Generation of Human-Induced Pluripotent Stem Cells in the Absence of Exogenous Sox2§


  • Author contributions: S.D.: conception of the study, manuscript writing, supervision of the study; W.L.: conception of the study, manuscript writing, experimentation; R.A.: manuscript writing, experimentation; H.Z., S.Z. and T.L.: experimentation; E.H.: generation of teratomas; A.H.: supervision of E.H.; J.Y.J.: generation of chimera data; H.R.S.: supervision of J.Y.J.

  • First published online in STEM CELLS EXPRESS October 16, 2009.

  • §

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


Induced pluripotent stem cell technology has attracted enormous interest for potential application in regenerative medicine. Here, we report that a specific glycogen synthase kinase 3 (GSK-3) inhibitor, CHIR99021, can induce the reprogramming of mouse embryonic fibroblasts transduced by only two factors, Oct4 and Klf4. When combined with Parnate (also named tranylcypromine), an inhibitor of lysine-specific demethylase 1, CHIR99021 can cause the reprogramming of human primary keratinocyte transduced with the two factors, Oct4 and Klf4. To our knowledge, this is the first time that human iPS cells have been generated from somatic cells without exogenous Sox2 expression. Our studies suggest that the GSK-3 inhibitor might have a general application to replace transcription factors in both mouse and human reprogramming. STEM CELLS 2009;27:2992–3000


Pioneering work showed that virus-mediated expression of four transcription factors, Oct4, Klf4, Sox2, and c-Myc [1], reprograms mouse somatic cells into induced pluripotent stem (iPS) cells, which closely resemble embryonic stem (ES) cells [1, 2]. Reprogramming human somatic cells had also been achieved through a similar strategy [3, 4]. The iPS cell technology has attracted enormous interest with respect to its potential practical applications. With reprogramming and differentiation processes, patient-specific pluripotent stem cells could be created and further differentiated into functional autologous cells for cell-based therapy with alleviated immunocompatibility issues and ethical concerns. However, iPS cell applications are hindered by safety concerns and its complexity as the generation of iPS cells typically involves integration of exogenous DNA sequences. The key advances aimed at overcoming these safety concerns have been achieved by using nonintegrating gene delivery approaches (such as adenovirus or episomal plasmid transfection) [5–7] or using cell membrane permeable proteins to trigger the reprogramming [8, 9]. However, reprogramming is extremely slow and inefficient under such conditions, which presents significant hurdles and potential risks to generate human iPS cells. Identification of small molecules or novel conditions that can enhance reprogramming or compensate the requirement of certain reprogramming factors will be highly desirable. We and others have shown it is possible to generate iPS cells with fewer factors by exploiting the endogenous gene expression [10–12]. Neural progenitor cells (NPCs) with endogenous Sox2 expression [13] could be reprogrammed into authentic iPS cells with only Oct4 and Klf4 transduction, but with a lower efficiency [10, 11]. With use of a chemical screen, a G9a histone methyltransferase inhibitor, BIX-01294 [14], was identified to enhance the reprogramming efficiency over eightfold or replace the requirement of Oct4 transduction in NPC reprogramming [11]. Importantly, BIX-01294 was also shown to enable the reprogramming of mouse embryonic fibroblasts (MEFs) (which do not express Sox2) under Oct4 and Klf4 two-factor conditions [15]. From a subsequent synergistic screen, other small molecules such as DNA methyltransferase (DNMT) inhibitor RG108 and L-type calcium channel agonist BayK8644 were identified to enhance MEF reprogramming. Similarly, another DNMT inhibitor, 5-AZA, was shown to improve the reprogramming efficiency in MEFs up to fourfold by transiting partially reprogrammed cells to become fully pluripotent. In another study, histone deacetylase inhibitors such as valproic acid (VPA) were shown to be able to enhance the reprogramming efficiency [16]. In particular, VPA enabled reprogramming of human fibroblasts with only Oct4 and Sox2 [17].

Previous studies showed that Wnt3a-conditioned media promotes reprogramming of MEF cells [18]. Wnt signaling entails inhibition of glycogen synthase kinase 3 (GSK-3) and stabilization of cytoplasmic β-catenin. Small-molecule inhibitors of GSK-3 can mimic the activation of Wnt signaling and maintain the pluripotent state of mouse embryonic stem (mES) cells [19–21]. Lluis et al. reported that BIO, a GSK-3 inhibitor, could promote the reprogramming of somatic cells after fusion with mES cells [22]. Silva et al. reported inhibition of mitogen-activated protein kinase Kinase (MEK) and GSK-3 (using PD0325901 and CHIR99021, respectively) could transit “pre-iPS cells” into fully reprogrammed pluripotent cells [23]. More recently, Lyssiotis et al. [24] identified another GSK-3/Cyclin-dependent kinase 2 (CDK2) inhibitor, kenpaullone, which could substitute Klf4 in reprogramming of MEFs in the presence of Oct4, Sox2, and cMyc [24]. However, as a more specific GSK-3 inhibitor, CHIR99021, failed in producing the same positive effects on inducing the reprogramming of MEF cells under the Oct/Sox2/c-Myc transduction, kenpaullone's effect may not result from its GSK-3 inhibition and its precise mechanism remains elusive.

Here, we reported that a specific GSK-3 inhibitor, CHIR99021, could allow the reprogramming of both mouse and human somatic cells without Sox2 transgene. Our studies suggest that the GSK-3 inhibitor might have a general application to replace transcription factors in both mouse and human somatic cell reprogramming.


Cell Culture and Viral Transduction

MEFs were derived from 129S2/SvPasCrlf and ROSA26+/−/OG2+/− mice according to the protocol reported on the WiCell Research Institute (Madison, WI, Web site: “Introduction to human embryonic stem cell culture methods”. ROSA26±/OG2± heterozygous transgenic mice carry GFP reporter gene under the control of the Oct4 promoter (Oct4-GFP) and the ubiquitously expressed neo/lacZ transgene [25]. Animal experiments were performed according to the Animal Protection Guidelines of the Max Planck Institute for Biomolecular Research, Germany. MEFs were transduced by Oct4, Klf4, and Sox2 three-factor or two-factor combinations of the pMXs-based retroviruses encoding mouse Oct4, Klf4, and Sox2 (Addgene Inc., Cambridge, MA, as previously described [1]. Twenty-four hours later, transduced MEFs were seeded in 6-well plates and incubated with mES cell growth medium: Knockout Dulbecco's modified Eagle's medium (DMEM), 7% ES cell-qualified fetal bovine serum (FBS), 10% Knockout serum replacement, 1% GlutaMAX, 1% nonessential amino acids, 1% penicillin/streptomycin, 0.1 mM β-mercaptoethanol, and 103 U/ml mouse leukemia inhibitory factor (Millipore, Billerica, MA, MEFs transduced with Oct4/Klf4/Sox2 (1 × 104 cells per well together with 105 cells per well CF1 feeders in 6-well plates) were then treated with GSK-3 inhibitor CHIR99021 (Stemgent, Cambridge, MA, http://www.stemgent. com) for 2 weeks, and EGFP-positive colonies were picked up at the third week after treatment. MEFs transduced with Oct4/Klf4 (1 × 105 cells per well in 6-well plates) were treated with 10 μM CHIR99021 for 4 weeks, and GFP-positive colonies (named miPSCs-OK) were picked up and expanded at the fourth to fifth week after treatment.

Neonatal human epidermal keratinocytes (NHEKs; Lonza Group Ltd, Basel, Switzerland, were cultured and transduced with two-factor combinations of lentiviruses encoding human Oct4, Sox2 (pSin-EF2-Puro-based) and mouse Klf4 (pLOVE-based) as previously described [4, 26]. Lentiviral vectors were obtained from Addgene. Twenty-four hours later, 1 × 105 transduced NHEKs were seeded on the irradiated x-ray inactivated CF1 MEF feeder cells in a 100-mm dish by keratinocyte medium (Lonza). One week after, the medium was changed to human ES cell medium: DMEM/F12, 20% Knockout serum replacement, 1% GlutaMAX, 1% nonessential amino acids, 1% penicillin/streptomycin, 0.1 mM β-mercaptoethanol, and 100 ng/ml basic fibroblast growth factor (bFGF) and treated with GSK-3 inhibitor CHIR99021 (Stemgent) (10 μM) alone or combined with valproic acid (0.5-2 mM), BIX-01294 (Stemgent) (1-2 μM), RG108 (Stemgent) (1-5 μM), Parnate (Sigma-Aldrich, St. Louis, (2-4 μM), PD0325901 (Stemgent) (0.5 μM), and SB431542 (Tocris Bioscience, Ellisville, MO, (2 μM). The media containing the above small-molecule combinations were changed every day. Two weeks after treatment, the cells were subcultured (1:1) on new feeder cells (PD0325901 and SB431542 were used only in the first 2-week treatment). After another 2 weeks, the small molecules were removed and the cells were stained with Alexa Fluor 555-conjugated mouse anti-human TRA-1-81 antibody (BD Pharmingen, San Diego, The positive colonies were marked and picked up for expansion on feeder cells in human ES cell medium about 7 weeks after transduction. The human iPS cells (named hiPSCs-OK) were subcultured regularly by Accutase (Chemicon, Temecula, CA, All cell culture products were from Invitrogen/Gibco-BRL (Invitrogen, Carlsbad, CA, http://www. except where mentioned.

Cytochemistry and Immunofluorescence Assay

Alkaline phosphatase staining was performed according to the manufacturer's protocol using the Alkaline Phosphatase Detection Kit (Millipore). For immunofluorescence assay, cells were fixed in 4% paraformaldehyde for 10 minutes and washed three times with phosphate-buffered saline (PBS) containing 0.1% Triton X-100 (Sigma-Aldrich). The fixed cells were then incubated in blocking buffer, 0.1% Triton X-100 and 10% normal donkey serum (Jackson Immunoresearch Laboratories, West Grove, PA, in PBS (Invitrogen/Gibco-BRL), for 30 minutes at room temperature. The cells were then incubated with primary antibody overnight at 4°C in blocking buffer. The day after, cells were washed with PBS and incubated with secondary antibody in PBS containing 0.1% Triton X-100 for 1 hour at room temperature. Mouse anti-Oct4 antibody (1:250) (Santa Cruz Biotechnology Inc., Santa Cruz, CA,, rabbit anti-Sox2 antibody (1:2000) (Chemicon), mouse anti-SSEA1 antibody (1:250) (Santa Cruz Biotechnology), rabbit anti-Nanog antibody (1:250) (Abcam, Cambridge, U.K.,, rat anti-SSEA3 antibody (1:1000) (Chemicon), mouse anti-SSEA4 antibody (1:1000) (Chemicon), mouse anti-TRA-1-81 antibody (1:1000) (Chemicon), goat anti-Sox17 (1:200) (R&D), mouse anti-βIII-Tubulin (Tuj1) antibody (1:1000) (Covance Research Products, Princeton, NJ,, and rabbit anti-Brachyury antibody (1:200) (Santa Cruz Biotechnology) were used as primary antibodies. Secondary antibodies were Alexa Fluor 486/555 donkey anti-mouse, anti-rat, anti-goat, or anti-rabbit IgG (1:500) (Invitrogen). Nuclei were visualized by 4′,6-diamidino-2-phenylindole staining (Sigma-Aldrich). Images were captured using a Nikon Eclipse TE2000-U microscope.

Differentiation of iPS Cells In Vitro

The in vitro differentiation of miPSCs-OK and hiPSCs-OK was carried out by the standard embryoid body (EB) differentiation method. The iPS cells were dissociated by either 0.05% Trypsin-EDTA (miPSCs-OK) or Accutase (hiPSCs-OK) and then cultured in ultra-low attachment 100-mm dish in DMEM medium supplemented with 10% FBS to form EBs. The medium was changed every other day. One week later, the EBs were harvested and transferred into Matrigel-coated 6-well plates in DMEM medium with 10% FBS. Three to seven days later, the cells were fixed for immunocytochemistry analysis.

Polymerase Chain Reaction Analysis

To detect the expression of pluripotency genes by MEFs and NHEKs that were treated with small molecules, nontransduced MEFs and NHEKs were treated for 3 days in mES cell growth medium with 10 μM CHIR99021 or in hES cell medium with either a combination of 10 μM CHIR99021 and 2 μM Parnate or a combination of 10 μM CHIR99021, 2 μM Parnate, 0.5 μM PD0325901, and 2 μM SB431542. For the semiquantitative and quantitative reverse transcription-polymerase chain reaction (RT-PCR) analyses, RNA was extracted from miPSCs-OK, hiPSCs-OK, MEFs, treated MEFs, and treated NHEKs using the RNeasy Plus Mini Kit in combination with QIAshredder (Qiagen, Hilden, Germany, Reverse transcription was performed with 1 μg of RNA using iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA, The expression of pluripotent markers by miPSCs-OK was analyzed by RT-PCR using Platinum PCR SuperMix (Invitrogen). The primers for the endogenous Oct4, Sox2, Klf4, and Nanog were as reported [1]. Amplification of viral transduced genes was done using the gene-specific forward primers (Klf4: 5′-GCG AAC TCA CAC AGG CGA GAA ACC-3′; Sox2: 5′-GGT TAC CTC TTC CTC CCA CTC CAG-3′; and Oct4: 5′-TTG GGC TAG AGA AGG ATG TGG TTC-3′) and common reverse primer pMXs-L3205 (5′-CCC TTT TTC TGG AGA CTA AAT AAA-3′) [3]. The RT-PCR was performed in 30 (amplification of pluripotent markers) or 35 (amplification of viral transduced genes) cycles (94°C for 30 seconds, annealing temperature for 30 seconds, and 72°C for 30 seconds). Real-time PCR was carried out using iQ SYBR Green Supermix (Bio-Rad). The primers for the human endogenous Oct4, total Oct4, endogenous Sox2, total Sox2, Nanog, Klf4, GDF-3, and Cripto were as reported [4, 27, 28]. The primer for viral Klf4 was 5′-CAC CTT GCC TTA CAC ATG AAG AGG-3′ and 5′-CGT AGA ATC GAG ACC GAG GAG A-3′. The primer for FGF-4 was 5′-GAC ACC CGC GAC AGC CT −3′ and 5′-TCA CCA CGC CCC GCT-3′. The expression of genes of interest was normalized to that of glyceraldehyde 3-phosphate dehydrogenase in all samples.

Genomic DNA was extracted from miPSCs-OK using DNeasy Blood & Tissue Kit (Qiagen). To analyze the viral integration in miPSCs-OK, the genomic DNA of miPSCs-OK was subjected to PCR analysis using the same primers employed to amplify the viral transduced genes in the RT-PCR experiments. For the methylation analysis of Oct4 promoter by bisulfite sequencing, DNA samples from hiPSCs-OK were isolated using the Non Organic DNA Isolation Kit (Millipore) and were then treated with the EZ DNA Methylation-Gold Kit (Zymo Research Corporation, Orange, CA, The treated DNA samples were then used as templates to amplify targets of interest. Primers used for the amplification of the Oct4 promoter fragment (406 bp, from −2192 to −1786) were 5′-GGA TGT TAT TAA GAT GAA GAT AGT TGG-3′ and 5′-CCT AAA CTC CCC TTC AAA ATC TAT T-3′ [29]. The resulting fragments were cloned using the TOPO TA Cloning Kit for sequencing (Invitrogen) and were then sequenced.

Aggregation of iPS Cells with Zona-free Embryos

miPSCs-OK were aggregated with denuded postcompacted eight-cell stage embryos to obtain aggregate chimeras. Eight-cell embryos (B6C3F1) were flushed from females at 2.5 dpc and cultured in microdrops of KSOM medium (10% fetal calf serum) under mineral oil. Clumps of iPS cells (10–20 cells) after short treatment of trypsin were chosen and transferred into microdrops containing zona-free eight-cell embryos. Eight-cell embryos aggregated with iPS cells were cultured overnight at 37°C, 5% CO2. Aggregated blastocysts that developed from eight-cell stage were transferred into one uterine horn of a 2.5 dpc pseudopregnant recipient. The recipient mice were sacrificed at embryonic day 13.5 days. The embryos were analyzed by x-gal staining to reveal the contribution of iPS cells.

Teratoma Formation

Three to five million hiPSCs-OK (passage 8, clone 1) were injected under the kidney capsule of SCID mice (n = 3). After 6-8 weeks, the neoplasm was removed and then histologically analyzed.


CHIR99021 Can Significantly Promote Reprogramming of MEFs Transduced by Oct4, Sox2, and Klf4

It had been shown that Oct4/Sox2/Klf4-infected MEFs could be reprogrammed into the pluripotent state with higher efficiency when cultured under Wnt3a-conditioned medium [18]. However, small-molecule activators of the Wnt signaling pathway were not found to have similar effects. A combination of CHIR99021, a GSK-3 inhibitor that can activate the Wnt signaling pathway, and PD0325901, a MEK inhibitor, was shown to promote partially reprogrammed iPS cells to full pluripotency [23]. Concurrent with those studies, we found that CHIR99021 could significantly promote reprogramming of murine fibroblasts. Treating Oct4/Sox2/Klf4-transduced MEFs with CHIR99021 for 2 weeks significantly increased the number of alkaline phosphase (AP)-positive mESC-like colonies in a dose-dependent manner (AP staining was done at the third week after treatment) (Fig. 1A). CHIR99021 treatment of Oct4/Sox2/Klf4-transduced MEFs (ROSA26±/OG2±), which express GFP under the control of Oct4 promoter and also ubiquitously LacZ, also increased the number of GFP-positive colonies, which could be observed as early as 2 weeks after treatment. CHIR99021 showed the greatest effects at about 10 μM, which increase efficiency from 0.03-0.08% to 0.2-0.4% of transduced MEFs. (Fig. 1B). Our results therefore suggest that CHIR99021 can significantly improve reprogramming efficiency of MEFs transduced with Oct4, Sox2, and Klf4. These mouse iPS cell colonies could be stably expanded under conventional mESC growth conditions and express typical pluripotency markers, such as AP, Oct4, Sox2, Nanog, SSEA1 by cytochemistry, and immunostaining.

Figure 1.

CHIR99021 promoted the reprogramming of mouse embryonic fibroblasts (MEFs) transduced by Oct4, Sox2, and Klf4. MEFs from the 129 strain were transduced with Oct4, Sox2, and Klf4 by retroviruses and treated the next day with increasing concentrations of CHIR99021 for 2 weeks. Three weeks later, alkaline phosphase was detected by staining cells in monolayer (A). ROSA26±/OG2± MEFs transduced with Oct4, Sox2, and Klf4 were seeded into 6-well plates at the density of 1 × 104 cells per well (together with 105 cells per well CF1 feeders) and treated with CHIR99021 for 2 weeks. After 3 weeks of treatment, the GFP-positive colonies in each well were counted (B). Error bars represent standard deviation for n = 3. Abbreviation: DMSO, dimethyl sulfoxide.

CHIR99021 Enabled Reprogramming of MEFs Transduced by Oct4/Klf4

We had previously identified BIX01294, a small-molecule inhibitor of a histone methyltransferase G9a, which enabled reprogramming of both mouse NPCs and MEFs infected by only Oct4 and Klf4 [11, 15]. We then investigated whether iPS cells could be generated from MEFs with fewer reprogramming factors in the presence of CHIR99021. OG2 MEFs transduced with different two-factor combinations (Oct4/Klf4, Oct4/Sox2, and Sox2/Klf4) were treated with 10 μM CHIR99021. GFP-positive iPS cell colonies were identified only when MEFs were transduced with the combination of Oct4 and Klf4, but not with any other combination. On average, about six GFP-positive colonies were identified out of 105 OG2 MEFs 4-5 weeks after Oct4/Klf4 transduction and CHIR99021 treatment. Stable iPS cell lines (miPSCs-OK) were established by picking up the GFP-positive colonies (Fig. 2A, 2B). Immunocytochemistry revealed that miPSCs-OK express typical pluripotency markers, such as Oct4, Sox2, Nanog, and SSEA-1 (Fig. 2C–2F). MEFs do not express Sox2 endogenously, and real-time PCR analysis revealed that CHIR99021 treatment did not induce the expression of Sox2 and Oct4 in MEFs (Fig. 2G). Therefore, the mechanisms by which CHIR99021 promotes the reprogramming of MEFs transduced by Oct4/Klf4 are independent of direct Sox2 induction.

Figure 2.

CHIR99021 enables the reprogramming of MEFs transduced with Oct4 and Klf4 only. ROSA26±/OG2± MEFs transduced with Oct4 and Klf4 were split into 6-well plates at the density of 105 cells per well and treated with 10 μM CHIR99021 for 4 weeks. Panel (A) shows GFP-positive colonies before picking. A total of four miPSCs-OK lines were established (B). The expression of Oct4 (C), Sox2 (D), Nanog (E), and SSEA-1 (F) by miPSCs-OK was detected by immunocytochemistry. The expression of pluripotency genes by MEFs after CHIR99021 treatment was analyzed by real-time polymerase chain reaction (G). Scale bars, 20 μm. Error bars represent standard deviation for n = 3. Abbreviations: MEFs, mouse embryonic fibroblasts.

RT-PCR analysis confirmed the reactivation and expression of the endogenous mouse Oct4, Sox2, Nanog, and Klf4 (Fig. 3A). With use of the specific primers for transgenes, RT-PCR analysis revealed that the viral genes were largely silenced (Fig. 3A). PCR of genomic DNA of miPSCs-OK confirmed the integration of retroviral Oct4 and Klf4, but no other reprogramming genes (Fig. 3B). To examine the developmental potential of miPSCs-OK, an in vitro differentiation assay was preformed. Immunostaining showed miPSCs-OK could differentiate into endoderm (Sox17), mesoderm (brachyury), and neuroectoderm (βIII-tubulin) derivatives under the standard embryoid body differentiation methods (Fig. 3C–3E). Most importantly, miPSCs-OK (clone 1 and clone 2) could efficiently incorporate into the inner cell mass of blastocysts after aggregation with eight-cell embryos, lead to mid-gestational chimerism (ED 13.5) after the aggregated embryos were transplanted into mice, and contribute to germ line cells in vivo (Fig. 3F–3H). However, no adult chimeric mice were found after 20 embryos aggregated with miPSCs-OK (clone 1) were transplanted. These in vitro and in vivo characterizations confirm that the miPSCs-OK are molecularly, morphologically, and functionally similar to the original four-factor iPS cells and the mouse ESCs.

Figure 3.

PCR analysis and in vitro differentiation of miPSCs-OK. The expression of typical endogenous pluripotency genes and Tg were analyzed by RT-PCR (A). Genomic PCR revealed integration of Oct4 and Klf4 retrovirues (B). 1∼4 refers to the four established miPSCs-OK lines. 100 bp DNA ladder (Invitrogen) was used as a marker. Rat iPS cells generated by Oct4/Klf4/Sox2 transductions were used as positive control (+) and MEFs were used as negative control (−). Under standard EB differentiation methods, the in vitro pluripotency of miPSCs-OK was analyzed by immunostaining (C–E). miPSCs-OK efficiently incorporated into the inner cell mass of a blastocyst after aggregation with an eight-cell embryo (F). Chimeric embryo (13.5 dpc) was obtained after the transfer of the aggregated embryos into a pseudo-pregnant mouse. LacZ staining showed the contribution of miPSCs-OK (G). miPSCs-OK contributed to the germ line cells (GFP-positive) in male gonad tissue isolated from chimeric embryos (H). Scale bars, 20 μm. Abbreviations: MEFs, mouse embryonic fibroblasts; RT-PCR, reverse transcription-polymerase chain reaction; Tg, transduced genes.

CHIR99021 Enabled Reprogramming of Human Neonatal Keratinocytes Transduced with Oct4/Klf4 When Combined with Parnate

We next investigated whether human iPS cells could be generated with fewer transcription factors in the presence of CHIR99021 and/or direct epigenetic modifiers including inhibitors of DNA methyltransferase (RG108), histone methyltransferase (BIX01294), histone deacetylase (valproic acid/VPA), and lysine-specific demethylase 1 (Parnate). To this end, we selected primary human neonatal epidermal keratinocytes (HNEKs), concurrent with recent studies suggesting that keratinocytes transduced with four factors could be reprogrammed into iPS cells more efficiently and rapidly in comparison to other somatic cell types [27]. Primary keratinocytes were transduced with different two-factor combinations (Oct4/Klf4, Oct4/Sox2, and Sox2/Klf4), treated with CHIR99021 alone, or combined with epigenetic modifiers and then stained with the human pluripotency cell-surface marker TRA-1-81 5 weeks postinfection. Tra-1-81-positive human ESC-like colonies could only be identified from culture infected by Oct4 and Klf4 in the presence of CHIR99021 and Parnate. On average, about 2 Tra-1-81-positive colonies could be identified out of 105 transduced HNEKs, which was at lease 100 times less efficient than four-factor-transduced keratinocytes. Stable human iPS cells could be established and long-term expanded by picking up these colonies (named hiPSCs-OK). In addition, we have also found that combined treatment using inhibitors of MEK (PD0325901) and TGFβ receptor (SB431542) could improve the reprogramming efficiency of human fibroblasts transduced by Oct4/Sox2/Klf4/c-Myc (unpublished data). With use of CHIR99021 (10 μM) and Parnate (2 μM) as the basal condition, addition of PD0325901 (0.5 μM) and SB431542 (2 μM) could further increase the TRA-1-81-positive colonies from human keratinocytes transduced with Oct4/Klf4 (about 5-10 Tra-1-81-positive colonies could be identified out of 105 transduced HNEKs), but the detailed mechanisms underlying this observation still need to be revealed. Nine TRA-1-81-positive colonies were expanded, and three stable human iPS cells (hiPSCs-OK), one from CHIR99021 and Parnate condition (hiPSCs-OK 1) and another two from CHIR99021/Parnate plus PD0325901/SB431542 condition (hiPSCs-OK 2, 3), were further studied and long-term-cultured for over 20 passages. hiPSCs-OK express typical pluripotency markers, such as AP, Oct4, Sox2, Nanog, TRA-1-81, SSEA3, and SSEA-4 (Fig. 4A–4D). Real-time PCR analysis confirmed expression of the endogenous human Oct4, Sox2, Nanog, Cripto, GDF-3, and FGF4 (Fig. 4E). Although the viral Oct4 and Klf4 expression was not completely silenced, bisulfite sequencing analysis revealed that the Oct4 promoter of hiPSCs-OK is largely demethylated (Fig. 4F). Similar to the CHIR99021 treatment of MEFs, real-time PCR analysis indicated neither CHIR99021/Parnate (two inhibitors) nor CHIR99021/Parnate/PD0325901/SB431542 (four inhibitors) treatment induced the expression of Sox2 and Oct4 in keratinocytes immediately (Fig. 5). The terminal differentiation of keratinocytes induced by the human ES cell culture media may result in the significant downregulation of c-Myc expression after treatment (Fig. 5).

Figure 4.

hiPSCs-OK are generated from primary human keratinocytes transduced by Oct4 and Klf4. Established hiPSCs-OK clones express pluripotency markers AP (A), SSEA3 (green)/Oct4 [red, (B)], TRA-1-81 (green)/Nanog [red, (C)], and SSEA4 (green)/Sox2 [red, (D)]. Expression of endogenous (endo) markers and viral transgenes in hiPSCs-OK 1, 2, and 3 was determined by real-time polymerase chain reaction (E). Primary human keratinocytes (K) and Hues9 human ES cells were used as controls. Error bars represent standard deviation for n = 3. The methylation status of the Oct4 promoter in primary human keratinocytes and hiPSCs-OK was analyzed using bisulfite sequencing. Open circles indicate unmethylated and filled circles indicate methylated CpG dinucleotides (F). Abbreviation: HNEKs, human neonatal epidermal keratinocytes.

Figure 5.

Real-time polymerase chain reaction analysis of the expression of pluripotency genes by keratinocytes after small-molecule treatment. Primary human keratinocytes (K), keratinocytes treated with either combination of 10 μM CHIR99021 and 2 μM Parnate (two inhibitors), or combination of 10 μM CHIR99021, 2 μM Parnate, 0.5 μM PD0325901, and 2 μM SB431542 (four inhibitors), and Hues9 human ES cells were analyzed. Error bars represent standard deviation for n = 3.

To examine developmental potentials of hiPSCs-OK, in vitro differentiation assays were preformed. Immunostaining confirmed that hiPSCs-OK could differentiate into endoderm (Sox17), mesoderm (brachyury), and neuroectoderm (βIII-tubulin) (Fig. 6A–6C) derivatives in vitro. Furthermore, after transplantation into the SCID mice, hiPSCs-OK formed teratoma consisting of representative derivatives of all three germ layers including epithelial tube structure (endoderm) (Fig. 6D), cartilage-like structure (mesoderm) (Fig. 6E), and neuroepithelium-like structure (ectoderm) (Fig. 6F). These in vitro and in vivo characterizations confirm that the human iPS cells generated by Oct4 and Klf4 viral transduction closely resemble human ES cells in terms of typical pluripotency marker expression and differentiation potential.

Figure 6.

hiPSCs-OK showed pluripotent potential in vitro and in vivo. With use of the standard EB differentiation method, the in vitro pluripotency of hiPSCs-OK was analyzed by immunostaining (A–C). hiPSCs-OK generated full teratoma in SCID mice. Hematoxylin and eosin staining of hiPSCs-OK teratoma sections showed epithelial tube structure (endoderm), cartilage-like structure (mesoderm), and neuroepithelium-like structure (ectoderm) appear in (D)–(F). Scale bars, 20 μm.


Reprogramming is a very slow and inefficient process. Such low efficiency and slow kinetics also present “hidden” risks in iPS cells, such as accumulated and selected subtle genetic and epigenetic abnormalities. Consequently, it is highly desirable to identify new conditions/small molecules that can promote reprogramming and/or replace certain factors.

In the present study we reported that the GSK-3 inhibitor CHIR99021 can significantly improve the reprogramming efficiency of MEFs transduced by Oct4/Sox2/Klf4 and also enable the reprogramming of MEFs transduced by only Oct4 and Klf4. When combined with Parnate, CHIR99021 can result in the reprogramming of human primary keratinocytes transduced with only Oct4 and Klf4 as well. Although previous studies showed that the activation of Wnt signaling promotes somatic cell reprogramming, this study is the first report to show GSK-3 inhibitor could allow the reprogramming of both mouse and human somatic cell without Sox2. Recently, it was reported that the target genes co-bounded by Oct4, Sox2, and Klf4 in ES cells showed a lower histone H3 lysine 4 (H3K4) trimethylation enrichment in partially reprogrammed cells than in ES/iPS cells, and this low histone H3K4 trimethylation may result in the lack of binding of many important regulators of pluripotency by Oct4, Sox2, and Klf4 [30]. Parnate, a monoamine oxidase inhibitor used as an antidepressant drug, showed potent inhibitory effect on lysine-specific demethylase 1 and inhibiting of the H3K4 demethylation, but does not influence the acetylation of H3K9/K14 [31]. Parnate may facilitate the full reprogramming of HNEKs transduced with only Oct4 and Klf4 by inhibiting H3K4 demethylation.

Particularly, this is also the first time human iPS cells have been generated from somatic cells without exogenous Sox2 expression. Both Oct4 and Sox2 are critical regulators in human/mouse ES cell pluripotency and also the only common reprogramming factors used for generation of human iPS cells. Replacement of Sox2 in human cell reprogramming represents an important step toward identifying a chemically defined condition that could allow reprogramming of human somatic cells by Oct4 only or without the forced expression of any exogenous factor. As HNEKs express Klf4 endogenously, it would be conceivable that HNEKs could perhaps be fully reprogrammed with only Oct4 transduction, but so far it was not achieved for unknown reasons. When Oct4-transduced HNEKs were treated under the same chemical condition, some human ES cell-like colonies were observed. After these colonies were picked up, stable lines were established that could be long-term-cultured under conventional human ES cell media. However, these cells are negative to AP staining, and expression of other pluripotency markers, such as Nanog and Sox2, could not be detected by immunostaining (data not shown).

Our studies underscore the unique advantage of the chemical approach for improving reprogramming that may ultimately allow the generation of iPS cells or multipotent tissue-specific cells in completely chemically defined conditions without any permanent genetic modification. Ultimately, a completely chemically defined condition for efficient reprogramming of somatic cells would be highly favorable for various iPS cell applications.


This work was supported by funding from Fate Therapeutics and CIRM to S.D. and Larry L Hillblom Foundation to A.H.


The authors indicate no potential conflicts of interest.