtfeld for help with FACS analysis. K.H. was supported by the NIH Director's Innovator Award, the Harvard Stem Cell Institute, the V Foundation, and the Kimmel Foundation. J.U. was supported by the Dr. Mildred Scheel Foundation for Cancer Research. This study has been approved by the Massachusetts General Hospital animal and biosafety committees.